PATHWAYS FOR DOUBLE-STRAND BREAK REPAIR IN GENETICALLY UNSTABLE Z-DNA-FORMING SEQUENCES

Abstract

DNA can adopt many structures that differ from the canonical B-form, and several of these non-canonical DNA structures have been implicated in genetic instability associated with human disease. Previously, we found that Z-DNA causes DNA double-strand breaks (DSBs) in mammalian cells that can result in large-scale deletions and rearrangements. In contrast, the same Z-DNA-forming CG repeat in E.coli resulted in only small contractions or expansions within the repeat. This difference in the Z-DNA-induced mutation spectrum between mammals and bacteria may be due to different mechanisms for DSB repair; in mammalian cells non-homologous end-joining (NHEJ) is a major DSB repair pathway, while E.coli do not contain this system and typically use homologous recombination (HR) to process DSBs. To test the extent to which the different DSB repair pathways influenced the Z-DNA-induced mutagenesis, we engineered bacterial E.coli strains to express an inducible NHEJ system, to mimic the situation in mammalian cells. Mycobacterium tuberculosis NHEJ proteins Ku and Ligase D (LigD) were expressed in E.coli cells in the presence or absence of HR, and the Z-DNA-induced mutations were characterized. We found that the presence of the NHEJ mechanism markedly shifted the mutation spectrum from small deletions/insertions to large-scale deletions (from 2% to 24%). Our results demonstrate that NHEJ plays a role in the generation of Z-DNA-induced large-scale deletions, suggesting that this pathway is associated with DNA structure-induced destabilization of genomes from prokaryotes to eukaryotes.

Introduction

Maintenance of genomic stability via DNA repair mechanisms is crucial for cells to survive exposure to exogenous and endogenous DNA damaging agents. Importantly, genetic instability can also occur in the absence of DNA damage per se. For example, naturally occurring non-canonical (i.e. non-B) DNA structures have been shown to induce genetic instability in mammalian cells and bacterial cells in the absence of exogenous sources of DNA damage^{1; 2; 3}. In fact, sequences that have the capacity to adopt non-B DNA structures (e.g. triplet repeats) have been implicated in many neurodegenerative and genomic disorders (for review see4).

Sequences at alternating purine-pyrimidine repeat sequences can adopt a Z-DNA structure that forms a left-handed, elongated helix relative to the traditional Watson-Crick right-handed B-DNA helical structure. Z-DNA can form in vivo under certain physiological conditions, and can function as an active element in many metabolic processes in the genome (for reviews see5; 6). Z-DNA-forming sequences often co-localize with breakpoint hotspots in disease-related genes, such as in the Alzheimer’s specific genes, amlyoid precursor protein, and Presenilin and ApoE (for reviews, see 7; 8). In addition, breakpoints in human translocation-related cancers (e.g. leukemias and lymphomas) often map to Z-DNA-forming sequences (for reviews, see 4; 9).

In cultured mammalian cells and in mouse chromosomes Z-DNA has been shown to induce large-scale deletions and complex rearrangements¹⁰ resulting from DSBs¹. The majority of the junctions of the Z-DNA-induced deletions contained microhomologies, implicating NHEJ-type mechanisms in the repair of the DSBs. However, the same Z-DNA-forming sequence in E.coli resulted in a strikingly different mutation pattern with the vast majority of deletions being small-scale deletions within the repetitive sequence^{1; 11}. There are a number of possible explanations for this interesting difference, including the possibility that the CG(14) sequence may form different secondary structures in mammalian and bacterial cells. Differences in the chromatin structures between mammalian cells and bacterial cells or DNA binding proteins ¹² may influence the CG(14) conformation. The Z-DNA structure might also be processed differently in mammalian cells compared to bacterial cells. Another possible explanation for the differences in Z-DNA-induced mutagenesis in mammalian versus bacterial cells could be in the processing of Z-DNA-induced DSBs¹. In mammalian cells, the usual error-prone NHEJ repair system generally predominates over the more error-free HR system in processing DSBs, with the exception of S-phase^{13; 14; 15; 16}; and NHEJ often results in mutations with loss of the sequences between the microhomologous regions (for review see17). In prokaryotes lacking NHEJ, such as E.coli, a generally error-free RecA-dependent HR mechanism is heavily relied upon for the repair of DSBs, although small misalignments can occur within repetitive sequences during the binding of single-stranded DNA (ssDNA) to the homologous template (for review see18). Based on this information, we speculated that the differences in the types of Z-DNA-induced mutations between mammals and E.coli may be the result of operative DSB repair pathway.

