High-throughput Loss-of-Heterozygosity Study of Chromosome 3p in Lung Cancer Using Single-Nucleotide Polymorphism Markers

Lung cancer is one of the most common malignancies in the world and one of the leading causes of cancer death in the United States, which accounts for ∼28% of all cancer death during 2001 (1). The overall 5-year survival rate of this prevalent cancer is <15% (2). Lung cancer can be classified into two major types: small-cell lung cancer and non–small-cell lung cancer (NSCLC), which is ∼25% and 75% of all lung cancers, respectively. The epidemiology of lung cancer is multifactorial with cigarette smoking as the major cause (3). Deletion of chromosome 3p is one of the most frequent allelic imbalances in various human tumors, including NSCLC (4, 5). Loss of heterozygosity (LOH) analysis is one of the most efficient methods to study chromosomal deletion and to narrow down the deleted region. High rates of LOH loci are commonly found to harbor putative tumor suppressor genes (TSG). Previously published LOH studies showed that 3p21.3 and 3p14.2 are the two most frequently deleted regions in lung cancer (4, 6). Several candidate TSGs, such as FHIT (7), RASSF1A (8), CACNA2D2 (9), and DLC1 (10), have been widely studied. However, comparative genomic hybridization results showed that the loss of 3p was often involved in the whole short arm in NSCLCs (5, 11). Therefore, the existence of some unknown TSGs at loci other than 3p21.3 and 3p14.2 regions is likely.

Conventional LOH uses limited microsatellite polymorphism markers (12), which is a time-consuming process with low-resolution genotyping. In addition, it is difficult to automate this method in large-scale LOH analysis. Because of these limitations, it is very hard to perform high-resolution LOH analysis using conventional LOH method. The progress of the human genome project makes it possible to apply single-nucleotide polymorphism (SNP) markers in LOH analysis. SNPs represent the most common form of sequence variation in the human genome, occurring ∼1 every 1,200 bp (13). SNPs can be used as high-density polymorphic markers for studying genetic variations, including LOH (14), although their heterozygosity is relatively low compared with microsatellites. In this study, a high-throughput and high-resolution LOH study of chromosome 3p in 30 NSCLCs was done using SNP-mass spectrometry-genotyping (SMSG) technology. Several high-LOH loci have been identified and three candidate TSGs were further studied.

Tumor samples. Thirty NSCLC samples were collected from the time of surgical resection at the Cancer Institute, Sun Yat-Sen University (Guangzhou, China), during the period of 1998 to 2002. The specimens were snap frozen and kept at −80°C until DNA extraction. All the samples were examined and macrodissected under a microscope by a pathologist to ensure the purity of tumor samples with <10% normal cell contamination.

SNP marker selection and primer design. A total of 386 SNP markers spanning over 0 to 70 Mb of 3p with minor allele frequency over 0.3 were selected, including 326 SNPs with minor allele frequency over 0.45, 41 SNPs between 0.4 and 0.44, 16 SNPs with between 0.35 and 0.43, and 4 SNPs between 0.3 and 0.34 (see Supplementary Table S1). All minor allele frequencies were calculated from Caucasian populations.

MassARRAY AssayDesign software (Sequenom, San Diego, CA) was used to design amplification and allele-specific extension primers for uniplexed or multiplexed assays. In the design, PCR primers have an additional 10-base tag (5′-ACGTTGGATG-3′) to prevent their interference in the resulting mass spectra. The designed assays were constrained to produce products of optimized size (90-120 bp) to maximize the PCR successful rate. The extension primer was designed to hybridize to the amplicon near the SNP locus for the extension of a single base or a few bases depending on the genotype of the allele (Supplementary Table S2).

PCR amplification and dephosphorylation. PCR reactions were done in a 384-well-plate format in a total volume of 6 μL per reaction with 5 ng genomic DNA, 0.3 pmol each of the specific forward and reverse primers, 200 μmol/L of each of the deoxynucleotide triphosphates, 3.25 mmol/L MgCl₂, and 0.2 unit of HotStarTaq polymerase (Qiagen, Valencia, CA). A total of 45 PCR cycles were carried out (95°C for 20 seconds, 56°C for 30 seconds, and 72°C for 1 minute) with an initial denaturation at 95°C for 15 minutes and final extension at 72°C for 3 minutes.

After PCR reaction, residual amplification nucleotides in the PCR products were dephosphorylated with alkaline phosphatase. A mixture of 0.2 μL hME buffer, 0.3 μL shrimp alkaline phosphatase (1 unit/μL; Sequenom), and 1.5 μL double-distilled water was added to the PCR products. The reaction solution was incubated at 37°C for 20 minutes, followed by 85°C for 5 minutes to inactivate the enzyme.

