The APC/β-catenin pathway in ulcerative colitis–related colorectal carcinomas
A mutational analysis
Although the APC/β-catenin pathway is known to play a crucial role in sporadic colorectal carcinogenesis, its influence on ulcerative colitis (UC)–related neoplastic progression is unknown. To elucidate the role of the APC-/β-catenin pathway in UC-related carcinogenesis, the authors identified APC and β-catenin mutations in a set of UC-related and sporadic colorectal carcinomas.
The mutational cluster region of APC (codon 1267 to 1529) and exon 3 of the β-catenin were directly sequenced.
Only 1 of 30 UC-related tumors (3%) showed an APC mutation whereas 11 of the 42 sporadic carcinomas (26%) had mutations within the mutational cluster region. Within the sporadic carcinoma group, only 8% of the right-sided carcinomas showed APC mutations whereas 50% of the left-sided carcinomas had mutations within the mutational cluster region. None of the tumors in either group showed a β-catenin mutation.
Mutations of the APC and β-catenin are rare in UC-related tumors. These genes may be altered because of mutations outside the regions studied, or by epigenetic silencing. Alternatively, other proteins involved in the APC/β-catenin signaling cascade may be altered, or this pathway may be involved infrequently in UC-related carcinogenesis. The significant difference in frequency of APC mutations between right- and left-sided sporadic tumors suggests different molecular pathways in these two tumor sites. Cancer 2002;94:1421–7. © 2002 American Cancer Society.
The APC/β-catenin pathway is known to play a crucial role in sporadic colorectal carcinogenesis. Alterations of this pathway are mostly caused by inactivation of the adenomatous polyposis coli (APC) gene, which is a frequent and early genetic event in sporadic colorectal carcinogenesis.1, 2 Most of the APC mutations (34–63%) occur within a 722–base pair (bp) region (codon 1281 to 1554) of exon 15.3–5APC mutations generally lead to truncation of the APC protein and loss of APC function as part of a protein complex targeting β-catenin for its degradation.5–7 The second APC allele often is lost, or it may be inactivated by a second mutation, or through other mechanisms, such as promoter methylation.5, 7–11 Loss of APC function results in elevated β-catenin signaling and transcription of candidate target genes such as cyclin D1 and c-myc.12–14 Mutations of the β-catenin gene itself rarely are observed in sporadic colorectal carcinomas.
The role of APC in ulcerative colitis (UC)–related carcinogenesis, however, is unclear. Although there is evidence of frequent loss of heterozygosity (LOH) at the site of the APC gene (chromosome 5q21) in UC-related carcinoma,15–17 the data regarding mutations of the APC gene in UC-related neoplasia are conflicting. The reported frequency of APC mutations in UC-related neoplasia varies from 0%18 to 6%19 and up to 50%20 in several small studies. Mutations of the β-catenin have not been found in UC-related neoplasia.21
Our previous studies have shown loss of chromosome 5q by comparative genomic hybridization (CGH) in 56% of the UC-related tumors studied,22 and abnormal APC and β-catenin protein expression in approximately 80% of UC-related carcinomas,23 indicating that the APC/β-catenin pathway may play a more important role in UC-related carcinogenesis than previously suggested.
The objective of this study was to identify mutations of the APC and β-catenin in UC-related carcinomas and compare their frequency and locations with to those in a set of stage-matched sporadic colorectal carcinomas.
Thirty-three UC-related carcinomas were identified from the pathology archives of the University of California San Francisco (15 cases) and Munich University (18 cases). All UC patients had a history of long-standing disease (>7 years), and their tumors had arisen from UC-affected mucosa. Forty-two archival sporadic carcinomas were selected from the 2 institutions (32 from UCSF, 10 from Munich) to match the stage distribution in the UC-related carcinoma group. Staging was performed according to International Union Against Cancer established criteria.24 Sporadic tumors were obtained from patients with no family history of colorectal carcinoma and no personal history of inflammatory bowel disease.