In mammalian cells, yeast, and some bacteria¹⁹, NHEJ and HR are both utilized to efficiently process DSBs. In Mycobacterium tuberculosis (Mt), a prokaryote that contains a NHEJ mechanism¹⁹, the DNA binding Ku-like protein (Mt-Ku) and the ATP-dependent ligase D (Mt-LigD), which are expressed from a single operon, have been identified as a two-component NHEJ system²⁰. The multi-domain protein Mt-LigD has DNA-dependent DNA polymerase, DNA-dependent RNA primase, 3’-5’ single stranded DNA exonuclease, terminal transferase, and DNA ligase activities^{20; 21}. A homodimer of Ku protein in bacteria binds to damaged DNA ends and interacts with the polymerase domain of Mt-LigD to specifically recruit and stimulate the end-joining activity²⁰. Although expression of Mt-Ku and Mt-LigD is sufficient to carry out DSB repair²⁰, thus suggesting the “two-component NHEJ” model, the outcomes of repair by Mt-LigD and Mt-Ku expressed in different genetic background can be different, indicating that other proteins provided by the host cells are also involved in the repair processes^{22; 23}. However, these data do suggest that the expression of LigD and Ku can initiate the processing of the damaged DNA ends to allow ligation of the two broken strands²⁰.

The HR repair system in E.coli initiates DSB repair using the nuclease activity of the RecBCD complex to process the DNA ends and load RecA onto the ssDNA²⁴. Further steps are required, and distinct from NHEJ, a homologous template is necessary for repair of the DSBs. The E.coli RecA protein participates in the homology search and strand-exchange reaction, which is then followed by the formation and resolution of a D-loop and other junctions (for more detail see25). Although mechanistically different, both of these DSB repair systems are employed to uphold and maintain the genomic integrity of the cell.

In this study, we examined the roles of these two DSB repair pathways, NHEJ and HR, in Z-DNA structure-induced genetic instability by determining mutation frequencies and spectra in modified E.coli that express the Mt-Ku and/or Mt-LigD proteins, thereby providing E.coli with NHEJ capacity. Twenty-four percent (12/50 total mutants) of the mutants generated in the strains expressing the Mt-NHEJ proteins were large-scale deletions [≥50 base pairs (bps)], which rarely occurred in the same strains when Mt-NHEJ proteins were not present (2%, 1/56 total mutants), indicating that the large-scale deletions were caused by the error-prone NHEJ system. Moreover, when Mt-Ku and Mt-LigD were expressed individually, we did not detect large-scale deletions in these mutants (0/40 total mutants), confirming that the NHEJ system, and not its individual components, was responsible for the generation of the large-scale deletions. These data suggest that Z-DNA-induced mutagenesis in eukaryotic and prokaryotic cells is affected by the different DSB repair pathways available. This new information provides a better understanding of the mechanistic basis for the large-scale deletions seen among the Z-DNA-induced mutants in mammalian cells.

Results and Discussion

To explore whether NHEJ can process Z-DNA-forming sequences into large deletions as seen in mammalian cells, which would provide at least one explanation or the different Z-DNA-induced mutation spectrum in bacteria, we determined the fate of the Z-DNA-forming CG(14) repeat in E.coli cells containing an inducible Mt-NHEJ system, which adds the NHEJ pathway to the existing pathways of DSB repair in E.coli, thus mimicking the situation in mammalian cells.