Allele-specific primer extension. Mass extend reactions (MassEXTEND Assay) were done in four groups of different terminations according to the design rationale (ddACG, ddACT, ddAGT, and ddCGT) in 10 μL reaction volume containing 1 unit of Thermosequenase (Sequenom), 50 μmol/L of the respective termination mix, and 0.6 pmol of each assay-specific extension primer. All the assays were done under the thermal cycle conditions of initial denaturation at 94°C for 2 minutes followed by 55 cycles of 94°C for 5 seconds, 52°C for 5 seconds, and 72°C for 5 seconds.

Genotype detection using matrix-assisted laser desorption/ionization-time of flight mass spectrometry. The final base extension products were treated with SpectroClean resin (Sequenom) to remove the salts in the reaction buffer. Briefly, 16 μL resin per water suspension was added to each base extension reaction and 10 nL of the reaction solution were dispensed onto a 384-well-plate format SpectroChip (Sequenom) prespotted with a matrix of 3-hydroxypicolinic acid using the MassARRAY Nanodispenser (Sequenom). The modified Brucker Autoflex matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometer (Brucker, Billerica, MA) was used for data acquisition. Data were automatically imported into the SpectroTYPER database (Sequenom) for the robust data analysis, including noise normalization and peak area analysis. The expected molecular weights of all relevant peaks were calculated by the MassARRAY AssayDesign Software before analysis and identified from the mass spectrum. In every assayed plate, one well for blank control and five wells for duplicate check on five samples for quality control were included. Twenty-four SpectroChips for 8-plexed reactions, in which each chip is composed of 3,072 genotypes, can be analyzed within 1 day. A high-throughput genotyping consisting of 73,728 genotypes is routinely done within 24 hours.

To test the effect of the tumor purity in LOH detection using SMSG, DNA from one tumor sample that contained <5% normal cells was used to determine the accuracy of genotyping. DNA samples from tumor, matched normal tissue, and series of mixtures (containing 5%, 10%, 20%, and 50% normal DNA, respectively) have been assessed with five SNP markers.

Comparative genomic hybridization. Comparative genomic hybridization was done as previously described (15). Briefly, 1 μg tumor test and normal reference DNA were labeled with Spectrum Green-dUTP and Spectrum Red-dUTP (Vysis, Downers Grove, IL), respectively, by nick translation at 15°C for 2 hours. The slide containing normal metaphase spreads was denatured and then hybridized with 200 ng of each probe at 37°C for 2 days. After hybridization, the slide was washed and counterstained with 1 μg/mL 4′,6-diamidino-2-phenylindole in an antifade solution. The image of hybridized metaphases was captured using a Zeiss Axioplan 2 microscope equipped with a Sensys cooled-charged device camera (Photometrics, Ltd., Tucson, AZ), analyzed using Quips comparative genomic hybridization program (Vysis), and interpreted according to the fluorescence intensity profile under program guidance. The threshold value for DNA copy number gain and loss was defined as intensity ratio of tumor/normal >1.2 and <0.8, respectively.

SMSG technology. In the SMSG system, a 90- to 120-bp DNA fragment containing the target SNP site was amplified by PCR with a pair of specific primers. The PCR product was then extended with another primer near the SNP site and extended sequence was automatically detected by MALDI-TOF. The diagrammatic representation of SMSG technology was shown in Fig. 1A. This revolutionary technology provides a useful tool for high-throughput and accurate genotyping analysis (16). A total of 386 SNP markers, spanning from 1 bp to70 Mb of chromosome 3p (3pter-p14.1), were selected for the LOH analysis in 30 primary NSCLC cases. The average intermarker distance in the present study is ∼180 kb.

LOH. The overall positive call rate (the percentage of SNP loci that genotypes can be assigned) was 91.0 ± 2.1%. For a given normal/tumor pair, each SNP can be identified as “LOH” (heterozygous in normal and LOH in tumor), “retention” (both normal and tumor are heterozygous), and “noninformative” (homozygous in normal; Fig. 1B). Among the 386 SNP markers used in this study, 29 markers showed a very low informative rate (<5%), which were excluded for the data analysis. The average heterozygosity for the remaining 357 SNP markers is 0.39.

To assess the effect of tumor purity on LOH detection, DNA samples from one case, including tumor, matched normal tissue, and series of mixtures (containing 5%, 10%, 20%, and 50% rnormal DNA, respectively), had been assessed with five SNP markers. Results indicated that LOH could be detected when the contamination of normal cell was <10%. Tumor samples with 80% tumor purity gave an increased false “retention” (Fig. 2).

The global pattern of LOH in 70 Mb of 3p in 30 NSCLC cases was shown in Fig. 3. Consistent with the findings of others, 3p21.3 and 3p14.2 are the two most frequently deleted regions in this study. One of the most interesting findings is that several other frequently deleted regions at 3p, including 3p26.3, 3p25.3, 3p24.1, and 3p23, were identified.

Validation of LOH results. LOH results in this study are generally agreed with the comparative genomic hybridization data. In comparative genomic hybridization study, 22 NSCLC cases showed DNA copy loss at 3p (first 22 cases in Fig. 3) and 8 cases showed nearly no loss (last 8 cases in Fig. 3). Figure 4A compared three comparative genomic hybridization results and their matched LOH results. Interestingly, an interstitial deletion of 3p at 3p14.3-p14.1 detected by SNP mass array in case L22 was perfectly confirmed by comparative genomic hybridization.

