Expression of a germline variant in the N-terminal domain of the human DNA glycosylase NTHL1 induces cellular transformation without impairing enzymatic function or substrate specificity

Oxidatively-induced DNA damage, widely accepted as a key player in the onset of cancer, is predominantly repaired by base excision repair (BER). BER is initiated by DNA glycosylases, which locate and remove damaged bases from DNA. NTHL1 is a bifunctional DNA glycosylase in mammalian cells that predominantly removes oxidized pyrimidines. In this study, we investigated a germline variant in the N-terminal domain of NTHL1, R33K. Expression of NTHL1 R33K in human MCF10A cells resulted in increased proliferation and anchorage-independent growth compared to NTHL1 WT-expressing cells. However, wt-NTHL1 and R33K-NTHL1 exhibited similar substrate specificity, excision kinetics, and enzyme turnover in vitro and in vivo. The results of this study indicate an important function of R33 in BER that is disrupted by the R33K mutation. Furthermore, the cellular transformation induced by R33K-NTHL1 expression suggests that humans harboring this germline variant may be at increased risk for cancer incidence.


INTRODUCTION
Maintenance of genomic integrity requires proper functioning of the base excision repair (BER) pathway, which is responsible for the repair of at least 30,000 lesions per cell per day arising from endogenous insults [1,2]. BER is initiated when a DNA glycosylase locates a lesion and catalyzes the cleavage of the N-glycosyl bond between the base and 2'-deoxyribose generating an AP (apurinic/apyrimidinic) site. Bifunctional DNA glycosylases, which generally repair oxidatively-induced DNA lesions, possess both glycosylase and lyase activity. Therefore, bifunctional DNA glycosylases can potentially cleave the DNA backbone (utilizing their lyase activity) generating DNA ends that are further processed by AP endonuclease I (APE1) or polynucleotide kinase (PNK).
Insertion of the missing base by DNA polymerase β (Pol β) and sealing of the nick by X-Ray Cross-Complimenting 1 (XRCC1) complexed with DNA ligase I or IIIα completes repair [2,3]. Consisting of only four-five reaction steps, BER has traditionally been considered one of the simplest DNA repair pathways; however, recent studies have begun to unveil the complexities of BER [2]. Although BER can be reconstituted in vitro by combining the enzymes necessary for each step in the pathway, in vivo studies have begun to unveil the complex interplay between BER enzymes and other repair and non-repair proteins driving accurate and efficient repair in the cellular environment.
NTHL1, the mammalian homolog of Escherichia coli (E. coli) endonuclease III, is a bifunctional DNA glycosylase that primarily recognizes and removes a wide range of modified pyrimidine derivatives, and purine-Priority Research Paper derived 4,6-diamino-5-formamidopyrimidine (FapyAde) and 2,6-diamino-4-hydroxy-5-formamidopyrimidines (FapyGua) from DNA [2,[4][5][6][7][8][9][10][11][12]. The core catalytic regions of NTHL1, including the Lys 220 and Asp 239 catalytic residues, as well as the iron-sulfur binding cluster (4Fe-4S cluster loop) are highly conserved with its bacterial homolog. However, mammalian NTHL1 contains a non-conserved N-terminal domain of 95 amino acids that is predicted to be disordered [11]. Despite conservation of catalytic domains/residues and DNAbinding motifs between NTHL1 and E. coli endoIII, NTHL1 has been shown to differ in substrate specificity and excision kinetics from its bacterial homologues [5,13,14]. Truncation of the N-terminal domain up to 80 amino acids has been shown to increase product release and enzyme turnover without impacting glycosylase or AP lyase activities [15]. Collectively, the evidence suggests a negative regulatory role of the N-terminal domain, possibly preventing premature product release by NTHL1 and exposure of cytotoxic and mutagenic BER intermediates. Interestingly, there are many DNA glycosylases and BER proteins that contain regions of disorder that are not conserved in their prokaryotic orthologs, including NEIL1, NEIL2, OGG1, MPG, APE1 and Pol β reviewed in [16]. Disordered domains are presumed to function in protein: protein and protein: DNA interactions given their inherent structural flexibility and plasticity, however the functions of these disordered domains have yet to be elucidated. Similarly, the functions of the N-terminal disordered region regarding NTHL1 regulation, BER coordination, various cellular processes, or genomic stability remains entirely unknown.
Several studies have linked alterations of NTHL1 function with carcinogenesis. Double-knockout Nthl1 -/-/Neil1 -/mice have a higher incidence of pulmonary and hepatocellular tumors compared to singleknockout strains, which may have been associated with accumulation of FapyAde and FapyGua [7], which are mutagenic, with FapyGua being even more mutagenic than 8-hydroxyguanine (8-OH-Gua) [17][18][19][20]. Mislocalization of NTHL1 to the cytoplasm as well as single nucleotide polymorphisms (SNPs) in the promoter region of NTHL1, resulting in decreased expression, have been linked to the etiology of a subset of human gastric tumors [21]. Two independent research groups have identified a germline nonsense mutation in NTHL1 in patients with multiple primary tumors, including colorectal cancer [22,23]. The broad increase in tumor incidence in those patients harboring the NTHL1 nonsense mutation established a novel "NTHL1 syndrome" [23,24]. More recently, work from our laboratory showed that expression of a germline variant that results in mutation of aspartic acid (D) 239 to tyrosine (Y) in NTHL1 induced cellular transformation and genomic instability in human nontransformed epithelial cells [4]. These studies demonstrate the inextricable link of NTHL1 dysfunction with cancer, providing impetus for investigations into the potential impact of additional germline mutations in the gene encoding for NTHL1 on genome integrity and cellular transformation.
In this study, we have characterized a missense germline mutation that results in the mutation of arginine 33 to lysine (R33K) (dsSNP: rs2302172) in the N-terminal domain of NTHL1. Although predicted to be non-damaging, expression of R33K-NTHL1 in MCF10A cells results in cellular transformation without impairing enzymatic function or substrate specificity. Collectively, our data provide evidence that expression of a conservative single amino acid mutation in the N-terminal domain of NTHL1 has significant deleterious effects on vital cellular processes, independent of its catalytic functions. These findings begin to elucidate the critical functions of the N-terminal domain in NTHL1 and further underscore the need for continuous investigations into the impact of germline mutations in genes that encode for BER proteins on genome integrity.

