TP53 mutations induced by BPDE in Xpa-WT and Xpa-Null human TP53 knock-in (Hupki) mouse embryo fibroblasts.

Somatic mutations in the tumour suppressor gene TP53 occur in more than 50% of human tumours; in some instances exposure to environmental carcinogens can be linked to characteristic mutational signatures. The Hupki (human TP53 knock-in) mouse embryo fibroblast (HUF) immortalization assay (HIMA) is a useful model for studying the impact of environmental carcinogens on TP53 mutagenesis. In an effort to increase the frequency of TP53-mutated clones achievable in the HIMA, we generated nucleotide excision repair (NER)-deficient HUFs by crossing the Hupki mouse with an Xpa-knockout (Xpa-Null) mouse. We hypothesized that carcinogen-induced DNA adducts would persist in the TP53 sequence of Xpa-Null HUFs leading to an increased propensity for mismatched base pairing and mutation during replication of adducted DNA. We found that Xpa-Null Hupki mice, and HUFs derived from them, were more sensitive to the environmental carcinogen benzo[a]pyrene (BaP) than their wild-type (Xpa-WT) counterparts. Following treatment with the reactive metabolite of BaP, benzo[a]pyrene-7,8-diol-9,10-epoxide (BPDE), Xpa-WT and Xpa-Null HUF cultures were subjected to the HIMA. A significant increase in TP53 mutations on the transcribed strand was detected in Xpa-Null HUFs compared to Xpa-WT HUFs, but the TP53-mutant frequency overall was not significantly different between the two genotypes. BPDE induced mutations primarily at G:C base pairs, with approximately half occurring at CpG sites, and the predominant mutation type was G:C>T:A in both Xpa-WT and Xpa-Null cells. Further, several of the TP53 mutation hotspots identified in smokers' lung cancer were mutated by BPDE in HUFs (codons 157, 158, 245, 248, 249, 273). Therefore, the pattern and spectrum of BPDE-induced TP53 mutations in the HIMA are consistent with TP53 mutations detected in lung tumours of smokers. While Xpa-Null HUFs exhibited increased sensitivity to BPDE-induced damage on the transcribed strand, NER-deficiency did not enhance TP53 mutagenesis resulting from damage on the non-transcribed strand in this model.

Introduction

The tumour suppressor p53 plays a crucial role in the DNA damage response, garnering the title ‘guardian of the genome’ [1]. A paramount function of p53 is to prevent DNA synthesis and cell division, or to promote apoptosis, following DNA damage, which it performs primarily by regulating a large network of transcriptional targets [2,3]. Disruption of the normal p53 response by TP53 mutation contributes to transformation by eliminating a key pathway of cellular growth control, enabling the survival and proliferation of stressed or damaged cells. Somatic mutations in TP53 occur in more than 50% of human cancers [4,5]. The majority of TP53 mutations are missense and occur between codons 125 and 300, corresponding to the coding region for the DNA binding domain [6]. Over 28,000 TP53 mutations from human tumours have been catalogued in the International Agency for Research on Cancer (IARC) TP53 mutation database, providing a key resource for studying the patterns and frequencies of these mutations in cancer [7]. Interestingly, exposure to some environmental carcinogens can be linked to characteristic signatures of mutations in TP53, which provide molecular clues to the aetiology of human tumours [8].

A useful model for studying human TP53 mutagenesis is the partial human TP53 knock-in (Hupki) mouse, in which exons 4-9 of human TP53 replace the corresponding mouse exons [9]. The Hupki mouse and Hupki mouse embryo fibroblasts (HUFs) have been used for both in vivo and in vitro studies of TP53 mutations induced by environmental carcinogens [10,11]. TP53 mutagenesis can be studied in cell culture using the HUF immortalization assay (HIMA). In this assay primary HUFs are first treated with a mutagen to induce mutations. The treated cultures, along with untreated control cultures, are then serially passaged under standard culture conditions, whereby the majority of HUFs will undergo p53-dependent senescent growth arrest, due to the sensitivity of mouse cells to atmospheric oxygen levels (20%). HUFs that have accumulated mutagen-induced or spontaneous mutations (e.g. in TP53) that enable bypass of senescence continue to proliferate and ultimately become established into immortalized cell lines. DNA from immortalized HUF clones is then sequenced to identify TP53 mutations. Environmental carcinogens that have been examined using the HIMA include ultraviolet (UV) radiation [12], benzo[a]pyrene (BaP) [13,14] and aristolochic acid I (AAI) [12,15]; in all cases the induced TP53 mutation pattern corresponded to the pattern found in human tumours from patients exposed to these mutagens.

To protect the genome from mutation, several efficient mechanisms exist in cells to repair damage to DNA. One key repair system responsible for removing damage induced by certain environmental carcinogens is the nucleotide excision repair (NER) pathway. NER removes several types of structurally distinct DNA lesions including UV-induced photolesions, intrastrand crosslinks and chemically-induced bulky DNA adducts, such as those formed after exposure to polycyclic aromatic hydrocarbons (PAHs) [16]. NER operates in two distinct subpathways: global genomic NER (GG-NER) that recognizes lesions that cause local structural distortions in the genome, and transcription-coupled NER (TC-NER) that responds to lesions that block the progression of RNA polymerase II (RNAPII) on the transcribed strand of transcriptionally active genes. Following damage recognition, a common set of factors are recruited that ultimately incise the DNA 5′ and 3′ to the lesion to remove a 24-32 nucleotide fragment. The undamaged strand serves as a template for replicative DNA polymerases to fill the gap, which is finally sealed by ligation [17].

Mouse models deficient in various NER components have been generated not only to study the role of NER in the repair of different types of damage and ascertain how this relates to cancer risk [18,19], but also to increase the sensitivity of carcinogenicity studies [20]. For example, Xpa-knockout (Xpa-Null) mice, or cells derived from them, are deficient in both GG-NER and TC-NER. Xpa-Null mice are highly sensitive to environmental carcinogens [18,21] and exhibit accelerated and enhanced tumour formation after treatment with carcinogens such as UV and PAHs like BaP, compared with wild-type (Xpa-WT) mice [19,22,23]. Increased mutation frequencies of a lacZ reporter gene have been measured in tissues from Xpa-Null mice treated with the aforementioned carcinogens, and an increased rate of p53-mutated foci was detected on the skin of Xpa-Null Trp53(+/−) mice exposed to UVB [21,22,24]. Further, in in vitro studies, cells with reduced or deficient repair capacity were also more sensitive to the lethal or mutagenic effects of DNA damage [18,25,26].

Here we have generated an Xpa-deficient Hupki mouse strain with the aim of increasing TP53 mutation frequency in the HIMA. As Xpa-Null cells are completely deficient in NER, we hypothesized that carcinogen-induced DNA adducts would persist in the TP53 sequence of Xpa-Null HUFs, leading to an increased propensity for mismatched base pairing and mutation during replication of adducted DNA [24,27]. In the present study primary Xpa-WT and Xpa-Null HUFs were treated with benzo[a]pyrene-7,8-diol-9,10-epoxide (BPDE), the activated metabolite of the human carcinogen BaP [28,29], which forms pre-mutagenic BPDE-DNA adducts (i.e. 10-(deoxyguanosin-N2-yl)-7,8,9-trihydroxy-7,8,9,10-tetrahydrobenzo[a]pyrene [BPDE-N2-dG]) that can be removed by NER (Fig. S1) [30]. BPDE-treated HUFs were subjected to the HIMA and TP53 mutations in immortalized clones were identified by direct dideoxy sequencing of exons 4-9. The induced TP53 mutation patterns and spectra were compared between the two Xpa genotypes and to mutations found in human tumours.

Materials and methods

Carcinogens

BaP (#B1760) was purchased from Sigma-Aldrich. For in vitro treatments, BaP was dissolved in DMSO (Sigma #D2650) to a stock concentration of 1 mM and stored at −20 °C. For in vivo treatments, BaP was dissolved in corn oil at a concentration of 12.5 mg/mL. BPDE was synthesized at the Institute of Cancer Research (London, UK) using a previously published method [31]. BPDE was dissolved in DMSO to a stock concentration of 2 mM under argon gas and stored at −80 °C in single-use aliquots.

Details of mouse strains and crossbreeding

Hupki mice (Trp53tm1/Holl (Arg/Arg codon 72), homozygous for a knock-in TP53 allele harbouring the wild-type human TP53 DNA sequence spanning exons 4–9) in the 129/Sv background [9] were kindly provided by Monica Hollstein (German Cancer Research Center; Heidelberg, Germany). Transgenic Xpa mice, heterozygous for the Xpa-knockout allele on a C57Bl/6 background [18,22] were obtained from the National Institute for Public Health and the Environment (Bilthoven, The Netherlands). In the Xpa-knockout allele, exon 3, intron 3 and exon 4 have been replaced by a neomycin resistance cassette with a PGK2 promoter. To generate Hupki mice carrying an Xpa-knockout allele, Hupki mice were first crossed with Xpa mice. Progeny with the Hupki;Xpa genotype were then backcrossed to Hupki stock to generate Hupki animals that were Xpa or Xpa. Hupki;Xpa and Hupki;Xpa offspring were thereafter intercrossed to maintain the colony and produce Xpa and Xpa (referred to here as Xpa-WT and Xpa-Null, respectively) mice and embryos for experiments. Animals were bred at the Institute of Cancer Research in Sutton, UK and kept under standard conditions with food and water ad libitum. All animal procedures were carried out under license in accordance with the law and following local ethical review.