LOH results in this study were also confirmed by sequencing analysis. Ten SNP markers were randomly selected and PCR products of a pair of tumor (with LOH) and its matched normal tissue (without LOH) for each SNP were sequenced. LOH was detected in all 10 tested tumor samples, whereas it was not observed in 10 matched normal tissues (Fig. 4B). About 10% to 38% of allelic loss was found at 357 SNP loci for the 22 cases with comparative genomic hybridization changes, whereas <7% of allelic loss was detected in those eight cases without comparative genomic hybridization change.

Candidate TSGs. The most frequently detected LOH was found at 22 SNP loci, which was shown in Fig. 5A and summarized in Table 1. Several candidate TSGs within these frequent LOH regions have been found, including APG7-like gene (APG7L) at 3p25.3, CLIP-associated protein 2 (CLASP2) at 3p23, and calcium channel, voltage-dependent, α-2/δ subunit 3 (CACNA2D3) at 3p21.1, besides FHIT and DLC1 (Table 1). The genomic size of CACNA2D3 gene is ∼1 Mb, which was covered by four SNP markers in this study. The overall LOH involving CACNA2D3 was detected in 19 cases. Similarly, FHIT was covered by two SNPs and the overall LOH of FHIT was observed in 15 cases. The expression levels of these candidate TSGs were studied by reverse transcription-PCR (RT-PCR) in 12 primary NSCLCs. Absent or down-regulated expressions of APG7L, CLASP2, and CACNA2D3 were observed in 10, 5, and 6 cases, respectively (Fig. 5B).

It is believed that the pathogenesis of lung cancer is a long-term process that involves multiple genetic alterations, including the loss of function of TSGs at 3p. Isolation and characterization of TSGs at 3p will significantly improve our knowledge in the pathogenesis of lung cancer. In the present study, a recently developed high-throughput and high-resolution technology, SNP mass array, was applied to identify frequent LOH loci at 3pter-p14.1 and isolate candidate TSGs within these loci.

Three candidate TSGs, including APG7L, CACNA2D2, and CLASP2, have been identified. APG7L is the human homologue of yeast Apg7, a key autophagy gene encoding an ubiquitin-E1-like enzyme essential for the APG12 conjugation system that mediates membrane fusion in autophagy (17). LOH involving CACNA2D3 gene was the most frequent genetic alteration in the present study. Although CACNA2D3 has not been associated with cancer development, exogenous expression of its family member CACNA2D2 could significantly inhibit tumor cell growth and induce apoptosis (18). Because both CACNA2D2 and CACNA2D3 are subunits of the Ca²⁺ channel complex, it is reasonable to believe that CACNA2D3 might also be correlated with pathogenesis of lung cancer. CLASP2 interacts with CLIP, binds to microtubules, and has microtubule-stabilizing effects (19). Increasing microtubule instability may cause genetic instability and loss of CLASP2 function may induce genetic instability and lead to the development of cancer.

High-throughput LOH study using SMSG provides a very useful tool for a rapid and high-resolution LOH investigation of interesting chromosomal regions. Recently, Lindbland-Toh et al. (20) applied a whole-genome HuSNP array (Affymetrix, Santa Clara, CA) with nearly 1,500 SNP markers for the LOH detection in small-cell lung cancer. Later, HuSNP array was used to study LOH in prostate (21) and bladder cancers (22). More recently, GeneChip human mapping 10 K array, which contains over 10,000 SNPs, has been used for genomewide genotyping and linkage analysis (23, 24).

Comparing with currently available HuSNP array, the SMSG technology used in this study is more rapid, less expensive, and more versatile. For example, investigation of 60 samples (30 pairs of NCSLCs) with 386 SNP markers (60 × 386 = 23,160 genotypes) can be finished within 24 hours. As no hybridization is included in the process, the bias toward genotype interpretation can be avoided. In addition, only a small amount of DNA (5 ng for each reaction) is required and that makes it possible to use purer DNA from tumor cells captured by laser dissection, which is the major confounded factor in LOH study. Another advantage of this method is its capability to provide a high-resolution analysis within a given chromosomal region. In most solid tumors, recurrent deletions are only detected in a few chromosomal regions by comparative genomic hybridization studies in a particular cancer. The strategy described here can focus on some of the given chromosomal regions with high resolution (20-100 kb) and less cost comparing with genomewide genetic characterization. In conclusion, the method described in this study extends the limits of conventional LOH analysis by providing a rapid source of high-throughput method and thus enabling the high-resolution LOH investigation of chromosomal regions. Also, this highly reproducible method allows the rapid identification of candidate TSGs in a given cancer.

Grant support: Chinese State Key Program for Basic Research grant G1998051207, Research Grant Council grant HKU7393/04 M, and Leung Kwok Tze Foundation.

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.