Mt-NHEJ in E.coli causes mutagenic DSB repair

In order to study the roles of the DSB repair pathways, particularly the effects of NHEJ, on Z-DNA-induced genetic instability, we utilized an established model system that allows for Mt-NHEJ expression in E.coli cells. In this system, the E.coli have been modified to contain the Mt-Ku and Mt-LigD expression vectors specifically integrated into designated sites of the E.coli chromosome²³. The Mt-NHEJ genes can be expressed by induction with L-arabinose in either HR-proficient, wild-type (WT/NHEJ+), or HR-deficient, RecB- (HR-/NHEJ+), bacterial strains. The strains generated and used in these experiments are listed in Table 1. The expression of the Mt-Ku and Mt-LigD genes were verified at the protein level prior to use in the mutagenesis assay via western blotting as described²³ (data not shown). Z-DNA-induced mutagenesis was detected using a mutation reporter shuttle vector containing a lacZ’ gene for facile blue-white screening. CG14 and its control sequences were inserted at an EcoRI-SalI cassette, between the promoter and the coding region of lacZ’. Z-DNA induced mutations, including deletions, insertions, and rearrangements will inactivate the lacZ reporter gene and result in a while colony after plasmid transformation into DH5α bacterial cells. Treatment with L-arabinose per se did not affect the mutation frequency of the lacZ reporter gene in the WT E.coli cells in the absence of the Mt-NHEJ expression vectors, neither on the control plasmid nor on the pUCG(14) plasmid that forms a Z-DNA structure [Figure 1(a)]. These data confirm that the Z-DNA-forming sequences are mutagenic in bacterial cells with mutation frequencies ranging from 14–17 x10⁻⁴, with or without L-arabinose, which is ~20-fold above that of the control plasmid, pUCON (~1 x10⁻⁴) (p value < 0.01) [Figure 1(a)]. As predicted, the majority of these mutants contained small-scale deletions within the CG repeats [Figure 1(b)], likely the result of misalignment during replication and/or HR repair of DSB generated by secondary DNA structures, e.g., Z-DNA.

Table 1.

Bacterial Strains Used In This Study

	HR RecA/RecB	Mt-NHEJ Ku/LigD
WT	+/+	−/−
RecB-/NHEJ +	+/−	+/+
RecA-/NHEJ +	−/+	+/+
WT/NHEJ+	+/+	+/+
Mt-LigD	+/+	−/+
Mt-Ku	+/+	+/−

Figure 1.

Mutation frequencies and spectra in the E.coli mutation-reporter system. (a) L-arabinose (Ara+) does not affect the spontaneous (pUCON) or the Z-DNA-induced (pUCG14) mutation frequency in bacterial cells in the absence of Mt-NHEJ expression vectors. The Z-DNA-forming CG14 sequence stimulated mutagenesis in WT cells. More than 100,000 colonies were counted in each group. Error bars show the standard deviation (StDev) of three independent experiments. (b) L-arabinose (Ara+) does not affect the spectrum of Z-DNA-induced mutation in bacterial cells in the absence of Mt-NHEJ expression vectors; the majority of Z-DNA-induced mutations are small deletions. Mutants were digested with EagI and BssSI, resulting in seven fragments that were separated by agarose gel electrophoresis. For each group, 20–30 mutants were analyzed; a representative gel is shown. The arrows refer to the fragments containing the Z-DNA-forming sequences; “*” refers to large-scale deletions (≥50 bps); “C“ refers to the control plasmid; and “M” refers to the marker. (c) 5’-overhanging “sticky” DSBs are processed in a mutagenic fashion in cells containing functional NHEJ. DSB-induced mutation frequencies in RecB- cells shown in the presence (NHEJ+) or absence (NHEJ-) of the Mt-NHEJ proteins. pUCG14 and the control plasmid pUCON were linearized by EcoRI digestion (pUCON 5’-OH and pUCG14 5’-OH) 4 bp from the lacZ reporter gene. Error bars show the StDev of two independent experiments. (d) The processing of the EcoRI linearized pUCG14 plasmid in the NHEJ+ cells results in large-scale deletions, as indicated by a “*”. EcoRI-induced mutants generated in the Mt-NHEJ-proficient RecB- strain were analyzed as in (b). For each group, 20–30 mutants were analyzed; a representative gel is shown. (e) Blunt-end DSBs are processed in a mutagenic fashion in bacterial cells. p2RT plasmid containing the lacZ reporter gene, as in pUCON, was linearized by EcoICRI digestion 4 bp from the lacZ reporter gene and mutants were screened in DH5α bacterial cells. Error bars show the StDev of two independent experiments.

