Oncogene (2013) 32, 242 --250 & 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE REV3L 30UTR 460 T4C polymorphism in microRNA target sites contributes to lung cancer susceptibility

S Zhang1,2,5, H Chen1,5, X Zhao1, J Cao2, J Tong2,JLu1,WWu1, H Shen3,QWei4 and D Lu1

REV3Lp, the catalytic subunit of DNA zeta, is the major participant in translesion DNA synthesis. Recent evidence suggests that REV3L has an important role in the maintenance of genome stability despite its mutagenic characteristics. Such a function makes it a cancer susceptibility candidate . To investigate association between REV3L polymorphisms and lung cancer risk in a Chinese population, we first genotyped 15 common polymorphisms of the REV3L gene and found that three single nucleotide polymorphisms (rs465646, rs459809 and rs1002481) were significantly associated with lung cancer risk. One of the strongest associations observed was for the 30-terminal untranslated region (30UTR) 460 T4C polymorphism (rs465646) (adjusted odds ratio (OR) ¼ 0.69 for TC/CC; P ¼ 0.007, compared with TT). Similar results were obtained in a subsequent replication study (adjusted OR ¼ 0.72; P ¼ 0.016). Combined data from the two studies of 1072 lung cancer patients and 1064 cancer-free controls generated an even stronger association (adjusted OR ¼ 0.71; P ¼ 3.04 Â 10À4). This 30UTR 460 T4C variant was predicted to modulate the binding of several micro RNAs. Surface plasmon resonance analysis and luciferase assays showed that the T allele demonstrated a stronger binding affinity for miR-25 and miR-32, resulting in significantly weaker reporter expression levels. Additional experiments revealed that miR-25/32 could downregulate endogenous REV3L. Furthermore, the tumor-suppressing role of REV3L was confirmed by the foci formation assay. These results support our hypothesis that the REV3L rs465646 variant modifies lung cancer susceptibility in Chinese Han population by affecting miRNA-mediated gene regulation.

Oncogene (2013) 32, 242--250; doi:10.1038/onc.2012.32; published online 20 February 2012 Keywords: genetic polymorphism; REV3L; lung cancer; microRNA

INTRODUCTION DNA synthesis.7 Nevertheless, these special are Lung cancer remains the leading cause of cancer-related deaths regarded as error-prone, because the fidelity of these special worldwide.1 Exposure to environmental carcinogens, especially polymerases has been estimated to be B1000--4000-fold lower tobacco smoke, has been proved to be the most important risk than the replicative DNA polymerases.6--8 The TLS processes factor for lung cancer. However, only a fraction of smokers sacrifice the fidelity of the to ensure the accom- develop lung cancer, suggesting interplay between genetic plishment of DNA replication.9 susceptibility and environmental exposure. Polymerase Zeta (Pol z or Pol Zeta) consists of two subunits, Cells constantly encounter various carcinogens or cytotoxic catalytic REV3 and structural REV7. This confers most of agents that directly or indirectly induce DNA damage, leading to spontaneous and induced mutagenesis found in yeast.10 The yeast DNA cross-links, DNA strand breaks or mutations that cause REV3 mutant strain exhibits reduced levels of mutagenesis when genomic instability. DNA repair mechanisms, including nucleotide treated with mutagens.11 The function of vertebrate REV3p shares excision repair, base excision repair, mismatch repair and double- some similarity with its yeast counterpart, despite that a region of strand break repair are responsible for the integrity of the B1400 amino acids does not exist in yeast. REV3-null mouse genome.2 Besides, eukaryotic cells from yeast to humans possess embryonic fibroblasts are more sensitive to UV irradiation, DNA post-replication repair or DNA damage tolerance system. mitomycin C and g irradiation with increased chromosomal Translesion DNA synthesis (TLS) and homologous DNA recombi- abnormalities.12 DT40 cells deficient in REV3 showed dramatic nation are two major post-replicational repair pathways.3 The sensitivity to cisplatin, mitomycin C and transplatin.3,13 These ubiquitous TLS consists of a series of specialized polymerases, results indicate that mammalian REV3L participates in cell including polymerase k, z, Z and i.4 In DNA replication, replication tolerance, which is consistent with its homolog in yeast. complex will be blocked at DNA lesions, where cells will initiate However, unlike the non-essential role in yeast, mammalian REV3 the TLS system to avoid disastrous collapses of the replication has a critical role in maintenance of genome stability. Human REV3L forks and double-strand DNA breaks.5 In TLS, the proliferating cell is ubiquitously expressed in normal and malignant human tissues, nuclear antigen recruits some low-fidelity DNA polymerases to suggesting its necessity. Disruption of mREV3 in mice lead to replace the replicative polymerase to incorporate nucleotides embryonic lethality;14,15 moreover, REV3LÀ/À showed dramatic opposite to damaged nucleotide residues rather than directly genome instability represented by chromosome rearrangements repairing them,6 which is followed by resumption of processive and translocations.3,12 The expression of human REV3L can be