Tissue Microdissection and DNA Extraction
One representative block of tumor was selected from each case. Hematoxylin and eosin–stained sections of the selected blocks were reviewed to identify the tumor region of interest, having minimal normal cell contamination and no necrosis. This region then was dissected from an adjacent methyl green–stained section. DNA extraction was as previously described with a 3-day proteinase K treatment.22, 25, 26
Polymerase chain reaction
Sequencing of the APC gene was limited to the mutational cluster region previously reported to contain up to 70% of the known mutations.4, 8 Seven overlapping sets of primers were used, spanning exon 15, codons 1267 to 1529. Each primer pair produced a polymerase chain reaction (PCR) product between 124 and 174 base pairs (bp) in length. Primer sequences are shown in Table 1.
Table 1. Polymerase Chain Reaction Primer Pairs for APC (Exon 15) and β-Catenin (Exon 3)
|APC 1 F||5′-ACTCCAATATGTTTTTCAAGATG-3′||1267||173|
|APC 1 R||5′-GGAACTTCGCTCACAGGAT-3′|
|APC 2 F||5′-GCAGATTCTGCTAATACCCT-3′||1296||171|
|APC 2 R||5′-AACAGCTTTGTGCCTGGCT-3′|
|APC 3 F||5′-CTGCAGGGTTCTAGTTTATC-3′||1337||174|
|APC 3 R||5′-ATCAAGTGAACTGACAGAAG-3′|
|APC 4 F||5′-GACCCCACTCATGTTTAGC-3′||1380||174|
|APC 4 R||5′-TTACTTCTGCTTGGTGGCAT-3′|
|APC 5 F||5′-GATCTTCCAGATAGCCCTGG-3′||1422||124|
|APC 5 R||5′-TCTTTTCAGCAGTAGGTGCTTT-3′|
|APC 6 F||5′-AAACAGCTCAAACCAAGCGA-3′||1444||163|
|APC 6 R||5′-TCTGGAGTACTTTCCGTGG-3′|
|APC 7 F||5′-CAGAGGGTCCAGGTTCTTCC-3′||1477||161|
|APC 7 R||5′-TCCTGAACTGGAGGCATTATTC-3′|
|β-cat 1 F||5′-TTGATGGAGTTGGACATGGC-3′||7||210|
|β-cat 1 R||5′-AGTGAAGGAATGAAGAAAATCC-3′|
Sequencing of the β-catenin gene was limited to exon 3, known to encode for the protein's phosphorylation sites.27 Primer sequences are shown in Table 1.
Reaction volume was 25 μL, consisting of 0.5 ng of DNA, 0.125 μL TaqGold (0.625U; Perkin-Elmer Biosystems, Foster City, CA), 2.0 μL dNTP (200μM; Boehringer Mannheim, Indianapolis, IN), 2.5 μL 10× PCR buffer (PE Biosystems, Foster City, CA), labeled-primer pairs (PE Biosystems), MgCl2 (PE Biosystems), and dH2O. The concentration of MgCl2 for all reactions was 3.0 mM with the exception of APC 2 and β-catenin, which were at 2.5 mM. Samples were processed at 95 °C for 12 minutes and then through 35 cycles of 95 °C for 30 seconds, annealing for 30 seconds at 50 °C (for APC 2, 4, 6), 55 °C (for APC 1, 3), or 59 °C (for APC 5 and 7 and β-catenin), and then 72 °C for 90 seconds using a PE Biosystems Thermal Cycler 9700. After extension at 72 °C for 10 minutes, each reaction was left at 4 °C until removal from the cycler.
Approximately 50 ng of the PCR product was treated with 20 U of exonuclease (Amersham, Piscataway, NJ) and 4 U of shrimp alkaline phosphatase (Amersham) for 15 minutes at 37 °C, and then 15 minutes at 80 °C. Forward or reverse primer (4.5 pmoles) was added to the mix before drying under vacuum.