To confirm that the modified E.coli cells that express the Mt-NHEJ genes actually exhibited functional NHEJ, DSB repair was tested using linearized plasmids, pUCON and pUCG14. The plasmids were first digested with EcoRI to generate “sticky-ended” DSBs with 5’-overhangs, and then the linearized plasmids were transformed into the RecB- strains with (NHEJ+) or without (NHEJ-) the expression of the Mt-NHEJ genes (with or without L-arabinose induction). Expression of the Mt-NHEJ proteins yielded approximately 20-fold more colonies than in the un-induced cells (estimated by the number of colonies formed after transformation), suggesting that the NHEJ mechanism, when expressed, was more receptive to functionally repairing the linearized plasmids. The repair of the DSBs resulted in approximately 1,000-fold higher mutation frequency in the NHEJ+ strain compared to the NHEJ- strain, indicating that the expression of both Mt-Ku and Mt-LigD initiated the repair of DSBs in a mutagenic manner, consistent with a functional NHEJ mechanism [Figure 1(c)]. Moreover, restriction analysis of the mutants generated from 5’-overhang containing “sticky-ended” DSBs revealed that the majority of the mutants generated in the NHEJ+ strain were large-scale deletions (7/10), as indicated by the length of heterogeneity of the restriction bands marked as “Z” in Figure 1(d), and were confirmed by sequencing analysis, which showed deletions greater than 50-bp, and ~80% of the junctions had 1–8 bp microhomologies or short additional sequences (data not shown). When RecB- cells were transformed with plasmids linearized by EcoICRI digestion which generated “blunt-ended” DSB four bps from the lacZ reporter gene, the repair was error-prone and the majority of colonies were white [Figure 1(e)]; meanwhile, expression of both Mt-Ku and Mt-LigD resulted in approximately 10 to 30-fold more colonies (estimated by the number of colonies formed after transformation, data not shown), and the vast majority of processing was still mutagenic [Figure 1(e)]. Our results shown in Figure 1 indicated that the Mt-NHEJ machinery was effectively induced in E.coli cells incubated with L-arabinose, and that these proteins can rejoin the DSBs in vivo in an error-prone manner, resulting in the production of large-scale deletions. Collectively, these results demonstrate the utility of this expression system to study DSB repair pathway choice in Z-DNA-induced genetic instability.

NHEJ stimulates Z-DNA-induced large-scale deletions in E.coli

CG repeats have been identified as hotspots for HR, and we posit that Z-DNA formed on CG repeats can induce DSBs, which can then be repaired by either HR or NHEJ pathways, providing both are available. To determine the effects of NHEJ on non-B DNA-induced mutagenesis, we used an E.coli strain that was deficient in HR and had an inducible NHEJ activity. Since RecB serves as a critical factor in processing breakpoints and facilitating RecA loading during HR²⁶, such that its deficiency reduces HR efficiency, and lack of RecBCD activity has no effect on the efficiency of Mt-NHEJ repair²², we first used the RecB- cell strain containing inducible Mt-NHEJ proteins. In the RecB- cells, the Z-DNA-induced mutation frequency was modestly increased in the presence of NHEJ [13 x10⁻⁴ in RecB-/NHEJ- cells versus 17 x10⁻⁴ in the RecB-/NHEJ+ cells (p value = 0.008); Figure 2(a)]. Although statistical analysis indicated that there was a significant increase in the Z-DNA-induced mutation frequency after Mt-NHEJ induction, the control plasmid pUCON also had a significant increase in mutation frequency from 1.5 x10⁻⁴ to 7 x10⁻⁴ after Mt-NHEJ induction (p value = 0.0003) [Figure 2(a)]. Thus, in RecB-deficient cells, the presence of Mt-NHEJ proteins increased mutation frequencies in a sequence-independent fashion. Meanwhile, expression of the Mt-NHEJ genes in HR-proficient WT E.coli cells did not significantly alter the background or the Z-DNA-induced mutation frequencies (data not shown). These results were surprising, since we expected the NHEJ and HR pathways to compete with each other in repairing the Z-DNA-induced DSBs when both were available, and that the absence of one pathway would shunt the repair of the DSBs into the other available pathway. If this was the case, then the presence of the more mutagenic NHEJ pathway should result in more mutations.

