Analysis of large deletion mutations induced by abasic site analog in human cells

Background Abasic sites are formed spontaneously and by nucleobase chemical modifications and base excision repair. A chemically stable abasic site analog was site-specifically introduced into replicable plasmid DNAs, which were transfected into human U2OS cells. The amplified DNAs were recovered from the cells and used for the transformation of a bacterial indicator strain. Results Large deletion mutations were induced by the analog, in addition to point mutations at the modified site. No apparent sequence homology at the deletion junctions was found. Conclusion These results suggested that the large deletions induced by the abasic site analog are formed by homology-independent events.


Background
Chemical modifications of nucleic acids have severe effects on organisms because they disturb replication, transcription, and translation, resulting in mutagenesis, carcinogenesis, and cell death. Various DNA lesions are formed spontaneously and by mutagens, such as reactive oxygen species and ultraviolet light [1][2][3][4][5]. An abasic (apurinic/apyrimidinic, AP) site is formed by the spontaneous hydrolysis of the N-glycosyl bonds of DNA, and this hydrolysis is accelerated by the chemical modifications of bases [6]. Moreover, this DNA lesion is produced as an intermediate of base excision repair when a monofunctional DNA glycosylase, such as uracil DNA glycosylase, acts on a damaged base. Approximately 10,000-50,000 abasic sites are reportedly present per mammalian cell under normal physiological conditions [7,8]. Thus, the abasic site is one of the important DNA lesions causing mutagenesis/carcinogenesis, due to its abundance and loss-of-base character.
In general, the incorporation of a 2′-deoxyribonucleotide opposite the abasic site and its analogs by mammalian replicative DNA polymerases (pols) is inefficient. For example, in vitro DNA synthesis by human DNA pol ε is severely blocked at the nucleotide preceding an abasic site analog [9]. Instead, translesion synthesis DNA pols are considered to conduct the bypass of the DNA lesion [10]. Due to the lack of a nucleobase, DNA pols "misincorporate" nucleotides opposite the DNA lesion. In living mammalian cells, various point mutations are induced by the abasic site (analog), depending on its opposite bases, flanking sequences, and cell type [11][12][13][14][15].
Previously, large deletion mutations were found when shuttle vectors containing ultraviolet-induced DNA lesions, cyclobutane and 6-4 thymine-thymine dimers, were replicated in simian cells [16]. Since these dimers block nucleotide incorporation by replicative DNA pols as with the abasic site [17], it may also cause large deletion mutations in addition to point mutations. In this study, we focused on deletion mutations upon the transfection of a shuttle plasmid DNA bearing a chemically stable, frequently used tetrahydrofuran analog (THF) of the abasic site. The analog actually caused large deletion mutations in living human cells, without any apparent homology at the deletion junctions.

Materials
The oligodeoxyribonucleotides (ODNs) used for plasmid construction and sequence analysis are listed in Table 1. ODN-1 and ODN-2 were previously synthesized [18,19].
ODN-3 and ODN-5 containing THF, and ODN-4 containing G instead of THF, were synthesized in this study. These ODNs were chemically phosphorylated on the support and purified by HPLC, as described previously [20]. PCR primers were purchased from Hokkaido System Science (Sapporo, Japan) and Eurofins Genomics (Tokyo, Japan) in purified forms. Escherichia coli KS40/pOF105, used as the indicator strain of the supF mutants, was provided by Professor Tatsuo Nunoshiba of the International Christian University [21].

Plasmid DNA containing THF
The double-stranded shuttle vectors containing THF were prepared from the single-stranded forms of pZ189-T_E107K/D402E (formerly pZ189-107 K/402E) and ODN-3 or ODN-5 by enzymatic reactions catalyzed by T4 DNA pol and T4 DNA ligase, as described [22,23]. To obtain highly pure DNAs, they were treated with Plasmid-Safe ATP-Dependent DNase (Epicentre, Madison, WI, USA). The DNAs were purified with a PureLink PCR Purification Kit (Thermo Fisher Scientific, Waltham, MA, USA) to remove proteins, ODNs, and mononucleotides before and after the DNase treatment. The plasmid DNAs bearing G or 8-oxo-7,8-dihydroguanine (G O ) were prepared by the same procedures.

