Polymorphic variant Asp239Tyr of human DNA glycosylase NTHL1 is inactive for removal of a variety of oxidatively-induced DNA base lesions from genomic DNA

Abstract

Base excision repair is the major pathway for the repair of oxidatively-induced DNA damage, with DNA glycosylases removing modified bases in the first step. Human NTHL1 is specific for excision of several pyrimidine- and purine-derived lesions from DNA, with loss of function NTHL1 showing a predisposition to carcinogenesis. A rare single nucleotide polymorphism of the Nthl1 gene leading to the substitution of Asp239 with Tyr within the active site, occurs within global populations. In this work, we overexpressed and purified the variant NTHL1-Asp239Tyr (NTHL1-D239Y) and determined the substrate specificity of this variant relative to wild-type NTHL1 using gas chromatography-tandem mass spectrometry with isotope-dilution, and oxidatively-damaged genomic DNA containing multiple pyrimidine- and purine-derived lesions. Wild-type NTHL1 excised seven DNA base lesions with different efficiencies, whereas NTHL1-D239Y exhibited no glycosylase activity for any of these lesions. We also measured the activities of human glycosylases OGG1 and NEIL1, and E. coli glycosylases Nth and Fpg under identical experimental conditions. Different substrate specificities among these DNA glycosylases were observed. When mixed with NTHL1-D239Y, the activity of NTHL1 was not reduced, indicating no substrate binding competition. These results and the inactivity of the variant D239Y toward the major oxidatively-induced DNA lesions points to the importance of the understanding of this variant’s role in carcinogenesis and the potential of individual susceptibility to cancer in individuals carrying this variant.

1. Introduction

Oxidatively-induced DNA damage, which is caused by endogenous and exogenous sources in living organisms (reviewed in [1–3]), is mainly repaired by base excision repair (BER), and, to a lesser extent, by nucleotide excision repair (reviewed in [4–7]). DNA glycosylases initiate BER by hydrolyzing the N-glycosidic bond between a modified DNA base and the 2^′-deoxyribose moiety in DNA, leaving behind an apyrimidinic/apurinic (AP) site. There are eleven human DNA glycosylases generated with various forms in targeted areas of cells such as the nucleus, mitochondria and cytoplasm [8,9]. Monofunctional DNA glycosylases remove a modified base only via an activated water molecule, whereas bifunctional DNA glycosylases use primary or secondary amines to carry out base release, with concomitant lyase activity generating β-elimination or β,δ-elimination strand breaks. Subsequently, the repair is completed by other enzymes (reviewed in [5,10–13]). Oxidatively-induced DNA base lesions are subject to repair by DNA glycosylases with different excision mechanisms, substrate specificities and excision kinetics (reviewed in [14–16].

Among the human bifunctional DNA glycosylases, NTHL1 (also called NTH1) consisting of 312 amino acids with an average molecular mass of 34.39 kDa exhibits an extensive sequence similarity with E. coli endonuclease III (E. coli Nth) and Schizosaccharomyces pombe Nth [17]. NTHL1 has been shown to excise urea, N-substituted urea lesions, thymine glycol (ThyGly) and 5,6-dihydrouracil (5,6-diHUra) (a deamination product of 5,6-dihydrocytosine) from double-stranded oligodeoxynucleotides, and to possess some incision activity on UV- or γ-irradiated plasmid DNA as well [17–23] (reviewed in [24,25]). When genomic DNA containing multiple DNA base lesions was used, NTHL1 excised 5-hydroxycytosine (5-OH-Cyt), 5-hydroxyuracil (5-OH-Ura), 5, 6-dihydroxycytosine, ThyGly, 5-hydroxy-6-hydrothymine, 5,6-dihydroxyuracil, 4,6-diamino-5-formamidopyrimidine (FapyAde), 5-hydroxy-5-methylhydantoin (5-OH-5-MeHyd), 8-hydroxyadenine (8-OH-Ade); however, it exhibited no activity on 8-hydroxyguanine (8-OH-Gua) [26, 27]. Using Nth1^−/− mice, mitochondrial and nuclear extracts, and oligodeoxynucleotides containing a single FapyAde or 2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyGua), subsequent studies provided the evidence that FapyAde and FapyGua are also the physiological substrates of NTHL1, but not 8-OH-Gua [28,29]. The importance of NTHL1 for protecting the genomic integrity and in the prevention of cancers is evidenced by the findings that the loss of its function underlies a predisposition to various types of cancers (reviewed in [6,24,25]). On the other hand, there is evidence that the overexpression of NTHL1 may also be caused by disruption of its interaction with other repair enzymes, leading to genetic instability [30]. Using the GENTS database (http://gent2.appex.kr/gent2/), the analysis with a large number of cancer patient samples showed the overexpression of NTHL1, with significantly different expression levels in a variety of tumor tissues including breast, colon, ovary and pancreatic tumors [25]. Interestingly, NTHL1 may also protect telomeric integrity [31], suggesting a role in aging. Since overexpression of DNA repair enzymes leads to the resistance to cancer therapy (reviewed in [32]), DNA glycosylases such as NTHL1, NEIL1, OGG1 and uracil DNA glycosylase have been investigated as possible targets in cancer therapy, and small molecule inhibitors of their glycosylase and/or lyase actions have been discovered as possible anticancer drugs (reviewed in [16]).