Genotyping

The Hupki and Xpa genotype was determined in mouse pups or embryos by PCR prior to experiments. To extract DNA for genotyping, ear clips or cells were suspended in 400 μL of 50 mM NaOH and heated to 95 °C for 15 min. Next, 35 μL of 1 M Tris-HCl (pH 8.0) was added to each sample, followed by centrifugation for 20 min at 13,000 rpm. The supernatant was used for genotyping. Primers and PCR reaction conditions for the Hupki, mouse Trp53 or Xpa alleles are described in Table S1.

In-vivo carcinogen treatment

Female Xpa-WT and Xpa-Null Hupki mice (∼3 months old) were treated with BaP as indicated below and sacrificed either 24 h or 5 days after the last administration following treatment regimens published previously [28,32]. Several organs (liver, lung, small intestine, spleen, colon and kidney) were removed, snap frozen in liquid N2, and stored at −80 °C until analysis.

In the first experiment, three groups of animals (n = 3, each group) were treated orally with 125 mg/kg bw of BaP. Groups 1 and 2 received a single dose and Group 3 was dosed once daily for 5 days. Groups 1 and 3 were sacrificed 24 h after the last administration, and Group 2 was sacrificed 5 days after the last administration. In the second experiment, two groups of animals (Group 4 and 5; n = 3, each group) were treated with 12.5 mg/kg bw BaP. Group 4 was treated with a single dose and sacrificed 24 h later. Group 5 was dosed once daily for 5 days and sacrificed 24 h after the last administration. Matched control mice (n = 3) for each group received corn oil only. DNA adduct formation was assessed as described below.

DNA adduct analysis by 32P-postlabelling

Genomic DNA was isolated from cells or tissue by a standard phenol/chloroform extraction method and stored at −20 °C. DNA adducts were measured in each DNA sample using the nuclease P1 enrichment version of the 32P-postlabelling method [33]. Briefly, DNA samples (4 μg) were digested with micrococcal nuclease (120 mU; Sigma, #N3755) and calf spleen phosphodiesterase (40 mU; Calbiochem, #524711), enriched and labelled as reported [28,34]. Solvent conditions for the resolution of 32P-labelled adducts on polyethyleneimine-cellulose (PEI) thin-layer chromatography (TLC) were: D1, 1.0 M sodium phosphate, pH 6; D3, 4.0 M lithium-formate, 7.0 M urea, pH 3.5; D4, 0.8 M lithium chloride, 0.5 M Tris, 8.5 M urea, pH 8. After chromatography TLC plates were scanned using a Packard Instant Imager (Dowers Grove, IL, USA). DNA adduct levels (RAL, relative adduct labelling) were calculated from adduct counts per minute (cpm), the specific activity of [γ-32P]ATP (Hartmann Analytic #HP601ND) and the amount of DNA (pmol) used. Results were expressed as DNA adducts/108 normal nucleotides (nt). An external BPDE-modified DNA standard was used for identification of BaP-DNA adducts.

Isolation of primary mouse embryonic fibroblasts

Xpa-WT and Xpa-Null Hupki mouse embryonic fibroblasts (HUFs) were isolated from day 13.5 embryos of intercrosses of Hupki;Xpa mice according to a standard procedure. Briefly, neural and haematopoietic tissues were removed from each embryo by dissection and the remaining tissue was digested in 1 mL of 0.05% trypsin-EDTA (Invitrogen #25300-062) at 37 °C for 30 min. The resulting cell suspension from each embryo was mixed with 9 mL of growth medium (Dulbecco's modified medium (Invitrogen #31966-021) supplemented with 10% fetal bovine serum (Invitrogen #26140-079) and 100 U/mL penicillin and streptomycin (Invitrogen #15140-130)), pelleted at 1000 rpm for 5 min, and then transferred into a 175-cm2 tissue-culture flask containing 35 mL of growth medium. Cells were cultured to 80–90% confluence at 37 °C/5% CO2/3% O2 before preparing frozen stocks (passage 0; approximately 3 days). Fibroblast cultures were genotyped as described above.

Culture of HUFs and growth curves

HUFs were cultured in growth medium (see 2.6) at 37 °C/5% CO2 with either 20% or 3% O2, adjusted using an incubator fitted with an oxygen sensor and a nitrogen source. All manipulations conducted outside the incubator were performed at 20% O2. For passaging, cells were detached with 0.05% trypsin-EDTA for 2–3 min, suspended in growth media and reseeded at the desired cell number or dilution. When required, cells were counted using an Improved Neubauer Hemacytometer according to the manufacturer's instructions.

Mammalian cell culture, including that of HUFs, typically takes place in incubators containing ambient air buffered with 5–10% CO2, which contains a much higher level of oxygen (20%) than the concentration to which tissues are exposed in vivo. The mean tissue level of oxygen is variable, but is typically about 3%, and mean oxygen tension in an embryo may be even less [35]. Mouse cells are more sensitive than human cells to growth at 20% O2, whereby they accumulate more DNA damage and respond by entering senescence within two weeks of culture [36,37]. While this is required for the selective growth of TP53-mutated clones in the HIMA, it also limits the number of primary cells available for experiments prior to initiating an assay (e.g. optimizing carcinogen treatment conditions). It was shown previously that culture-induced senescence of primary mouse embryo fibroblasts (MEFs) could be inhibited by growing cells in 3% (physiological) oxygen [37].

To compare the growth of HUFs at atmospheric or physiological oxygen levels, passage 0 primary HUFs (Xpa-WT and Xpa-Null) were thawed and cultured in either 20% O2 or 3% O2. After 48 h, the cells were trypsinized and counted. Cells (2.5 × 105) were reseeded into 25-cm2 flasks and again incubated at either 20% or 3% O2. Cells were counted every 3–4 days and reseeded at 2.5 × 105 cells/25-cm2 flask for several weeks. Cultures in 3% O2 began to proliferate more rapidly after ∼3 weeks in culture and were subsequently reseeded at 0.8–1.2 × 105 cells/25-cm2 flask. Cultures in 20% O2 began to populate with spontaneously immortalized, faster growing cells after 30–40 days and were subsequently reseeded at 0.8–1.2 × 105 cells/25-cm2 flask.

The fold-population increase (PI) was calculated each time cells were counted (# of cells counted/# of cells seeded = PI, PI, etc.) and was used to calculate the cumulative population increase (CumPI): CumPI = PI, CumPI = PI * PI, CumPI = PI * PI * PI, etc. The cumulative population increase was then used to calculate the cumulative population doubling (CumPD): CumPD = log2 (CumPI), CumPD = log2 (CumPI), etc.

Crystal violet staining assay for cell survival

The crystal violet staining assay [38] was used to determine relative cell survival following BaP or BPDE treatment, compared with control (untreated) cells. Cells were seeded on 96-well plates at 2.5–5.0 × 103/well and treated the following day with BaP (24 or 48 h) or BPDE (2 h) diluted in growth medium to a highest final concentration of 1 μM (0.1% DMSO final). BPDE treatment media was replaced after 2 h with normal growth media. Treatment was performed in 5 replicate wells per condition at 37 °C/5% CO2/3% O2. At 24 or 48 h following initiation of treatment, cells were rinsed with PBS and adherent cells were fixed and stained for 15 min with 0.1% (w/v) crystal violet (Sigma #C3886) in 10% ethanol. Cells were gently washed with PBS to remove excess crystal violet and allowed to dry. For quantification, the dye was solubilized in 50% ethanol (100 μL per well) and absorbance at 595 nm was determined using a plate reader. Data are presented as the amount of absorbance in wells of treated cells relative to that of DMSO-treated cells and are representative of at least three independent experiments. The linearity of this method was confirmed for subconfluent HUF cultures (data not shown).

Carcinogen treatment of HUFs for DNA adduct analysis

The day prior to treatment, primary HUFs (Xpa-WT and Xpa-Null) were seeded so as to be sub-confluent at the time of harvest. For BaP treatment, 1.5 × 106 or 1.0 × 106 cells were seeded into 75-cm2 flasks for treatments of 24 or 48 h, respectively. For BPDE treatment (up to 2 h), cells were seeded at 2.0–2.5 × 106 cells into 75-cm2 flasks. Duplicate flasks were treated for each condition. BaP and BPDE were diluted in growth medium to a highest final concentration of 1 μM (0.1% DMSO final). Cells were incubated with BaP at 37 °C/5% CO2, in either 20% or 3% O2, or with BPDE at 37 °C/5% CO2/3% O2. Cells grown in medium containing 0.1% DMSO served as control. In the BPDE-DNA adduct removal experiment, treatment medium was removed after 2 h and replaced with fresh growth medium. Cells were harvested following the indicated incubation time (BaP: 24 or 48 h; BPDE: 0.5, 2 or 6 h) and stored as pellets at −20 °C until analysis. DNA adduct formation was assessed as described above.