Expression of NTHL1 R33K in a human epithelial cell line induces cellular transformation
To determine the cellular effects of R33K-NTHL1 expression, we generated MCF10A cells (a non-transformed human breast epithelial cell line) with stable and equivalent expression of HA-tagged wt-NTHL1 or HA-tagged R33K-NTHL1 as previously described ( Figure 1A) [4]. As a human germline mutation, we were first interested in testing whether R33K-NTHL1 expression in MCF10A cells could induce phenotypes indicative of cellular transformation. Therefore, we measured proliferation (using the CyQUANT NF Cell Proliferation Assay) and anchorage-independent growth (using the CytoSelect 96-well cell transformation Assay) of wt-NTHL1 and R33K-NTHL1 pools. As shown in Figure 1B-1D expression of R33K-NTHL1 resulted in increased anchorage independent growth and increased proliferation as compared to wt-NTHL1-expressing cells. R33K-NTHL1-expressing cells exhibited increased anchorage-independent growth by passage 14 ( Figure  1B, 1C) and increased proliferation by passage 18 ( Figure  1D). The highly conservative nature of the mutation (Arg → Lys), which maintains overall bulkiness and charge at amino acid position 33, prompted the additional generation of MCF10A HA-tagged wt-NTHL1 or HA-tagged R33K-NTHL1-expressing cells with knockdown of endogenous NTHL1 (wt/R33K-NTHL1 shNTH) to reduce possible compensation from the endogenous protein and consequent masking of potential cellular phenotypes. Two short-hairpin RNAs (shRNA) with similar sequence homology to the target sequence in NTHL1 (with binding sites in NTHL1 that overlap) were utilized to increase the efficacy of knockdown and provide an additional level of control of expression. The shRNAs were stably expressed in MCF10A cells and expression of endogenous NTHL1 was measured by Western blot. As shown in Figure 2A, expression of endogenous NTHL1 was efficiently reduced. To achieve knockdown of endogenous NTHL1 without impacting expression of exogenously expressed wt-NTHL1 or R33K-NTHL1, two silent mutations were generated in the open reading frame (ORF) of wt-NTHL1 and R33K-NTHL1 (labeled as NTHL1 g48a) within the sequence targeted by two overlapping shRNAs ( Figure 2B). Therefore, the exogenous NTHL1 proteins were resistant to silencing by the shRNAs while achieving efficient knockdown of endogenous NTHL1 protein expression ( Figure 2B, labeled herein as wt-NTHL1 shNTH or R33K-NTHL1 shNTH). As with the dominant pools, we measured proliferation and anchorage-independent growth of wt-NTHL1 shNTH R33K-NTHL1 shNTH pools beginning at passage 2 with equivalent expression of exogenous HA-tagged wt-NTHL1 or R33K-NTHL1 ( Figure 2C). As shown in Figure 2D-2F, expression of R33K-NTHL1 shNTHL1 resulted in increased anchorage independent growth and increased proliferation, respectively, as compared to wt-NTHL1-expressing cells. R33K-NTHL1 shNTH-expressing cells exhibited increased anchorage independent growth ( Figure 2E) and proliferation ( Figure 2F) as early as passage 6. Furthermore, we show that expression of NTHL1 R33K is required for increased anchorage-independent growth. Importantly, shNTH clonal cell lines with R33K-NTHL1 expression exhibit significantly increased levels of colonies growing in an anchorage independent manner when compared to shNTH cells that do not express R33K-NTHL1 ( Figure 3A, 3B). Notably, knockdown of endogenous NTHL1 alone does not induce cellular transformation up to passage 16 ( Figure  3C). Therefore, the expression of R33K-NTHL1 induces cellular transformation, which is significantly accelerated in the absence of endogenous NTHL1.

