Recurrent APC Splice Variant c.835-8A>G in Patients With Unexplained Colorectal Polyposis Fulfilling the Colibactin Mutational Signature

Abstract

Despite the clear autosomal dominant inheritance of germline APC variants causing familial adenomatous polyposis, carriers can still present with a negative family history suggesting a de novo variant. Depending on the exact temporal occurrence of the de novo variant, all or only a subset of cells in the body will be affected. Presence of a de novo variant in only a subset of cells is called mosaicism. Jansen et al1Jansen A.M. et al.Gastroenterology. 2017; 152: 546-549 e3Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar reported that APC mosaicism can be detected using next-generation sequencing in DNA isolated from formalin-fixed paraffin-embedded adenoma tissue. These variants were often not found in leukocyte DNA. APC analysis in adenomas is part of our regular diagnostics for unexplained polyposis patients. The identification of possible hotspot variants in APC will help to interpret findings suggestive of mosaicism. Does a finding of 2 colonic lesions sharing the same variant indicate mosaicism or is it coincidental? This question is considered in Jansen et al1Jansen A.M. et al.Gastroenterology. 2017; 152: 546-549 e3Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar with a patient carrying the same APC variant in 10 of 16 lesions. Formalin-fixed paraffin-embedded tissue blocks from colorectal adenomas and carcinomas were collected from 201 unexplained polyposis patients. In total, 872 colorectal lesions were sequenced using next-generation sequencing. The detected variants were categorized by pathogenicity and loss of heterozygosity was determined. A more detailed description is provided in the Supplementary Methods. In 11.9% (24 of 201) of patients, true APC mosaicism was identified, meaning the same APC variant present in all analyzed lesions. After excluding the lesions of true mosaic cases, 763 lesions remained, consisting of 61 carcinomas and 702 adenomas. In 72% of these lesions at least 1 pathogenic APC variant was detected. In carcinomas, the frequency of APC variants was 69% and in adenomas was 72%. In total, 108 APC variants occurred more than once in nonmosaic colorectal lesions. The most frequently observed APC variant, occurring in 7% of lesions, was a splice variant located in intron 8; NM_000038.5: c.835-8A>G. Two patients showed the c.835-8A>G in a true mosaic pattern. However, it was not observed in any of the normal tissues tested (n = 7; Supplementary Table 1). Moreover, in 44% of patients (16 of 36) with the c.835-8A>G variant, a subset (more than 1, ranging from 2 of 9 to 6 of 10, but not all) of lesions harbored this specific variant, a so-called hybrid mosaic pattern. Also in these patients, none of the normal tissues tested positive for the variant (n = 16). The c.835-8A>G variant was observed in both adenomas (n = 61) and carcinomas (n = 6). The majority (46 of 67 [69%]) of lesions containing the variant were located in the distal colon. Furthermore, in 54% (36 of 67) of lesions, 1 or more other pathogenic variant was detected in the APC gene, in 26 (72%) of these lesions the c.835-8A>G variant showed the highest variant allele frequency. In 28% (19 of 67), loss of heterozygosity was observed, the remaining lesions (18%) did not show any second hit. Recently, a mutational signature caused by pks+ Escherichia coli was identified.2Pleguezuelos-Manzano C. et al.Nature. 2020; 580: 269-273Crossref PubMed Scopus (350) Google Scholar,3Boot A. et al.Genome Res. 2020; 30: 803-813Crossref PubMed Scopus (17) Google Scholar This signature is characterized by single base substitutions T>N mostly in ATN and TTT context with strong enrichment of adenines 3 and 4 base pairs 5′ of the mutation site and a strong transcriptional strand bias.3Boot A. et al.Genome Res. 2020; 30: 803-813Crossref PubMed Scopus (17) Google Scholar Interestingly, the c.835-8A>G variant has a sequence context of TTAATTTTT (Figure 1A), where the underlined adenine is substituted by a guanine. Transformed in a T>N orientation (Figure 1B), the context perfectly fulfills the mutational signature caused by pks+ E coli with the hexanucleotide AAAATT as predominant sequencing context (Figure 1C). Furthermore, fulfilling the signature means that the c.835-8A>G variant is suggested to arise from adducts