Hupki mouse embryo fibroblast immortalization assay (HIMA)

Immortalization of primary Hupki mouse embryo fibroblasts treated with BPDE was performed twice (HIMA 1 and 2), according to previously published protocols [15,39]. Frozen Xpa-WT and Xpa-Null primary HUFs (passage 0) were thawed and seeded into 175-cm2 flasks at 37 °C/5% CO2/3% O2 for expansion. After 3 days, cells were trypsinized, counted and seeded (passage 1) at 2.0 × 105 cells/well into 6-well Corning CellBind® plates. Cells were treated the following day with 0.5 μM BPDE (for each Xpa genotype, HIMA 1: 48 cultures; HIMA 2: 54 cultures) or 0.1% DMSO (for each Xpa genotype, HIMA 1: 24 cultures; HIMA 2: 30 cultures) for 2 h.

Following treatment, as the cells approached confluence, the HUFs were subcultured on 6-well Corning CellBind® plates at a dilution of 1:2–1:4. After a further 4 days, all cultures were transferred to 20% O2 to select for senescence bypass. Cultures were passaged once or twice more at 1:2–1:4 before the cultures slowed significantly in their rate of proliferation, and began to show signs of senescence (large, flattened morphology). During senescent crisis, cultures were not passaged again until regions of dividing cells or clones had emerged, and were not diluted more than 1:2 until cells were able to repopulate a culture dish in less than 5 days after splitting. Cultures that did not contain dividing cells were passaged 1:1 every 2 weeks until clones developed. When immortal cells emerged from the senescent cultures and expanded into clones, serial passaging was resumed at dilutions of at least 1:2–1:4 for several passages, followed by further passaging at dilutions up to 1:20. Once a culture achieved a doubling rate of ≤48 h and appeared homogeneous, it was progressively expanded from a 6-well plate to larger flasks (25-, 75-cm2), frozen stocks were prepared and a portion of cells was pelleted for DNA extraction (≥passage 12; 8–16 weeks).

TP53 mutation analysis

DNA was isolated from cell pellets using the Gentra Puregene Cell Kit B (Qiagen, #158745), according to the manufacturer's instructions. Human TP53 sequences (exon 4 to exon 9, including introns) were amplified from each sample using the human TP53-specific primers and cycling conditions described in Table S2. Amplification products were assessed by electrophoresis on 2% agarose gels (containing 0.5 μg/mL ethidium bromide) in TBE buffer. Band size and intensity were monitored by loading 4 μL of Gel Pilot 100 bp Plus marker (Qiagen, #239045) onto each gel. To remove primers and deoxynucleoside triphosphates prior to sequencing, PCR reactions (12 μL) were digested with 2 U exonuclease I (New England Biolabs, UK, #M0293S) and 10 U shrimp alkaline phosphatase (USB Products, USA, #70092Y) for 20 min at 37 °C followed by an inactivation step at 80 °C for 15 min.

Samples were submitted to Beckman Coulter Genomics (Takeley, UK) for Sanger dideoxy fluorescent sequencing using the sequencing primers indicated in Table S2. Chromas software was used to export the FASTA sequence from the chromatograms which were also visually inspected. FASTA sequences were analyzed by alignment against a human TP53 reference sequence, NC_000017.9 from Genbank, using the Basic Local Alignment Search Tool (BLAST) from the National Center for Biotechnology Information (NCBI) (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Variations (e.g. single base substitutions, deletions) were assessed using the mutation validation tool available at the IARC TP53 mutation database (http://www-p53.iarc.fr/MutationValidationCriteria.asp), and could be classified as either homo-/hemi-zygous or heterozygous. Mutations were confirmed by sequencing DNA from an independent sample of cells from the same clone.

Statistical analysis

All statistical analyses were carried out using SAS v. 9.3 for Windows XP (SAS Institute, Cary, NC). The effects of Xpa status and chemical treatment (i.e. BaP or BPDE) on adduct formation were examined using ordinary least-squares 2-factor Analysis of Variance (ANOVA) followed by Bonferroni's post-hoc contrasts. Homogeneity of variance across the treatment groups was examined using the Bartlett test. Alternatively, pair-wise comparisons employed the Student's t-test. The effects of Xpa status and chemical treatment (i.e. BPDE) on the frequency of TP53 mutant clones or TP53 mutation frequency were analyzed using 2 × 2 × 2 contingency table analysis. The Chi-square test was used to test the null hypothesis that row (i.e. chemical treatment or Xpa status) and column (i.e. TP53 mutant) variables are not significantly associated. Odds Ratio values were employed to assess the relative risk of a given outcome (e.g. TP53 mutant) for paired levels of the chemical treatment or Xpa genotype (e.g. Xpa-Null versus Xpa-WT or BPDE-treated versus solvent control). The exact statement in Proc Freq provided exact tests and confidence limits for the Pearson Chi-square and Odds Ratio values. Since a small proportion of TP53 mutant clones contained more than one mutation, the TP53 mutation response was treated as an ordinally-scaled dependent variable (i.e. a multinomial response with outcome none, single or double). The effects of Xpa genotype and chemical treatment on mutation response were determined using ordinal logistic regression (i.e. cumulative logit model) in SAS Proc Catmod.

Pair-wise statistical comparisons of mutation patterns (i.e. type of mutations) employed a variation on the algorithm originally published by [40]; later modified by [41]. Briefly, statistical comparisons of mutation patterns for two conditions (e.g. Xpa-Null versus Xpa-WT for BPDE treated) were assessed using the Fisher's exact test with P values estimation determined using Monte Carlo simulation with 50,000 iterations.

Results

Creation of Xpa-deficient Hupki mice and embryos

Hupki mice deficient in NER were generated by crossing the Hupki strain with transgenic mice harbouring an Xpa-knockout allele. The Hupki;Xpa offspring were healthy and did not show any obvious phenotypic differences from Hupki;Xpa mice within the timeframe of these studies (up to 6 months). Likewise, Xpa-WT and Xpa-Null HUFs were morphologically similar.

BaP-induced DNA adduct formation in Xpa-WT and Xpa-Null Hupki mice

DNA adduct formation after treatment with BaP was initially assessed in vivo (Fig. 1, Fig. S2 and Fig. S3). One day following a single BaP treatment at 125 mg/kg bw, DNA adduct levels were significantly higher in three of six tissues examined (spleen, colon and kidney) from Xpa-Null mice compared with their WT littermates, ranging from 1.4- to 3.7-fold (Fig. 1A). Unexpectedly, all Xpa-Null mice died within 2–3 days of BaP (125 mg/kg bw) treatment (Fig. 1B and Fig. S2). In the Xpa-WT mice, BaP-DNA adduct levels following a single treatment persisted 5 days later in all tissues except the small intestine where there was a 2.5-fold decrease (Fig. S2). Further, DNA adduct levels greatly increased in Xpa-WT animals that received BaP daily for 5 days (14-, 12-, and 4-fold in liver, lung and small intestine, respectively) (Fig. 1B).

Fig. 1

DNA adduct formation in Xpa-WT and Xpa-Null Hupki mice treated with BaP. Mice were treated with BaP (125 or 12.5 mg/kg bw) as indicated; (A) and (C) were treated once, (B) and (D) were treated once a day for 5 days. DNA adduct levels in different tissues were assessed by 32P-postlabelling. Values represent means ± SD from 3 animals, and each DNA sample was measured by two independent 32P-postlabelling analyses. ND = not detected; † = not determined due to death of Xpa-Null animals. See Fig. S3 for representative autoradiograms showing the adduct profiles in the different tissues examined. Statistical analysis, comparing adduct levels in tissues from Xpa-WT and Xpa-Null mice, was performed using the Student's t-test; *P < 0.05, **P < 0.01, ***P < 0.001.

Due to the acute toxicity induced by BaP at 125 mg/kg bw in Xpa-Null mice, animals were treated with a 10-fold lower dose (i.e. 12.5 mg/kg bw) in a subsequent experiment (Fig. 1C and D). A single low dose of BaP (12.5 mg/kg bw) resulted in detectable DNA adducts in all tissues examined which exhibited a trend towards being higher in Xpa-Null mice compared to Xpa-WT mice (ranging from 1.8-fold [spleen] to 2.7-fold [small intestine]) (Fig. 1C) in line with the results obtained after a single administration of 125 mg/kg bw (compare to Fig. 1A). Xpa-Null mice were able to tolerate 5 daily treatments with the lower dose of BaP. Interestingly, after 5 daily treatments, DNA adduct levels were about the same in Xpa-WT and Xpa-Null animals (Fig. 1D). Taken together, these experiments indicate that BaP-DNA adduct removal is dependent on Xpa/NER within 1 day of treatment, although NER fails to remove these adducts following 5 days of BaP exposure even in Xpa-WT mice. Further, the inability of Xpa-Null mice to repair BaP-DNA adducts can result in lethal toxicity if the DNA damage exceeds a certain threshold.