NTHL1 R33K does not alter cellular localization
The R33K germline mutation is located between the mitochondrial and nuclear localization signals in the N-terminal domain of NTHL1. Therefore, we tested whether cellular localization of NTHL1 is disrupted by the R33K mutation by preparing cytoplasmic and nuclear extracts then blotting for NTHL1 and alpha tubulin (as a control for separation of the cellular compartments). As shown in Figure 4, wt-NTHL1 and R33K-NTHL1 exhibit similar cellular localization. These data demonstrate that R33K has no impact on cellular localization of NTHL1 and therefore the transformative phenotypes observed in R33K-NTHL1 shNTH-expressing cells cannot be attributed to changes in cellular localization.

NTHL1 R33K does not impact enzymatic functions
Although previous studies have shown that the N-terminal domain (up to 80 amino acids) is dispensable for glycosylase/lyase functions [15], we tested whether R33K-NTHL1 disrupts excision of ThyGly from a double-stranded oligodeoxynucleotide (dsOligo) substrate under multiple turnover conditions using bacterially-expressed wt-NTHL1 and R33K-NTHL1. As expected, bacterially-expressed R33K-NTHL1 exhibited excision kinetics equivalent to that of wt-NTHL1 ( Figure 5A). A previous study showed that truncation of the N-terminal domain (up to 80 amino acids) decreases product inhibition of NTHL1, leading to increased enzymatic turnover [15]. Given the location of R33K in the N-terminal domain of NTHL1, we tested whether this mutation altered product inhibition similar to the N-terminal truncation mutant, Δ63-NTHL1. As show in Figure 5B, R33K-NTHL1 does not increase enzyme turnover as compared to Δ63-NTHL1, included as a positive control. We also tested for differences in enzymatic activity of DYKDDDDK tag (FLAG)-purified wt-NTHL1 and R33K-NTHL1 from human embryonic kidney cells (HEK293T cells) using a dsOligo ThyGlysubstrate. However, there were no differences in excision kinetics detected between FLAG-purified wt-NTHL1 and R33K-NTHL1 proteins ( Figure 5C, 5D). To test for changes in substrate specificity, γ-irradiated calf thymus DNA (containing a broad range of DNA lesions) was incubated with bacterially-expressed wt-NTHL1, R33K-NTHL1 or Δ63-NTHL1, and the number of DNA base lesions excised from the DNA were identified and quantified by GC-MS/MS. Wt-NTHL1 and R33K-NTHL1 excised similar levels of nine DNA base lesions ( Figure 6), exhibiting similar substrate specificities. These results are consistent with the results presented in Figure 5B, showing similar excision kinetics of ThyGly by wt-NTHL1 and R33K-NTHL1. No significant excision of 8-OH-Gua was observed in agreement with previous results [5,6]. The greater excision of most of the lesions by Δ63-NTHL1 than the excision by wt-NTHL1 or R33K-NTHL1 is on a par with the data in Figure 5B in terms of the greater excision of ThyGly by Δ63-NTHL1 from the dsOligo ThyGly-substrate than the excision by wt-NTHL1 or R33K-NTHL1. (A) Western blot demonstrating that shRNA expression constructs effectively knockdown endogenous NTHL1 in MCF10A cells. Image has been cropped and contrast/brightness has been adjusted for both channels for visualization purposes only. Quantitation of bands performed using data embedded in image file at time of collection. (B) Western blot demonstrating that NTHL1 g48a construct is resistant to targeting by overlapping shRNA constructs with overlapping sequence homology in NTHL1 (binding to similar sequence in NTHL1) and shGFP construct does not reduce exogenous or endogenous NTHL1 expression. Image has been cropped and contrast/brightness has been adjusted for both channels for visualization purposes only. Quantitation of bands performed using data embedded in image file at time of collection.

