Tissue samples were obtained from 32 patients with lung cancer (19 with adenocarcinoma, 12 with squamous cell carcinoma and one with large cell carcinoma). Sections of carcinoma tissues and non-tumor lung tissue were removed and frozen immediately following surgery at Kanazawa University Cancer Research Hospital (Kanazawa, Japan). The tissues were stored at −80°C until further analysis.

Frozen 10-μm tissue sections were fixed with 70% ethanol for 10 min and stained with Kernechtrot (Merck, Darmstadt, Germany). A total of ~1,000 carcinoma cells in the frozen sections were punched out using a LCM system (LM100; Olympus Corporation, Tokyo, Japan) and collected on a plastic cap (CapSureTM LCM Caps; Arcturus, Mountain View, CA, USA). All specimens were assessed histologically by a pathologist (Professor E. Kawahara).

Genomic DNA from the frozen sections of the normal tissues was extracted using a Wizard^® SV Genomic DNA Purification system (Promega Corporation, Fitchburg, WI, USA). To extract total RNA from the microdissected carcinoma cells, a PicoPure™ RNA isolation kit (Arcturus) was used. The extracted RNA was further purified using the acid guanidinium phenol chloroform method, and possible contaminating DNAs were digested with DNase I (Takara, Tokyo, Japan). Aliquots of 1 μl extracted total RNA in a volume of 20 μl were converted to cDNA using AMV reverse transcriptase (Promega Corporation). The reverse transcription reaction was performed for 30 min at 42°C and then the sample was heated for 5 min at 99°C to inactivate the enzyme. Genomic DNA from normal whole tissues was amplified by PCR using the following primer pair for exon 9 of human IGF2: Forward, 5′-CTTGGACTTGAGTCAAATTGG-3′ and reverse, 5′-GGTCGTGCCAATTACATTTCA-3′. The tumor cDNAs obtained from the microdissection samples were amplified by nested PCR using the same primer pair followed by the second primer pair: Forward, 5′-CTTGGACTTTGAGTCAAATTGG-3′ and reverse, 5′-GGTTTTCATGCTCTGTCCTC-3′. The amplified 292 and 257 base-pair fragments were digested by ApaI (Toyobo, Osaka, Japan), which recognizes the C/T single nucleotide polymorphism site in the amplified area. Successful digestion of contaminating DNA was confirmed by PCR designed to amplify the exon-intron spanning region of the integrin β5 gene. Analysis of the polymorphism in the IGF2 gene was then performed. The digested and undigested DNAs were run on 2% agarose gels containing ethidium bromide. The gel revealed three bands for heterozygotes, in which the restriction enzyme recognition sequence was present in one allele and absent in the other. The PCR products were also sequenced using Dye Terminator 1.1 (Applied Biosystems, Carlsbad, CA, USA) and ABI PRISM^® 3130×l Genetic Analyzer (Applied Biosystems).

Digital images of the bands scanned into a computer were analyzed using the Image J software (National Institutes of Health, Bethesda, MD, USA). The band densities were measured and the ratio of the two bands was calculated. Similarly, the ratio of the two C and T peaks was calculated. The imprinting status was determined according to the criterion described in our previous study (10). In the case that two clear bands were detected with PCR-RFLP (ratio = 10–90%) and the sequence pattern revealed two clear peaks (ratio = 20–80%), the status was considered to be LOI; otherwise, the status was considered as imprinted.

Statistical analyses were performed using StatMate software version 4.0 (ATMS, Co., Ltd., Tokyo, Japan) to perform the χ² test. P<0.05 was considered to indicate a statistically significant difference.

The study was approved by the Instititional Review Board for Analytical Research for Human Genome and Gene at Kanazawa University. All patients provided informed consent prior to sample collection according to the institutional guidelines.

The genomic DNA of whole non-tumor mucosa was employed to screen for whether the samples were informative or not, using the PCR-based, IGF2 gene ApaI polymorphism. Of the 32 lung cancer cases, 23 (72%) were informative. All informative cases were examined for imprinting status in the microdissected carcinoma cells. The PCR products from the cDNA of the carcinoma cells revealed either two bands or a single band following ApaI analysis, and either two C and T peaks or one peak with a possible background peak subsequent to sequence analysis (Fig. 1). Prior to determining whether the status was imprinted or LOI, the density ratio between the top and bottom bands, determined using PCR-RFLP, and the height ratio of the C and T peaks, examined using sequence analysis, were measured and calculated (Table I). Each case was identified as either imprinted or LOI, according to the criteria described above (Table I). LOI was detected in the adenocarcinoma and large cell carcinoma specimens, but was not detected in the squamous cell carcinoma samples. The percentage of LOI was 39% in total lung carcinomas (9 out of 23 cases) and 62% (8 out of 13 cases) in adenocarcinomas.

Since a previous study concerning IGF2 LOI in lung adenocarcinoma refers to histological type (7), the association between the histological grade of the adenocarcinoma sample and LOI was also analyzed in the present study. Four cases of well-differentiated adenocarcinoma were found to be LOI and one case of poorly differentiated adenocarcinoma was imprinted (Table I). These results do not support the previous suggestion that IGF2 LOI occurs more frequently in poorly differentiated adenocarcinomas (7). Statistical analysis revealed no significant association between the degree of differentiation and LOI.