Growth of Xpa-WT and Xpa-Null HUFs at 20% and 3% oxygen

Due to the previously reported inhibition of culture-induced senescence of MEFs at 3% O2, we sought to determine whether the growth of primary HUFs could also be enhanced and/or extended in 3% oxygen. Growth curves were generated over 6–9 weeks of culture in order to establish the growth characteristics of both Xpa-WT and Xpa-Null HUFs in 20% and 3% oxygen (Fig. 2A and B).

Fig. 2

Growth and BaP-induced DNA adduct formation in HUFs cultured at 20% or 3% O2. (A) and (B): Xpa-WT or Xpa-Null HUFs (2.5 × 105 cells/25-cm2 flask) were cultured for up to 50 days in 20% O2 (A) or 3% O2 (B). Cells were counted every 3–4 days, diluted and reseeded to determine the cumulative population doublings over time. (C): Xpa-WT HUFs cultured at 20% or 3% O2 were treated for 24 or 48 h with 1 μM BaP. DNA adduct levels were assessed by 32P-postlabelling. Values represent means ± SD of two biological replicates where each DNA sample was measured by two independent 32P-postlabelling analyses.

In 20% O2 the primary HUF cultures (Xpa-WT and Xpa-Null) rapidly proliferated for the first several days in culture (Fig. 2A). After 5 days the original populations had increased approximately 40-fold, and after 11 days the populations had increased about 200-fold. The proliferation of HUFs in 20% O2 markedly decreased after 11–15 days, as cells began to senesce. The cultures resumed proliferation after 35–45 days as immortal cells emerged.

The HUF cultures grew rapidly in 3% O2 for the first 11 days, at a slightly increased rate compared with cells grown in 20% O2, doubling every 24–30 h (Fig. 2B). After 5 days, the original populations had increased by 70-fold, and after 11 days they had increased by 2100-fold. After this point the cultures temporarily proliferated at a reduced rate (to varying degrees) and the cells appeared morphologically heterogeneous. By 25 days in culture in 3% O2, cultures were again rapidly proliferating and were homogeneous in appearance.

Another set of HUF cultures was grown in 3% O2 for 1 week and then transferred to 20% O2, in order to determine whether cultures grown temporarily at 3% O2 would still be capable of senescence and immortalization at 20% O2 (Fig. S4). Following transfer to 20% O2, the cultures underwent 2 population doublings in the first 3 days of culture, but then slowed and began to senesce by the next passage. Similarly to cells grown continuously at 20% O2, immortalized cells emerged in the cultures after 35–40 days.

Effect of oxygen on DNA adduct formation in HUFs treated with BaP

The growth curves generated in 20% and 3% O2 indicated a clear growth advantage for primary HUFs grown in 3% O2, at least prior to 2 weeks in culture. However, the impact of 3% versus 20% O2 on the metabolic activation of carcinogens (i.e. BaP) has not been previously examined. Primary Xpa-WT HUFs grown in 20% and 3% O2 (passage 1, ≤7 days in culture) were treated with 1 μM BaP for 24 or 48 hr to assess DNA adduct formation (Fig. 2C). Interestingly, DNA adduct levels were markedly higher in HUFs treated in 3% O2 than in cells treated in 20% O2. After 24 h treatment, a 4-fold higher level of DNA adducts was detected in HUFs treated in 3% O2 (513 ± 34 versus 129 ± 10 adducts per 108 nt) and after 48 h treatment DNA adduct levels were 2-fold higher in 3% O2 (839 ± 294 versus 391 ± 27 adducts per 108 nt) than in cells treated in 20% O2. Therefore, growing HUFs temporarily in 3% O2 not only provides a substantial increase in cells available for experiments, but may enhance the formation of DNA-reactive metabolites following treatment with environmental carcinogens. From these observations, all subsequent experiments were performed in 3% O2.

Survival of Xpa-WT and Xpa-Null HUFs treated with BaP or its reactive intermediate BPDE

Previous studies using MEFs from Xpa-knockout mice (with WT Trp53) showed that Xpa-deficient cells are highly sensitive to DNA damage that is normally repaired by NER, including that induced by BaP [20,42,43]. Here, we have compared Xpa-Null HUFs with Xpa-WT HUFs for their sensitivity to BaP and its reactive intermediate BPDE. Xpa-Null HUFs were indeed more sensitive to treatment with both compounds, although the difference was more pronounced after 48 h (Fig. 3A and B). Following treatment with 1 μM BaP, 63% of Xpa-Null cells had survived after 24 h (91% for Xpa-WT), but by 48 h BaP-treated Xpa-Null cells were 23% of control (60% for Xpa-WT) (Fig. 3A). Upon 1 μM BPDE treatment, 41% of Xpa-Null cells had survived after 24 h (59% for Xpa-WT), but this had decreased to 6% at 48 h (47% for Xpa-WT) (Fig. 3B). Interestingly, surviving Xpa-Null cells treated with ≥0.5 μM BaP did not resume proliferation, whereas those treated with ≤0.25 μM BaP resumed proliferation within 1–2 days of treatment (monitored visually with a microscope for up to two weeks; data not shown).

Fig. 3

Survival of Xpa-WT and Xpa-Null HUFs and DNA adduct formation following treatment with BaP or BPDE. (A) and (B) Cells were treated with the indicated doses of BaP (24 or 48 h) or BPDE (2 h); survival was measured at 24 or 48 h following initiation of treatment using a crystal violet staining assay. Cells treated with 0.1% DMSO served as solvent control. Mean values are shown as % of control ± SD of 5 replicate wells. The data are representative of at least three independent experiments (variation ≤ 15%). (C) and (D) Cells were treated with the indicated doses of BaP (24 h) or BPDE (2 h). DNA adduct levels were assessed by 32P-postlabelling. Values represent means ± SD of two biological replicates where each DNA sample was measured by two independent 32P-postlabelling analyses. Statistical analysis, comparing DNA adduct levels in Xpa-WT and Xpa-Null HUFs, was performed by 2-factor ANOVA followed by Bonferroni's post-hoc contrasts; *P < 0.05. Insets: Autoradiographic profiles of DNA adducts obtained in HUFs treated with BaP or BPDE, as indicated; the origins, in the bottom left-hand corner, were cut off before exposure; arrow shows 10-(deoxyguanosin-N2-yl)-7,8,9-trihydroxy-7,8,9,10-tetrahydrobenzo[a]pyrene (BPDE-N-dG).

BaP- and BPDE-induced DNA adduct formation in Xpa-WT and Xpa-Null HUFs

Xpa-Null HUFs were shown to be highly sensitive to treatment with BaP and BPDE. Next, the level of DNA adducts induced by these compounds was assessed in Xpa-Null and Xpa-WT HUFs (Fig. 3C and D). Cells were treated with 0.05–1.0 μM of BaP for 24 h or 0.125–1.0 μM BPDE for 2 h. A concentration-dependent increase in DNA adduct formation was found after treatment with both compounds.

Interestingly, the Xpa-Null HUFs, despite being deficient in NER, accumulated similar or slightly lower levels of BaP-induced DNA adducts to Xpa-WT HUFs, reaching up to 370 ± 111 adducts per 108 nt at 1 μM BaP versus 513 ± 34 adducts per 108 nt in Xpa-WT HUFs. On the other hand, DNA adduct formation following BPDE treatment was slightly higher in Xpa-Null HUFs than in Xpa-WT HUFs, although the difference was significant only at the highest concentration of BPDE (1 μM) where DNA adduct levels reached 566 ± 88 adducts per 108 nt in Xpa-Null HUFs versus 475 ± 40 adducts per 108 nt in Xpa-WT HUFs.

Additionally, we examined a time course of BPDE adduct formation and removal in Xpa-WT and Xpa-Null HUFs. Previous studies have shown that the half-life of BPDE in cells is ∼12 minutes and peak adduct formation appears to vary from 20 min to 2 h, perhaps depending on cell type and experimental design [44-46]. Here, HUFs were treated with 0.25 μM BPDE for up to 2 h, and one set of cultures was further incubated in normal medium for 4 h after BPDE was removed. Cells were harvested to assess DNA adduct levels at 30 min, 2 h and 6 h (Fig. S5). Longer incubation times were not included to avoid effects caused by proliferation. After 30 min incubation with BPDE, Xpa-Null and Xpa-WT HUFs accumulated the same level of DNA adducts (138 ± 1 adducts per 108 nt in both Xpa-Null and Xpa-WT HUFs). This initial DNA adduct level progressively declined in Xpa-WT HUFs, by 18% at 2 h (114 ± 7 adducts per 108 nt) and by 30% at 6 h (96 ± 12 adducts per 108 nt). In Xpa-Null HUFs, however, DNA adduct levels peaked at 2 h (161 ± 19 adducts per 108 nucleotides), and were similar at 6 h to the levels detected at 0.5 h (132 ± 3 adducts per 108 nt). These results demonstrate that Xpa-WT HUFs are able to repair BPDE-DNA adducts over time, while the repair capacity of the Xpa-Null HUFs is impaired.