R33K-NTHL1 does not impact enzymatic functions in vivo under normal or oxidativelystressed conditions
It has been proposed that initial scanning by DNA glycosylases for DNA base lesions may be mediated by electrostatic, non-specific and mostly transient interactions between the DNA and disordered domains present in eukaryotic DNA glycosylases, given the inherent flexibility of disordered regions as well as the clustering of positively charged and basic residues (as seen in the disordered domain of NTHL1) [for a review see [16]]. Therefore, the R33K mutation may disrupt the scanning behavior of NTHL1 within the context of the cell. Additionally, DNA in the cell is organized into chromatin which can impede access to lesions by DNA glycosylases and thus inhibit repair. However, lesion recognition and excision kinetics has thus far been tested on dsOligos and de-proteinized calf thymus DNA. To test this hypothesis, asynchronous MCF10A wt-NTHL1-and R33K-NTHL1 shNTH-expressing cells were treated with 10 mmol/L H 2 O 2 . Genomic DNA was isolated from control and treated cells, and DNA base lesions were identified and quantified by GC-MS/MS. As Figure 7 shows, the background levels and oxidatively-induced levels of each of the six base lesions were similar in both untreated wt-NTHL1 and R33K-NTHL1 shNTH-expressing cells (no significant differences detected). Following the H 2 O 2treatment, significant increases in the levels of DNA lesions were observed (except for in the level of 5-OH-Cyt) in wt-NTHL1 and R33K-NTHL1 shNTH-expressing cells. The levels of 8-OH-Ade and 5,6-diOH-Ura were also measured; however, no increases in their levels were observed over the background levels in either cell lines (data not shown). These results indicate that R33K-NTHL1 does not disrupt DNA scanning, initial lesion recognition, or lesion excision by NTHL1 broadly in vivo. They are also on a par with the results shown in Figure 6. Collectively, these data demonstrate that R33K has no impact on enzymatic function or substrate specificity of NTHL1 and therefore the transformative phenotypes observed in R33K-NTHL1 shNTH-expressing cells cannot be attributed to disruption of enzymatic repair functions.

DISCUSSION
In this study, we have shown that expression of the NTHL1 germline variant, R33K-NTHL1, in MCF10A cells induces cellular transformation as measured Figure 4: R33K does not impact cellular localization. Fractionation of cytoplasmic and nuclear compartments in MCF10A wt-NTHL1 or R33K-NTHL1 shNTH cell lysates demonstrates nuclear localization of wt-and R33K-NTHL1. Nuclear localization of wt-NTHL1 serves as a positive control for nuclear localization of R33K-NTHL1. Image was obtained using an Odyssey CLx Infrared Imaging System. Fluorescent signals were measured using a 700 Channel laser source (solid state diode laser at 785 nm) and 800 Channel laser source (solid state diode laser at 785 nm). Band intensities were quantified using the data recorded during image capture using Image Studio™ Software (LI-COR). by increased proliferation and increased anchorageindependent growth. Investigations into the molecular mechanisms driving cellular transformation in R33K-NTHL1 shNTH-expressing cells indicate that R33K mutation does not impact NTHL1 enzymatic functions, substrate specificity or coordination of BER in the cell. The cellular transformation induced by the expression of R33K-NTHL1 implicate a potential role for this germline mutation in the initiation of human cancer and therefore individuals harboring this mutation may be at a higher risk for lifetime cancer incidence.
Mutation of arginine 33 to a lysine is a highly conservative amino acid change, maintaining charge and overall bulkiness at the residue. Folding or stability of the protein would not be impacted given the location of R33K in the N-terminal domain of NTHL1, which is predicted to be disordered (Polyphen). As we have shown in this study, the R33K mutation has no observable impact on glycosylase/lyase functions or substrate specificity. This is not altogether surprising given a previous study that showed truncation of the N-terminal domain of NTHL1 (up to 80 amino acids) had no effect on glycosylase/lyase activity but did increase product release and enzyme turnover, thus resulting in enhanced activity under enzyme limiting conditions [15]. Although located in the N-terminal domain, R33K-NTHL1 does not increase enzyme turnover as observed with Δ63-NTHL1. Despite no observed changes in enzymatic activity or substrate recognition/specificity, expression of R33K-NTHL1 in MCF10A cells results in increased proliferation and anchorage-independent growth which are indicative of cellular transformation. Therefore, the question remains how a highly conservative mutation in the disordered domain of NTHL1 that is dispensable for its enzymatic functions induces cellular transformation. The disordered region in NTHL1 is not conserved with its bacterial homologue, which suggests an evolutionary adaptation for the complexity of repair mechanisms in mammalian systems. NTHL1 has been shown to differ in substrate specificity and excision kinetics from its bacterial homolog [5,6,13,14]. As previously discussed, the N-terminal domain in NTHL1 was shown to decrease enzyme turnover in vitro [15]. These studies suggest that the N-terminal domain of NTHL1 negatively regulates NTHL1 function, however no studies to date have elucidated the functions of the N-terminal domain of NTHL1. In fact, our study provides the first evidence for a critical role of the N-terminal domain in NTHL1 in cellular processes that prevent cellular transformation. Therefore, characterization of NTHL1 as a tumor suppressor does not depend only on the catalytic functions (glycosylase/ lyase activities) but also on the functions of the N-terminal disordered region. One likely function of the disordered N-terminal domain in NTHL1 is to serve as an interface for interactions with other BER and non-BER proteins. One possible hypothesis we propose is that R33 is a key amino acid for specific protein interactions and R33K may disrupt those NTHL1-protein interactions resulting in altered NTHL1 function and the cellular outcomes observed. Investigations are ongoing to identify potential NTHL1-protein interactions by the R33K mutation.