Transfection of plasmid DNAs and mutant frequency determination
U2OS cells (5.0 × 10 4 cells) were cultured in Dulbecco's modified Eagle's medium, supplemented with 10% fetal bovine serum. After 24 h, 100 ng (29 fmol) of the plasmid DNA bearing THF was transfected with Lipofectamine (Thermo Fisher Scientific), and the cells were cultured for 48 h. The plasmid molecules propagated in the U2OS cells was isolated by the Hirt procedure [24]. The DNA was extensively digested by DpnI to break down the unreplicated plasmid. E. coli KS40/pOF105 cells were electroporated with the plasmid, and the supF mutant frequency was calculated [21,25]. The plasmids bearing G and G O , instead of THF, were transfected as controls. The sequence of the mutant supF plasmid was analyzed, using the ODN-seq1 and ODN-seq2 primers (Table 1).

Results
Induction of large deletions by an abasic site analog in the supF gene (experiment 1) First, we incorporated THF into position 122 of the supF gene. The mutational properties of other DNA lesions, G O and O 6 -methylguanine, in this position have previously been examined [18,19,26]. The plasmid bearing the unique THF was prepared by enzyme reactions using ODN-3 ( Table 1). The DNA was transfected into human U2OS cells, recovered from the cells, and introduced into E. coli KS40/ pOF105 cells. The numbers of colonies on the titer plate and the selection plates containing nalidixic acid, streptomycin, and X-gal were counted [21,25]. Mutations in the supF gene confer resistance to the two antibiotics and the lacZ − phenotype to E. coli. The plasmid DNAs containing G and G O , instead of THF, were used as controls.
The numbers of colonies on the titer plates, which semi-quantitatively reflect the amounts of plasmid DNA amplified in the U2OS cells, were comparable for all progeny plasmids, indicating the similar replication efficiencies of the G-, G O -, and THF-plasmids and THF removal by DNA repair. The supF mutant cells formed white or pale blue colonies on the selection plates. The supF mutant frequency was calculated by dividing the numbers of supF mutant colonies on the selection plates by those of the total colonies on the titer plates.
The supF mutant frequency was 1.1 × 10 − 3 when the control plasmid containing G, instead of THF, was transfected ( Fig. 1A). This value was similar to the expected one calculated by the error frequency of the DNA pol used in the plasmid preparation and the length of the supF gene, suggesting that a significant portion of the mutations was due to nucleotide misincorporation during the plasmid preparation [27]. The replacement of the G with THF at position 122 greatly enhanced the supF mutant frequency (1.2 × 10 − 2 ). This value was much higher than the mutant frequency for another major DNA lesion, G O .
We next analyzed the supF plasmids in the mutant colonies (Tables 2 and 3). Previously, THF has been shown to induce base-substitution mutations [15]. These types of mutations were also found in the experiment. THF➔C and THF➔T mutations were observed as targeted substitutions in this study. In addition, a targeted − 1 deletion was detected. Moreover, as we expected, large deletion mutations were present among the mutants. The frequencies of the large deletion mutations were calculated as the products of the total supF mutant frequencies and the ratios of the large deletions. The frequencies were 4.8 × 10 − 4 and 2.5 × 10 − 3 in the G and THF experimental groups, respectively. In the case of G O , the frequency was 6.4 × 10 − 4 , indicating that not Induction of large deletions including the supF gene by an abasic site analog located outside the gene (experiment 2) Next, we incorporated THF outside the supF gene, to detect large deletion mutations more easily. The analog was introduced into the site located 9 "bases" downstream of the gene, which has a length of 162 bases. We named this site "position 171", although it is outside of the gene. Point mutations at this site alone do not inactivate the supF gene, but large deletions containing a part of the gene produce supF mutants. ODN-5 was used for plasmid construction ( Table 1). The supF mutant frequencies were 1.1 and 3.6 × 10 − 3 when the plasmid DNAs without and with THF, respectively, were transfected (Fig. 1B). Sequence analyses (Tables 4  and 5) and calculations of the large deletion frequency revealed that THF also induced large deletions in this case: the frequencies were 2.5 × 10 − 4 and 2.8 × 10 − 3 for the control and THF plasmid DNAs, respectively (P < 0.01, Student's t-test). Thus, we confirmed that the presence of the abasic site analog induces large deletions.