A rare, single nucleotide polymorphism (SNP) (rs3087468) of the NTHL1 gene, which is defined by a guanine to thymine substitution, leading to the substitution of Asp239 with Tyr within the active site of the protein (www.ncbi.nlm.nih.cov/SNP), is widely distributed throughout Europe, Asia, and sub-Saharan Africa. The resulting variant NTHL1-Asp239Tyr (NTHL1-D239Y) has been expressed and purified from E. coli [33]. Asp239 is a highly conserved aspartic acid residue of NTHL1 and its homologs in organisms ranging from Escherichia coli, Bacillus stereothermophilus to Homo sapiens [34]. The location of Asp 239 in crystal structures indicates that, in its unprotonated form, it serves as the protein’s active site nucleophile. Thus, the alteration of Asp 239 to Tyr results in an inactive enzyme likely due to loss of the nucleophilic activity. Biochemical analyses using site-specifically modified oligodeoxynucleotides demonstrated that it was inactive for excision of ThyGly and 5,6-diHUra paired with adenine and guanine, respectively [33]. In addition, AP-lyase activity was greatly diminished on substrates containing an AP site paired with adenine or guanine. Further analyses revealed that, when wild-type NTHL1 was mixed with NTH1-D239Y, a decrease in ThyGly release was observed, potentially suggesting competition for binding to this lesion. Furthermore, the expression of NTH1-D239Y in human cells also expressing wild-type NTHL1 resulted in genomic instability, cellular transformation, chromosomal aberrations, and sensitivity to ionizing radiation- and H₂O₂-treatment that resulted in accumulation of DNA double-strand breaks. The DNA damage response was also activated in human cells expressing NTHL1-D239Y. These results suggested that individuals carrying the germline NTHL1-D239Y variant might be at risk for genomic instability and carcinogenesis. Previous studies on NTHL1-D239Y were limited to one DNA base lesion and a deamination product of another DNA base lesion in a single sequence context. However, oxidatively-induced damage produces a plethora of lesions from all four DNA bases in different sequence contexts within genomic DNA in vitro and in vivo (reviewed in [2]). Given the design limitations of the DNA substrates in previous studies and the knowledge that wild-type NTHL1 possesses a wide substrate specificity, we sought to investigate the activity of NTHL1-D239Y along with those of the wild-type NTHL1 and other major human and E. coli DNA glycosylases for other pyrimidine- and purine-derived DNA lesions.

2. Materials and methods

2.1. Materials

Cloning, expression and purification of human NTHL1, NTHL1-D239Y, NEIL1 and OGG1 were performed as describes previously [33, 35–37]. E. coli Nth and E.coli Fpg were purchased from New England Biolabs (Ipswich, MA). High-molecular calf thymus genomic DNA was purchased from Sigma-Aldrich, Inc. (St. Louis, MO).

2.2. DNA glycosylase-catalyzed reactions and analysis of excised DNA lesions by gas chromatography-tandem mass spectrometry with isotope-dilution