TP53 mutations induced by BPDE in Xpa-WT and Xpa-Null HUFs

Mutation frequency

Primary Xpa-WT and Xpa-Null HUF cultures (102 per genotype) were exposed for 2 h to 0.5 μM BPDE and then serially passaged for 8–16 weeks (≥12 passages) in 20% O2, resulting in one immortalized cell line per culture. Untreated cells of each Xpa genotype (54 cultures) were immortalized in parallel. Mutations in the human TP53 sequence of immortalized HUF lines were identified by PCR amplification of exons 4–9 (along with adjacent introns) and direct dideoxy sequencing (Tables 1 and 2). From untreated HUF cultures, only two spontaneously immortalized lines of each Xpa genotype were found to contain mutated TP53 (3.7%). Treatment with BPDE markedly increased the frequency of TP53 mutations over that observed in untreated cultures. Of the 102 immortalized cell lines derived from BPDE-exposed Xpa-WT HUFs, 16 clones harboured a total of 20 mutations (four clones carried two mutations each), while 23 immortalized cell lines derived from BPDE-exposed Xpa-Null HUFs harboured a total of 29 mutations (six clones contained two mutations each). Statistical data analyses initially examined the effect of BPDE treatment on the frequency of TP53 mutant clones, and confirmed a statistically significant effect for both Xpa-WT cells (i.e. Chi-squared = 5.0, P < 0.04) and Xpa-Null cells (i.e. Chi-squared = 9.3, P < 0.003). Similarly, the analyses showed a significant effect of BPDE treatment on TP53 mutation frequency for Xpa-WT cells (i.e. Chi-squared = 4.1, P < 0.05) as well as Xpa-Null cells (i.e. Chi-squared = 7.1, P < 0.008). Furthermore, these data suggest a trend for an increased frequency of TP53 mutagenesis in BPDE-exposed Xpa-Null HUFs (22.5%) compared with Xpa-WT HUFs (15.7%) that was confirmed by statistical analyses. More specifically, Odds Ratio values confirmed that Xpa-Null cells are more susceptible to the effects of BPDE treatment (i.e. OR = 7.6, 95% confidence interval = 1.7–33.5) as compared with Xpa-WT cells (i.e. OR = 4.8, 95% confidence interval = 1.1–21.9). However, the increase in the relative risk of TP53 mutation between Xpa-Null and Xpa-WT HUFs is not statistically significant due to the relatively small number of mutants obtained and the consequently low statistical power. Indeed, separate statistical analysis that examined the impact of Xpa status on TP53 mutation frequency for BPDE treated cells only failed to detect a significant effect (i.e. Chi-squared = 1.9, P = 0.19, OR = 1.6 with 95% confidence interval = 0.83–3.0).

Mutation pattern

Most mutations induced by BPDE occurred at G:C base pairs (90% Xpa-WT; 83% Xpa-Null), predominantly consisting of single base substitutions (Table 2 and Fig. 5). The most frequent mutation type was a G:C > T:A transversion (40% Xpa-WT; 31% Xpa-Null), the signature mutation of BaP/BPDE, followed by G:C > C:G transversions (25% Xpa-WT; 21% Xpa-Null), and G:C > A:T transitions (20% Xpa-WT; 17% Xpa-Null). Single or tandem deletions of guanines, leading to a frameshift, were also observed but were more frequent in Xpa-Null clones (5% Xpa-WT; 14% Xpa-Null). Approximately half of the mutations at G:C base pairs occurred at CpG sites (56% Xpa-WT; 46% Xpa-Null). Out of 33 CpG sites between exons 4–9 in TP53, 11 were mutated by BPDE, most commonly resulting in G:C > C:G or G:C > T:A transversions. Of the four mutations found in untreated control cultures, one was a G:C > C:G transversion and three were A:T > C:G transversions. The A:T > C:G mutation type did not occur in BPDE-treated Xpa-WT HUFs but was found in three BPDE-treated Xpa-Null clones (XN-BP-278, -299, -300).

Table 2

TP53 mutations in immortalized clones of Xpa-WT and Xpa-Null HUFs treated with BPDE and from spontaneously immortalized controls. SA = splice acceptor site; CpG indicates the presence of a mutation at a methylated CpG site; NTS = non-transcribed strand; TS = transcribed strand.

Xpa status	Clone ID*	Codon #	Exon	Mutation type	Strand	WT codon	MUT codon	Coding change	CpG	Zygosity	Contact (C), Structure (S), Zinc (Z)	Activity (Kato)**
TP53-mutated immortalized clones from HUFs treated with 0.5 μM BPDE
WT	XW-BP-91	91	4	G:C > A:T	NTS	TGG	TGA	W91stop		Homo-/hemi-		NA
WT	XW-BP-10	132	5	A:T > T:A	NTS	AAG	ATG	R132 M		Homo-/hemi-		NF
WT	XW-BP-50	143	5	G:C > A:T	NTS	GTG	ATG	V143 M		Hetero-		NF
WT	XW-BP-83	154	5	G:C > T:A	NTS	GGC	GTC	G154V		Hetero-		NF
WT	XW-BP-73	158	5	del. G	TS	CGC	_GC	frameshift	CpG	Homo-/hemi-		NA
WT	XW-BP-26	181	5	G:C > C:G	NTS	CGC	CCC	R181P	CpG	Hetero-		NF
WT	XW-BP-17	195	6	G:C > C:G	TS	ATC	ATG	I195 M		Homo-/hemi-		PF
WT	XW-BP-17	196	6	G:C > T:A	NTS	CGA	CTA	R196L	CpG	Homo-/hemi-		PF
WT	XW-BP-2	203	6	G:C > T:A	NTS	GTG	TTG	V203L		Hetero-		PF
WT	XW-BP-50	213	6	G:C > T:A	NTS	CGA	CTA	R213L	CpG	Hetero-		NF
WT	XW-BP-83	224	6	G:C > C:G	NTS	GAG	GAC	E224D		Hetero-		F
WT	XW-BP-42	245	7	G:C > C:G	NTS	GGC	CGC	G245R	CpG	Hetero-	S	NF
WT	XW-BP-16, -55	248	7	G:C > A:T	NTS	CGG	CAG	R248Q	CpG	Homo-/hemi-	C	NF
WT	XW-BP-95	248	7	G:C > T:A	NTS	CGG	CTG	R248L	CpG	Hetero-	C	NF
WT	XW-BP-2	249	7	G:C > C:G	NTS	AGG	AGC	R249S		Hetero-	S	NF
WT	XW-BP-6, -63	273	8	G:C > T:A	NTS	CGT	CTT	R273L	CpG	Homo-/hemi-	C	NF
WT	XW-BP-9	275	8	G:C > T:A	NTS	TGT	TTT	C275F		Homo-/hemi-		NF
WT	XW-BP-38	286	8	A:T > G:C	NTS	GAA	GGA	E286G		Hetero-		NF
Null	XN-BP-229	34	4	G:C > T:A	TS	CCC	CCA	(silent) P34P		Hetero-		NA
Null	XN-BP-292	65	4	A:T > T:A	NTS	AGA	TGA	R65stop		Homo-/hemi-		NA
Null	XN-BP-201	110	4	G:C > C:G	NTS	CGT	CCT	R110P	CpG	Hetero-		NF
Null	XN-BP-294	127	5	G:C > T:A	TS	TCC	TAC	S127Y		Homo-/hemi-		NF
Null	XN-BP-300	131	5	A:T > C:G	NTS	AAC	ACC	N131 T		Hetero-		PF
Null	XN-BP-300	135	5	G:C > C:G	TS	TGC	TGG	C135 W		Hetero-		PF
Null	XN-BP-204	155	5	G:C > A:T	TS	ACC	ATC	T155I		Hetero-		NF
Null	XN-BP-225	157	5	G:C > A:T	TS	GTC	GTT	(silent) V157V		Homo-/hemi-		NA
Null	XN-BP-268	157	5	G:C > T:A	NTS	GTC	TTC	V157F	CpG	Homo-/hemi-		NF
Null	XN-BP-225, -228	158	5	G:C > T:A	NTS	CGC	CTC	R158L	CpG	Homo-/hemi-		NF
Null	XN-BP-238	171	5	G:C > T:A	NTS	GAG	TAG	E171stop		Hetero-		NA
Null	XN-BP-206	188	6	G:C > T:A	NTS	CTG	CTT	(silent) L188L		Homo-/hemi-		NA
Null	XN-BP-206	189	6	del. G	NTS	GCC	_CC	frameshift		Homo-/hemi-		NA
Null	XN-BP-273	196	6	G:C > C:G	TS	CGA	GGA	R196G	CpG	Homo-/hemi-		PF
Null	XN-BP-278	211	6	A:T > C:G	NTS	ACT	CCT	T211P		Homo-/hemi-		NF
Null	XN-BP-265	242	7	G:C > A:T	NTS	TGC	TAC	C242Y		Homo-/hemi-	Z	NF
Null	XN-BP-229	253	7	A:T > T:A	NTS	ACC	TCC	T253S		Hetero-		PF
Null	XN-BP-299	265	8	A:T > C:G	TS	CTG	CGG	L265R		Hetero-		NF
Null	XN-BP-296	267	8	del. G	NTS	CGG	C_G	Frameshift	CpG	Homo-/hemi-		NA
Null	XN-BP-257	267	8	del. GG	NTS	CGG	C__	Frameshift	CpG	Homo-/hemi-		NA
Null	XN-BP-210	272	8	G:C > A:T	NTS	GTG	ATG	V272 M		Hetero-		NF
Null	XN-BP-237	273	8	G:C > C:G	TS	CGT	GGT	R273G	CpG	Homo-/hemi-	C	NF
Null	XN-BP-274	273	8	G:C > C:G	NTS	CGT	CCT	R273P	CpG	Homo-/hemi-	C	NF
Null	XN-BP-251	273	8	G:C > T:A	TS	CGT	AGT	R273S	CpG	Hetero-	C	NF
Null	XN-BP-254	274	8	G:C > T:A	NTS	GTT	TTT	V274F		Homo-/hemi-		NF
Null	XN-BP-238	282	8	G:C > C:G	TS	CGG	GGG	R282G	CpG	Hetero-	S	NF
Null	XN-BP-255	330	9	del. G	TS	CTT	_TT	Frameshift		Hetero-		NA
Null	XN-BP-204	in. 6 (SA)		G:C > A:T	NTS			Splice		Hetero-		NA
TP53-mutated spontaneously immortalized clones from control HUFs (treated with 0.1% DMSO)
WT	XW-C-115	138	5	G:C > C:G	NTS	GCC	CCC	A138P		Homo-/hemi-		NF
WT	XW-C-137	173	5	A:T > C:G	TS	GTG	GGG	V173G		Homo-/hemi-		NF
Null	XN-C-325	113	4	A:T > C:G	TS	TTC	GTC	F113V		Homo-/hemi-		NF
Null	XN-C-338	113	4	A:T > C:G	TS	TTC	GTC	F113V		Hetero-		NF