Analysis of deleted sequences and junctions
Finally, we analyzed the deleted sequences and junctions. Sequence homology at the junctions of deleted regions should be found, if the deletions occurred in a homology-dependent manner. However, we did not find apparent similarity in any case (Table 6). Thus, homology-mediated events were not involved in the deletion formation.
Twelve mutants obtained in the THF (position 122) experiment contained large deletions. Among them, position 122 was included in the deleted portions for 10 mutants ( Table 6

Discussion
In this study, we examined the large deletion mutations induced by the abasic site analog THF in human cells. Unsurprisingly, substantial fractions of the progenies of THF-bearing vectors contained large deletions (Tables 3  and 5). In the case of THF at position 122 (experiment 1), most of the deleted regions were upstream of the modified site, although the downstream deletion produces supF mutants and thus is detectable. Since THF was introduced into the upper strand, the region corresponding to the 5′-side of the lesion was lost. A similar tendency was observed for THF at position 171 (experiment 2), although the lesion was located downstream of the supF gene and thus the deletion of the upstream region is essential to inactivate the gene.
When we consider replication, the 5′-side of the lesion is equivalent to the downstream region of the opposite site in the nascent strand. As described in the Introduction section, large deletion mutations were also induced by cyclobutane and 6-4 thymine-thymine dimers in mammalian cells [16]. Since the photodimers and THF are replication blocking lesions, we hypothesize that the inefficient bypass of THF causes the large deletion generated during replication.
Alternatively, the large deletion might be formed as a byproduct of DNA repair. APE1 (APEX, HAP, Ref-1) is the major apurinic/apyrimidinic endonuclease in human cells [28][29][30][31]. THF is a substrate of APE1 [32] and the enzyme incises DNA 5′ to THF, resulting in single-strand break (SSB) formation. Op het Veld et al. reported that large deletions were frequently observed in XRCC1-deficient Chinese hamster cells, after a treatment with methyl methanesulfonate [33]. Since the protein is involved in SSB repair and     [34]. Although they also found that large deletion mutations (8-168 bases) were induced by the DNA lesion, no information about the deletion junctions was reported. We detected the absence of homology in the deletion junctions we analyzed ( Table 6) and concluded that the deletions were due to homology-independent events, such as DNA replication block and/or SSB formation, as described above.
A deaminated base, hypoxanthine, was previously shown to cause an A➔G mutation in mammalian cells [35]. However, DeVito et al. recently reported that the DNA lesion induced large deletions in human HEK293 and HCT116 cells [36]. N-methylpurine DNA glycosylase excises this modified base, leaving an abasic site [37]. Since hypoxanthine does not seem to be a replication blocking lesion due to its structural similarity to G, the abasic sites formed by the DNA glycosylase possibly contributed to the deletion mutations, as DeVito et al. discussed. In contrast, G O did not induce large deletions in U2OS cells (experiment 1) although the major repair enzyme OGG1 produces an abasic site. This discrepancy might be due to balance of DNA glycosylase and apurinic/apyrimidinic endonuclease activities. The amounts of the repair proteins would depend on cell types and the abasic sites formed by N-methylpurine DNA glycosylase might not be efficiently removed in HEK293 and HCT116 cells. In line with this possibility, G O induces large deletions in HCT116 cells [36]. The repair by monofunctional DNA glycosylases possibly acts as a double-edged sword in cells.
In this study, we found THF➔C and THF➔T mutations at position 122 (Table 3). These substitution mutations seemed to be caused by dGMP and dAMP incorporations, respectively, opposite the lesion. In contrast to E. coli, no clear rule of nucleotide incorporation was present in mammalian cells [11,15]. Interestingly, the − 1 deletion at the THF site (5′-G-THF-T-3′) was observed, in contrast to the previous reports. This mutation might be triggered by dCMP incorporation opposite THF, followed by mispairing of the 5′-flanking G and the incorporated C after looping-out of THF. Alternatively, it might be initiated by dCMP incorporation opposite the 5′-flanking G after looping-out of THF, followed by the extension of the G:C pair.
The detection and quantitation of abasic sites are important research, and useful chemical probes have been    [38,39]. Investigations of the mutational properties of abasic sites, as well as their quantitation, are necessary for understanding the contributions of this important DNA lesion to mutagenesis, carcinogenesis, aging, and neurodegeneration.

Conclusions
In conclusion, the abasic site analog THF induced large deletion mutations in human cells. The lack of homology at their junction sites indicated that the deletions are caused by homology-independent events. Investigations of their molecular mechanism are quite pivotal, and further experiments are currently in progress. Authors' contributions TS and HK conceived of the study, participated in its design and coordination, and drafted the manuscript. YKa performed the mutagenesis studies, the sequence analysis, and the data analysis. YKo synthesized and purified ODNs containing THF. All authors read and approved the final manuscript.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.