1State Key Laboratory of Genetic Engineering and MOE Key Laboratory of Contemporary Anthropology, Institute of Genetics, School of Life Sciences, Fudan University, Shanghai, China; 2School of Radiation Medicine and Protection, Medical College of Soochow University, Suzhou, China; 3Department of Epidemiology and Biostatistics, Cancer Center of Nanjing Medical University, Nanjing, China and 4Department of Epidemiology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 5These authors contributed equally to this work. Correspondence: Professor D Lu, State Key Laboratory of Genetic Engineering and MOE Key Laboratory of Contemporary Anthropology, Institute of Genetics, School of Life Sciences, Fudan University, 220 Handan Rd., Shanghai 200433, China. E-mail: [email protected] Received 30 July 2011; revised 20 December 2011; accepted 21 December 2011; published online 20 February 2012 microRNA polymorphism in lung cancer susceptibility S Zhang et al 243 upregulated at the transcriptional level in response to N-methyl-N’- Because of the relatively small number of variant homozygotes nitro-Nnitrosoguanidine. Although N-methyl-N’-nitro-Nnitrosoguani- (o5%) observed in both cases and controls, there was not enough dine-induced mutation was REV3 dependent, suppression of REV3L statistical power to detect any effect among variant homozygotes. by small interfering RNA had no effect on spontaneous mutation.16 To enhance statistical power, autosomal SNPs were modeled in a Recent findings indicate that REV3L can attenuate accumulation of dominant mode of inheritance (heterozytotes þ mutant homo- spontaneous double strand break (DSB) and prevent tumorigen- zygotes vs wild homozygotes), if the frequency of variant esis.17 Thus, mammalian REV3L shoulders more important functions homozygotes was o10%, and codominant mode of inheritance besides handling point mutations. Constitutive and induced expres- (heterozytotes vs wild homozygotes, variant homozygotes vs wild sion of this gene is associated with levels of genome stability. homozygotes), if the variant homozygotes frequency was at least Pan et al.18 reported human Pol Zeta is significantly down- 10%. Further logistic regression analyses revealed that after regulated in human lung, stomach and colorectal cancers. Given adjustment for confounding factors, significantly decreased lung the essentiality of REV3L in genomic stability, we therefore cancer risk was observed for the rs465646 TC and CC genotypes hypothesize that variants in REV3L, the human catalytic subunit (adjusted OR ¼ 0.69, 95% confidence interval (CI) 0.53--0.90; of Pol zeta, are associated with the lung cancer susceptibility. To P ¼ 0.007) and the rs1002481TA and AA genotypes (adjusted investigate the association of REV3L with lung cancer risk as well OR ¼ 0.68, 95% CI 0.52--0.90; P ¼ 0.006), compared with their wild- as gene--environment interactions, we genotyped 15 polymorph- type genotypes. Moreover, the CC genotype of rs459809 showed a isms in 517 cancer-free controls and 500 lung cancer cases in a significantly increased risk of lung cancer with OR of 1.44 Chinese population. Three single nucleotide polymorphisms compared with the TT genotype (Supplementary Table S1). (SNPs) were significantly associated with lung cancer risk. One of Of the three SNPs significantly associated with lung cancer risk, the polymorphisms located in the 30-terminal untranslated region rs465646 was of the most interest because this polymorphism (30UTR) has been proved to be functional, participating in the showed the potential to be functional for the following reasons: (1) miRNA-mediated interplay. We further demonstrated that REV3L this SNP is located in the 30UTR, which is considered a regulatory was downregulated in human lung cancer tissues, acting as a region. (2) This 30UTR 460T4C SNP resides in a rather conserved tumor suppressor gene. region among species (Supplementary Figure S1). Replacement of the allele T to C may completely or partially impair the function of this region. (3) Bioinformatics tools, Targetscan (http://www.targets- can.org) and Pictar (http://www.pictar.mdc-berlin.de), both predicted RESULTS that the REV3L 30UTR harbors two binding sites for miR-25/32/92/ REV3L polymorphisms are associated with lung cancer risk 363/367 for the following reasons (Figure 1a). Interestingly, this We investigated and validated the association between REV3L SNP is located within one of these binding sites. Therefore, this variants and risk of lung cancer in two independent case--control polymorphism was most likely to be functional in modulating studies in a Chinese Han population. The characteristics of lung mRNA::miRNA interaction (Supplementary Figure S2). cancer patients and controls in study 1 and study 2 were The functional significance of the 30UTR 460 T4C polymorph- previously described.19,20 All markers were in Hardy--Weinberg ism prompted us to validate the observed association between equilibrium in the control subjects (P40.05). The minor allele this polymorphism and risk of lung cancer in another case--control frequency for each polymorphism in our control subjects was also study. The replication study (study 2) used another 572 lung similar to that published for the Chinese population included cancer cases and 547 controls. As shown in Table 2, we found a in the HapMap database or dbSNP database (Table 1). We similar protective effect of TC and CC genotype against lung genotyped 15 SNPs of REV3L gene in study 1. Three SNPs showed cancer risk (adjusted OR ¼ 0.72, 95% CI 0.55--0.94; P ¼ 0.016), significant differences in minor allele frequencies between the further confirming our results from study 1. There was no cases and the controls (P ¼ 0.015 of rs465646, P ¼ 0.049 of statistical heterogeneity between the two studies (P ¼ 0.767), rs459809 and P ¼ 0.009 of rs1002481) (Table 1). and combined results by pooling studies 1 and 2 generated a

Table 1. Information about 15 genotyped SNPs of REV3L gene

Gene name (OMIM No.) No. NCBI Genomic Location Base MAF P-valueb P-value % and locus SNP ID positiona in gene change for HWEc Genotyped In databased Cases Controls

REV3L 1 rs181294 15789631 3’ UTR G4A 0.17 0.187 0.209 0.216 0.775 100 OMIM:602776 2 rs465646 15790187 3’ UTR T4C 0.15 0.155 0.196 0.014 0.843 100 3 rs457497 15791978 Intron 33 T4C 0.37 0.347 0.348 0.978 0.910 99.8 6q21 4 rs240998 15798308 Intron 32 C4T 0.33 0.340 0.345 0.803 1.000 100 5 rs240969 15805237 Intron 29 C4T 0.33 0.339 0.345 0.766 1.000 100 6 rs240966 15814094 Intron 27 C4T 0.18 0.187 0.212 0.162 0.906 100 7 rs459809 15822596 Intron 25 T4C 0.48 0.496 0.460 0.049 0.588 100 8 rs464401 15838436 Intron 22 C4T 0.46 0.467 0.442 0.257 0.910 100 9 rs240993 15843143 Intron 20 T4C 0.44 0.470 0.446 0.276 0.878 99.9 10 rs462779 15865316 Exon 15 C4T 0.48 0.499 0.458 0.055 0.726 100 11 rs1002481 15881451 Intron 7 T4A 0.14 0.156 0.200 0.009 0.801 100 12 rs6922226 15893913 Intron 6 T4C 0.49 0.499 0.458 0.055 0.862 100 13 rs6937734 15906071 Intron 4 C4T 0.48 0.499 0.458 0.055 0.862 100 14 rs6935759 15948149 Intron 2 C4T 0.49 0.499 0.459 0.059 0.822 99.6 15 rs6913769 15962326 Intron 2 C4T 0.45 0.465 0.440 0.260 0.931 99.9 Abbreviations: HWE, Hardy-Weinberg equilibrium; MAF, minor allele frequency; OMIM, online Mendelian inheritance in man (http://www.ncbi.nlm.nih.gov/sites/ entrez?db ¼ OMIM); SNP,single nucleotide polymorphism. aSNP position in NCBI dbSNP database. bP-value for difference in allele distributions between cases and controls. cHWE P-value in the control group. dMAF from both HapMap and dbSNP databases. Bold characters indicate corresponding P values are o0.05.

