It is currently widely accepted that genetic instability is key to cancer development. Many types of cancers arise as a consequence of a gradual accumulation of nucleotide aberrations, each mutation conferring growth and/or survival advantage. Genetic instability could also proceed in sudden bursts leading to a more drastic upheaval of structure and organization of the genome. Genetic instability, as an operative force, will produce genetic variants and the greater the instability, the larger the number of variants. We report here that the overexpression of human DNA polymerase κ, an error-prone enzyme that is up-regulated in lung cancers, induces DNA breaks and stimulates DNA exchanges as well as aneuploidy. Probably as the result of so many perturbations, excess polymerase κ favors the proliferation of competent tumor cells as observed in immunodeficient mice. These data suggest that altered regulation of DNA metabolism might be related to cancer-associated genetic changes and phenotype.
Genetic instability in cancer has a wide range of expression modes. It may affect nucleotide proofreading (leading to base substitutions, deletions, or additions), chromosomal structure (producing translocations, sequence gains, or losses), or karyotypic integrity (resulting in aneuploidy). Whereas the impact of such genetic abnormalities in cancer progression remains controversial (1), it is now accepted that stochastic occurrence of genetic disorders and its gradual increase in frequency along the natural history of the disease is often promoted by either early mutations in genes that maintain genetic stability in normal cells and/or bigger genetic changes such as gains or losses of chromosomes (2, 3).
Together the highly conserved pathways involved in genome supervision act to limit cancer risk. This is clearly illustrated by patients bearing a germ line mutation in genes encoding “guardians” of genome stability, who show vastly increased risk of cancer development. Hereditary forms of colon, breast, ovary and skin cancers are caused by mutations in mismatch repair (e.g., hMLH1), DNA break repair (e.g., BRCA1), nucleotide excision repair pathways (e.g., XP proteins), or affecting the capacity to replicate through DNA damage (e.g., polη), respectively (4–7). In somatic cancers, such early mutations become “diluted” in the alterations that follow, making the relationship less obvious. However, it is very likely that genetic instability either induces or accelerates the proliferation of cancer cells by favoring the emergence of variant cells. Indeed, a controlled alteration of genes involved in genome maintenance promotes or favors carcinogenesis (8, 9).
The accurate maintenance of undamaged genomic DNA requires the action of the error-free DNA polymerases polδ and polε. When DNA is damaged, specialized error-prone DNA polymerases, including polζ, polη, polι, polβ, and polκ are believed to take part in the replication repair of damage that otherwise would not be tolerated (10). Polκ, the product of the human dinB1 gene, bypasses in vitro thymine glycols (11) or benzo[a]pyrene-N(2)-dG (12) lesions by preferentially incorporating correct nucleotides. These features suggest a specialized and adaptative role of Polκ towards such lesions.
Polκ is targeted to the replication machinery probably because of its interaction with the proliferative cell nuclear antigen (13, 14) and the Rev1 protein (15). Because it lacks proofreading activity (16)and replicates DNA with limited processivity (17), Polκ manifests an error rate of about 5 × 10−3 per nucleotide incorporated when copying undamaged DNA in vitro (12, 17). In vivo this enzyme confers a mutator phenotype when overexpressed (14, 18). Besides its specialized role this enzyme could thus promote untargeted mutagenesis when up-regulated. Here we wondered whether the human DNA polymerase Polκ, which is overexpressed in non–small cell lung cancer (NSCLC, ref. 19), could be a candidate likely to play a role in genetic instability and cancer progression.
We found that the ectopic expression of Polκ promotes homologous events, aneuploidy as well as tumorigenesis in nude mice. Moreover, we found that of eight polκ NSCLC biopsies, seven displayed losses of heterozygosity (LOH) compared with adjacent nontumoral tissues. Taken together, these data suggest that misregulation of this error-prone enzyme directly promotes or accelerates the emergence of a large spectrum of genetic disorders associated with a malignant phenotype.
