SEARCH

SEARCH BY CITATION

Keywords:

  • cyclin A2;
  • single nucleotide polymorphism;
  • cancer;
  • susceptibility;
  • tissue specificity

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

BACKGROUND:

The objective of this was to identify functional single nucleotide polymorphisms (SNPs) in cyclin-dependent kinases (CDKs) and cyclins that are associated with risk of human cancer.

METHODS:

First, 45 SNPs in CDKs and cyclins were analyzed in 106 lung cancers and 108 controls for a pilot study. One SNP (reference SNP [rs] 769236, +1 guanine to adenine [G[RIGHTWARDS ARROW]A]) at the promoter region of cyclin A2 (CCNA2) also was analyzed in 1989 cancers (300 breast cancers, 450 colorectal cancers, 450 gastric cancers, 367 hepatocellular carcinomas, and 422 lung cancers) and in 1096 controls. Genotyping was performed using matrix-assisted laser desorption-ionization/time-of-flight mass spectrometry. Transcriptional activity of the SNP according to the cell cycle was analyzed by using a luciferase reporter assay and fluorescence-activated cell sorting analysis in NIH3T3 cells.

RESULTS:

In the pilot study, the SNP (rs769236) was associated significantly with the risk of lung cancer. In the expanded study, multivariate logistic regression indicated that the AA homozygous variant of the SNP was associated significantly with the development of lung cancer (P < .0001; codominant model), colorectal cancer (P < .0001), and hepatocellular carcinoma (P = .02) but not with breast cancer or gastric cancer. The luciferase activity of a 300-base pair construct that contained the A allele was 1.5-fold greater than the activity of a construct with the G allele in NIH3T3 cells. The high luciferase activity of constructs that contained the A allele did not change with cell cycle progression.

CONCLUSIONS:

The current results suggested that an SNP (rs769236) at the promoter of CCNA2 may be associated significantly with increased risk of colon, liver, and lung cancers. Cancer 2011;. © 2011 American Cancer Society.

Cell cycle progression through different stages of cell division is regulated by sequential activation and inactivation of the cyclin-dependent kinases (CDKs). Cyclins, as regulatory subunits of CDKs, modulate temporal transitions between various phases of the cell cycle by controlling the activation of CDKs. Single nucleotide polymorphisms (SNPs) are the most common types of genetic variation in humans and are related to the risk of a variety of cancers.1, 2 Knowledge of SNP patterns in the human genome has contributed to our definition of the genetic basis of complex diseases like cancer.3

Two members of the mammalian A-type cyclin family have been identified in humans: embryonic-specific cyclin A1 and somatic cyclin A2. Cyclin A1 is expressed during meiosis in very early embryos, whereas cyclin A2 is present in proliferating somatic cells and plays a critical role in cell cycle progression. Cyclin A2 is produced at the onset of DNA synthesis in proliferating somatic cells and promotes passage through G1 to S phase by formation of a complex with CDK2 kinase.4 Cyclin A2 also plays a role in the G2/M phase transition of the cell cycle by activating CDK1 kinase and is degraded by the ubiquitin-dependent proteolysis pathway in the early phases of mitosis.5

Cyclin A2 is expressed ubiquitously in various tissues, and its overexpression is associated with carcinogenesis of different types of cancers. Cyclin A2 expression at the surface of Barrett esophagus is associated with progression of esophageal dysplasia to adenocarcinoma.6 The bottom-up type of tumorigenesis with ulcerative colitis-associated dysplasia is related to altered expression of cyclin A2.7 It is known that cyclin A2 expression gradually increases from normal tissue, through hyperplasia, to adenoma, and to early colorectal carcinoma.8 The amount of cyclin A2 protein increases significantly in the advanced stage of esophageal carcinoma9 and is high in a poorly differentiated type of cancer with strong invasive ability.10 Although gene amplification has been reported in breast cancers with elevated expression of cyclin A2,11 and although it is known that expression of cyclin A2 is induced by many factors (ie, c-Myc,12 p53,13 p21waf1/cip1,13, 14 p57kip2,14 p107,14-16 ras,16, 17 and raf16), the mechanisms underlying overexpression of cyclin A2 in most human cancers remain to be elucidated.