Calf thymus DNA in a N₂O-saturated buffered aqueous solution was γ-irradiated in a ⁶⁰Co-γ source at a dose of 5 Gy and then dialyzed as described previously [38]. Unirradiated control samples were also dialyzed. Aliquots of 50 μg of DNA samples were dried in a SpeedVac. For each data point, a triplicate of 50 μg of DNA samples were supplemented with aliquots of ThyGly-²H₄, 5-OH-Cyt-¹³C,¹⁵N₂, 5-OH-5-MeHyd-¹³C,¹⁵N₂, FapyAde-¹³C,¹⁵N₂, 8-OH-Ade-¹³C,¹⁵N₂, FapyGua-¹³C,¹⁵N₂ and 8-OH-Gua-¹⁵N₅ as internal standards, which are a part of the National Institute of Standards and Technology Standard Reference Material 2396 Oxidative DNA Damage Mass Spectrometry Standards (NIST SRM 2396) (for details see http://www.nist.gov/srm/index.cfm and https://www-s.nist.gov/srmors/view_detail.cfm?srm=2396). The samples were then dissolved in 50 μL of an incubation buffer consisting of 50 mmol/L phosphate buffer (pH 7.4), 100 mmol/L KCl, 1 mmol/L ethylenediaminetetraacetic acid (EDTA), and 0.1 mmol/L dithiothreitol. The samples were incubated with 2 μg of NTHL1, NTHL1-D239Y, NEIL1, OGG1, E. coli Nth or E.coli Fpg at 37 ^◦C for 60 min. For time-course experiments, incubation times of 10 min, 20 min, 30 min, 45 min and 60 min were used. For the competition assay, DNA samples were pre-incubated with 1 μg, 2 μg, 3 μg or 4 μg of NTHL1-D239Y for 60 min and then incubated with 2 μg of wild-type NTHL1 for another 60 min. Control samples without added DNA glycosylases were also incubated for 60 min. For another set of the competition assays, DNA samples were pre-incubated with 2 μg of NTHL1-D239Y for 10 min, and then incubated with 2 μg of wild-type NTHL1 for 1, 2, 3, 4, and 5 min. Without the pre-incubation with NTHL1-D239Y, another set of DNA samples were incubated with 2 μg of wild-type NTHL1 for the same period of times. After incubations, an aliquot of 100 μL ethanol was added to the samples to precipitate DNA and to stop the reaction. After centrifugation, the supernatant fractions were separated, lyophilized, derivatized by trimethylsilylation and then analyzed by GC-MS/MS using multiple reaction monitoring as described previously [38–40]. The following mass/charge (m/z) transitions were used for identification and quantification: m/z 448 → m/z 259 and m/z 452 → m/z 262 for both cis- and trans-diastereomers of ThyGly and ThyGly-²H₄, respectively; m/z 343 → m/z 342, and m/z 346 → m/z 345 for 5-OH-Cyt and 5-OH-Cyt-¹³C,¹⁵N₂, respectively; m/z 331 → m/z 316 and m/z 334 → m/z 319 for 5-OH-5-MeHyd and 5-OH-5-MeHyd-¹³C,¹⁵N₂, respectively; m/z 369 → m/z 368 and m/z 372 → m/z 371 for FapyAde and FapyAde-¹³C,¹⁵N₂, respectively; m/z 367 → m/z 352 and m/z 370 → m/z 355 for 8-OH-Ade and 8-OH-Ade-¹³C,¹⁵N₂, respectively; m/z 457 → m/z 368 and m/z 460 → m/z 371 for FapyGua and FapyGua-¹³C,¹⁵N₂, respectively; m/z 455 → m/z 440 and m/z 460 → m/z 445 for 8-OH-Gua and 8-OH-Gua-¹⁵N₅, respectively. The quantification was achieved using the integrated areas of the m/z transitions of the monitored DNA base lesions and those of their stable isotope-labeled analogues.

2.3. Statistical analysis

Three independently prepared DNA samples were used for each data point. Statistical analyses of the data were performed using the GraphPad Prism 8.4.3 software (GraphPad Software, LLC, La Jolla, California) ordinary one-way ANOVA and Tukey’s multiple comparisons test.