Each PCR product was redissolved to a final volume of 15 μL in 6 μL of Big Dye Terminator Ready Reaction Mix (PE Biosystems), and 9 μL of dH2O. Samples were heated at 95 °C for 10 minutes, then amplified through 25 cycles of 96 °C for 10 seconds, 55 °C for 5 seconds, 60 °C for 5 minutes. Each reaction was left at 4 °C until removal from the cycler. Samples were purified in a G-50 Sephadex spin column for 5 minutes at 900×g to eliminate any unincorporated nucleotides and then dried and resuspended in dye-formamide solution and loaded onto an ABI Prism 377 DNA Sequencer (Applied Biosystems, Foster City, CA). Each mutation detected was confirmed by reverse sequencing of the same product. ABI Sequencing Analysis Software Version 3.0 was used to detect heterozygotes, using a 30% cutoff value.
DNA from all of the sporadic carcinoma and 30 of the UC-related carcinomas were sequenced successfully. The amount of DNA from three of the UC-related cases was inadequate for sequencing, and these three cases were excluded from analysis.
Cell lines with known APC (SW480, SW837, Lovo, SW403) or β-catenin (HCT-116, SW48) mutations were used as positive controls, and normal colonic mucosa was used as a negative control.
The normal colonic mucosa did not show any mutations of APC or β-catenin, although a reported polymorphism at codon 1493 of APC (ACA/ACG) was observed. SW837 and Lovo both showed a CGA to TGA transition at APC codon 1450; SW480 showed a CAG to TAG transition at codon 1338 of APC. SW403, despite a reported mutation at codon 1197 of APC, did not show any mutations within the mutational cluster region. HCT-116 showed a TCT deletion at codon 45 of β-catenin, whereas SW48 showed a C to A mutation at codon 33 of β-catenin.
Comparisons between tumor groups and correlation between mutational status and clinical parameters were tested using the nonparametric Fisher exact test.
Sequencing of the APC mutational cluster region (exon 15, codons 1267 to 1529) and β-catenin (exon 3) was performed on DNA extracted from 30 UC-related and 42 stage-matched sporadic colorectal carcinoma. Clinical characteristics associated with these patients are listed in Table 2. Patients with UC-related carcinomas were younger than patients with sporadic carcinomas (average, 48 vs. 68 years). Also, the matched set of sporadic tumors showed a relatively high number of right-sided lesions compared with the UC-related tumors.
Table 2. Clinical Characteristics
|Male gender (%)||52||52|
|Age1 (yrs ± SD)||48 ± 16||68 ± 13|
|Tumor location (%)|
| Right colona||43||57|
| Left colon||30||26|
|UICC stage (%)|
Only one UC-related tumor showed a hemizygous mutation (no normal APC allele sequence seen) within the mutational cluster region of the APC gene (Table 3). It was an insertion of 5 bp at codon 1488 leading to a frameshift and introduction of a stop codon. The UC-related tumor that was found to have the somatic APC mutation was a moderately differentiated (G2), Stage III adenocarcinoma located in the ascending colon. At codon 1493, a known polymorphic site, 60% of UC-related tumors were ACA, 20% of tumors were ACG, and 20% were heterozygous for ACA/ACG. All polymorphic codons encode for threonine. None of the UC-related tumors showed a mutation within exon 3 of β-catenin.