Many SNPs at the promoter or exon of cyclins have been reported, but their relation with the risk of human cancer has not been studied extensively. An SNP in cyclin D1 (CCND1), a guanine-to adenine change at codon 870 at the exon 4 splice site (G870A), has been associated with increased risk for several solid tumors, including cancers of the colon, lung, and stomach.18-21 To validate the SNP (reference SNP rs769236; +1 G[RIGHTWARDS ARROW]A) that was discovered in our pilot study, we expanded the study and analyzed 10 SNPs at the promoter region of the cyclin A2 gene (CCNA2) in 5 different types of human cancers. Our functional study of the SNP also demonstrated that a difference in the expression of CCNA2 depended on the genotypes of the SNP in vitro.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Study Population

The study population consisted of 1096 controls and 422 patients with lung cancer, 300 patients with breast cancer, 450 patients with colorectal cancer, 367 patients with hepatocellular carcinoma, 450 patients with gastric cancer, and 596 individuals from the general population for calculation of allele frequencies of specific SNPs and for the construction of haplotypes. All patients with cancer underwent curative surgical resection at the Samsung Medical Center, a major referral center for cancer in Seoul, Korea, during the period from February 1999 to November 2007. Age-matched and sex-matched control participants for 5 cancers were selected from among healthy individuals who visited the Samsung Medical Center for regular health checkups. Peripheral blood samples from all participants were collected according to ethical guidelines on the use of human tissues and biologic samples in research after obtaining appropriate institutional review board permission and written informed consent from all participants. Information regarding demographics was obtained by using an interviewer-administered questionnaire.

Matrix-Assisted Laser Desorption-Ionization/Time-of-Flight Mass Spectrometry

Genomic DNA from peripheral blood lymphocytes was extracted using a QIAamp Blood Kit (Qiagen, Valencia, Calif). The DNA quality was checked using an ultraviolet (UV) spectrophotometer (Pharmacia Biotech, Cambridge, United Kingdom) and the PicoGreenTM double-stranded DNA quantitation kit (Molecular Probes, Eugene, Ore) on a SpectraMax Gemini UV spectrometer (Molecular Devices, Sunnyvale, Calif). Genotyping of all SNPs in this study was performed by using the homogenous MassExtend assay (Sequenom, San Diego, Calif), as described previously.22 Both the amplification primer and the extension primer were designed using the Sequenom Assay Design program. The polymerase chain reaction (PCR) and primer extension reactions were performed according to the protocol for hME procedure (MassARRAY System Training User's Guide; Sequenom).

Cell Culture and Transfection

Murine fibroblast NIH3T3 cells (CRL-1658; obtained from the American Type Culture Collection, Manassas, Va) were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 100 U of penicillin/mL, and 100 μg of streptomycin/mL at 37°C in a 5% CO2 atmosphere. NIH3T3 cells were plated at 1 × 105 cells per well in 12-well plates in DMEM supplemented with 10% FBS and were transfected 24 hours later with 0.7 μg of pGL3-Basic (Promega, San Luis Obispo, Calif), pGL3-0.3AA (p0.3A) or pGL3-0.3GG (p0.3G) and 7 ng of pRL-luciferase internal control plasmid. For transfection, we used lipofectamine reagent (Invitrogen, Carlsbad, Calif) and followed the manufacturer's protocol. Cells were harvested 24 hours after transfection and were lysed with 250 μL of passive lysis buffer (Promega) followed by 3 cycles of freezing and thawing, and the whole cell lysates were used for luciferase activity assays.

Plasmid Constructs and Luciferase Reporter Assay

To determine the functional consequences of the SNP (rs769236; +1 G[RIGHTWARDS ARROW]A) at the promoter of CCNA2, a reporter assay was performed in constructs that contained the SNP. Genomic DNA was prepared from each genotype of normal lymphocyte DNA and was used as a template for PCR to amplify fragments that contained the CCNA2 promoter region using the forward primer 5′-CCCAAGCTTCATCCCTTTACCCGTCTCGT and the reverse primer 5′-GGGGTACCCTTCTGAAAGGAACATAATTAT. The 1.3-kb PCR product was cloned into the SmaI site of pGL3-Basic vector (Promega). The 1.3-kb fragment was divided further into the 300-base-pair (bp) fragment, blunted with Klenow, and cloned directionally into the SmaI site of pGL3-Basic vector (Promega). Each CCNA2 promoter construct that contained enhancers and the core promoter elements of the CCNA2 promoter region was 300 bp in size and was named either p0.3G or p0.3A, depending on its sequence. The fragments were subcloned in the plasmid pGL3-Basic vector (Promega) that lacked promoter and enhancer and contained firefly luciferase (Luc) as a reporter. Then, the fragments were transfected transiently into NIH3T3 cells along with Renilla luciferase as an internal control. Each construct was sequenced to verify the sequence of the variant.