Figure 2.

Spontaneous or Z-DNA-induced mutation frequencies and spectra in NHEJ-proficient versus NHEJ-deficient E.coli strains. (a) Spontaneous (pUCON) and Z-DNA-induced (pUCG14) mutation frequencies in RecB- cells in the presence (NHEJ+) or absence (NHEJ-) of the Mt-NHEJ proteins. More than 100,000 colonies were counted in each group. Error bars show the StDev of three independent experiments. (b) pUCG14 mutants generated from NHEJ- (top panel) or NHEJ+ (bottom panel) RecB- strains were digested with EagI and BssSI, resulting in seven fragments that were separated by agarose gel electrophoresis as shown. The arrows refer to the fragments containing the Z-DNA-forming sequences; “*” refers to large-scale deletions (≥50 bps); “C” refers to the control plasmid; and “M“ refers to the marker. For each group, 50–60 mutants were analyzed; a representative gel is shown. (c) Percentage of large deletions from 20–30 total Z-DNA-induced mutants that were selected from RecA or RecB-deficient bacterial strains in the presence (NHEJ+) or absence (NHEJ−) of NHEJ proteins. (d) Sequences of small deletion mutants generated in RecB-/NHEJ+ cells, labeled as “M”. “WT” refers to the wild-type sequence. (e) Diagram of large deletion mutants generated in RecB-/NHEJ+ cells. The blank areas between the lines indicate deletions. Letters at the ends of the blank areas indicate microhomologies at the junctions.

HR repair of DSBs is generally accurate and error-free with the use of a homologous sequence as a template. However, small expansions and deletions can occur at simple repeat sequences, such as with the CG(14) repeat used in this study, due to misalignments during single-strand invasion and homologous pairing. These events are usually restricted to the simple repetitive regions and do not affect the adjacent sequences. On the contrary, NHEJ repair generally results in larger deletions of sequences between two homologous regions. In order to determine if the presence of the Mt-NHEJ proteins altered the types of mutations induced by Z-DNA in E.coli, we characterized the mutants by restriction digestion and sequencing analyses. DNA from the mutants generated in the RecB- strain in the absence of Mt-NHEJ induction [Figure 2(b), top panel], and in the presence of the Mt-NHEJ proteins [Figure 2(b), bottom panel] were digested with EagI and BssSI and electrophoresed on a 1.4% agarose gel. Strikingly, the presence of Mt-NHEJ had a substantial affect on the mutation spectrum; the proportion of mutants with large-scale deletions increased from 2% (1/56 total mutants, in the absence of NHEJ) to 24% (12/50 total mutants, in the presence of NHEJ). Sequencing analysis also confirmed that the majority of mutations were small-scale deletions/insertions in the cells lacking NHEJ, with 2–24 bps missing or inserted within the Z-DNA-forming CG(14) repeat [Figure 2(d)]. In contrast, a significant fraction of the mutants (12/50 total mutants) had undergone large-scale deletions (≥50 bps) in cells containing functional NHEJ, with most having complete loss of the Z-DNA-forming repeat and some adjacent sequences. Most of the junctions contained 2–4 bps of microhomologies, similar to that found in mammalian cells¹ [Figure 2(e)]. These data demonstrate that Z-DNA-induced mutations generated in NHEJ-proficient E.coli cells were very similar to those generated in mammalian cells, which contain endogenous NHEJ mechanisms. Although large-scale deletions were also seen on the control plasmid, pUCON, in NHEJ-proficient cells, sequencing data revealed that the deleted areas were distributed randomly, rather than centered around the inserted CG repeats as seen in the pUCG(14) plasmid (data not shown). We performed similar experiments in several other bacterial strains with various genetic backgrounds, such as RecA-deficient E.coli [Figure 2(c)] or wild-type cells that contained functional RecA and RecB-dependent HR, and similar results were obtained (data not shown). These results implicate DSB formation and the NHEJ repair pathway in the Z-DNA-induced large-scale deletions.