3. Results and discussion

In the present work, we used an approach different from the previous work to test the activities of wild-type NTHL1 and NTHL1-D239Y using high-molecular genomic DNA containing multiple DNA base lesions formed from all four DNA bases. The technique of gas chromatographyisotope-dilution tandem mass spectrometry (GC-MS/MS) was used to identify and quantify DNA base lesions. GC-MS/MS simultaneously identifies and quantifies DNA base lesions in a given DNA sample, thus enabling the determination of which damaged bases are or are not excised from DNA by a DNA glycosylase (reviewed in [14]). High-molecular genomic calf thymus DNA was irradiated at 5 Gy of ionizing radiation and used for all experiments. This low dose of γ-radiation in aqueous solution of DNA generates lesions from all DNA bases. The levels of eight DNA base lesions were measured. Fig. 1 shows the structures of these lesions. Seven lesions were excised by wild-type NTHL1 with different efficiencies, namely, cis-ThyGly, trans-ThyGly, 5-OH-Cyt, 5-OH-5-MeHyd, FapyAde, 8-OH-Ade and FapyGua. Time dependence of excision was measured at incubation times from 10 min to 60 min (Fig. 2A-H). No excision of 8-OH-Gua was observed (Fig. 2H). These results are comparable with those of previous studies performed with oligodeoxynucleotides and genomic DNA containing multiple base lesions in vitro, and with knockout animals in vivo [17–23,26–29,33,41]. The excised levels of both cis-ThyGly and trans-ThyGly were the greatest among the measured lesions with the level of the latter being approximately 2-fold greater than that of the former (Fig. 2A and B). Although there are four diastereomers of 2^′-deoxythymidine glycol, i.e., 5R,6S and 5S,6R (cis-diastereomers), and 5R,6R and 5S,6S (trans--diastereomers) [42–45], GC-MS/MS cannot separate the cis-pairs and trans-pairs of the excised ThyGly as a free base. Thus, each pair is detected as one separate signal, and designated as cis-ThyGly or trans--ThyGly. This is the first report of the excision by NTHL1 of both cis- and trans-diastereomers of ThyGly from genomic DNA containing multiple lesions. Previously, the stereoselective excision of the diastereomers of ThyGly by mouse NTHL1 and human NTHL1 from oligodeoxynucleotides containing a single ThyGly has been reported with the excision of trans-ThyGly being more efficient than that of cis-ThyGly [22,46]. Our results are on a par with these previous observations on the excision of the diastereomers of ThyGly. In contrast to wild-type NTHL1, no excision by NTHL1-D239Y was detected for any of the seven lesions (Fig. 2A to G). The failure of excision of cis-ThyGly or trans-ThyGly is consistent with the previous work that used an oligodeoxynucleotide containing a single ThyGly without a distinction between its diastereomers [33]. It should be pointed out that DNA lesions investigated in this work possess mutagenic/cytotoxic properties (reviewed in [47,48]). Thymine glycol, the major substrate of NTHL1, is poorly mutagenic, and constitutes a strong block to DNA polymerases and thus, a lethal lesion [49]. In E. coli, ThyGly yielded a low T → C transition mutation; however, it was not mutagenic in duplex DNA [50]. 5-OH-5-MeHyd is also a strong blocking lesion to various DNA polymerases. On the other hand, 5-OH-Cyt and the purine-derived lesions are bypassed by DNA polymerases and paired with non-cognate DNA bases, leading to various types of mutations [47]. For example, FapyGua leads to G → T transversion mutations as it pairs with Ade, and is more mutagenic than the mostly investigated 8-OH-Gua [51,52], which is not a substrate of NTHL1 [26–28] (Fig. 2H). Moreover, FapyGua is formed in cellular DNA in comparable or greater yields than 8-OH-Gua [28,53]).

Fig. 1.

Structures of the DNA base lesions measured in this work.

Fig. 2.

Time course of excision of eight DNA base lesions by six DNA glycosylases. Uncertainties are standard deviations.

We also tested two other major human DNA glycosylases, NEIL1 and OGG1, and two E. coli DNA glycosylases, Fpg and Nth, which are homologs of OGG1 and NTHL1, respectively, to compare their activities for the excision of the eight lesions with that of NTHL1. These experiments were also necessary to ensure the capacity of the assays to detect the substrates of other major DNA glycosylases under identical experimental conditions, and thus rigorously define the limit and preference of substrates for wild-type NTHL1. The results are also shown in Fig. 2A-H. NTHL1 and E. coli Nth exhibited similar activities for the excision of cis-ThyGly and trans-ThyGly (Fig. 2A and B), FapyAde (Fig. 2E) and 8-OH Ade (Fig. 2F). However, the activity of NTHL1 for the excision of FapyGua was at least 5-fold greater than that of E. coli Nth (Fig. 2G). (The fold comparisons are expressed for excisions at 60 min.) In contrast, E. coli Nth exhibited significantly greater activity for 5-OH-Cyt and 5-OH-5-MeHyd than NTHL1 (Fig. 2C and D). The activities of NTHL1 and NEIL1 were similar for 5-OH-5-MeHyd, FapyAde and FapyGua and (Fig. 2D, E and G, respectively). For the other lesions, the activity of NTHL1 was greater than that of NEIL1 (Fig. 2A, B, C and F). OGG1 excised FapyGua and 8-OH-Gua only (Fig. 2G and H), whereas E. coli Fpg exhibited an efficient excision of FapyAde, 8-OH-Ade, FapyGua and 8-OH-Gua, with the excision of 8-OH-Gua being the most efficient one followed by that of FapyGua (Fig. 2E, F, G and H, respectively). Both enzymes excised FapyGua with similar activities (Fig. 2G), whereas the excision of 8-OH-Gua by E. coli Fpg was almost two-fold greater than that of OGG1 (Fig. 2H).