Fluorescence-Activated Cell Sorting

Cells grown on 6-well plates were collected by trypsinization and centrifuged at 1200 revolutions per minute for 3 minutes. The cells were resuspended, washed twice in phosphate-buffered saline (PBS), and finally fixed in 70% ice-cold (−20°C) ethanol overnight in suspension. After fixing, the cells were spun down and washed in PBS that contained 0.5% bovine serum albumin (BSA). Then, the cells were incubated with a propidium iodide (PI) solution that contained 50 μg/mL PI, 0.1% sodium citrate, 0.3% NP-40, 50 μg/mL RNase A, and 1 times PBS for 30 minutes at room temperature. DNA profiles of PI-stained cells were analyzed using a FACSCalibur system (Becton Dickinson, Franklin Lakes, NJ) and CELLQuest software (version 3.3; Becton Dickinson).

Statistical Analysis

The chi-square test was used to assess deviation from Hardy-Weinberg equilibrium (HWE) for the distribution of genotypes. Linkage disequilibrium (LD) between all pairs of biallelic loci was measured by calculating Lewontin D′ (|D′|) and r2 values. Haplotypes and their frequencies were constructed using an expectation-maximization (EM) algorithm. The association between genotypes and the prevalence of each cancer was analyzed using the Pearson chi-square test and the Cochran-Armitage test for trend. Multivariate logistic regression was conducted to determine the relation between genotype and cancer risk after adjusting for confounding factors, such as age and sex. The additive model, in which associations depend additively on the minor allele, and the codominant model, in which every genotype gives a different and nonadditive risk, were tested in the analysis. All statistical analyses were 2-sided with a 5% type I error rate.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Association Between 45 SNPs and Lung Cancer

To understand the effects of genetic variants on altered expression of CDKs and cyclins in human cancers, we selected 45 SNPs around the promoter regions and exons of cyclins and CDKs from dbSNP (http://www.ncbi.nlm.nih.gov/SNP accessed July 10, 2009) based on SNP information, including information regarding the flanking sequence, validation status, heterozygosity, and genomic position of the SNPs. First, we analyzed the association of 45 SNPs in cyclins and CDKs between 106 lung cancers and 108 controls (Table 1) and observed that an SNP (rs769236; +1 G[RIGHTWARDS ARROW]A) at the promoter region of the CCNA2 gene was associated significantly with the risk of lung cancer. To further analyze haplotypes, to increase statistical power, and to determine whether the SNP (rs769236; +1 G[RIGHTWARDS ARROW]A) also was involved in other cancers, we selected another 7 SNPs at the promoter region of the CCNA2 and compared the prevalence of genotypes at a total of 10 SNPs between a group of 1989 patients who had 5 different types of cancers and a control group of 1096 individuals. Five-hundred ninety-six individuals from the general population also were recruited for calculation of allele frequencies of 10 SNPs and for the construction of haplotypes.

Table 1. Comparison of Minor Allele Frequencies of 45 Single Nucleotide Polymorphisms in Cyclins and Cyclin-Dependant Kinases Between 106 Lung Cancers and 108 Controls
    Minor Allele Frequency
Gene NameSNP Database Reference No.GenotypeLocusLung CancerControl
  1. SNP indicates single nucleotide polymorphism;, CCN, cyclin; rs, reference SNP; C, cytosine; G, guanine; UTR, untranslated region; A, adenine; T, thymine; CDK, cyclin-dependent kinase.