Supporting this notion, we confirmed that the CG(14) sequence did indeed lead to DSBs in E. coli by using Linker-mediated PCR (LMPCR). We extracted plasmid DNA from MBM7070 bacterial cells according the method described by Zhang et al (1995)²⁷, and the DSBs on the recovered DNA were determined using LMPCR as we have described¹. As expected, we saw a DSB hotspot at the CG(14) sequence that forms Z-DNA, but not on the control plasmid pUCON (data not shown). These data further demonstrate that Z-DNA causes DSBs in bacteria, similar to that seen in mammalian cells.

Functional NHEJ, and not Ku activity or LigD activity alone, is involved in Z-DNA-induced large-scale deletions

Mt-LigD contains a ligase domain at the C-terminus, a polymerase domain at the N-terminus, and a nuclease/phosphoesterase domain centrally located, which contains 3’ to 5’ exonuclease activity^{28; 29}. Therefore, it is possible that Mt-LigD per se, rather than functional NHEJ, is involved in DNA breakpoint processing and is responsible for the large-scale deletions induced by Z-DNA. The Mt-Ku homodimer binds to DNA breaks and is also capable of altering the type of DSB repair employed in processing DSBs^{20; 21; 23}. To determine whether the NHEJ pathway or the Mt-LigD/Mt-Ku protein alone, was responsible for the Z-DNA-induced large-scale deletions, the Z-DNA-induced mutation frequencies and spectra were determined in BW35 bacterial strains (both RecA and RecB proficient) that expressed Mt-Ku or Mt-LigD separately, where the NHEJ pathway was inactive. The expression of Mt-LigD only in bacterial cells did not significantly alter the Z-DNA-induced mutation frequency (21 x10⁻⁴ in the absence of Mt-LigD and 17 x10⁻⁴ in the presence of Mt-LigD, respectively) or the background mutation frequency seen on the control plasmid [2 x10⁻⁴ in the absence of Mt-LigD and 1 x10⁻⁴ in the presence of Mt-LigD, respectively; Figure 3(a)]. More importantly, the types of mutations induced by Z-DNA were not altered by expressing Mt-LigD only; i.e., the vast majority of mutations were small deletions within the repeats [Figure 3(b)], as seen in wild-type cells that do not express Mt-LigD. These data agree with observations from Glickman and Shuman’s group that the nuclease activity of the Mt-LigD was not involved in breakpoint processing²². Similarly, in cells that expressed Mt-Ku only, the Z-DNA-induced mutation frequency (32 x10⁻⁴) was not significantly higher than that detected in the same cells in the absence of Mt-Ku expression (17 x10⁻⁴) (p = 0.16) after screening over 100,000 total colonies in each group [Figure 3(c)]. Again, the Z-DNA-induced mutations were predominantly small deletions within the repeat, regardless of the expression status of Mt-Ku [Figure 3(d)]. These data indicate that functional NHEJ rather than either Mt-Ku or Mt-LigD alone, is required to generate large-scale deletions on Z-DNA-forming plasmids in bacterial cells.

Figure 3.

Spontaneous or Z-DNA-induced mutation frequencies and spectra of the Mt-Ku or Mt-LigD expressing bacterial strains. (a) Background (pUCON) and Z-DNA-induced (pUCG14) mutation frequencies in bacterial cells with (Mt-LigD+) or without (Mt-LigD-) Mt-LigD expression. (b) pUCG14 mutants generated in cells containing Mt-LigD (Mt-LigD+) were digested with EagI and BssSI, resulting in seven fragments that were separated by agarose gel electrophoresis, as shown. For each group, 20–30 mutants were analyzed; a representative gel is shown. (c) Background and Z-DNA-induced mutation frequencies in cells with (Mt-Ku+) or without (Mt-Ku−) Mt-Ku expression. More than 70,000 colonies were counted in each group. Error bars show the StDev of three independent experiments. (d) pUCG14 mutants generated in cells containing Mt-Ku (Mt-Ku+) were digested with EagI and BssSI, resulting in the seven fragments that were separated by agarose gel electrophoresis. The arrows refer to the fragments containing the Z-DNA-forming sequences; “C” refers to the control plasmid; and “M” refers to the marker. For each group, 20–30 mutants were analyzed; a representative gel is shown.