& 2013 Macmillan Publishers Limited Oncogene (2013) 242 --250 microRNA polymorphism in lung cancer susceptibility S Zhang et al 244

Figure 1. Identification of two miRNA-binding sites in REV3L 30UTR and the effection of T/C polymorphism on expression. (a) Schema of REV3L 30UTR and 460 T4C SNP. The nucleotide number is relative to the translational stop codon. (b) Schema of the four constructs harboring different alleles or deletions of miRNA-binding sites. (c) Relative luciferase activity of pGL-T versus pGL-DmiR and pGL-DDmiR in HELF, H1299 and H460 cells. At 24 h after transfection with pGL-T, pGL-DmiR or pGL-DDmiR, luciferase assays were performed. Relative luciferase activities were assayed by using Renilla luciferase activity for normalization. (d) Influence of miR-25/32 or their inhibitors on pGL-T, pGL-DmiR and pGL-DDmiR expression. (e) Relative luciferase activity of pGL-T vs pGL-C in the HELF and H1299 cells. The plasmid was mock co-transfected or co-transfected with miR-25 or miR-32. Values are means±s.d. from more than three separate experiments, each performed in triplicate. *Po0.05, **Po0.01.

Table 2. Genotype frequencies of the rs465646 among lung cancer patients and controls, and its association with lung cancer risk

Logistic regression SNP ID Genotype Cases Controls P-value (2df)a No. (%) No. (%) Adjusted OR (95%CI)b P-valueb

Study 1 TT 362 (72.4) 335 (64.8) 0.032 1.00 (referent) rs465646 TC 121 (24.2) 161 (31.1) 0.68 (0.51--0.90) 0.008 CC 17 (3.4) 21 (4.1) 0.76 (0.39--1.47) 0.41 TC+CC 138 (27.6) 182 (35.2) 0.69 (0.53--0.90) 0.007 Study 2 572 547 rs465646 TT 402 (70.3) 356 (65.1) 0.014 1.00 (referent) TC 157 (27.4) 174 (31.8) 0.80 (0.62--1.04) 0.1 CC 13 (2.3) 17 (3.1) 0.68 (0.32--1.44) 0.32 TC+CC 170 (29.7) 191 (34.9) 0.72 (0.55--0.94) 0.016 All combined 1072 1064 rs465646 TT 764 (71.3) 691 (64.9) 2.40 Â 10À4 1.00 (referent) TC 278 (25.9) 335 (31.5) 0.68 (0.52--0.90) 0.008 CC 30 (3.0) 38 (3.6) 0.76 (0.39--1.47) 0.42 TC+CC 308 (28.9) 373 (35.1) 0.71 (0.59--0.85) 3.04 Â 10À4 Abbreviations: CI, confidence interval; OR, odds ratio. aP (2df), genotype frequencies in cases and controls were compared using a w2 test with 2df. bAdjusted for age, sex, smoking status and family history of cancer. Bold characters indicate corresponding P values are o0.05.

much smaller P value of 3.04 Â 10À4 (adjusted OR ¼ 0.71, 95% CI enhanced the reporter gene activity by 1.9--2.2-fold, compared 0.59--0.85) (Table 2). with control (pGL-T) in human embryonic lung fibroblast cell line (HELF) (with the TT genotype), H1299 (with the CC genotype) and H460 cells (with the TT genotype, Supplementary Figure S3); in Micro RNAs (miRNAs) participate in the REV3L 30UTR regulation addition, loss of both miRNA-binding sites (pGL-DDmiR) signifi- To investigate whether the predicted two miRNA-binding sites cantly elevated levels of the reporter by 2.5--4.3- participate in REV3L expression regulation, luciferase assays were fold, compared with control vector in three cell lines (Figure 1c). performed. Downstream of the luciferase gene was inserted by Among predicted miRNAs, both miR-25 and miR-32 were the the REV3L complete 30UTR (pGL-T), 30UTR with a 11-bp deletion in most common miRNAs in miRNA expression profiles of the lung or the upstream predicted miRNA-binding site (pGL-DmiR) or 30UTR lung cancer.21 To further explore whether these two miRNAs with both deletions for the predicted two miRNA-binding sites participated in REV3L expression regulation, we contransfected (pGL-DDmiR) (Figure 1b). After transfection with pGL-T, pGL-DmiR with miR-25 or miR-32 and the luciferase reporter plasmid pGL-T, or pGL-DDmiR for 24 h, the relative luciferase activities were pGL-DmiR or pGL-DDmiR in HELF cells. Luciferase activity was assayed using Renilla luciferase activity for normalization. We significantly reduced by miR-25 or miR-32 in pGL-T-transfected found that loss of the upstream miRNA-binding site (pGL-DmiR) cells, compared with the control (cotransfected with an unrelated

Oncogene (2013) 242 --250 & 2013 Macmillan Publishers Limited microRNA polymorphism in lung cancer susceptibility S Zhang et al 245 ab RU RU ′ ′ 200 REV3L 3 UTR-T-allele mRNA 200 REV3L 3 UTR-T-allele mRNA

150 150

100 100

50 50

0 0

–50 Sec –50 Sec –200 –80 40 160 280 400 –200 –80 120 280 400 RU RU 200 REV3L 3′ UTR-C-allele mRNA 200 REV3L 3′ UTR-C-allele mRNA

150 150

100 100

50 50

0 0

–50 Sec –50 Sec –200 –80 40 160 280 400 –200 –80 120 280 400

Binding constants for REV3L 3′UTR mRNA-miR25 Binding constants for REV3L 3′UTR mRNA-miR32 × 5 -1 -1 × -4 -1 × -9 × 5 -1 -1 × -4 -1 × -9 Interaction Ka ( 10 M s ) Kd ( 10 s ) KD ( 10 M) Interaction Ka ( 10 M s ) Kd ( 10 s ) KD ( 10 M) 3′UTR-T-allele mRNA 2.39 6.38 2.66 3′UTR-T-allele mRNA 1.17 4.45 3.80 3′UTR-C-allele mRNA 1.81 5.74 3.18 3′UTR-C-allele mRNA 0.92 4.09 4.45

Figure 2. Comparison of the binding affinity of RNA--miRNA by SPR. Biotin-labeled single-stranded RNA harboring 32 bp 30UTR wide-type-T- allele RNA or 30UTR mutant-C-allele RNA was immobilized to a streptavidin-modified sensor chip. Cytoplasmic extracts from the H1299 cells were incubated with miR-25 or miR-32 at different concentrations to compare the relative binding affinities. The SPR measurements were conducted using a 6 Â 6 channels ProteOn XPR36 protein interaction array system (Biorad). (a) Binding affinity of REV3L 30UTR T or C containing RNA to miR-25 was monitored. The SPR-binding signal of the T-allele seuquence interaction with miR25 was higher than that of C-allele with miR-25. (b) Binding affinity of REV3L 30UTR T- or C-allele containing RNA to miR-32 was monitored. The SPR-binding signal of the T-allele seuquence interaction with miR32 was higher than that of C-allele with miR-32.