CCNA1rs3814805C>G5′UTR0.060.08
CCNA1rs3814803G>A5′UTR0.060.08
CCNA1rs3829944C>A5′UTR0.060.08
CCNA2rs3217756C>T5′UTR0.060.06
CCNA2rs769235T>G5′UTR0.010.00
CCNA2rs769236C>T5′UTR0.590.44
CCND1rs1944129A>G5′UTR0.160.15
CCND1rs3212859C>A5′UTR0.000.00
CCND2rs1049606T>CExon0.310.36
CCND2rs3217925C>TExon0.330.33
CCND2rs3217926A>GExon0.090.06
CCND2rs3217933T>CExon0.080.13
CCNE1rs3218045A>GExon0.000.00
CDK10rs466001C>T5′UTR0.130.13
CDK10rs258324G>T5′UTR0.250.32
CDK10rs2075880C>T5′UTR0.430.50
CDK2rs2069413G>CExon0.010.02
CDK2rs2069415C>TExon0.170.15
CDK2rs1045435C>GExon0.010.03
CDK4rs238532C>TExon0.000.01
CDK4rs2856298C>GExon0.000.00
CDK4rs2650063C>G5′UTR0.000.00
CDK4rs2072052G>T5′UTR0.280.32
CDK6rs3731267C>T5′UTR0.080.06
CDK6rs3731268G>A5′UTR0.060.05
CDK6rs992519A>G5′UTR0.080.06
CDK6rs2158134T>C5′UTR0.010.01
CDK6rs2374589A>G5′UTR0.070.06
CDK6rs3731283G>A5′UTR0.140.14
CDK6rs3731288T>G5′UTR0.140.14
CDK6rs3731295G>A5′UTR0.030.02
CDK6rs2237576T>C5′UTR0.010.02
CDK6rs2282995T>C5′UTR0.140.14
CDK6rs445C>T5′UTR0.310.33
CDK7rs2972357T>C5′UTR0.460.44
CDK7rs2932779C>T5′UTR0.170.20
CDK8rs4770964T>GIntron0.380.44
CDK8rs623251T>CIntron0.080.04
CDK8rs676150A>TIntron0.470.48
CDK8rs2117929T>GIntron0.040.03
CDK8rs2769317A>GIntron0.080.04
CDK8rs559665A>CIntron0.070.04
CDK8rs478752A>CIntron0.470.48
CDK8rs1175723T>CIntron0.380.44
CDK9rs8083T>CExon0.430.39

Demographic Distribution

The mean age and sex of the 1989 patients with cancer and the 1096 controls are listed in Table 2. In the control group, men accounted for 4 times the number of women and twice the number of women compared with the group of cancer patients, except for those with breast cancer. The overall mean age of cancer patients was 59 years, which was similar to the mean age in the control group. The mean age (±standard deviation) of patients with breast cancer was 51 ± 9 years. The patients with colorectal cancer included 243 men (54%) and 207 women (46%), and these patients ranged in age from 28 years to 83 years. The 450 patients with gastric cancer included 274 men (61%) and 176 women (39%). The 367 patients with hepatocellular carcinoma included 294 men (80%) and 73 women (20%) who ranged in age from 29 years to 76 years. The mean age of the patients with lung cancer was 63 ± 10 years, and 308 of those patients (73%) were men.

Table 2. Demographic Distribution (N=3085)
  Sex
VariableAge: Mean±SD, yMenWomen
  1. SD indicates standard deviation.

Control60±14855241
Cancer
 Breast51±90300
 Colorectal63±11243207
 Gastric60±12274176
 Hepatocellular57±1029473
 Lung63±10308114

Minor Allele Frequencies and Haplotypes

The locations of 10 candidate SNPs reported in the National Center for Biotechnology Information database are listed in relation to the genomic structure of CCNA2 in Figure 1A. Seven SNPs were located in the 5′-untranslated region (UTR), 2 SNPs in were located exon 1, and 1 SNP was located in intron 1. Ten candidate SNPs were genotyped without assay failure, and all SNPs were identified as polymorphic in the Korean sample that we studied. Ages of the general population examined ranged from 14 years to 106 years. Frequencies of rare alleles of the 10 polymorphic SNPs in the Korean sample (N = 596) were calculated (data not shown), and 1 SNP (−1577 adenine to thymine [A[RIGHTWARDS ARROW]T]) was not analyzed statistically in further study because it deviated from HWE. A subset of 6 common SNPs in this target region using the criteria for SNPs with minor allele frequencies ≥5% was selected to construct the haplotype. One haplotype block based on pair-wise LD analysis was identified (Fig. 1B), 8 haplotypes were identified of 64 (26) possible haplotypes, and 4 major haplotypes accounted for >86% of haplotype distribution. Haplotype frequencies estimated by EM algorithm are illustrated in Figure 1C. We examined whether the LD structure in Koreans differed from that presented in HapMap. It turned out that the LD block we identified using Korean samples in the promoter region of cyclin A2 (CCNA2) belonged to the broader range of an LD block that included CCNA2 and neighboring genes in HapMap populations. We examined the LD structure of the target region in HapMap populations (phase 3) of Han Chinese in Beijing and Japanese in Tokyo (CHB + JPT) (data not shown) and Utah residents with Northern and Western European ancestry and Toscans in Italy (CEU + TSI) (Fig. 1D), respectively, by using Haploview and compared that structure with the LD block that we identified in the Korean population. We observed that the promoter region of CCNA2 was in the LD block that encompassed the exonuclase C-S9 (EXOCS9), CCNA2, and Bardet-Biedel syndrome 7 (BBS7) genes, which are not limited to ethnicity. It revealed that there was no Korean-specific LD structure in the region of interest.