Materials and Methods
Cell Lines and Plasmids. MRC5 human fibroblasts (American Type Culture Collection, Rockville, MD, AA8 Chinese hamster ovary (CHO) hamster cells (American Type Culture Collection) and DRA10 CHO hamster cells (provided by M. Jasin, Memorial Sloan-Kettering Cancer Center, New York, NY) were grown in supplemented MEMα (Life Technologies, Gaithersburg, MD) medium. hdinB cDNA was PCR-amplified using the pHSE2 DNA (provided by H. Ohmori, Faculty of Medicine, Kyoto University, Japan) as a template and CTCGCTAGCCCATGGATAGCACAAAAGAA and GCGGGATCCTTACTTAAAAAATATATCAA as primers. The PCR product was digested using NheI/BamHI and then inserted into the restricted pIRESpuro2 plasmid (Invitrogen, San Diego, CA) to produce the pIRESpolκ (pIK) vector. To generate the pGFPpolκ (pGK) plasmid we used pHSE2 as a template and GGGCTCGAGCTCGATAGCACAAAGGAGAAGTGTGACAG and GGGGATCCTTACTTAAAAAATATATCAAGGGTATGTTTGGG as primers. PCR products were digested with XhoI/BamHI and then inserted into the restricted pEGFPC3 (Clontech, Palo Alto, CA). Cells were transfected using either Fugene6 (Roche, Nutley, NJ) or JetPEI (QBiogen, Illkirch, France). Polκ overexpression was checked by immunoblotting nuclear proteins with anti-hpol κ serum (1/2,000, provided by H.Ohmori) and normalizing to proliferative cell nuclear antigen expression. When real time-PCR (RT-PCR) was used, RNA was extracted using the RNeasy Minikit (Qiagen, Chatsworth, CA) then checked by capillary electrophoresis (RNA 6000 nanochips kit, Agilent, Palo Alto, CA). Complementary DNAs were synthetized using the ThermoScript RT-PCR system (Invitrogen) following the manufacturer instructions and RT-PCR when then done by using probes specific to human polκ (Assay-On-demand Hs00211963, Applied Biosystems, Foster City, CA).
γ-H2AX Detection. For immunoblotting, 250,000 cells seeded 24 hours before were washed with PBS, harvested by scraping in 150 μL of PBS 1% SDS, and sonicated. Twelve microliters of the total protein extract were resolved by electrophoresis on a 15% SDS-polyacrylamide gel. The gel was transferred to a polyvinylidene difluoride Hybond-P membrane (Amersham, Arlington Heights, IL), which was probed with monoclonal anti γ-H2AX antibody (Upstate, Lake Placid, NY) at room temperature for 1 hour. Horseradish peroxidase-goat anti-mouse antibody was used as the secondary antibody (Sigma, St. Louis, MO). The membrane was then incubated with enhanced chemiluminescence reagent (Amersham) and exposed to X-ray film. Equivalent amounts of loaded extracts were assessed by immunoblotting the membrane with an anti-actin antibody. For immunofluorescent staining, 50,000 cells were grown on coverslips for 24 hours, fixed 15 minutes in 4% PFA, permeabilized 5 minutes in PBS-TritonX100 0.2%, blocked 30 minutes in PBS 3% bovine serum albumin, and then incubated 1 hour with the monoclonal anti-γ-H2AX antibody (1:500). Asa positive control, 0.03% methyl methane sulfonate was added for 1 hour. Anti-mouse fluorescein antibody (Molecular Probes, Eugene, OR) was used as secondary antibody for 1 hour. Coverslips were prepared with Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole. Fluorescence images were captured by using a Zeiss Axioskop fluorescence microscope (100×).
Homologous Recombination Assay. A few passages after obtaining stable clones, cells (106 per dish) were plated and incubated for 10 days with G418 (1 mg/mL). Stained colonies of more than 50 cells were scored. The recombination rate was calculated by taking into account the plating efficiency determined by plating 300 cells in the drug-free medium.