XW = Xpa-WT; XN = Xpa-Null; BP = BPDE-treated; C = control.

The overall transactivation activity of the mutant in a yeast functional assay published by Kato et al. [50]. NF = non-functional, PF = partially functional, F = functional, NA = not assessed.

Fig. 5

The codon distribution of TP53 mutations induced by BPDE in Xpa-WT and Xpa-Null HUFs compared with the spectrum found in human tumours. Shown are (A) the number of mutations at each codon (within exons 4–9) in the TP53 gene induced by BPDE in HUFs compared with the frequency of mutation at each codon in (B) lung cancer of smokers (exclusions: radon, asbestos, mustard gas, and coal [74,75]), (C) lung cancer of non- and passive-smokers (exclusions: radon, asbestos, mustard gas, and coal [74,75]) and (D) cancer overall. Reference for human cancer mutation codon distribution: IARC TP53 Mutation Database, R17 (November 2013). Mutation hotspots are indicated in grey.

Strand bias

It has been shown previously that DNA damage induced by BPDE is repaired more rapidly if it occurs on the transcribed strand of TP53 compared with the non-transcribed strand [47,48]. This is thought to explain the strand bias of G to T mutations in TP53 found in lung tumours of tobacco smokers, where mutations are preferentially found on the non-transcribed strand [49]. In contrast, in TC-NER-deficient cells mutations are biased in favour of the transcribed strand [50,51]. Indeed, here we found an increased number of BPDE-induced mutations on the transcribed strand in Xpa-Null HUFs (38%) relative to Xpa-WT HUFs (10%) (Table 1 and Fig. 4). Statistical analysis to examine the influence of Xpa status on the transcribed and non-transcribed strand BPDE-induced TP53 mutation frequencies, respectively, showed a statistically significant effect for the former, but not the latter. More specifically, Xpa status had a significant effect on the frequency of BPDE-induced mutations on the transcribed strand (i.e. Chi-squared = 6.5, P < 0.02). Moreover, the Odds Ratio confirmed a 6-fold increase in the average likelihood of BPDE-induced transcribed strand mutations for Xpa-Null cells compared to Xpa-WT cells (i.e. OR = 6.0, 95% confidence interval = 1.3–27.8). No such effect of Xpa status was observed for BPDE-induced TP53 mutations on the non-transcribed strand (i.e. Chi-squared = 0.05, P = 0.86). All mutation types detected on the non-transcribed strand of Xpa-Null clones were also found on the transcribed strand, with the exception of A:T > T:A.

Table 1

Summary of TP53 mutation frequency and patterns in immortalized clones of Xpa-WT and Xpa-Null HUFs treated with BPDE and from spontaneously immortalized controls.

BPDE	Xpa-WT	Xpa-Null	Control	Xpa-WT	Xpa-Null
Total BPDE-treated HUF cultures (#)	102	102	Total 0.1% DMSO-treated HUF cultures (#)	54	54
TP53-mutant immortalized clones (#)	16	23	TP53-mutant immortalized clones (#)	2	2
Total TP53 mutations detected (#)	20	29	Total TP53 mutations detected (#)	2	2
Frequency of TP53-mutant clones	15.7% (16/102)	22.5% (23/102)	Frequency of TP53-mutant clones	3.7% (2/54)	3.7% (2/54)
Mutations on the transcribed strand	10% (2/20)	38% (11/29)	Mutations on the transcribed strand	50% (1/2)	100% (2/2)
TP53 mutation pattern			TP53 mutation pattern
G to A	20% (4/20)	17% (5/29)	G to A	0% (0/2)	0% (0/2)
G to C	25% (5/20)	21% (6/29)	G to C	50% (1/2)	0% (0/2)
G to T	40% (8/20)	31% (9/29)	G to T	0% (0/2)	0% (0/2)
A to C	0% (0/20)	10% (3/29)	A to C	50% (1/2)	100% (2/2)
A to G	5% (1/20)	0% (0/29)	A to G	0% (0/2)	0% (0/2)
A to T	5% (1/20)	7% (2/29)	A to T	0% (0/2)	0% (0/2)
del. G/GG	5% (1/20)	14% (4/29)	del. G/GG	0% (0/2)	0% (0/2)
TP53 mutation pattern at CpG sites			TP53 mutation pattern at CpG sites
Total G mutations at CpG	56% (10/18)	46% (11/24)	Total G mutations at CpG	0% (0/1)	0% (0/0)
G to A at CpG	50% (2/4)	0% (0/5)	G to A at CpG	0% (0/0)	0% (0/0)
G to C at CpG	40% (2/5)	83% (5/6)	G to C at CpG	0% (0/1)	0% (0/0)
G to T at CpG	63% (5/8)	44% (4/9)	G to T at CpG	0% (0/0)	0% (0/0)
del. G/GG at CpG	100% (1/1)	50% (2/4)	del. G/GG at CpG	0% (0/0)	0% (0/0)

Fig. 4

The pattern and strand bias of TP53 mutations in immortalized clones of Xpa-WT and Xpa-Null HUFs treated with BPDE. (A) The percentage of each type of single base substitution or deletion is shown. (B) The number of each type of single base substitution or deletion occurring on the non-transcribed strand (NTS) or transcribed strand (TS) is shown.

Effect of mutations on p53 protein function

The majority of BPDE-induced TP53 mutations were missense (37/49). Additionally, three nonsense (R65X, W91X, E171X), three silent (P34P, V157V, L188L), one splice (acceptor site, intron 6) and five frameshift (codons 158, 189, 267, 330) mutations were induced, most of which occurred in Xpa-Null clones. All of the silent mutations occurred in clones that harboured a second mutation. Most of the missense mutations found in the immortalized HUF clones could be classified as ‘non-functional’ (NF), or defective in transactivation activity, as determined by a comprehensive yeast functional study [52]. This indicates, as shown previously, that loss of transactivation activity is an important aspect of senescence bypass by p53 inactivation [53]. However, eight missense mutations were classified as ‘partially functional’ (PF) or functional (F), whereby these p53 mutants retained some or all of their transactivation activity. Notably, all but one (R196G) of the PF/F mutants occurred in clones that also contained a second mutation, suggesting that partial loss of p53 transactivation function is not sufficient for senescence bypass.