thumbnail image

Figure 1. The locations of cyclin A2 (CCNA2) Single nucleotide polymorphisms (SNPs) and a linkage disequilibrium (LD) plot are shown. (A) The locations of 10 SNPs that were analyzed in the current study are illustrated. The first base of the transcription start site is counted as position +1 (GenBank accession no. NG_009111). Coding exons are marked by shaded blocks, and the 5′ and 3′ untranslated regions (UTRs) are represented by white blocks. A[RIGHTWARDS ARROW]T indicates adenine to thymine; T[RIGHTWARDS ARROW]C, thymine to cytosine; G[RIGHTWARDS ARROW]A, guanine to adenine. (B) LD strength between the 6 common SNPs with a minor allele frequency ≥5% was determined by the LD coefficient Lewontin D′ (|D′|) or by correlation coefficient (r2) values between each locus. Data from |D′| values are not shown. (C) Haplotype frequencies were estimated from genotype data from the 6 common SNPs within the LD block. (D) The LD structure of the HapMap CEU + TSI populations (Utah residents with Northern and Western European ancestry and Toscans in Italy; phase 3) is shown. The LD structure that was identified in the Korean population was not different from that presented in HapMap. EXOCS9 indicates exonuclase C-S9; BBS7, Bardet-Biedel syndrome 7.

Download figure to PowerPoint

Univariate Analysis of Cancer Risk

We also evaluated the associations between genotypes at 9 polymorphic loci and the prevalence of human cancer. One variant (rs769236; +1 G[RIGHTWARDS ARROW]A) of 9 analyzed SNPs differed significantly in genotypic distribution in a tissue-specific manner even after Bonferroni correction for multiple testing. In patients with breast cancer and gastric cancer, the distribution was similar to that observed in the control group; however, in patients with lung cancer, colorectal cancer, and hepatocellular carcinoma, the distribution differed significantly compared with that observed in the control group (Table 3). The frequency of the AA and GG genotypes in control samples was 20% and 27%, respectively. One hundred twenty-five (28%) of 450 patients with colorectal cancer had the AA genotype, and 115 (26%) had the GG genotype; this distribution differed significantly from that observed in the control group (Pearson chi-square test: P = .002; Cochran-Armitage test for trend: P = .01). Patients with hepatocellular carcinoma also had the AA homozygote more frequently than the GG genotype (28% vs 23%; P = .006). The AA genotype also had a more highly significant and susceptible effect on the development of lung cancer than the GG genotype (32% vs 21%; P < .0001). The AA genotype was identified in 24% of breast cancers and in 21% of gastric cancers. These frequencies did not differ significantly from genotype rates in the control group. Associations between haplotypes and cancer risk also were analyzed (Table 4). The common haplotype (Ht_1) (TAGCAC) with the A allele at the +1 locus was associated significantly with an increased risk of lung cancer (P < .0001), colorectal cancer (Pearson chi-square test: P = .004; Cochran-Armitage test for trend: P = .008), and hepatocellular carcinoma (P < .001).

Table 3. Association of +1 Guanine-to-Adenine Polymorphism With the Prevalence of Cancers
 No. (%)
  Cancer
GenotypeControlBreastColorectalGastricHepatocellularLung
  • G indicates guanine; A, adenine.

  • a

    P values were based on the association between controls and each cancer.

  • b

    Pearson chi-square test.

  • c

    Cochran-Armitage trend test.

GG295 (27)84 (28)115 (26)114 (25)86 (23)89 (21)
GA586 (53)145 (48)210 (47)240 (53)180 (49)200 (47)
AA215 (20)71 (24)125 (28)96 (21)101 (28)133 (32)
Total no.1096300450450367422
Pa
 Chi-squareb.20.002.68.006<.0001
 Trendc.51.01.39.006<.0001
Table 4. Association of Haplotype With the Prevalence of Cancer
 Haplotype: No. (%)Pa
VariableHt_1/Ht_1Ht_1/OtherbOther/OtherChi-SquarecTrendd
  • a

    P values were based on the association between controls and each cancer.

  • b

    b“Other” indicates all haplotypes except Ht_1.

  • c

    Pearson chi-square test.

  • d

    Cochran-Armitage trend test.