Loss of Heterozygosity. Genomic DNA was extracted from surgically resected tumors and adjacent normal tissues (>3 cm away from the tumor) at Chiba Cancer Center (Japan) and Institut Curie (Paris, France) and was used for LOH analysis. Ten DNA repeats [i.e., D1 (AAGG 1p33.6), D2 (AAT 2q24), D3 (GATA 3q27), D5 (GATA 5q21), D10 (GATA 10q24), D11(GATA 11p13), D15 (GATA 15q26), D16 (TAGA 16p13.3), D21 (TAAA 21q22), D22 (GATA 22q11)] dispersed throughout the genome were simultaneously amplified as described previously (20) and analyzed by electrophoresis in a MegaBace capillary automated DNA sequencer (Amersham Biosciences, Buckinghamshire, United Kingdom) using the Genetic Profiler 1.5 software (Amersham Biosciences). Loss of one allele was visualized as an imbalance in the proportions of the two peaks.
Karyotypes. Nocodazole (10 μmol/L, Sigma) was added to the medium for 2 hours and cells were trypsinized. After hypotonic treatment (50mmol/L KCl), the pellet was fixed in ethanol/acetic acid (3:1). The metaphasic nuclei were then spread on slides and stained with Giemsa (Sigma). Chromosomal distributions included the analysis of metaphase spreads.
Tumorigenicity Assays. Immunodeficient mice (Swiss nu/nu males, 4 weeks old, Charles River France, Les Oncins, France) received 0.5 × 106cells by s.c. injection in the neck, after being anaesthetized with isofurane. Mice were analyzed every 2 days and sacrificed when the diameter of the tumor reached 1 cm.
Ectopic Expression of Human Polκ. Overexpression of human Polκ in CHO cells used in this study was checked at the protein level (Fig. 1). It was also verified by RT-PCR using primers designed to match specifically human pol cDNA. For untransfected cells that only express endogenous hamster polκ, a signal was detected after 40 cycles, whereas only 30 cycles were required for detection of ectopic human pol mRNA in all the independent polκ-overexpressing clones (data not shown). These clones were further characterized and displayed high mutation frequency (data not shown), confirming previous results in murine (18) and human (14) cells.
Ectopic Polκ Expression Induces DNA Breaks. Because of its distributive mode of DNA synthesis, we hypothesized that excess Polκ may affect DNA elongation during the course of DNA replication and therefore generate breaks. To investigate the presence of DNA breaks we analyzed the presence of γ-H2AX, the phosphorylated form of the H2AX histone detected in response to double-strand breaks (21). H2AX is a core component of chromatin that is phosphorylated in chromatin flanking DNA double strand breaks (22). As positive controls, we treated cells with methyl methane sulfonate, which induces phosphorylation and accumulation into nuclear foci of H2AX (23). Polκ cells displayed more foci, and with higher intensity, than control cells, 47.6% (pIK2) and 60% (pIK8) displaying foci, whereas only 13.8% control cells were positive (Fig. 2B). Western blot analysis with γ-H2AX antibodies (Fig. 2C confirmed the immunofluorescence data, showing that polκ cells expressed higher levels of γ-H2AX protein than controls. Since it is widely accepted that γ-H2AX represent remodeled chromatin in response to DNA double-strand breaks (24), these data support the idea that accumulation of double-strand breaks has occurred in these cells.