Previously, it was found that both NTHL1 and NTHL1-D239Y bind with similar affinities to an oligodeoxynucleotide containing a single ThyGly [33]. When added to NTHL1, NTHL1-D239Y decreased overall glycosylase activity of NTHL1 for ThyGly, potentially suggesting both proteins compete for binding to this lesion. Interestingly, NTHL1-D239Y has been found to bind to tetrahydrofuran with affinity similar to that of the wild-type enzyme [33]. To test whether this type of a competition occurs when high-molecular genomic DNA containing multiple lesions is used as a substrate, the DNA samples were preincubated with 1, 2, 3, and 4 μg of NTHL1-D239Y, and then incubated with 2 μg of NTHL1. No statistically significant changes in excised levels of DNA lesions were observed as shown in Fig. 3A-D in the case of the four major substrates of NTHL1. When 1 μg of NTHL1 was used, no significant differences between the excised levels of DNA lesions were observed, either (data not shown). We also tested whether the competition occurs at lower incubation times with the lowest possible detection level of excision in a time-course experiment. For this purpose, we pre-incubated the DNA samples with NTHL1-D239Y and then incubated them with NTHL1 from 1 to 5 min. For comparison, another set of DNA samples were incubated with NTHL1 only at the same time periods. The results are shown in Fig. 4. Again, no significance difference was observed between the excised levels of the lesions under these two different conditions for any of the seven substrates of NTHL1. These results suggest that, when present with the wild-type NTHL1, the variant NTHL1-D239Y may not prevent the excision of the substrates of NTHL1 from genomic DNA containing multiple lesions, possibly indicating that no competition for binding to the substrates takes place between the two forms of the protein under the experimental conditions used in the present work. Alternatively, the lesions recognized by NTHL1, and other DNA glycosylases, may not be visible to these enzymes at all times. This is likely, given that DNA is packaged into chromatin inside cells. For example, recent experimental findings indicate that NTHL1 is unable to excise ThyGly from sterically occluded sites in model nucleosomes unless magnesium and ATP are present [54]. This indicates that living cells have a factor that facilitates chromatin remodeling with the purpose of enhancing the initiation of BER by DNA glycosylases. The interaction of DNA glycosylases with various substrates and how they locate these substrates among many undamaged DNA bases has been the subject of extensive studies in recent years (reviewed in [13]).

Fig. 3.

Excised levels of four DNA base lesions with 2 μg of NTHL1 following the pre-incubation of DNA samples with 1 μg, 2 μg, 3 μg, or 4 μg of NTHL1-D239Y. The level of ThyGly represents the total of the levels of cis- and trans-diastereomers. Uncertainties are standard deviations.

Fig. 4.

Time course of excision of seven DNA base lesions with 2 μg of wild-type NTHL1 from 1 min to 5 min following the pre-incubation of DNA samples with 2 μg of NTHL1-D239Y for 10 min ●: incubated with wild-type NTHL1 only; ■: preincubated with NTHL1-D239Y and then incubated with wild-type NTHL1. Uncertainties are standard deviations.

In conclusion, our study presents the first evidence that the variant of NTHL1 with the germline mutation D239Y in its active site is completely inactive for removal of the known seven pyrimidine- and purine-derived substrates of NTHL1 from high-molecular genomic DNA containing multiple DNA lesions. The substrates of NTHL1 are known to possess mutagenic/cytotoxic properties. The absolute inactivity of the variant NTHL1-D239Y toward these major oxidatively-induced DNA lesions points to the importance of the understanding of this variant’s role in carcinogenesis and, also in ageing. Future investigations may take into consideration these DNA lesions as the substrates of NTHL1 in the determination of the underlying mechanisms and the potential of individual susceptibility to cancer.