Control201 (18)581 (53)314 (29)  
Cancer
 Breast69 (23)147 (49)84 (28).18.23
 Colorectal121 (27)218 (48)111 (25).004.008
 Gastric95 (21)241 (54)114 (25).28.11
 Hepatocellular100 (27)179 (49)88 (24).001.001
 Lung126 (29)205 (49)91 (22)<.0001<.0001

Multivariate Logistic Regression Analysis

Multivariate logistic regression analysis was conducted to determine the relation between human cancers and the SNP (rs769236) at the transcription start site and to calculate the odds ratios (ORs) after adjusting for potential confounding factors (Table 5). A critical issue was to determine whether the strength of the association between the SNP and cancer susceptibility was uniform across cancer type. We examined whether the population ORs were constant across cancer type. The Breslow-Day test for homogeneity of the OR was applied, and there was evidence that the OR was heterogeneous (P = .02). Therefore, we determined that the variable “cancer type” could be an effect modifier in this study, and we treated the data in the various contingency tables as if they were drawn from distinct populations. Instead of applying polytomous logistic regression, we computed a different OR for each cancer type rather than computing a single summary value for the overall relative odds.

Table 5. Multivariate Logistic Regression Analysis of the Association Between Genotypes at the +1 Locus and the Risk of Cancera
 Codominant Modelb
 Additive ModelGAAA
  • G indicates guanine; A, adenine; OR, odds ratio; CI, confidence interval.

  • a

    Adjusted for age and sex.

  • b

    GG was the reference in the codominant model.

CancerOR (95% CI)POR (95% CI)POR (95% CI)P
Breast1.09(0.91-1.32).351.14(0.52-2.50).750.71(0.30-1.66).43
Colorectal1.29(1.05-1.58).021.71(1.20-2.43).0031.67(1.12-2.49)<.0001
Gastric1.11(0.92-1.33).271.11(0.80-1.53).531.23(0.85-1.77).27
Hepatocellular1.31(1.04-1.64).021.69(1.16-2.47).0071.71(1.10-2.64).02
Lung1.52(1.26-1.84)<.00011.95(1.42-2.68)<.00012.28(1.57-3.32)<.0001

Individuals who had the AA genotype had a 1.67 times greater risk of colorectal cancer (codominant model: 95% confidence interval [CI], 1.12-2.49; P < .0001) and a 1.31 times greater risk of hepatocellular carcinoma (additive model: 95% CI, 1.04-1.64; P = .02) compared with individuals who had the GG genotype after adjusting for age and sex, The AA genotype also was associated significantly with an increased risk of lung cancer (additive model: adjusted OR, 1.52; 95% CI, 1.26-1.84; P < .0001; codominant model: adjusted OR, 2.28; 95% CI, 1.57-3.32; P < .0001). However, this genotype was not associated with an increased risk of breast cancer (additive model: adjusted OR, 1.09; 95% CI, 0.91-1.32; P = .35) or gastric cancer (additive model: adjusted OR, 1.11; 95% CI, 0.92-1.33; P = .27).

Promoter Assay of CCNA2

Reporter assays were performed to determine the functional consequences of the SNP (rs769236; +1 G[RIGHTWARDS ARROW]A). Several constructs of different sizes were made, and the luciferase activity of each construct was compared. Among these constructs, the 0.3-kb fragment (Fig. 2A), which contains the major transcription start site, several protein-binding elements, and cell cycle-regulated elements, appeared to be the core promoter of the TATA-less CCNA2 promoter, because constructs that contained the 0.3-kb fragment produced high luciferase activity in several repeated experiments (Fig. 2B). The 0.3-kb fragment that contained the A allele (p0.3A) had transcriptional activity of approximately 120-fold, whereas the 0.3-kb fragment that contained the G allele (p0.3G) had an approximately 80-fold increase in its activation relative to the pGL3-Basic vector alone (set at 1). After conducting several separated-transfection experiments, we determined that the transcriptional activity of the p0.3A fragment was 1.5-fold greater than that of the p0.3G fragment, suggesting that the variant at the transcription start site of CCNA2 may affect transcriptional regulation of the CCNA2 gene.

thumbnail image

Figure 2. Plasmid construct and transcriptional activity of the cyclin A2 (CCNA2) promoter are illustrated. (A) The region marked by the dashed box shows the 0.3-kb fragment of luciferase reporter cassettes. An asterisk indicates the major transcription start site, and the rectangular box indicates the single nucleotide polymorphism (SNP) site (reference SNP [rs] 769236). ATF indicates activating transcription factor; CREB, cyclic adenosine monophosphate response element binding; NF-Y, nuclear transcription factor; CBF, core binding factor; CDE/CHR, cell cycle-dependent element/cell cycle gene homology region; E2F, eukaryote transcription factor; p53, tumor protein 53. (B) NIH3T3 cells were transfected transiently with either p0.3G, p0.3A, or a Renilla luciferase vector as an internal control. Cells were harvested and analyzed for luciferase activity 24 hours after transfection. The construct that contained the adenine (A) allele at the transcription start site of the CCNA2 gene had greater transcriptional activity than the construct that contained the guanine (G) allele. Values shown are the means ± standard error relative luciferase activity.