Ectopic Polκ Expression Results in Increased Homologous Recombination. Since DNA breaks may act as a source of DNA exchanges, we then investigated the impact of Polκ overexpression on homologous recombination (HR) pathway. We transfected the pIres-polκ (pIK) plasmid into a CHO DRA10 cell line that contains a unique copy of an intrachromosomal recombination substrate composed of two inactive copies of the neomycin-resistant gene (25). Intrachromosomal HR events can restore a functional neomycin gene rendering cells resistant to G418 (25). Two independent clones DRA10 pIK2 and DRA10 pIK8 were selected because of a 2- and 2.1-fold overexpression, respectively (Fig. 1. Table 1 shows that the HR frequency is significantly increased in these Polκ overexpressing clones compared with the control cell lines (the parental cell line and a clone transfected with the empty vector). HR is stimulated by at least 41- and 7.7-fold in pIK2 and pIK8 clones, respectively. Such increase was also found (data not shown) by using independent recombination marker cells (i.e., V79-derived Chinese hamster SPD8 cells, which contains a duplication in the endogenous hprt gene that is resolved by HR; ref. 26).
|Clone .||Homologous recombination (×10−6) frequency .||SD (×10−6) .|
|Clone .||Homologous recombination (×10−6) frequency .||SD (×10−6) .|
P < 0.001 when compared with DRA10; P < 0.05 when compared with CT2B.
NOTE: The frequency of homologous recombination in control (DRA10, empty pI2) and Polκ (pIK2 and pIK8) cells is the number of G418-resistant colonies per million cells, considering the plating efficiency of each clone. The data indicate the mean of three to five independent experiments.
Abbreviation: ND, not determined
Polκ Overexpression Results in LOH. It is admitted that LOH most likely results from homologous recombination events(27). Therefore, we investigated whether up-regulated Polκ, which promotes recombination, also induces LOH in MRC5 human fibroblasts, by using a set of PCR primers matching human DNA repeats dispersed along the genome. We did not find any LOH in all isogenic control MRC5 clones that we tested (data not shown). In contrast, LOH was observed in pGFPpolκ-transfected polκconst cells that stably overexpress Polκ (Fig. 1, with the loss of an allele at the D11 (GATA 11p13) locus (Table 2). These findings suggest that recombination stimulated by excess Polκ could induce LOH.
|Cells or tumors .||Age .||Sex .||Smoke .||Stage .||LOH .||Polκ excess .|
|Cells or tumors .||Age .||Sex .||Smoke .||Stage .||LOH .||Polκ excess .|
NOTE: polκconst cells are MRC5 cells constitutively expressing Polκ
Ni/Ti are coupled Pol κ–overexpressing NSCLC biopsies from Japanese patients. ML4 and ML5 are coupled nonoverexpressing NSCLC control biopsies.
The 10 DNA repeats dispersed throughout the genome (AFL10 system) and simultaneously amplified were D1 (AAGG 1p33.6), D2 (AAT 2q24), D3 (GATA 3q27), D5 (GATA 5q21), D10 (GATA 10q24), D11 (GATA 11p13), D15 (GATA 15q26), D16 (TAGA 16p13.3), D21 (TAAA 21q22), and D22 (GATA 22q11)
—, no clinical information.
Abbreviations: N, nontumoral; T, tumoral; ND, not detected.
To assess the physiologic relevance of these data we investigated LOH in Polκ-overexpressing cancer biopsies by using the same set of PCR primers. We analyzed eight paired tumor and normal biopsies from Japanese patients suffering from NSCLC. The whole biopsies were shown as overexpressing Polκ both at the full-length protein (ref. 19 and data not shown) and transcript (RT-PCR data not shown) levels. As additional negative controls we analyzed NSCLC coupled biopsies (ML4 and ML5), which do not overexpress Polκ at either level (data not shown). Table 2 shows that of the eight paired polκ biopsies, seven displayed LOH in D1, D3, D5, D10, or D11 loci, D10 (GATA repeat in 10q24 region) and D3 (GATA repeat in 3q27 region) being particularly concerned. In contrast, we found such kind of LOH neither in ML4/ML5 (Table 2) nor in additional RT-PCR negative controls (data not shown). No correlation was found between the presence of LOH and either age, sex, pathology stage, or smoking habits (Table 2). Taken together these data suggest that up-regulation of Polκ could play a role in the generation or acceleration of genetic instability in NSCLC.
Excess Polκ Promotes Aneuploidy. We then analyzed and compared the chromosomal distribution in the parental AA8, a clone transfected with an empty vector and two independent polκ cell lines (pIK9 and pIK10). We found that ectopic expression of Polκ induced aneuploidy by mostly promoting loss of chromosomes (Fig. 3), demonstrating that excess Polκ induces genetic instability also at the chromosomal level.
Formation of Tumours in Immunodeficient Nude Mice. Inline with our observations that elevated expression of Polκ promotes a pleiotropic genetic instability, we investigated the tumor incidence by inoculating nude mice with polκ AA8 or K1-derived DRA10 CHO cells and various controls. Less than 17% (4 of 24) control clones induced formation of tumors after 4 weeks (Table 3). In contrast, when we injected cells from five independent stably transfected Polκ clones we observed that 45% (19 of 42) of the injections induced tumors. Regarding the high number of mice analyzed, these data significantly indicate (P < 0.01) that excess Polκ amplifies the malignant phenotype of CHO cells, which already express a mutated p53 protein (28).
During the long time interval between carcinogen exposure and clinical detection, cancer cells acquire the possibility to proliferate by adapting to organic barriers and therapeutic treatments. Such adaptation could be facilitated by the intrinsic genetic variability of tumor cells (3). Aneuploidy, microsatellite instability, frequent allelic loss, and multiple small-scale insertions and deletions have indeed all been found in common types of cancers, the total number of mutations being likely to reach 1012 (29). Normal mutation rates being insufficient to account for so many mutations, it has been proposed that these disorders could be in part generated by early mutations in genes that maintain genome integrity (3). At least three cancer predisposition syndromes in humans result from early defects in DNA repair pathways: xeroderma pigmenstosum (caused by defective processing of UV damage), hereditary nonpolyposis colon cancer (a consequence of defective mismatch repair), and hereditary breast cancer (because of a defective repair of DNA breaks; refs. (4–7).
The mutator phenotype of cancer cells could also result directly from early alterations of genes that control the accuracy of DNA replication (8). Consistent with this hypothesis, a point mutation in the proofreading domain of polδ causes a mutator and cancer phenotype in mice, strongly suggesting that unrepaired DNA polymerase errors contribute to carcinogenesis (30). Because of their relaxed fidelity, we proposed that up-regulation of the error-prone DNA polymerases presumed to be specialized in the bypass of DNA lesions can lead to untargeted mutagenesis along the undamaged regions of the genome and contribute to tumorigenesis (31). In this respect, we have previously reported that the up-regulation of Polβ could contribute to the emergence of a mutator phenotype (32) by interfering with error-free DNA transactions such as DNA repair or DNA replication (33, 34), rendering them mutagenic. As a consequence of such deleterious actions, excess Polβ favors tumorigenesis in mice (35). Recently, it has been shown that Polιexpression is elevated in breast cancer cells (36), authors suggesting that it could be involved in the generation of increased spontaneous mutations during DNA replication, thereby contributing to the accumulation of genetic damage. Our working model is that an even slight misregulation of specialized error-prone DNA polymerases could play an enhancer role in the accumulation of genetic disorders. It has been previously suggested that an x-fold increase in an in vivo mutation rate could increase cancer incidence by a factor of xn, where n is the number of mutations required to develop a tumor (37).