miRNA, Let-7i), but moderately decreased in pGL-DmiR. Further- whether or not this variation could have an effect on gene more, mixed inhibitors of miR-25/32 could restore the expression expression, we evaluated expression differences between the two of the reporter in pGL-T and pGL-DmiR-transfected cells alleles. Significant differences in expression levels were observed (Figure 3d). These results indicated that the REV3L 30UTR harbored between the major T- and minor C-allele-containing vectors two miR-25/32-binding sites. (Figure 1e). The plasmid containing the minor C allele displayed 35% significantly higher levels of luciferase expression than that of T-allele 30UTR RNA shows a stronger binding affinity to miRNAs the plasmid with the T allele in the HELF and H1299 cells. Forced expression of miR-25 or miR-32 downregulated both pGL-T and As bioinformatics tools predict that the T to C change may alter pGL-C vectors, whereas the C-allele-containing plasmids still the binding motif between REV3L mRNA and miRNAs (Supple- showed a significantly higher reporter activity (Figure 1e). These mentary Figure S2), to determine the affinity between miR-25/32 results indicated that individuals with the protective C allele are and the wild-type or variant RNA in cytoplasmic extracts, we used likely to express an adequate amount of REV3Lp, which may surface plasmon resonance (SPR) to monitor the mRNA::miRNA protect cells against genome instability. interaction signals. Single-stranded RNA oligonucleotides carrying the wild-type (the T allele) or variant (the C allele) sequence were each immobilized to a separate channel. Synthesized miR-25 or miRNAs can downregulate endogenous REV3L expression via miR-32 at different concentrations was subjected in a single run. translation inhibition Kinetic binding constants were determined directly from the SPR Even though we have demonstrated that miR-25/32 could traces using the Biorad ProteOn manager software (Bio-Rad downregulate the reporters carrying REV3L 30UTR, the influence Laboratories, Hercules, CA, USA). Sensorgrams for the two of miR-25/32 on endogenous REV3L was still unclear. As the full- channels that contained immobilized T-allele 30UTR RNA and 0 length REV3L cannot be detected by western blotting, we C-allele 3 UTR RNA revealed that they interacted significantly with determined the REV3L protein by direct immunofluorescence, as miR-25 and miR-32, respectively (Figure 2). We observed a described by Brondello et al.17 Chemically synthesized RNAs (miR- stronger binding of mRNA carrying the T-allele, than that of the 0 À9 À9 25, miR-32 and 32bp REV3L 3 UTR RNAs) were transfected into C-allele, to both miR-25 (2.66 Â 10 M vs 3.18 Â 10 M) and miR- À9 À9 H1299 (with CC genotype) and HELF (with TT genotype) cells, 32 (3.80 Â 10 M vs 4.45 Â 10 M). These results indicated higher respectively. Results of immunofluorescent staining showed that binding affinities for the wild-type T-allele RNA to bind to miR-25 the REV3L protein was mainly localized in the nucleus. Fluorescent and miR-32. signals were reduced in miR-25 or miR-32-transfected cells (Figure 3a and Supplementary Figure S4). However, the mRNA The 30UTR 460 T4C polymorphism enhances reporter gene level of REV3L or luciferase was not decreased after transfection of expression miR-25/32 (Figures 3bi and 3bii). In addition, the strongest As mentioned above, the rs465646 SNP is located within the fluorescent signal was observed in cells transfected with a predicted upstream miRNA-binding site. The T to C change may synthesized short REV3L 30UTR RNA, which could competitively modulate the REV3L mRNA::miRNA imperfect complementarities bind to the endogenous miR-25/32 (Figure 3a and Supplementary (Supplementary Figure S2), and thus may alter gene expression. Figure S4). These results demonstrated that miR-25/32 could The binding affinity between REV3L mRNA and miRNAs was downregulate endogenous REV3L expression by translation attenuated because of the T4C allelic change. To illustrate inhibition.

& 2013 Macmillan Publishers Limited Oncogene (2013) 242 --250 microRNA polymorphism in lung cancer susceptibility S Zhang et al 246

Figure 3. Influence of miR-25/32 on endogenous REV3L expression. (a) REV3L protein was detected by immunofluorescence in H1299 cells without or with synthenic RNA transfection. All the fluorescence pictures were acquired under the same exposure time. Fluorescent signals were reduced in miR-25 or miR-32-transfected cells and rescued by transfection of RNA complementary to the seed sequence of miR-25/32. (bi) H1299 cells were transfected with a control miRNA (Let-7i), miR-25 or miR-32. Twenty-four hours after transfection, relative REV3L mRNA level was measured by (RT)-PCR. (bii) Equivalent pGL-T or pGL-C vector was co-transfected with rellina luciferase in H1299 cells and. At 24 h after transfection, Firefly luciferase and Renilla luciferase mRNA levels were measured by RT-PCR.

REV3L overexpression diminishes foci formation We then evaluated the influence of the T4C SNP on the foci To further determine the role of REV3L in cell transformation, we formation. Synthesized partial REV3L 30UTR RNA (32 bp) carrying performed foci formation assays. As shown in Figure 4a and either T- or C-allele was transfected into HELF and H1299 cells, in Supplementary Figure S5, HELF and H1299 cells transfected with a which they may bind competitively to the endogenous miRNAs in REV3L expression vector formed less foci than cells transfected individual manner. As short half-time (5 days in the cytoplasm) of with the control pcDNA3.1 vector. Furthermore, cell foci were synthesized 20-methoxy-modified RNA oligonucleotides (32 bp), significantly increased in cells transfected with a REV3L-RNAi we also performed a second transfection 5--7 days after the first vector compared with cells transfected with control plasmid. transfection. For once-transfected or twice-transfected experi- Similar results were also obtained by stable cell lines. These data ments, the results are similar. As demonstrated by SPR that the suggested that overexpression of REV3L could attenuate cellular exogenous T-allele RNA competed more efficiently for the binding proliferation and transformation. to the endogenous miRNAs, cells transfected with partial 30UTR

Oncogene (2013) 242 --250 & 2013 Macmillan Publishers Limited microRNA polymorphism in lung cancer susceptibility S Zhang et al 247

Figure 4. Cell focus formation assay. 1000 cells were seeded on each plate and transfected with pcDNA3.1, pcDNA3.1-REV3L, REV3L-RNAi and 30UTR wide-type-T-allele RNA or 30UTR mutant-C-allele RNA. After 10 days, cells were stained by crystal violet. Colonies consisting of 450 cells were counted. (a) Influence of REV3L expression on cell focus formation. Overexpression of REV3L (transfected with pcDNA3.1-REV3L) formed less foci, furthermore, reduce REV3L expression (transfected with REV3L-RNAi) generated more foci compare with control (transfected with pcDNA3.1). (b) Influence of 32bp REV3L 30UTR T- or C-allele-containing RNA on cell focus formation. Cells transfected with 30UTR T-allele-mRNA generated fewer colonies compared with C-allele-mRNA.

was more negatively regulated by miRNAs than the C-allele counterpart, resulting in less REV3L expression.