Table 3. APC Mutations in UC-Related and Sporadic Tumors
| Case 7||righta||1488||ins TACTT||Frameshift|
| Case 8||left||1356||TCA → TAA||Stop|
| Case 10||left||1380||del CC||Frameshift|
| Case 17||right||1307||del AAAAG||Frameshift|
| Case 24||left||1337||del TG||Frameshift|
| Case 26||left||1406||CAG → TAG||Stop|
| Case 30||left||1328||CAG → TAG||Stop|
| Case 31||left||1301||ins CCCT||Frameshift|
| Case 32||left||1322||GAA → TAA||Stop|
| Case 35||left||1307||del AAAAG||Frameshift|
| Case 36||left||1291||CAG → TAG||Stop|
| Case 39||right||1450||CAG → TAG||Stop|
APC mutations were significantly more frequent in sporadic than in UC-related carcinomas (P = 0.02). A mutation within the mutational cluster region was detected in 11 sporadic tumors (26%). Nine of these tumors were located in the left, and two were located in the right colon. One case showed a 4-bp insertion, 4 cases showed deletions of 2–5 bp, and 6 cases had nonsense mutations. Two tumors showed the same mutation: a 5-bp deletion spanning codon 1307 to 1309. Seven of the mutations were hemizygous, and four were heterozygous (normal APC allele sequence observed in addition to the mutant sequence). All the mutations led to either frameshift or immediate introduction of a stop codon (Table 3). At codon 1493, a known site for polymorphisms, we found 52% of sporadic tumors with ACA, 17% of tumors with ACG, and 31% were heterozygous for ACA/ACG.
APC mutational status did not correlate with tumor stage, tumor grade, or patient age in this tumor group. However, there was a significant difference in APC mutation frequency between left- and right-sided sporadic carcinomas. The frequency of APC mutations was 50% (9 of 18) in left-sided tumors, but only 8% (2 of 24) in right-sided tumors (P < 0.01). None of the sporadic tumors showed a mutation within exon 3 of β-catenin.
This study used direct sequencing of APC and β-catenin in a set of UC-related colorectal carcinomas to elucidate the role of the APC/β-catenin pathway in these tumors. The previously reported data regarding APC in UC-related neoplasia are conflicting. Although LOH at the site of APC (chromosome 5q21) has been reported in several studies,15–17 mutation of APC appears to be a rare event in UC-related neoplasia.18, 19 There are very few data on alterations of β-catenin in UC-related carcinoma. Loss of heterozygosity at the site of β-catenin (3p22) was observed in 30% of UC-related and 30% of sporadic carcinomas,17 whereas mutations of the gene were not found in a small study of UC-related tumors.21
We previously have shown loss of chromosome 5q21 by CGH in greater than 50% of these same UC-related carcinomas.22 Abnormal APC and β-catenin protein expression was observed by immunohistochemistry in approximately 80% of these tumors.23 Cytoplasmic APC staining was reduced and was coupled with an increase in cytoplasmic and nuclear β-catenin.23 However, mutational analysis of APC and β-catenin in these 30 cancers detected only 1 mutation within the mutational cluster region of APC, and no mutations within exon 3 of β-catenin. The tumor with the APC mutation did show abnormal staining for the APC and β-catenin proteins. The mutation that was detected, an insertion of 5 bp at codon 1488, has not been reported previously to our knowledge.
The absence of mutations in APC and β-catenin in the UC-related carcinomas does not preclude the possibility that the APC/β-catenin pathway plays an important role in UC-related colorectal carcinoma carcinogenesis. Altered APC/β-catenin protein expression suggests that this pathway is important in UC-related carcinogenesis. Mutations of APC may be located outside the studied gene regions, or APC may be inactivated by other mechanisms, such as epigenetic silencing of the gene via gene promoter methylation. Also, the APC/β-catenin pathway may be deranged because of alterations of other binding partners involved in this signaling cascade (GSK-3β, TCF/Lef-1),28, 29 or because of abnormal cross-talk with other signaling pathways.30–32 That none of the UC-related carcinomas with CGH-detected 5q losses had an APC mutation raises the possibility that genes other than APC on chromosome 5q are involved in UC-related carcinogenesis.