Influence of sequence context

We also examined how sequence context and the presence of methylated CpG sites influenced the pattern of G:C base pair mutations induced by BPDE. In Table S3 mutations of each type were sorted by the bases 5′ and 3′ to the mutated base. For G:C > A:T transitions, mutations occurred at CpG sites (2/9), at GG pairs (7/9; two of which were also CpG sites), or at G bases with a 5′ or 3′ T (4/9; two of which were also GG pairs). For G:C > C:G transversions, mutations occurred at CpG sites (7/11) or at GG pairs (7/11; three of which were also CpG sites). For G:C > T:A transversions, most mutations occurred at CpG sites (9/17), and the remaining mutations arose at GG pairs (6/17; one also being a CpG site) or in a 5′T-G-T3′ context (3/17). Similarly, deletions at G:C base pairs occurred either at CpG sites (3/5) or at GG pairs. In the case of mutations occurring at GG pairs, the second G was 5′ or 3′ to the mutated base.

Codon distribution and comparison to human cancer mutation spectra

A total of 46 unique BPDE-induced mutations were detected in the sequenced exons (4–9), occurring in 38 codons overall. Three codons (158, 196, 273) were mutated in both Xpa-WT and Xpa-Null clones, but unique mutations were induced in each case. Mutations were found in codons for two key residues that make direct contact with the DNA (R248, R273), three that support the structure of the DNA binding surface (G245, R249, R282) and one that is important for coordinating Zn(2+) binding (C242).

The mutations identified in BPDE-exposed HUF clones were compared with TP53 mutations found in human cancer across all tumour types, as well as specifically in lung cancer from smokers and non-smokers, using the IARC TP53 mutation database, version R17 (Fig. 5). All but five of the 46 TP53 mutations found in BPDE-exposed HUFs have been detected in at least one human tumour (Table S4). Mutations that were infrequently or not found in human tumours included silent mutations, frameshifts, and mutations that resulted in a partially functional mutant p53 protein. Of the six hotspots most frequently mutated across all cancer types (R175, G245, R248, R249, R273, and R282), mutations were induced by BPDE at each, with the exception of R175. Further, BPDE also targeted two lung cancer-specific hotspots, codons V157 and R158. All of these hotspots, with the exception of codon 249, contain CpG sites.

At codon 157 (GTC), BPDE induced one G:C > T:A mutation at the 1st position and one G:C > A:T mutation at the 3rd position. At codon 158 (CGC), BPDE induced a G:C base pair deletion at the 1st position and two G:C > T:A mutations at the 2nd position. Codons 157 and 158 are more frequently targeted in smokers’ lung cancer compared with cancer overall, with G:C > T:A transversions predominating (codon 157: 24/30; codon 158: 32/44). In cancer overall, G:C > T:A transversions are also the most common mutation type at codon 157 (210/313), but are less frequent at codon 158 (102/326), where G:C > A:T transitions at the second position are more common (113/326). No mutations at codons 157 or 158 have ever been detected in spontaneously immortalized HUFs (Table S5).

The most frequently mutated TP53 codons in cancer overall and in smokers’ lung cancer are 248 and 273 (Fig. 5). Likewise, these two codons were hotspots for BPDE-induced mutations in the current HIMA. In cancer overall and nonsmokers’ lung cancer, the most frequent mutation type at codon 248 (CGG) is G:C > A:T (∼90%), and indeed two of the mutations induced by BPDE were G:C > A:T transitions at the 2nd position. Additionally, one G:C > T:A mutation at the 2nd position was detected in the BPDE-treated HUFs. G:C > T:A transversions at the 2nd position in codon 248 are much more frequent in smokers’ lung cancer (23/56) compared with all cancer (121/1880). Notably, mutation at codon 248 has not been detected in untreated, spontaneously immortalized HUFs (Table S5). With regards to codon 273 (CGT), BPDE-induced mutations included one G:C > T:A transversion at the first position, one G:C > C:G transversion at each of the 1st and 2nd positions, and two G:C > T:A mutations at the 2nd position. The most common mutation type found in human cancer at codon 273 is G:C > A:T (∼90%); G:C > T:A transversions at the 2nd position are much more frequent in smokers’ lung cancer (30/62). G:C > C:G mutations at codon 273 occur in 1-2% of cancer overall and 3-5% of smoker's lung cancer.

Comparison to mutations induced in previous HIMAs

BPDE-induced TP53 mutations were compared further with mutations detected in previous HIMAs (Table S5), including HUFs treated with BaP, 3-NBA, AAI, UV and MNNG and untreated controls. Seven codons mutated by BPDE were also mutated in cells treated with BaP (135, 157, 158, 224, 248, 273, 282), and six identical mutations were induced by both compounds. Codons 157 and 224 were not mutated in HIMAs using other compounds. One mutation at codon 158 was induced by AAI, one mutation at codon 248 was induced by UV, and codon 273 was targeted once each by 3-NBA and AAI.

Discussion

We have generated an NER-deficient Hupki model by crossing the Hupki mouse with an Xpa-deficient mouse. We hypothesized that Xpa-deficiency would increase the sensitivity of the Hupki model to DNA damage normally repaired by NER and thereby increase the frequency of carcinogen-induced TP53 mutations in immortalized HUFs. Xpa-WT and Xpa-Null mice and HUFs were treated with the ubiquitous environmental carcinogen BaP, or its reactive metabolite BPDE, which form DNA adducts (BPDE-N-dG) that have been shown to be repaired by the NER pathway. We found that Xpa-Null Hupki mice and HUFs were more sensitive than their Xpa-WT counterparts to the DNA damage induced by BaP or BPDE, exhibiting pronounced mortality at the highest doses tested. Further, we observed a bias for BPDE-induced mutations on the transcribed strand of TP53 in immortal clones derived from Xpa-Null HUFs, although TP53 mutation frequency overall was not significantly increased in Xpa-Null HUFs compared to Xpa-WT HUFs.

Although BaP- and BPDE-induced DNA adduct levels were generally similar between Xpa-WT and Xpa-Null HUFs, Xpa-Null cells were less able to survive the DNA damage. This suggests that the sensitivity of Xpa-Null cells was not due to retention of more DNA adducts. The sensitivity of Xpa-Null HUFs to BaP and BPDE is more likely caused by the blockage of RNAPII by BPDE-N-dG adducts in actively transcribed genes. The persistence of DNA lesions on the transcribed strand of active genes in TC-NER-deficient cells is a strong trigger for apoptosis induction [54]. It has been observed that Xpa-Null cells, deficient in both GG- and TC-NER, undergo apoptosis after DNA damage induced by carcinogens such as UV or BaP, whereas Xpc-Null cells, deficient only in GG-NER, do not, although this may be cell-type specific [55]. TC-NER-deficient cells are unable to repair RNAPII-blocking lesions; subsequent induction of apoptosis appears to occur during replication, possibly due to collision of DNA replication forks with stalled transcription complexes during S phase [56].

Xpa-Null Hupki mice were also highly sensitive to treatment with BaP; while treatment was well tolerated by Xpa-WT Hupki mice, Xpa-Null mice died within 2–3 days of receiving the highest dose tested (i.e. 125 mg/kg bw BaP). This sensitivity to genotoxins has been shown previously for Xpa-Null mice with WT Trp53 after exposure to UV, 7,12-dimethylbenz[a]anthracene (DMBA), BaP and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) [18,57]. As discussed above for HUFs, the sensitivity of Xpa-Null mice to these carcinogens is likely due to TC-NER deficiency and blockage of RNAPII by unrepaired DNA adducts. Xpc-Null mice, deficient only in GG-NER, do not share the same sensitivity [58]. One day following a single treatment with BaP, several tissues analyzed from Xpa-Null Hupki mice had a higher level of BPDE-N-dG adducts compared with Xpa-WT mice. When the animals were treated with 5 daily doses of BaP (12.5 mg/kg bw), however, similar DNA adduct levels were detected in Xpa-WT and Xpa-Null mice. This suggests that GG-NER is activated initially following BaP treatment in Xpa-WT mice, but is unable to deal with continuing damage. Interestingly, when BaP was previously tested in Xpa-Null and Xpa-WT mice with WT Tp53 (3 doses of 13 mg/kg bw per week for 13 weeks), 9 weeks of treatment were required before DNA adduct levels in Xpa-Null mice surpassed those of Xpa-WT mice [22]; in that experiment DNA adduct formation was not assessed following a single treatment. Our results and those of others [22,59] suggest that GG-NER kinetics of BPDE-N-dG adducts in NER-proficient mice is dose- and time-dependent. It is apparent that further investigations are required to explain these observations.

In addition to increased sensitivity to BPDE-N-dG adducts, Xpa-Null HUFs also exhibited enhanced BPDE-induced mutagenesis on the transcribed strand of TP53 compared with Xpa-WT HUFs following BPDE treatment. These data further suggest that Xpa-Null HUFs are unable to repair BPDE-N-dG adducts on the transcribed strand; adducts that do not induce apoptosis may be converted to mutations. While the number of immortal Xpa-WT and Xpa-Null clones harbouring TP53 mutations on the non-transcribed strand was nearly the same, 5.5-fold more Xpa-Null clones contained mutations on the transcribed strand compared to Xpa-WT clones. Further, the number of additional mutations on the transcribed strand induced by BPDE in Xpa-Null HUFs was equal to the overall increase in TP53 mutations in Xpa-Null HUFs compared to Xpa-WT cells (see Table 1). This, and the accompanying statistical analyses, suggests that the increase in BPDE-induced TP53-mutagenesis in Xpa-Null HUFs compared to Xpa-WT cells can be primarily be accounted for by the inability of Xpa-Null HUFs to repair damage on the transcribed strand.