REV3L mRNA is downregulated in lung cancer tissues To assess the expression of REV3L in normal lung and lung cancer tissues, we quantified the levels of REV3L transcripts in tumor samples from 46 lung cancer patients. We found that cancerous tissues exhibited significantly fewer REV3L transcripts compared with corresponding adjacent normal counterparts (P ¼ 0.017, Figure 5. Relative expression ratios of REV3L mRNA in 46 pairs of lung Figure 5), which led us to conclude that REV3L expression was cancer biopsies. The 46 coupled biopsies were obtained from fresh decreased in lung cancer. This result was consistent with the surgical resection. The expression level of REV3L was measured in report from Pan et al.,18 though inconsistent with our findings in triplicate by real-time PCR analysis. T (tumor)/N (non-tumor) relative gliomas.22 Therefore, dysregulation of REV3L expression may be miRNA expression ratios between normalized values were calculated. tissue-specific. T/N values o1 were transformed to the inverse value N/T. þ and À indicate that the expression is higher or lower, respectively, in the tumor sample than in the respective adjacent control tissue. DISCUSSION We identified a mechanism to explain the association of 30UTR 460 T-allele-RNA generated fewer colonies compared with those with SNP of REV3L with lung cancer susceptibility through post- the C-allele-RNA (o0.5 fold, Figure 4b), indicating a higher transcriptional gene regulation by miRNAs::mRNA interaction. Our endogenous REV3L expression in T-allele-RNA transfected-cells. first association study explored 15 common SNPs and identified Therefore, it is likely that the endogenous REV3L T-allele-mRNA three polymorphisms associated with lung cancer risk. The

& 2013 Macmillan Publishers Limited Oncogene (2013) 242 --250 microRNA polymorphism in lung cancer susceptibility S Zhang et al 248 association between 30UTR 460 SNP and lung cancer risk was growth factor 20 is associated with Parkinson’s disease and validated in another independent case--control study. These disrupts a binding site for microRNA-433.37 A30UTR SNP in results strongly support that genetic variation of the REV3L gene human dihydrofolate reductase (DHFR) affects its expression may modulate the lung cancer risk. Addiditional luciferase assays by interfering with miR-24 function, resulting in DHFR over- experiments further showed that the 30UTR 460 SNP was located expression and methotrexate resistance. However, this SNP is within a miRNAs-binding site. It appeared that the T4C change located out of the miR-24-binding site.38 In our study, the 30UTR caused relaxed mRNA::miRNA binding and consequently led to an 460 T4C SNP at least modulates the interaction between mRNA elevated levels of REV3L expression. and miR-25/32 rather than only one miRNA. As for the other Among the TLS polymerases, only Pol Z was found to be predicted miRNAs, including miR-92, miR-363 and miR-367, their responsible for bypassing UV lesions in DNA, and its absence may roles need to be further investigated. The final REV3L expression result in the variant form of the genetic disorder, xeroderma appears to be comprehensively regulated by these miRNAs, but pigmentosum variants.7,23 Other TLS polymerases have specificities the competitive relationship among these miRNAs remains elusive. for different types of DNA damage, but their precise roles inside the Having extended our previous association study, we success- cell have not yet been established. The REV3L gene, encoding the fully identified the risk-contributing polymorphism in the REV3L catalytic subunit of human polymerase z, has a significant role in gene and portrayed a possible underlying mechanism in mutagenicity. Comparatively, Pol z is much more efficient at carcinogenesis. Larger studies involving ethnically diverse popula- extending from base mispairs than it is at creating the mispairs.5 tions are warranted to confirm our findings. However, there has been no association study or pedigree study on this gene and its association with risk of cancers. Genome instability is a hallmark of all tumors, including lung MATERIALS AND METHODS cancer, and genomic instability is characterized by an increased Study population and sampling frequency of chromosomal aberrations and gene mutations in The first association study (study 1) included 500 lung cancer patients and cells, which result from an elevated rate of DNA damage or an 19 attenuated capability of DNA repair.24 Mammalian cells with the 517 cancer-free controls. As previously described, all subjects were homologous REV3L deletion did not proliferate in vitro, perhaps genetically unrelated ethnic Han Chinese. The case patients were recruited because of unrepaired aberrations activating the checkpoint between July 2002 and December 2004 in Nanjing, Jiangsu province, control.12 hREV3L gene encodes a 353-kDa protein, which is twice China. Cancer-free controls were enrolled from a pool of 10 500 individuals the size of yeast REV3 (173 kDa).25,26 Much of this difference is from communities in Jiangsu province. Study 2 included an additional 572 because of the exon 13, which encodes 1388 amino-acid lung cancer patients and 547 controls. Eligible cases with incident lung residues.9 This uncharacterized region is believed to render this cancer were consecutively recruited between October 2003 and July 2006 large molecule with new functions. Our results in colony formation from four hospitals in Shanghai whereas the controls were from Yangpu indicated that REV3L could attenuate the proliferation of cells, and Jingan District, Shanghai, China. The detailed population information and population features of study 2 are summarized in our previously which reduced the possibility of cancer development. In addition, 20 results of the real-time PCR also disclosed a significant reduction published paper. In both studies, the controls were frequency-matched of REV3L expression in 46 lung carcinomas, compared with that of to the cases by age (±5 years), sex and residential area (urban or the normal adjacent tissues. The protective role of REV3L conflicts countryside). After a written informed consent, each participant was with the traditional concept that the low-fidelity enzyme scheduled for an interview, where a structured questionnaire was accelerates cancer development by causing mutations.10,27 Recent administered by interviewers to collect information on demographic data findings also support the tumor suppressor role of REV3L,17 and environmental exposure history, including tobacco smoke. Lung cancer consistent with our results. tissues and corresponding adjacent normal lung tissues were obtained from Regulation of the REV3L expression has been focused on its 46 patients receiving surgery at Changhai Hospital, Shanghai, China. All transcription. REV3L (mice REV3) is inducible by N-methyl-N’-nitro- patients who provided samples gave a written informed consent for their Nnitrosoguanidine, adriamycin and pentylenetetrazole.16,28,29 tissues to be used for scientific research. These tissues were immediately Little is known about the posttranscriptional regulation of REV3L. frozen and stored at À80 1C for the real-time PCR analysis. MicroRNAs are B22 nucleotide small non-coding RNAs that All of the above studies were approved by the institutional review board regulate gene expression by targeting their 30UTRs.30 miRNAs of Fudan University and Nanjing Medical University. that only partially complement their corresponding mRNAs are important in regulating gene expression. The fate of the Selection of REV3L SNPs targeted mRNA is determined by the extent of complementary 31 The human REV3L gene encompasses 184.7 kb, located on chromosome pairing between the miRNA and its targeted mRNA. A high 6q21. As this study was initiated in January 2005, when the SNP density of degree of complementary binding increases the chance for 32 the phase I HapMap SNP database was not adequate, we chose SNPs from cleavage that leads to digestion of target mRNAs. Our data in both the HapMap and dbSNP databases. The genetic model of REV3L 0 SPR showed a higher affinity for the T allele of 3 UTR 460 SNP to reveals 33 exons and 32 introns with 902 SNPs, as listed in the dbSNP Build bind to miRNAs, resulting in reduced expression, which is in line 121. To capture the majority of common variants across this gene, an 0 with this concept. Therefore, the 3 UTR 460T4C change modulates algorithm to score the SNPs was developed, and a set of SNPs were the binding pattern and consequently leads to the altered selected on the basis of their final scores. The detailed procedure has been expression. We confirmed our prediction that miR-25 and miR-32 described elsewhere.39 In brief, for a SNP, the following criteria were 0 interacted with REV3L mRNA. 3 UTR polymorphisms at miRNA considered: (i) interdistance between two adjacent SNPs; (ii) heterozygos- target sites can modulate gene expression through alteration of ity; (iii) functional relevance and (iv) compatibility with the genotyping miRNA target-binding capability, ultimately leading to differences platform. As a result, 15 SNPs (rs181294, rs465646, rs457497, rs240998, in the susceptibility to complex genetic diseases, such as cancer. rs240969, rs240966, rs459809, rs464401, rs240993, rs462779, rs1002481, However, because the prevalence and importance of miRNAs rs6922226, rs6937734, rs6935759 and rs6913769) were selected for have been recognized in recent years, only a few studies genotyping in this study. have described the polymorphisms associated with binding miRNAs.33 --35 Clop et al.36 reported a mutation in the 30UTR of the myostatin gene that is responsible for the characteristic Genotyping muscular hypertrophy of Texel sheep owing to the creation of a de The genomic DNA was extracted from venous blood samples taken from novo miRNA-binding site A variation in the 30UTR of fibroblast every participant, using the kits of guanidine hydrochloride bought from