The possibility also exists that UC-related carcinogenesis may be driven by other molecular pathways. Several investigators have suggested that mismatch repair deficiency may be relatively common in UC-related carcinomas.33–36 The role of the APC/β-catenin pathway in mismatch repair deficiency–driven colorectal carcinogenesis remains controversial. Some authors report a similar frequency of APC mutations in mismatch repair–deficient and sporadic tumors37–39; others report a much lower frequency of APC mutations in mismatch repair–deficient tumors.40, 41 We previously found high frequency microsatellite instability in only 2 of 14 (14%) UC-related tumors (a subset of the current cases).26 Therefore, mismatch repair deficiency is highly unlikely to play a major role in UC-related carcinogenesis.
Alterations of the APC/β-catenin pathway are known to play an important role in sporadic colorectal carcinogenesis.1, 2 Mutation of APC is a frequent and early event.1, 2, 5, 42 Most (34–63%) APC mutations are reported to cluster within a 722-bp region of exon 15.3, 5, 8 The APC mutations found in our set of sporadic tumors all were located at codons previously known to be altered in sporadic or hereditary colorectal carcinoma.43, 44 In agreement with other studies, a 5-bp deletion at codon 1307 to 1309 accounted for 20% of mutations.8, 45 Mutations in the β-catenin gene are rare in sporadic carcinoma, and β-catenin dysregulation appears secondary to APC inactivation.27
The striking difference in the frequency of APC mutations in right- versus left-sided sporadic tumors suggests either a different molecular pathway or a difference in the mechanism of APC inactivation between the two sites. For example, APC inactivation may be important and frequent in both left- and right-sided carcinomas, but in left-sided carcinomas, APC may be most commonly inactivated by somatic mutation, whereas in right-sided carcinomas the APC gene may be silenced through promoter methylation. Promoter methylation has been reported to be more frequent in right-sided than left-sided colorectal carcinomas.43 Alternatively, mismatch repair deficiency is known to be more frequent in right-sided tumors. Although we have not tested for microsatellite instability in this set of sporadic carcinomas, only four of these tumors showed normal profiles by CGH and therefore are more likely to be mismatch repair-deficient. Thus, as in our UC-related carcinomas, it is unlikely that mismatch repair deficiency is driving the molecular progression of more than a few of these sporadic carcinomas.
A limitation of our study was that APC sequencing was restricted to the mutation cluster region (codons 1267 to 1529). Comprehensive sequencing of the entire genomic region, extending beyond the previously defined mutational cluster region,8, 9 was not feasible because of the large size of the gene. This may have reduced the sensitivity of mutation detection in both UC-related and sporadic tumors. Our overall mutation frequency of 26% in this set of sporadic tumors is lower than expected, even with sequencing limited to the mutational cluster region. This may in part be because of the relatively high percentage of right-sided carcinomas, which had a very low frequency of APC mutations (8%). The left-sided carcinomas showed a mutation rate of 50% within the mutation cluster region, which is in agreement with other studies of sporadic colorectal carcinoma.8, 9, 46 Also, a few mutations may have been missed during sequencing because only the forward reaction was used for screening.
We did not find β-catenin mutations in any of the sporadic carcinomas, corroborating prior reports that β-catenin mutations are an infrequent event in sporadic carcinogenesis, and agreeing with the conclusion that alterations of the APC/β-catenin pathway are primarily because of inactivation of the APC gene.
In summary, we have shown that mutations of the APC gene are significantly less frequent in UC-related than in sporadic carcinomas, whereas β-catenin mutations are equally infrequent in both tumor groups. Although previous LOH, CGH, and immunohistochemistry data suggested that alterations of APC are important in UC-related carcinomas, our mutational data do not corroborate these findings. However, our data do support the importance of APC/β-catenin mutations in sporadic colorectal carcinomas. Of note, we found a significant disparity in the frequency of APC mutations between left- and right-sided sporadic tumors, suggesting different molecular progression pathways are at play at these two tumor sites.
The authors thank Dr. David Ginzinger for his advice and assistance with sequencing.