Download figure to PowerPoint

Distinctive CCNA2 Transcriptional Activity by Allelic Imbalance

To determine whether the effect of the constructs on transcriptional activity depended on the cell cycle, the constructs p0.3A and p0.3G were transfected transiently into synchronized NIH3T3 cells. We observed that the p0.3A construct resulted in greater transcriptional activity independent of cell cycle progression (Fig. 3). After 24 hours of serum starvation, most of the cells were in G1-phase of the cell cycle, as indicated in Figure 3A. Next, serum was added to the cultures, and the cells were analyzed for DNA content by fluorescence-activated cell sorting analysis at 3-hour intervals, starting at 15 hours after addition of serum. Relative luciferase activity was determined at 3-hour intervals. At 15 hours after serum addition, the majority of cells had entered S-phase, and transcription of the CCNA2 promoter was activated. Throughout the entire cell cycle, the p0.3A construct produced greater transcriptional activity than that produced by the p0.3G construct (Fig. 3B). Therefore, these results demonstrated that the single G-to-A substitution in the CCNA2 promoter affects transcriptional regulation, resulting in higher transcriptional activity, and that this effect does not depend on the cell cycle.

thumbnail image

Figure 3. Cell cycle independence of transcriptional activity by allelic imbalance is illustrated. (A) NIH3T3 cells were transfected transiently with either p0.3G, p0.3A, or a Renilla luciferase vector as an internal control. Twenty-four hours after transfection, serum was depleted for 24 hours; then, the cells were treated with media containing 10% fetal bovine serum and were harvested at each time point for fluorescence-activated cell sorting analysis. The times indicated are the hours (hrs) after release from serum starvation. (B) Throughout the entire cell cycle, the p0.3A construct had greater transcriptional activity than the p0.3G construct through the luciferase reporter assay. Values shown are the means ± standard error of relative luciferase activity and were calculated from 3 separate experiments.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

In this study, a genetic variant at the transcription start site of CCNA2 was associated significantly with an increased risk of lung cancer, hepatocellular carcinoma, and colorectal cancer but not with increases in the risk of breast cancer or gastric cancer, suggesting that the effect of the genetic variant of CCNA2 on human cancer development is tissue-specific. However, it is not clear how the variant influences disease susceptibility in a tissue-specific manner. It is known that genetic variants affect the function of proteins by altering the structure of the encoded protein or by changing gene expression. Furthermore, the effect of a variant may be influenced by tissue-specific factors. Substantial proportions of tissue-specific differences in transcript levels result from genetic variations in the regulatory machinery of genes23; and, generally, tissue specificity is not the result of a lack of expression in certain tissues but, rather, is caused by allelic imbalances in expression. Accordingly, alleles of regulatory polymorphisms that are proximal to a gene (cis) activate the gene at different levels; therefore, the 2 copies of the gene undergo different levels of transcription.24 The transcriptional activity of the p0.3A fragment in the current study was approximately 1.5-fold greater than that of the p0.3G fragment in NIH3T3 cells, suggesting that the cis-acting variant at the transcription start site of CCNA2 contributes to tumorigenesis by affecting transcriptional regulation of CCNA2.

Cyclin A2 expression is regulated through tight control of its promoter activity and, depending on the cell cycle, is modulated through periodic relief of transcriptional repression.16 Cell cycle-regulatory elements (cell cycle-responsive elements and cell cycle-dependent elements) are involved in the periodic regulation of CCNA2 transcription. In addition, several protein binding elements, such as the CRE and the CAAT boxes, also are involved in the regulation of cyclin A2 expression through several signal transduction pathways. These elements may contribute to periodic modulation of CCNA2 transcription through interplay with cell cycle-dependent recruiting proteins.16 We reanalyzed transcriptional activity through the cell cycle in NIH3T3 cells that were transfected with the A allele or the G allele to determine whether or not the effect of the variant on transcription of cyclin A2 was cell cycle-dependent. The construct that contained the p0.3A fragment had greater luciferase activity than the construct that contained the p0.3G fragment at 15 hours, 18 hours, and 21 hours after stimulation. These observations indicate that the effect of the variant on transcriptional regulation does not depend on cell cycle progression.