Oncogene (2013) 242 --250 & 2013 Macmillan Publishers Limited microRNA polymorphism in lung cancer susceptibility S Zhang et al 249 the company of QIANGEN (Hilden, Germany). The 15 SNPs were genotyped quantitative reverse transcriptase PCR in triplicates on a Prism 7900 Real- by using the Illumina Genotyping Facility, which is a highly efficient Time PCR machine (Applied Biosystems, Foster City, CA, USA) with primers genotyping assay that combines Illumina Golden Gate assay, Sentrix array as follows: for glyceraldehyde 3-phosphate dehydrogenase,50-GAAGGTGAAG matrices and Sherlock scanner technology (Illumina Corp, Foster City, CA, GTCGGAGTC-30 (sense), 50-GAAGATGGTGATGGGATTTC-30 (antisense); for USA) at the Chinese National Center in Shanghai, China. REV3L,50-CGCGTCAGTTGGGACTTAAG-30 (sense), 50-ACTATCGCCAACCTCA More detailed description of each of the steps performed by the Illumina ATGC-30 (antisense). Genotyping Facility is available in our previously published paper.19 In study 2, the samples were genotyped using SNPScan technology (Generay, Shanghai, China). Of all assays, 5% of samples were randomly selected and SPR experiments tested by sequencing, and the reproducibility was 100%. The SPR measurements were conducted using a 6 Â 6 channels ProteOn XPR36 protein interaction array system (Biorad, Hercules, CA, USA). Biotin- 0 Construction of promoter reporter plasmids labeled single-stranded RNA harboring 32 bp 3 UTR wild-type-T-allele RNA (50-CUUUUCAAACGAGTAAAAUAAUGUGCAAUGAU-30)or30UTR mutant-C- For the effect of the 30UTR 460 T/C SNP on posttranscriptional regulation of allele RNA (50-CUUUUCAAACGAGCAAAAUAAUGUGCAAUGAU-30) (400 mM) REV3L, a fragment containing the REV3L 30UTR ( 1to 837, relative to þ þ was immobilized to a streptavidin-modified sensor chip. Cytoplasmic translational stop codon) carrying the major T allele at the position extracts from the H1299 cells were incubated with miR-25 or miR-32 at þ 460 bp (pGL-T) was amplified by PCR, using the following primers: different concentrations to compare the relative binding affinities. The flow 0 0 5 -GCTCTAGAGGGGTACCCTCTTCCGCCAGCCAGATCCG-3 (forward) and rate was 50 ml/min, and the binding between RNAs was monitored for 50-GCTCTAGACCCAAGCTTCGCGTGAGCCCACCTGGAG-30 (reverse). The mutant 0 400 s. The SPR experiment was conducted at room temperature, and the (pGL-C) was made using the following primers: 5 -CTTTTCAAACGAG data was evaluated using the Biorad ProteOn manager software. For 0 0 0 CAAAATAATGTGC-3 (forward) and 5 -GCACATTATTTTGCTCGTTTGAAAAG -3 regeneration of the immobilized RNA, 0.1% SDS was applied for 2--3 min (reverse). In addition, the pGL-DmiR construct, which lost an 11-bp putative after each cycle of protein binding. miRNA seed sequence was generated using the following primers: 50-TTCAAACGAGTAAAAGATTTTTATACAAATG-30 (forward) and 50-CAT TTGTATAAAAATCTTTTACTCGTTTGAA-30 (reverse). The pGL-DDmiR Immunostaining construct, which lost another 10 bp putative miRNA seed sequence H1299 and HELF cells were fixed with 4% paraformaldehyde, washed with 0 was generated using the following primers: 5 -TGTTACAAACCTGTGGG phosphate-buffered saline and permeabilized with 1% Triton X-100 in 0 0 CACTTTAAAAAT-3 (forward) and 5 -GAATAAAATTTTTTATTTTTAAA phosphate-buffered saline. Cells were blocked with blocking buffer 0 GTGCCCACAGGT-3 (reverse). Each amplified fragment was digested (phosphate-buffered saline, 1% Triton X-100 and 5% bovine serum by XbaI and inserted downstream of the firefly luciferase gene in a albumin) and incubated at 4 1C with the REV3L antibody (1:1000; Abnova, pGL3-promoter vector (Promega, Madison, WI, USA). These recombi- Taiwan, China) overnight. FITC-conjugated goat anti-mouse (1:100) was nation plasmids were sequenced to confirm their orientation. incubated for 30 min at room temperature. Nuclear counterstaining was performed using 4,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich, Cell culture and transfection St Louis, MO, USA). Cells without the treatment of the primary antibody HELF human embryonic lung fibroblasts were cultured in Dulbecco’s served as negative control. modified Eagle medium supplemented with 10% fetal bovine serum (Gibco, Grand Island, NY, USA). H460 human lung adenocarcinoma cells Statistical analysis and H1299 human non-small cell lung cancer cells were cultured in RPMI 1640 with 10% fetal bovine serum. For H1299 cells, medium was The Hardy--Weinberg equilibrium test was performed for each genotyped additionally supplemented with 10 mM HEPES ((4-(2-hydroxyethyl)-1- SNP among controls. Differences in demographic characteristics, selected piperazineethanesulfonic acid) and 1 mM sodium pyruvate. Cells were variables and frequencies of the genotypes of REV3L variants between the cases and the controls were evaluated using the Pearson’s w2 testing. The grown in a 37 1C incubator with 5% CO2. For transfection, cells were grown on 24-well culture plates with associations between REV3L variants and lung cancer risk were estimated 70--80% confluence. Cells were transfected by Lipofectamine 2000 by computing the ORs and their 95% CIs from both univariate and transfection reagent (Invitrogen, Carlsbad, CA, USA) with 1 mg of the multivariate logistic regression analyses with adjustment for age, sex, pGL-T, pGL-C, pGL-DmiR or pGL-DDmiR vector containing REV3L 30UTR smoke status and family history of cancer. ± fragment coupled to the firefly luciferase reporter gene, respectively. Other data were expressed as the mean s.e. from at least three Chemically synthesized RNAs, including miR-25, miR-32, miR-25/32 independent experiments, unless indicated otherwise. s.e. bars were inhibitors, 30UTR-T-allele RNA and 30UTR-C-allele RNA were obtained included for all data points. The relative mRNA level between lung cancer from GenePharma (GenePharma, Shanghai, China). For co-transfection, tissues and normal tumor-adjacent tissues was calculated by Student’s cells were transfected with 0.5 mg of the pGL-T, pGL-C, pGL-DmiR or two-tailed, paired t-test. Other statistical differences between test and pGL-DDmiR vector and 0.5 mg of the chemically synthesized RNA. In control values were analyzed by means of Student’s two-tailed, unpaired each transfection, 50 ng of pRL-TK (Promega) was used to correct for t-test. All the statistical analyses were performed with the SPSS 15.0 transfection efficiency. After incubation for 24 h, cells were broken using software (SPSS, Chicago, IL, USA). Data were considered significant and passive lysis buffer. were indicated by ‘*’ if Po0.05 vs control and ‘**’ if Po0.01.

Luciferase assays CONFLICT OF INTEREST Luciferase activity was measured with the Dual-Luciferase Reporter Assay System (Promega). Promoter activities were expressed as the ratio The authors declare no conflict of interest. between Firefly luciferase and Renilla luciferase activities.

Quantitative analysis of REV3L mRNA and miRNAs ACKNOWLEDGEMENTS This work was partially supported by the Shanghai Science and Technology Research RNA was extracted and converted to complementary DNA using an Program (Grants 09JC1402200 and 10410709100), the Natural Science Foundation of oligo(dT)12 primer and Superscript II (Invitrogen). The SYBR green China (Grants 30800622, 81001114 and 81020108028), the Scientific Research dye (Takara Bio Inc., Shiga, Japan) was used for the amplification Foundation for the Returned Overseas Chinese Scholars (State Education Ministry), of complementary DNA. REV3L and internal standard glyceraldehyde the Doctoral Fund of the Ministry of Education of China and the Shanghai Key 3-phosphate dehydrogenase (GAPDH) RNA were measured by the real-time Subject Project for Public Health (Grant 08GWZX0301).

& 2013 Macmillan Publishers Limited Oncogene (2013) 242 --250 microRNA polymorphism in lung cancer susceptibility S Zhang et al 250 REFERENCES 20 Lu J, Zhang S, Chen D, Wang H, Wu W, Wang X et al. Functional characterization of 1 Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. a promoter polymorphism in APE1/Ref-1 that contributes to reduced lung cancer CA Cancer J Clin 2011; 61: 69 --90. susceptibility. FASEB J 2009; 23: 3459 --3469. 2 Hoeijmakers JH. Genome maintenance mechanisms are critical for preventing 21 Wang QZ, Xu W, Habib N, Xu R. Potential uses of microRNA in lung cancer cancer as well as other aging-associated diseases. Mech Ageing Dev 2007; 128: diagnosis, prognosis, and therapy. Curr Cancer Drug Targets 2009; 9: 572 --594. 460 --462. 22 Wang HB, Zhang SY, Wang S, Lv J, Wu WT, Weng L et al. REV3L confers 3 Sonoda E, Okada T, Zhao GY, Tateishi S, Araki K, Yamaizumi M et al. Multiple roles chemoresistance to cisplatin in human gliomas: The potential of its RNAi for of Rev3, the catalytic subunit of polzeta in maintaining genome stability in synergistic therapy. Neuro Oncol 2009; 11: 790 --802. vertebrates. Embo J 2003; 22: 3188 --3197. 23 Lehmann AR. Replication of damaged DNA by translesion synthesis in human 4 Friedberg EC, Lehmann AR, Fuchs RP. Trading places: how do DNA polymerases cells. FEBS Lett 2005; 579: 873 --876. switch during translesion DNA synthesis? Mol Cell 2005; 18: 499 --505. 24 Rockwell S, Yuan J, Peretz S, Glazer PM. Genomic instability in cancer. Novartis 5 Venkatesan RN, Bielas JH, Loeb LA. Generation of mutator mutants during Found Symp 2001; 240: 133 --142; discussion 142-151. carcinogenesis. DNA Repai (Amst) 2006; 5: 294 --302. 25 Gibbs PE, McGregor WG, Maher VM, Nisson P, Lawrence CW. A human homolog of 6 Matsuda T, Bebenek K, Masutani C, Hanaoka F, Kunkel TA. Low fidelity DNA the Saccharomyces cerevisiae REV3 gene, which encodes the catalytic subunit of synthesis by human DNA polymerase-eta. Nature 2000; 404: 1011 --1013. DNA polymerase zeta. Proc Natl Acad Sci USA 1998; 95: 6876 --6880. 7 Johnson RE, Washington MT, Haracska L, Prakash S, Prakash L. Eukaryotic 26 Lin W, Wu X, Wang Z. A full-length cDNA of hREV3 is predicted to encode DNA polymerases iota and zeta act sequentially to bypass DNA lesions. Nature 2000; polymerase zeta for damage-induced mutagenesis in humans. Mutat Res 1999; 406: 1015 --1019. 433: 89 --98. 8 Zhang Y, Yuan F, Wu X, Wang Z. Preferential incorporation of G opposite 27 Datta A, Jinks-Robertson S. Association of increased spontaneous mutation rates template T by the low-fidelity human DNA polymerase iota. Mol Cell Biol 2000; 20: with high levels of transcription in yeast. Science 1995; 268: 1616 --1619. 7099 --7108. 28 Kajiwara K, Nagawawa H, Shimizu-Nishikawa S, Ookuri T, Kimura M, Sugaya E. 9 Gan GN, Wittschieben JP, Wittschieben BØ, Wood RD. DNA polymerase zeta (pol Molecular characterization of seizure-related isolated by differential zeta) in higher eukaryotes. Cell Res 2008; 18: 174 --183. screening. Biochem Biophys Res Commun 1996; 219: 795 --799. 10 Lawrence CW, Maher VM. Mutagenesis in eukaryotes dependent on DNA 29 Krieg AJ, Hammond EM, Giaccia AJ. Functional analysis of p53 binding under polymerase zeta and Rev1p. Philos Trans R Soc Lond B Biol Sci 2001; 356:41--46. differential stresses. Mol Cell Biol 2006; 26: 7030 --7045. 11 Lemontt JF. Induction of forward mutations in mutationally defective yeast. Mol 30 Kim VN, Nam JW. Genomics of microRNA. Trends Genet 2006; 22: 165 --173. Gen Genet 1972; 119:27--42. 31 Doench JG, Sharp PA. Specificity of microRNA target selection in translational 12 Wittschieben JP, Reshmi SC, Gollin SM, Wood RD. Loss of DNA polymerase zeta repression. Genes Dev 2004; 18: 504 --511. causes chromosomal instability in mammalian cells. Cancer Res 2006; 66:134--142. 32 Vatolin S, Navaratne K, Weil RJ. A novel method to detect functional microRNA 13 Nojima K, Hochegger H, Saberi A, Fukushima T, Kikuchi K, Yoshimura M et al. targets. J Mol Biol 2006; 358: 983 --996. Multiple repair pathways mediate tolerance to chemotherapeutic cross-linking 33 Saunders MA, Liang H, Li WH. Human polymorphism at microRNAs and microRNA agents in vertebrate cells. Cancer Res 2005; 65: 11704 --11711. target sites. Proc Natl Acad Sci USA 2007; 104: 3300 --3305. 14 Bemark M, Khamlichi AA, Davies SL, Neuberger MS. Disruption of mouse 34 Yu Z, Li Z, Jolicoeur N, Zhang L, Fortin Y, Wang E et al. Aberrant allele frequencies polymerase zeta (Rev3) leads to embryonic lethality and impairs blastocyst of the SNPs located in microRNA target sites are potentially associated with development in vitro. Curr Biol 2000; 10: 1213 --1216. human cancers. Nucleic Acids Res 2007; 35: 4535 --4541. 15 Esposito G, Godindagger I, Klein U, Yaspo ML, Cumano A, Rajewsky K. Disruption 35 Nicoloso MS, Sun H, Spizzo R, Kim H, Wickramasinghe P, Shimizu M et al. Single- of the Rev3l-encoded catalytic subunit of polymerase zeta in mice results in early nucleotide polymorphisms inside microRNA target sites influence tumor embryonic lethality. Curr Biol 2000; 10: 1221 --1224. susceptibility. Cancer Res 2010; 70: 2789 --2798. 16 Zhu F, Jin CX, Song T, Yang J, Guo L, Yu YN. Response of human REV3 gene to 36 Clop A, Marcq F, Takeda H, Pirottin D, Tordoir X, Bibe B et al. A mutation creating a gastric cancer inducing carcinogen N-methyl-N0-nitro-N-nitrosoguanidine and its potential illegitimate microRNA target site in the myostatin gene affects role in mutagenesis. World J Gastroenterol 2003; 9: 888 --893. muscularity in sheep. Nat Genet 2006; 38: 813 --818. 17 Brondello JM, Pillaire MJ, Rodriguez C, Gourraud PA, Selves J, Cazaux C et al. Novel 37 Mishra PJ, Humeniuk R, Mishra PJ, Longo-Sorbello GS, Banerjee D, Bertino JR. A evidences for a tumor suppressor role of Rev3, the catalytic subunit of Pol zeta. miR-24 microRNA binding-site polymorphism in dihydrofolate reductase gene Oncogene 2008; 27: 6093 --6101. leads to methotrexate resistance. Proc Natl Acad Sci USA 2007; 104: 13513 --13518. 18 Pan Q, Fang Y, Xu Y, Zhang K, Hu X. Down-regulation of DNA polymerases kappa, 38 Wang G, van der Walt JM, Mayhew G, Li YJ, Zuchner S, Scott WK et al. Variation in eta, iota, and zeta in human lung, stomach, and colorectal cancers. Cancer Lett the miRNA-433 binding site of FGF20 confers risk for Parkinson disease by 2005; 217: 139 --147. overexpression of alpha-synuclein. Am J Hum Genet 2008; 82: 283 --289. 19 Hu Z, Wang H, Shao M, Jin G, Sun W, Wang Y et al. Genetic variants in MGMT and 39 Chen D, Jin G, Wang Y, Wang H, Liu H, Liu Y et al. Genetic variants in peroxisome risk of lung cancer in Southeastern Chinese: a haplotype-based analysis. Hum proliferator-activated receptor-gamma gene are associated with risk of lung Mutat 2007; 28: 431 --440. cancer in a Chinese population. Carcinogenesis 2008; 29: 342 --350.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2013) 242 --250 & 2013 Macmillan Publishers Limited