To gain further insight into the physiologic relevance of an imbalance between error-free and error-prone DNA polymerases, we analyzed here the genetic and cancer consequences of ectopic expression of the error-prone DNA Polκ, which is up-regulated in human lung cancer (19). We report that a slight increase of Polκ promotes DNA recombination probably because of a perturbation of the replication machinery and the subsequent generation of DNA breaks. High steady state levels of polκ transcripts, described in mouse testis and ovaries (38) further suggest a role in meiosis and thus possibly in recombination. We also found that such ectopic expression induces LOH, the most commonly reported alteration taking place in tumor suppressors, mostly induced by recombination (27). We propose that the genetic exchanges stimulated by excess Polκ result from a competition between the error-prone Polκ and error-free Polδ and/or Polϵ. Indeed we and others (13, 14) already showed that Polκ colocalizes or interacts with the proliferative cell nuclear antigen replicative cofactor, which plays a key role in coordinating and organizing the replication forks. However, we cannot rule out that the LOH observed in lung biopsies could be independent from Polκ involvement. LOH, which is one of the genetic alterations commonly observed in sporadic tumors, can indeed arise from several pathways, including not only mitotic recombination but also chromosomal deletion, mitotic nondisjunction or reduplication and point mutation. Misregulation of genes other than polκ could thus be involved in such alternative pathways and induce LOH in the tumors analyzed here.
Replication slippage and loss of chromosomes are two additional basic mechanisms likely to produce LOH (27). Polκ being a polymerase with low processivity (17) and replication slippage being admitted as involving DNA polymerase pausing and dissociation (39), excess Polκ could induce frameshift mutations. Indeed, purified Polκ promotes in vitro −1 and −2 nucleotides deletions by replicating an oligonucleotide and in vivo experiments showed that 6TGR polκ cells displayed −1 nucleotide deletions in the sequenced hprt gene (18),(40). These data suggest that LOH found in MRC5 cells could be in part attributed to enhanced replication slippage. Finally, we also found that excess Polκ confers hypoploidy in CHO cells, suggesting that an impact on chromosome partitioning could thus also explain LOH. We previously showed such an aneuploidy with Polβ whose overexpression leads to a perturbed G2-M checkpoint (35).
Taken together, the data presented here suggest a possible incidence of excess Polκ in cancer-associated LOH, probably because of a pleiotropic impact on the fidelity of DNA replication, DNA recombination and chromosome partitioning. This was corroborated with the analysis of polκ NSCLC lung tumors. By comparing tumoral and adjacent nontumoral tissues, we detected LOH in 7 of 8 polκ NSCLC biopsies, contrary to nonoverexpressing control samples. These results suggest that genetic factors such as Polκ up-regulation could predisposeto lung cancer. To further investigate tumor incidence that might arise from these major genetic disorders, we inoculated immunodeficient mice with polκ cells. Our data show that excessPolκ favors the proliferation of competent tumor cells.
Our working model is that aberrant expression of specialized DNA polymerases can, in cancer cells, along with defective cell cycle control, be a motor for genetic instability. This work shows the importance of a tight regulation of the cellular expression of specialized error-prone DNA polymerases such as Polκ for the maintenance of genome integrity. Reduced fidelity of the replication machinery because of an overrepresentation of Polκ might indeed accelerate tumorigenesis by conferring a selective growth advantage during cancer cell evolution. More generally, an increased knowledge on the involvement of misfunctioning DNA replication in cancer could lead to the definition of novel cancer prognostic markers and the design of new anticancer drugs for individual therapy orientation.
Grant support: Ligue Nationale Contre le Cancer (IGC labellisée), Cancéropôle Grand Sud Ouest (réseau labellisé “L' instabilité génétique comme signature péjorative de la maladie”, coordonnateur C. Cazaux), “Association Contre le Cancer Grand-Sud” (RT-PCR platform), and CAPES/COFECUB.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank A. Lehmann (University of Sussex, Brighton, United Kingdom) for critical reading of the article, M. Jasin (Memorial Sloan-Kettering Cancer Center, New York, NY) for the gift of the DRA10 cell line, H. Ohmori (Kyoto University, Japan) for providing the dinB cDNA and anti-Polκ antibodies, and A. Kumari (University of Sheffield, United Kingdom) for help with recombination assays.