HR and NHEJ are not the only mechanisms to repair DSBs in vivo; other pathways such as the alternative RecFOR HR pathway is also available and functions independently of RecBCD³⁰. Interestingly, unverified DNA end rejoining processes showed high accuracy in rejoining the 5’-overhang containing “sticky-ends” generated by EcoRI digestion in our assay, and was very mutagenic in repairing the “blunt-ends” generated by EcoICRI (>60% of the colonies contained a mutation in the lacZ gene 4-bps from the DSB); thus, we speculate that a direct ligation of DSBs might also be involved. Nevertheless, these processes are not as efficient as HR or NHEJ repair and play a minor role in processing DSBs, as evidenced by the reduced number of colonies formed after transforming the linearized plasmid into the RecB- strain in the absence of NHEJ compared to the RecBCD and RecA-proficient cells or cells containing Mt-NHEJ proteins. Thus, we expected that when both major DSB repair mechanisms were available, such as in the E.coli cells containing the Mt-NHEJ proteins, HR and NHEJ pathways might compete with each other in processing the (Z-DNA-induced) DSBs. For example, this type of competition for a DSB intermediate has been seen after UV-A damage, when both systems are available³¹. However, if NHEJ is not available (as in wild-type E.coli), then the repair of the DSBs should be shunted into the other available pathways (Figure 4), which can affect both the mutation frequency and spectrum. In this study, we detected a significant increase in large-scale deletions when NHEJ was made available to E.coli, but the Z-DNA-induced mutation frequency was not significantly different in the bacteria in the presence or absence of NHEJ. Moreover, contraction/expansion of CG repeats can also occur due to slippage events during DNA replication¹ independent of DSBs and/or repair processing. This replication-related contraction/expansion of CG repeats may overshadow the HR-related small deletions or expansions initiated by DSBs, such that a change in mutation frequency was not detectable.

Figure 4.

Model of Z-DNA-induced DSB processing by HR and/or NHEJ pathways. (a) Z-DNA-induced DSBs may be processed by several different mechanisms, such as HR, NHEJ, and/or other processes (e.g. direct ligation of DNA breaks). “Del/GR” is deletion/gross rearrangement, “Del/Ins” is deletion/insertion. (b) In cells where NHEJ is available and HR is suppressed (due to RecB deficiency in our experimental system, or from a lack of homologous donor sequence in mammalian cells at G1 or resting phase), the Z-DNA-induced DSBs can be processed into large deletions. (c) In the absence of NHEJ, as is the situation in E.coli, DSBs are largely repaired via HR mechanisms, which may result in small expansions/contractions within the repeat due to misalignment events.

To summarize, by exploring a possible explanation for the different mutation patterns stimulated by Z-DNA-forming repeats in mammalian cells versus bacterial cells, we found that, if available, NHEJ repair of the Z-DNA-induced DSBs is responsible for the large-scale deletions and rearrangements (Figure 4), providing new and useful information regarding the mechanism of Z-DNA-induced genetic instability in various species. Further studies to elucidate mechanisms of DNA structure-induced genetic instability are warranted to further our understanding of the roles of DNA structure in genetic evolution, genetic instability, and human disease.

Materials and Methods

Plasmid substrates

All plasmids were constructed from a 7,075 bp shuttle vector plasmid, pUCNIM, containing an SV40 replication origin, a neomycin resistance gene, an ampicillin resistance gene, and the lacZ’ gene expressing the amino-terminal fragment of β-galactosidase as the mutation-reporter gene¹. The CG(14) Z-DNA-forming sequence and the control sequence CON were synthesized by Midland Certified Reagent Co (Midland, TX), and cloned into the shuttle vector pUCNIM at an EcoRI-SalI cassette, between the promoter and the coding region of the lacZ’ gene. The sequences used have been studied extensively and are described by Wang et al. (2006)¹. The following reporter plasmids with their specified inserts were constructed from the pUCNIM backbone and named accordingly; pUCG14, and pUCON. To determine the repair efficiency of DSBs via the expression of Mt-NHEJ vectors, pUCON and pUCG14 were linearized by digestion with EcoRI.