Plasmids, cloning and recombinant proteins
A short hairpin RNA (shRNA)-resistant human wt-NTHL1 vector (referred to as NTHL1 g48a) was generated by introducing two silent mutations [25] using site-directed mutagenesis within the overlapping binding regions of TRC human NTHL1 shRNA constructs (CloneID: TRC0000007915 and CloneID: TRC0000007916) (Dharmacon). NTHL1 R33K was generated using standard methods of site-directed mutagenesis using the shRNAresistant wt-NTHL1 g48a construct as a template. Wt-NTHL1 and R33K-NTHL1 were expressed and purified from E.coli as previously described [4].

Thymine glycol excision assay
The active site fraction of the recombinant wt-NTHL1 and R33K-NTHL1 were equivalent [assessed using a Schiff-base trap assay [26]]. All enzyme concentrations indicated in the text and figure legends refer to the active enzyme concentration. An oligodeoxynucleotide (oligo) (35 bases) with a centrally positioned thymine glycol (ThyGly) was 5ʹ end labeled with 32 P as previously described [26]. The labeled oligo was annealed to a compliment containing an adenosine opposite the ThyGly. Excision of ThyGly was assayed at 25°C in 100 mmol/L NaCl, 25 mmol/L 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 8.0, 2 mmol/L ethylenediaminetetraacetic acid (EDTA), and 100 µg/mL bovine serum albumin (BSA). At various time-points aliquots were taken and quenched in 400 mmol/L NaOH and heated to 95°C for 2 min. Two volumes of formamide (96%), 20 mmol/L EDTA was added. Substrate was separated from product using 12% denaturing polyacrylamide gel electrophoresis (PAGE) in 1× Tris/Borate/EDTA (TBE) buffer. Bands were visualized and quantified using the Pharos FX plus Phosphoimager (BioRad). Product was graphed as a function of time and data were fit to a double exponential equation (GraphPad Prism). Enzymes purified from HEK293 cells were quantified by western blot, diluted to equivalent concentration and assayed as above.

Cell fractionation
Cytoplasmic and nuclear fractions were isolated from cell lysates using the NE-PER Nuclear and Cytoplasmic Extraction kit (Thermo Scientific) as per the manufacturers' instructions.