It is unclear why Xpa-deficiency did not also increase TP53 mutagenesis on the non-transcribed strand. It is known that repair of BPDE-DNA adducts is slower on the non-transcribed strand compared to the transcribed strand in the TP53 gene of normal human fibroblasts, creating a bias of mutations on the non-transcribed strand in NER-proficient cells [47]. Despite the relative inefficiency of GG-NER compared to TC-NER, BPDE-DNA adduct removal from the bulk of the genome has been shown, to varying extents, in multiple studies. The amount of removal within 8 h of BPDE exposure ranged between 5 and 60% in normal human fibroblasts [60], to 50% in V79-XEM2 cells [45], to 75% removal in A549 lung carcinoma cells [44]. In the current study, we found that Xpa-WT HUFs removed 30% of BPDE-N-dG adducts within 6 h of treatment. It is not known what percentage of BPDE-N-dG adducts may have persisted in the HUF genomes beyond this time-point.

Few studies have compared BaP/BPDE-induced mutagenesis in NER-proficient and NER-deficient cells. Xpa-Null mouse embryonic stem cells treated with BaP exhibited a higher rate of Hprt mutations than their WT counterparts, although the Xpa-Null cells also had a higher rate of spontaneous mutagenesis [43]. Further, more Hprt mutations were induced by BPDE in an NER-defective Chinese hamster ovary cell line (UV5) relative to a WT line (AA8) [45]. On the other hand, in vivo, similar mutation frequencies at a lacZ reporter gene were detected in the liver and lung of BaP-treated Xpa-Null and Xpa-WT mice; lacZ mutation frequencies did increase in the spleens of Xpa-Null mice, but only after 13 weeks of BaP treatment [22]. Thus, the impact of NER-deficiency on mutagenesis resulting from BPDE-N-dG adducts may be cell-type specific or dependent on the target gene of interest and whether or not the gene is subject to TC-NER (e.g. lacZ is not transcribed by mammalian cells).

In agreement with previous studies, the majority of BPDE-induced TP53 mutations occurred at G:C base pairs in both Xpa-WT and Xpa-Null HUFs, with G:C > T:A transversions being the predominant mutation type [14,61,62]. A high percentage (46–56%) of the mutations at G:C base pairs occurred at CpG sites; G:C > C:G and G:C > T:A transversions were more common at these sites than G:C > A:T transitions. Further, we found that BPDE induced mutations at several sites that are hotspots for mutation in cancer overall (codons 245, 248, 249, 273, 282), or smokers’ lung cancer specifically (codons 157 and 158, in addition to those listed for cancer overall). Codons 157, 158 and 273 were also mutated in prior HIMAs with BaP-treated HUFs [14].

The pattern and spectrum of TP53 mutagenesis can be influenced by a number of factors. In previous studies DNA adduct formation by BPDE was enhanced at methylated CpG sites in TP53 hotspot codons 157, 158, 245, 248, and 273 on the non-transcribed strand [63,64]; the precise mechanism underlying this phenomenon is not yet understood. It has been proposed that the methyl group of 5-methylcytosine allows increased intercalation of BPDE at methylated CpG sites and that this increase in BPDE intercalative binding subsequently results in increased covalent interaction [65,66]. Others have suggested that the methylation of cytosine enhances the nucleophilicity of the exocyclic amino group of the base paired guanine (electronic effect) [67]. All of the CpG sites in Hupki TP53 are methylated [68]. Interestingly, codon 179 (CAT), which is a mutation hotspot in smokers’ lung cancer but does not contain a CpG site or a G on the non-transcribed strand, was not mutated by BPDE in our study and was not targeted by BPDE in normal human bronchial epithelial (NHBE) cells [69]. On the other hand, codon 267 (CGG), which is infrequently mutated in lung cancer but does harbour a CpG site, was mutated by BPDE in two HUF clones and exhibited pronounced BPDE-DNA adduct formation in NHBE cells [69]. Our data provide additional support for the idea that certain TP53 mutation hotspots (i.e. codons 157, 158, 248, 273) act as selective BPDE binding sites.

Additional factors such as sequence context, efficiency of lesion repair (as discussed above), and fidelity of translesion synthesis polymerases also play important roles in TP53 mutagenesis. We found that a common context for BPDE-induced single base substitutions or deletions at G:C base pairs was GG dinucleotide sequences; mutation hotspots for BPDE-N-dG adducts have previously been found in such sequences [70,71]. Sequence context likely influences adduct conformation which may result in different sequence-dependent removal rates of the lesion and also control mutagenic specificity.

Furthermore, the TP53 mutations ultimately observed in human cancers are strongly influenced by functional selection for mutants that have a deleterious impact on the normal function of p53 or that acquire gain-of-function properties [6]. For example, many mutation hotspots occur at codons for amino acids that are essential for DNA contact (248, 273) or structure of the DNA binding domain (175, 245, 249, 282); mutations at these sites create a mutant protein that lacks the ability to transactivate the normal suite of p53 target genes. With the exception of codon 175, all of these hotspots were mutated by BPDE in our study. Further, most of the missense TP53 mutations detected in our study were classified as non-functional and, with one exception, the mutations that retained some functionality occurred only in clones that also harboured a non-functional mutation. Taken together, the pattern and spectrum of mutations generated in this study indicate that similar factors influence TP53 mutagenesis in the HUF immortalization assay and human cancer, further supporting the strength of this model for assessing the effects of carcinogens on this tumour suppressor gene.

We also showed that the replicative capacity of primary HUFs could be extended by culturing the cells at 3% O2. After 11 days of culture, the population increase of HUFs grown at 3% O2 was 10-fold higher than that of HUFs grown at 20% O2. The enhanced growth permitted by temporary culture in 3% O2 provides a substantial increase in cells available for further experiments. To select for TP53-mutated cells, HUFs must eventually be transferred to 20% O2, where the ability of cells to bypass senescence serves as the selection pressure for mutants. Importantly, we found that untreated HUFs grown at 3% O2 for one week were still able to senesce when transferred to 20% O2, and immortal variants that bypassed senescence developed in a similar timeframe to cells maintained at 20% O2. Parrinello et al. found that MEFs cultured at 3% O2 for more than 15 population doublings (≥30 days) lost their propensity to senesce at 20% O2, which they speculated may be due to a mutagenic event or adaptive response [37]. It may therefore only be beneficial to culture HUFs for 1–2 weeks at 3% O2.

In addition to a clear growth advantage, we found that HUFs accumulated a higher level (2–4 fold) of DNA adducts at 3% O2 relative to 20% O2 following treatment with BaP. We did not determine the mechanism for this in our study, but future work could examine the expression/activity of enzymes required for BaP activation (i.e. Cyp1a1 and Cyp1b1) at 3% O2 and 20% O2. Recent work by van Schooten et al., using the human lung carcinoma cell line A549 treated with BaP, has shown that the level of DNA adducts and the gene expression of CYP1A1 and CYP1B1 was increased under hypoxic conditions (0.2% O2), while gene expression of UDP-glucuronosyltransferase detoxifying enzymes UGT1A6 and UGT2B7 decreased [72]. The authors concluded that the balance of metabolism of BaP shifts towards activation instead of detoxification under low oxygen conditions. These results corroborate our findings that altered oxygen levels can influence the metabolic activation of compounds such as BaP.

Although we were unable to detect a significant increase in BPDE-induced TP53 mutations overall in Xpa-Null HUFs compared to Xpa-WT HUFs, perhaps a divergence in mutation frequency would be evident at the genome-wide level. Less than 25% of HUF clones were immortalized by TP53 mutation, thus other genes are clearly involved in senescence bypass by HUFs and could also be targeted by BPDE [53]. Recently, exome sequencing of DNA from HUFs treated with various mutagens (e.g. BaP and AAI) was used to extract genome-wide mutational signatures of mutagen exposure [73]. This study demonstrates that a single mutagen-treated immortal HUF clone harbours hundreds of single base substitutions at the exome level, likely consisting of both immortalization driver mutations and passenger mutations. Therefore, whole-genome sequence analysis of BPDE-treated Xpa-WT and Xpa-Null clones may allow differences in mutation frequency in the context of the genome to be detected that are not observed in an assay for a single gene. Furthermore, beyond simply increasing mutation frequencies, Xpa-Null HUFs may be useful in the future to enhance our understanding of the role of NER in shaping carcinogen-induced mutagenesis of both TP53 and the genome.

Conflict of interest

None.