Transcriptional activity of CCNA2 was greater in NIH3T3 cells that contained the A allele than in the cells that contained the G allele, irrespective of the cell cycle. However, the mechanism by which the variant affects transcriptional activity is not clear. In general, initiation of the transcription processes requires the assembly of associated factors, such as general transcription factors IID (TFIID), TFIIB, and TFIIF, on the promoter region. The binding of TFIID to core promoter elements is essential to initiate formation of the RNA polymerase II preinitiation complex.25-27 Approximately 30% to 50% of all known promoters contain a TATA-box located from 45 bp to 25 bp upstream of the transcription start site; however, CCNA2 has a TATA-less promoter.28 In TATA-less promoters, the exact position of the transcription start point is controlled by nucleotide sequences of the transcription initiation region (Inr) or downstream promoter element (DPE), which typically is observed 30 bp downstream of the transcription start site. It is interesting to note that the variant that was associated with cancer susceptibility in the current study was located at the major transcription start site. TATA-binding protein-associated factors (TAFIIs), which are subunits of the TFIID complex, play a critical role in the recognition of core promoters,29 particularly on TATA-less promoters,30, 31 the downstream promoter element (DPE),32 and the initiator element (Inr).33 Significantly, TAFII250, the largest subunit of TFIID, is involved in the regulation of transcriptional activity of the CCNA2 core promoter34 and the cyclin D1 gene.35, 36 Human TAFII150 also functions in the recognition and selection of core promoters, including cell cycle-specific genes, such as cyclin A and cyclin B1.37 Accordingly, it is possible that the variant modulates transcriptional activity of CCNA2 by affecting interactions between TAFIIs and the core promoter of CCNA2.

With regard to cancer susceptibility, the factors responsible for tissue specificity of the cis-acting variant at the transcription start site of CCNA2 were not investigated in this study. In general, tissue specificity of genetic influences on human cancer may result from different levels of underlying expression in each tissue. Furthermore, trans-acting effects of tissue-specific components of the general transcription machinery may play important roles at different levels of gene expression by genetic variants during transformation. Alternate forms of the general transcription machinery also have been described in several tissues or cell types, and epigenetic events may influence allelic imbalances in expression. In addition, genetic variation may influence different levels of tissue-specific gene expression through interaction with environmental factors or other proteins. To gain a better understanding of tissue-specific differences in disease susceptibility of the variant, a large study with different types of cancer tissues and tissue-specific characterization is warranted. Furthermore, interaction between the genetic variant and environmental factors associated with development of each cancer will need to be evaluated. An epigenetic study also would be helpful to gain an understanding of the relation between genetic variants and cyclin A2 expression.

The general population and control groups in the current study were compared for genotype distribution of the SNP (rs769236; +1 G[RIGHTWARDS ARROW]A) within sex and age strata to ensure that the control group represented general residents. No difference was observed in the distribution of genotype frequency between those groups (data not shown). In addition, the SNP in the control groups was not in HWE (P = .01). Therefore, we analyzed whether the controls were in HWE by sex and age group. The variable “age” was divided into 2 groups at a similar frequency. The SNP (rs769236; +1 G[RIGHTWARDS ARROW]A) was in HWE for the groups aged <60 years (P = .09) and aged ≥60 years (P = .07). The SNP also was in HWE for men (P = .06) and for women (P = .07). It was possible that the SNP (rs769236; +1 G[RIGHTWARDS ARROW]A) was associated more with later disease stage at presentation, especially given the strong associations for liver and lung cancers. Therefore, we analyzed the association between the SNP and stage, but no association was observed (data not shown), possibly because of the small number of samples from patients with later stage disease. Further study with a large sample will be needed to elucidate the effects of stage.

In conclusion, the current findings suggest that an SNP at the transcription start site of CCNA2 may be associated functionally with the development of human cancer in a tissue-specific manner. This variant may prove to be a valuable biomarker for evaluating the susceptibility of individual patients to cancer.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

This work was supported by grants from the National Research Foundation of Korea funded by the Korea government (Ministry of Education, Science, and Technology; R11-2005-017-06002-0) and the Research Program of Dual Regulation Mechanisms of Aging and Cancer from the Korea Science and Engineering Foundation (grant 20090093587).

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES