Mutations of Adenomatous Polyposis Coli and β-Catenin Genes during Progression of Lung Tumors Induced by N-Nitrosobis(2-hydroxypropyl)amine in Rats¹

The majority of lung cancers, one of the most common forms of cancer in humans, are NSCLCs³. It is now widely accepted that carcinogenesis is a multistep process involving sequential accumulation of changes at the genome level(1). Although there are a number of reports on gene alterations in human lung cancers, rate-limiting events have yet to be established. Previously, we described a model for the development of NSCLCs in rats given BHP in drinking water with high yields of adenomatous lesions, including adenocarcinomas (2, 3). This model is useful for investigation of molecular mechanisms involved in step-by-step development of lung adenocarcinomas from adenomatous preneoplastic lesions. The molecular events demonstrated thus far are a high frequency of Ki-ras mutations, but no mutations of Ha-ras and p53, in relatively early lesions(4). We also found overexpression of VEGF to be related to up-regulation of VEGF receptor-1/fms-like tyrosine kinase-1 and VEGF receptor-2/fetal liver kinase-1 expression (5) and that of midkine (6) during BHP-induced lung carcinogenesis.

The protein β-catenin, a submembranous component of the cadherin-mediated cell-cell adhesion system, has been demonstrated to be the downstream activator of Wnt signal transduction (7, 8). The amounts of cytoplasmic β-catenin are mainly regulated by interaction with APC protein, the adenomatous polyposis coli gene product, and by phosphorylation at serine and/or threonine residues through the action of GSK-3β (9). Mutations of APC orβ-catenin affect the degradation of β-catenin protein by the ubiquitin/proteasome system and result in its stabilization and accumulation within the cell (10, 11). Association with members of the Tcf family (12, 13) and their complexes then causes transactivation of growth-promoting genes, such as c-myc and cyclin D1 (14, 15). It has been suggested that the APC-β-catenin-Tcf signaling pathway plays a major role in colon cancer of humans, through mutational inactivation of APC or activation of β-catenin (10, 11, 16). With regard to human lung cancers, several reports have documented deletions at 5q of the APC gene (17, 18, 19) without evidence of APC point mutations (19, 20), and it has been suggested that APC itself is unlikely to play a significant role in human lung cancer (20). In rodents, decreased expression of the APC gene in mouse lung neoplasias has been reported, whereas mutation and loss of heterozygosity of APC were not found (21). Recently, β-cateninmutations have been described in several cancers of humans and rodents(10, 11, 22, 23, 24, 25). However, to our knowledge there have been no reports of such alteration in lung cancer. When studied by immunohistochemistry, β-catenin showed no evidence of oncogenic activation in human lung cancer (26).

In this present study, to clarify the involvement of APC-β-catenin-Tcf signaling pathway in lung carcinogenesis, we therefore investigated APC and β-catenin gene mutations in lung lesions induced by BHP in rats.

Male Wistar rats, 5 weeks of age, were purchased from Japan SLC, Inc.(Shizuoka, Japan) and housed three to five to a plastic cage in an air-conditioned room, with a constant temperature of 25°C with a 12-h light/dark cycle. Food and water were given ad libitumthroughout the study. After a 1-week acclimation period on a basal diet in pellet form (Oriental MF Diet; Oriental Yeast Co. Ltd., Tokyo,Japan), the animals received 2000 ppm BHP (Nakalai Tesque Co. Ltd.,Kyoto, Japan) in their drinking water for 12 weeks and then drinking water without BHP. The animals were killed under ether anesthesia 25 weeks after the beginning of the experiment.

The lungs were immediately removed, fixed in formalin at 4°C, and routinely processed for paraffin embedding. Three serial thin sections were made. Two cut at 3-μm thickness were stained with H&E for histological examination and for immunohistochemical analysis ofβ-catenin. The other section at 5-μm thickness was used for DNA extraction. Lung lesions were classified according to the diagnostic criteria described previously (2, 3).

DNA extraction from paraffin-embedded sections of 17 hyperplasias, 15 adenomas, 20 adenocarcinomas, and 1 normal lung tissue was performed as described previously (27), followed by PCR-SSCP analysis for APC and β-catenin mutations.

For the APC gene, PCR-SSCP and PCR-restriction-SSCP analysis were carried out with methods reported previously (28, 29). The primers used in this study were chosen to amplify exon 1 through the MCR in exon 15 of the APC gene with intron sequences flanking coding exons as described previously (Refs.28 and 29; Table 1). Briefly, PCR for SSCP was performed in 5 μl of reaction mixture containing 4 pmol of each primer, 1× PCR buffer (Perkin-Elmer Corp.,Applied Biosystems Division, Foster City, CA), 200μ m of each deoxynucleotide triphosphate, 68 nm of [α-³²P]dCTP, 0.5 unit of AmpliTaq (Perkin-Elmer), and 50 ng of template DNA. The mixture was subjected to 35 cycles of amplification, each consisting of 0.5 min at 94°C for denaturation, 1 min at 55°C for annealing, and 2 min at 72°C for extension. Amplified products longer than 300 bp were digested with restriction enzyme before electrophoresis (Refs.28 and 29; Table 1). To rule out the PCR artifacts, PCR amplification was repeated from individual original DNA at least once. The samples were applied to 6 or 10% polyacrylamide gels with or without 5% glycerol after denaturation at 90°C for 3 min. Electrophoresis was performed at 40 W for 2–4 h at 20°C. The gels were dried on filter paper and used to expose X-ray film at−80°C.

For the β-catenin gene, primers of appropriate oligonucleotide sequences, 5′-GCTGACCTGATGGAGTTGGA-3′ and 5′-GCTACTTGCTCTTGCGTGAA-3′, were used for specific amplification of the consensus sequence for GSK-3β phosphorylation as described previously(Table 1; Ref. 23). PCR amplification for SSCP was performed under the following reaction conditions: denaturation step for 5 min at 95°C, 35 cycles for 1 min at 95°C, 1 min at 60°C and 2 min at 72°C, and extension for 10 min at 72°C (25). The resultant PCR products were applied to 6% polyacrylamide gels with or without 5% glycerol. Electrophoresis was performed at 40 W for∼2.5 h at 20°C, and gels were subjected to autoradiography.

The DNA fragments of mobility-shifted bands in the polyacrylamide gel were extracted and reamplified. The PCR products obtained were cloned with a TOPO TA cloning kit (Invitrogen, San Diego, CA), and recombinant plasmid DNA clones were sequenced using Sequencing Pro(TOYOBO Co. Ltd., Tokyo, Japan). In each experiment, 5–10 clones from different bacterial colonies were investigated.

Immunohistochemical staining for β-catenin was studied for 216 hyperplasias, 37 adenomas, and 40 adenocarcinomas,including the samples examined for mutation analysis of β-catenin and Ki-ras. Demonstration of anti-β-catenin binding was performed according to a standard protocol for DAKO ENVISION + (Dako Japan, Kyoto, Japan). Briefly, deparaffinized tissue sections were incubated in 3%H₂O₂ dissolved in methanol for 30 min and then autoclaved in 10 mm citrate buffer (pH 6.0) for 10 min twice. Normal goat serum (Sigma Chemical Co., St. Louis, MO) was diluted at 4% in Tris-buffered saline and used to block nonspecific cross-reactions by incubation for 30 min at room temperature. The mouse monoclonal anti-β-catenin antibody(Transduction Laboratories, Lexington, KY) was used at the concentration of 0.05 μg/ml in 4% goat serum added to Tris-buffered saline and incubated for 30 min at 37°C. 3,3′-Diaminobenzidine tetrahydrochloride (Nacalai Tesque Co. Ltd., Kyoto, Japan) was used for visualization of binding. The specificity of the binding was confirmed by negative control staining using mouse nonimmune serum instead of the primary antibody (data not shown).

Representative results of PCR-SSCP and PCR-restriction-SSCP analysis and sequencing analysis for APC mutations are shown in Fig. 1. None of 17 hyperplasias showed bandshifts in exon 1 through the MCR in exon 15. Two of 15 adenomas showed bandshifts in exons 12 and 15A, respectively (13.3%), whereas 8 of 20 adenocarcinomas showed bandshifts in exons 3, 8, 9, 10, 11, 14, 15BD, and 15FH. Among those 8,2 cases had bandshifts in both exons 3 and 9, and exons 15BD and 15FH,respectively. Two adenomas showed GCT to CGT (Ala to Arg) transversions at codon 515 and TGT to TGC (Cys to Cys) transition at codon 679,respectively, the former leading to a base substitution and the latter to no amino acid alteration. In the 8 adenocarcinomas, one showed a GAA to AAA (Glu to Lys) transition at codon 292 and a TGC to TGT (Cys to Cys) transition at codon 415, and another a GTT to GTC (Val to Val)transition at codon 819 and an AGT to AAT (Ser to Asn) transition at codon 1392. One of the two mutations in both cases resulted in no amino acid alteration. The other 5 cases showed a GTT to ATT (Val to Ile)transition at codon 440, a GGG to AGG (Gly to Arg) transition at codon 469, a TGG to TGA (Trp to Stop) transition at codon 591, an ATG to GTG(Met to Val) transition at codon 817, and a GAC to GGC (Asp to Gly)transition at codon 1422. That involving codon 298 was without amino acid alteration. No deletions or LOH were found for the 17 hyperplasias, 15 adenomas, and 20 adenocarcinomas.

Results of PCR-SSCP analysis and sequencing analysis for β-catenin mutations are illustrated in Fig. 2. In 17 hyperplasias, no fragments showed bandshifts. However, in 3 of 15 adenomas (20.0%) and 5 of 20 adenocarcinomas (25.0%), bandshifts were apparent indicative of β-cateninmutations. In the 3 adenomas, there were two GAT to GTT (Asp to Val)transversions at codon 32 and one GGA to GAA (Gly to Glu) transition at codon 34. The 5 adenocarcinomas had two GAT to GTT (Asp to Val)transversions at codon 32, one TCT to TTT (Ser to Phe) transition at codon 33, one ATC to GTC (Ile to Val) transition at codon 35, and one ACC to ATC (Thr to Ile) transition at codon 41. The patterns and incidence of APC and β-cateninmutations in adenomas and adenocarcinomas are summarized in Table 2.

Representative micrographs of lesions stained with anti-β-catenin antibody are shown in Fig. 3. β-Catenin protein was found to be localized in cell membranes in background tissue and in hyperplasias of the lungs of rats. In adenomas and adenocarcinomas, in contrast, staining was mainly in the cytoplasm and/or nucleus. Data for the localization of β-catenin protein are summarized in Table 3. β-Catenin protein was localized at the cell membranes in all hyperplasias, whereas it was localized in the cytoplasm in 5 of 37 adenomas (13.5%) and 6 of 40 adenocarcinomas (15.0%) and in the cytoplasm and nucleus in 5 of 37 adenomas (13.5%) and 15 of 40 adenocarcinomas (37.5%). All cases that demonstrated APCand/or β-catenin mutations with amino acid substitution also featured localization of β-catenin protein in the cytoplasm and/or nucleus (Table 2).

Previously, we reported relatively high frequencies of Ki-ras mutations in lung lesions induced by BHP in rats;40.0% of hyperplasias, 35.7% of adenomas, and 72.0% of adenocarcinomas were positive, suggesting that Ki-rasmutation is an early event of BHP-induced lung carcinogenesis in rats(4). In this study, we found APC mutations in 13.3% (2 of 15) of adenomas and 40.0% (8 of 20) of adenocarcinomas,along with β-catenin mutations in 20.0% (3 of 15) of adenomas and 23.0% (5 of 20) of adenocarcinomas. However, no mutations of APC and β-catenin were detected in 17 hyperplasias. Therefore, the present results suggest that the APC-β-catenin-T-cell factor signaling pathway is involved in the acquisition of growth advantage from adenomas to adenocarcinomas in BHP-induced rat lung carcinogenesis.

In human colon tumors, >95% of the APC mutations are frameshift or nonsense mutations resulting in a truncated protein(30, 31). There have been few reports of APCmutations in rodents (28, 29, 32). In rat colon tumors induced by heterocyclic amines, such as 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine and 2-amino-3-methylimidazo[4,5-f]quinoline,2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine caused 5′-GGGA-3′ to 5′-GGA-3′ frameshift mutations of APC(50.0%), whereas 2-amino-3-methylimidazo[4,5-f]quinoline was associated with missense (7.7%) and nonsense mutations (7.7%;Ref. 28). In contrast, no APC mutations were apparent in ulcerative colitis-associated rat colon carcinogenesis induced by 1-hydroxyanthraquinone and methylazoxymethanol acetate(29). Recently, missense mutations have been reported to be a feature of rat colorectal carcinogenesis induced by dimethylhydrazine (32).

In the present study, among 2 adenomas and 8 adenocarcinomas with APC mutations, 9 cases were missense or silent and only 1 case was nonsense, with no frameshift mutations being found. Therefore,it seems that the frequency and the mode of APC mutation may depend on the inducing carcinogen or species. It is unknown whether missense mutations of the APC gene have any significance. In this study, the nonsense mutation at codon 591 would be expected to cause truncation of the ACP protein. A repeated 20-amino acid sequence in the central region of APC protein includes the phosphorylation site and binding of β-catenin to this region is dependent on phosphorylation by GSK-3β (33, 34). This region also includes the binding site for Axin and coductin (35, 36). Missense mutations at codons 1392 and 1422 in this region may therefore have influenced phosphorylation by GSK-3β or the binding of these proteins. The first 1000 amino acids of APC protein contain heptad repeat motifs, including the Armadillo repeat (33, 37). The heptad repeats are believed to be capable of facilitating protein-protein interaction through formation of extended α-helices that present a series hydrophobic residues extending along one side. These hydrophobic regions can stabilize the interaction of two suchα-helices to form either homo- or heterodimers through generation of a coiled coil (33, 37). Codons 292, 440, 469, 515, 591,and 817 are included in this region, and missense mutations in this region may affect protein-protein interactions. We have showed a correlation between APC mutations with amino acid substitution and the accumulation of β-catenin protein. No accumulations were found in the cases without APC or β-catenin mutations (Table 2). These findings provide support for the possibility that some missense mutations,though presumably not all cases, may contribute to accumulation ofβ-catenin protein. In this study, we did not investigate the expression levels of APC and wnt genes in lung lesions induced by BHP in rat. Decreased expression of APC(20) or activation of Wnt signaling (7, 8)could also affect the stabilization and accumulation of β-catenin protein. Whereas 50–60% of the somatic mutations of APC in human colon tumors are clustered in a 700-bp region, designated as the MCR, in the middle part of exon 15 (30), in this study,such APC changes accounted for only 2 of 10 in our genetic alterations(adenocarcinomas 3 and 5; 20%).

Mutations of β-catenin have been reported at codons 32, 33, 34, 35,37, and 41 in colon tumors and hepatocellular carcinomas induced by various carcinogens in rats (23, 24, 25). It is considered that the serine and threonine sites located in codons 33, 37, and 41 in β-catenin are important for GSK-3βphosphorylation, and codons 32, 34 and 35, which neighbor a serine, are also supposed to be necessary for the ubiquitin-dependent proteolysis system (9, 10, 11). Therefore, β-catenin mutations at these sites result in the stabilization and accumulation of β-catenin protein. In the present study, β-catenin mutations were detected at the same codons, with the exception of codon 37, but no specific site was evident.

It is considered that the G/C to A/T transition is a common mutation induced by ethylating N-nitroso compounds (38). However, in this study, neither G/C to A/T nor C/G to T/A transitions of APC were found in 2 adenomas and 5 adenocarcinomas with APC mutations. Among 8 β-cateninmutations, G/C to A/T or C/G to T/A transitions were detected in 1 of 3 adenomas and 2 of 5 adenocarcinomas. The Ki-ras mutations in contrast were all G/C to A/T transitions at codon 12(4). Therefore, it seems that the Ki-rasmutations were caused by BHP per se and that APCand β-catenin mutations might have been attributable to some other factors, such as DNA damage caused by chronic oxidative stress, acting during lung carcinogenesis initiated by BHP.

The present immunohistochemical study revealed a frequent shift inβ-catenin protein localization from the cell membranes to the cytoplasm and nucleus in adenomas and adenocarcinomas but not in hyperplasias, suggesting accumulation in tumor cells. The incidence of protein localization in the cytoplasm and/or nucleus generally corresponded with a finding for APC and/or β-catenin mutations, providing supporting evidence for a causal role in the stabilization and accumulation ofβ-catenin protein within the cells. Accumulated β-catenin could interact with Tcf family members in the nucleus, resulting in acquisition of growth advantage by activation of target genes.

In human lung cancers, no mutations of APC or oncogenic activation of β-catenin have been reported(19, 20, 26). Therefore, the molecular pathways underlying human NSCLCs and BHP-induced lung tumors appear to be quite different. Additional studies are now necessary to determine the involvement of specific growth-promoting genes regulated by β-catenin protein complexes and members of the Tcf family.

We express our gratitude to Dr. Malcolm Moore for help with the English language and careful reading of the manuscript, and we also express our gratitude to Rie Maeda and Yumi Horikawa for assistance in the preparation of the manuscript.