Measurement of DNA base lesions in wt-NTHL1 and NTHL1 R33K-expressing cells
Genomic DNA was isolated from asynchronous MCF10A wt-NTHL1 or R33K-NTHL1 shNTHexpressing cells either untreated or treated with 10 mmol/L H 2 O 2 for 30 min at 4°C followed by a 30 min recovery using the Qiagen Blood and Cell Culture DNA Maxi kit without the use of phenol as per the manufacturers' instructions. The eluted genomic DNA was ethanol-precipitated then washed 3 times with ethanol. Ethanol was removed, and the DNA was dissolved in water for 18 h at 4°C, and then quantified using an absorption spectrophotometer (absorption of 1 = 50 μg DNA). Aliquots (50 μg) of the DNA samples were dried in a SpeedVac under vacuum. DNA samples were 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 EDTA and 0.1 mmol/L dithiothreitol. Aliquots of FapyAde-13 C, 15 N 2 , FapyGua-13 C, 15 N 2 , 8-OH-Ade-13 C, 15 N 2 , 8-OH-Gua-15 N 5 , 5-OH-Cyt-13 C, 15 N 2 , 5-OH-Ura-13 C 4 , 15 N 2 , ThyGly-2 H 4 , 5-OH-5-MeHyd-13 C, 15 N 2 and 5,6-diOH-Ura-13 C, 15 N 2 (isodialuric acid-13 C, 15 N 2 ) were added as internal standards. Samples were incubated with 1 μg of E. coli Nth plus 1 μg of E. coli Fpg for 1 h at 37°C in a water bath to release DNA lesions from DNA samples. After incubation, 125 μL of cold ethanol (kept at -20°C) were added to the samples to stop the reaction and to precipitate DNA. The samples were kept at -20°C for 1 h, and then centrifuged at 15000 g for 30 min at 4°C. DNA pellets and supernatant fractions were separated. Ethanol was removed from the supernatant fractions under vacuum in a SpeedVac. Aqueous supernatant fractions were frozen at -80°C for 1 h and then lyophilized to dryness for 18 h. For all measurements, 3 independently prepared DNA samples were used. Dried samples were derivatized and analyzed by gas chromatography-tandem mass spectrometry (GC-MS/MS) as described previously [27]. www.oncotarget.com Determination of the substrate specificities of wt-NTHL1, R33K-NTHL1 and ∆63-NTHL1 Commercially available calf thymus DNA was dissolved in phosphate buffer (pH 7.4) at a concentration of 0.3 mg/mL. This solution was bubbled with N 2 O for 30 min and irradiated with γ-rays in a 60 Co γ ray-source at a dose of 10 Gy (dose rate 5.2 Gy/min). Irradiated DNA sample was dialyzed against water at 4°C for 18 h. Water outside the dialysis tube was changed three times during the dialysis. After dialysis, the concentration of DNA in the dialyzed sample was measured by absorption spectrophotometry. Aliquots (50 μg) of the DNA sample were dried in a SpeedVac under vacuum. Unirradiated DNA samples as controls were treated in the same manner. DNA samples were 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 EDTA and 0.1 mmol/L dithiothreitol. Aliquots of the stable isotopelabeled internal standards were added as described above. Samples were incubated with 1 μg of wt-NTHL1, 1 μg of R33K-NTHL1, or 1 μg of ∆63-NTHL1 for 1 h at 37°C in a water bath. Three other DNA samples to be used as controls were incubated without any enzyme. After incubation, 125 μL of cold ethanol (kept at -20°C) were added to the samples to stop the reaction and to precipitate DNA. The samples were kept at -20°C for 1 h, and then centrifuged at 15000 g for 30 min at 4°C. DNA pellets and supernatant fractions were separated. Ethanol was removed from the supernatant fractions under vacuum in a SpeedVac. Aqueous supernatant fractions were frozen at -80°C for 1 h and then lyophilized to dryness for 18 h. For all measurements, 3 independently prepared DNA samples were used. Dried samples were derivatized and analyzed by GC-MS/MS as described previously [27].

CONCLUSIONS
Our study provides evidence that the germline mutation, R33K-NTHL1, induces cellular transformation. Importantly, this study further underscores the importance for investigations into germline mutations in NTHL1 and other BER proteins in order to identify mutations that increase the risk for cellular transformation and may lead to increased cancer incidence in humans harboring these mutations. Ultimately, the dysfunctional phenotypes exhibited by R33K-NTHL1 in this study and in future studies of NTHL1 variants will provide invaluable clues into the broad functions of NTHL1, and more generally BER, in diverse cellular processes.

ACKNOWLEDGMENTS AND FUNDING
We would like to acknowledge members of the Sweasy and Pederson labs for their discussions and advice. Certain commercial equipment or materials are identified in this paper in order to specify adequately the experimental procedure. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose. DNA samples isolated from human MCF10A cells at the University of Vermont were provided to the National Institute of Standards and Technology pursuant to a Material Transfer Agreement between the two institutions. This work was supported in part by P01CA98993 from the National Cancer Institute.