Bacterial strains

Genetic backgrounds of all the bacterial strains used are listed in Table 1, with their corresponding NHEJ or HR deficiencies and/or proficiencies. The bacterial strains were constructed from wild-type (WT) E.coli (Hfr KL16 (PO-45) thi-1 relA1 spoT1e14-λ-) with varying deficiencies and proficiencies of the HR protein, recB, and carrying single-copy integration of the vector expressing the Mt-NHEJ repair proteins, Ku (Mt-Ku) and/or DNA Ligase D (Mt-LigD) from M. tuberculosis. For more information on how these strains were constructed and the location of the Mt-Ku and Mt-LigD expression vectors on the chromosome, see ref.23. Mt-Ku and Mt-LigD each contain a His-tag on their N-terminus that does not interfere with the activity of the protein²³. In Mt-NHEJ competent strains, L-arabinose induces the expression of the Mt-NHEJ proteins, Mt-Ku and Mt-LigD. The HR-deficient strain containing a mutation in recB (recB268::Tn10) is labeled the RecB- strain. The WT/NHEJ+ strain contains the wild-type proficiency of the E.coli HR repair system and the expression of the Mt-NHEJ proteins after their induction by L-arabinose (i.e. functional NHEJ). Mt-Ku and Mt-LigD are expressed separately by L-arabinose in the Mt-Ku only and Mt-LigD only strains.

Mutagenesis assay

Specified bacterial strains listed in Table 1 expressing Mt-NHEJ proteins, Mt-Ku and/or Mt-LigD, were treated with a final concentration of 0.2% L-arabinose 1 hour prior to and after transformation. 50 ng of the Z-DNA-containing plasmid or control plasmid DNA was transformed into the desired bacterial strains via electroporation using the Bio-Rad Gene Pulser II (Bio-Rad, Hercules, CA) with 1.7 kV per reaction and for overnight growth at 37°C at 250 rpm in LB with ampicillin (100 μg/ml) and kanamycin (50 μg/ml) antibiotics. After 1 hour of recovery, a final concentration of 0.2% L-arabinose was added into the LB antibiotic mixture to maintain expression of the Mt-NHEJ genes. After overnight incubation, the plasmids were isolated by use of the Qiagen QIAprep Spin Miniprep Kit (QIAGEN Inc., Valencia, CA) according to the manufacturer’s recommendations. LacZ mutants derived during the overnight incubation were screened in commercial DH5α-derived cells, NEB 5-alpha Electrocompetent E.coli (New England Biolabs, Ipswich, MA), using a blue/white screen; bacteria were plated onto LB-ampicillin-kanamycin (same concentrations as described above) plates with 5-bromo-4-chloro-3-indolyl β-D-galactoside (X-gal) (50 mg/ml) and isopropyl β-D-thiogalactoside (IPTG) (160 mg/ml). Mutation frequencies were calculated as the number of white (mutant) colonies divided by the total (blue + white) colonies, and statistical analysis performed using a pairwise t-test. For further analysis of the Z-DNA-induced mutants, some selected mutants were double digested with EagI and BssSI [Figures 1(b), 1(d), 2(b), 3(b) and 3(d)]; these linearized fragments of mutant plasmids and of control plasmids were then separated on a 1.4% agarose gel. Large deletions caused by Z-DNA-forming sequences were identified by a small restriction band in the gel, which represents a deletion of greater than 50 bp, as confirmed by sequencing analysis. The difference in the percentage of large-scale deletions in total mutants for the bacteria containing Mt-NHEJ versus the NHEJ-deficient bacteria was statistically analyzed using a Fisher exact test.

Abbreviations

NHEJ

non-homologous end-joining

HR

homologous recombination

ssDNA

single-stranded DNA

DSB

DNA double-strand break

LMPCR

Linker-mediated PCR

Mt

Mycobacterium tuberculosis