on the untranscribed strand and is therefore not removed by transcription-coupled nucleotide excision repair.3Boot A. et al.Genome Res. 2020; 30: 803-813Crossref PubMed Scopus (17) Google Scholar Of the other recurrent variants, 7 fulfill the pks+ E coli mutational signature (Supplementary Table 2). Remarkably, in 13 patients, ≥50% (up to 100%) of lesions carried an APC variant fulfilling the pks+ E coli mutational signature (Supplementary Table 1). In total, the majority (54 of 79 [68%]) of lesions with such a variant was located distally. Performing APC mosaicism analysis in patients with unexplained polyposis provided an opportunity to study the occurrence and frequency of pathogenic APC variants in colorectal lesions. The most frequently observed APC variant, c.835-8A>G, has been described as a germline variant twice.4Fostira F. et al.BMC Cancer. 2010; 10: 389Crossref PubMed Scopus (20) Google Scholar,5Jarry J. et al.Fam Cancer. 2011; 10: 659-665Crossref PubMed Scopus (11) Google Scholar Complementary DNA analysis showed that the variant creates a new splice acceptor site causing a frameshift leading to a premature stop codon. Although the variant is located in a region associated with classical familial adenomatous polyposis, the patient presented with a medium polyp burden, suggesting the variant to have a mild impact on the APC gene or the original splice site is still partly active, leading to some normal protein. Jarry et al5Jarry J. et al.Fam Cancer. 2011; 10: 659-665Crossref PubMed Scopus (11) Google Scholar predicted the protein change to be p.Gly279Phefs∗11. The c.835-8A>G variant has also been described somatically in 3% of sporadic colorectal cancers.6Yaeger R. et al.Cancer Cell. 2018; 33: 125-136.e3Abstract Full Text Full Text PDF PubMed Scopus (413) Google Scholar Interestingly, 45% of c.835-8A>G carcinomas did not carry a second hit. Moreover, the carcinomas exhibit nuclear β-catenin staining,6Yaeger R. et al.Cancer Cell. 2018; 33: 125-136.e3Abstract Full Text Full Text PDF PubMed Scopus (413) Google Scholar this might suggest that the splice variant provides a growth advantage to the colon crypt cell even with an intact second allele. In 2 patients in our cohort, the c.835-8A>G variant was identified in a true mosaic pattern. One mosaic patient developed adenomas at the age of 24 years and was diagnosed with ulcerative colitis. Ulcerative colitis is known to be associated with "field cancerization" in which premalignant areas in the colon share the same dysplastic changes simultaneously through repopulation of destroyed crypts.7Baker K.T. et al.Carcinogenesis. 2018; 39: 11-20PubMed Google Scholar This phenomenon might be an explanation for the detected mosaicism and development of adenomas at a young age in patients with inflammatory bowel disease. The presence of pks+ E coli, causing a specific mutational signature (Figure 1), might be an additional explanation for unexplained polyposis patients. This especially applies to the large proportion of patients carrying the c.835-8A>G variant and other pks+ E coli variants in multiple lesions. Remarkably, the pks+ E coli mutational signature seems to predominantly affect the distal colon,8Lee-Six H. et al.Nature. 2019; 574: 532-537Crossref PubMed Scopus (273) Google Scholar as confirmed by the location of lesions with pks+ E coli variants in our cohort. These findings show the relevance of further research into the presence and influence of pks+ E coli in our cohort and other unexplained polyposis patients. Collaborators: Dr Alexandra M.J. Langers, Department of Gastroenterology, Leiden University Medical Center, Leiden; Dr Carli M. Tops, Department of Clinical Genetics, Leiden University Medical Center, Leiden; Dr Dina Ruano, Department of Pathology, Leiden University Medical Center, Leiden. The authors acknowledge the work of Dr Dina Ruano in the bioinformatical support, Dr Carli M. Tops in assisting in the interpretation of the variants, and Dr Alexandra M.J. Langers in acquistion of the patients. Diantha Terlouw, MSc (Data curation: Lead; Formal analysis: Lead; Writing – original draft: Lead). Manon Suerink, MSc (Writing – review & editing: Supporting). Arnoud Boot, PhD (Writing – review & editing: Supporting). Tom van Wezel, PhD (Conceptualization: Supporting; Funding acquisition: Supporting; Supervision: Supporting; Writing – review & editing: Supporting). Maartje Nielsen, PhD (Conceptualization: Supporting; Funding acquisition: Supporting; Supervision: Supporting; Writing – review & editing: Supporting). Hans Morreau, PhD (Conceptualization: Lead; Funding acquisition: Lead; Supervision: Lead; Writing – review & editing: Lead). As part of diagnostics, patients with multiple colorectal adenomas and/or carcinomas were sent in for APC mosaicism testing either via their clinical geneticist or gastroenterologist at the Leiden University Medical Center. Of 201 patients without a germline explanation for the polyposis coli, colonic adenoma and/or carcinoma material was collected. The study protocol was approved by the local ethics committee (LUMC B18.042). The material collected comprised formalin-fixed paraffin-embedded tissue blocks and H&E slides. The H&E slides were used to examine the region of interest and determine tumor percentages (preferably >30%). When possible, formalin-fixed paraffin-embedded tissue blocks were punched to collect tumor cells. Otherwise, whenever enough lesional cells were present, complete 10-μm hematoxylin-stained sections were taken (whole section). If tumor cell percentage was too low, an inverted microscope was used to scratch the tumor cells from the sections (microdissection). DNA from collected tumor cells was isolated using the automated Tissue Preparation System, as described previously.1Jansen A.M. et al.Gastroenterology. 2017; 152: 546-549 e3Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar Next-generation sequencing of the APC gene was performed. Most (n = 538) of the lesions were sequenced for APC only, and 334 lesions were sequenced using a custom-designed mosaic colorectal cancer panel. This panel includes 20 colorectal cancer and polyposis-associated genes, but also hotspots of the CTNNB1 gene. The AmpliSeq (ThermoFisher Scientific, Waltham, MA) NGS libraries were prepared following manufacturer's instructions. In short, 2 AmpliSeq primer pools were prepared, isolated DNA was added, and a first amplification polymerase chain reaction was performed. Next, the 3 primer pools were combined, after which the primers were partly digested during a second polymerase chain reaction. To ligate the barcodes, another polymerase chain reaction run was performed. Lastly, the libraries were purified with AMPureXP beads (Beckman Coulter Life Sciences, Indianapolis, IN) and the 2 pools were combined. After loading the samples on the chip using the Ion Chef (ThermoFisher Scientific), sequencing was performed in an Ion GeneStudio S5 Series sequencer (ThermoFisher Scientific). The sequencer's output of unaligned reads was mapped against the human reference genome (hg19) using Burrows-Wheeler aligner. Multiple different softwares (VarScan, ANNOVAR, and Integrative Genomics Viewer) were used for variant calling, annotation, and visualization of the alignment and variants, respectively. Whenever variant interpretation was desired, Alamut software (Interactive Biosoftware, Louen, France) was used. Detected variants were categorized by pathogenicity: 1 = benign, 2 = likely benign, 3 = uncertain significance, 4 = likely pathogenic and 5 = pathogenic using the Leiden Open Variant Database and ClinVar. Variants were reported whenever the read count was at least 100 and variant allele frequency at least 10%. Loss of heterozygosity was determined based on allele frequencies of heterozygous single nucleotide polymorphisms and, when present, the (pathogenic) variant.Supplementary Table 1Patients With the c.835-8A>G Variant in at Least 1 LesionVariablePatientCRC age, yAdaCumulative number of adenomas.Age, ybAge at first adenoma.c.835-8A>GcNumber of lesions with c.835-8A>G variant/total number of lesions tested; number of CRC with c.835-8A>G.ndNumber of normal tissue with c.835-8A>G variant/total number of normal tissue tested.Other pks+ E coli variants, neNumber of lesions with 1 of the other 7 recurrent variants fulfilling the pks+ E coli mutational signature/total number of lesions tested (Supplementary Table 2).Total pks+ E coli variants, n fNumber of lesions with a pks+ E coli variant/total number of lesions tested. (%)Mosaic106925693/3; 1x CRC0/3 (Leu, Ur and BS)—3/3 (100)165—3253/30/4 (C, Leu, Ur and BS)—3/3 (100)Hybrid342gCRC not sequenced for APC.22692/7——2/7 (28.6)11—15503/7—3/76/7 (85.7)1266hCRC sequenced for APC but negative for the c.835-8A>G variant.13652/90/4 (C)—2/9 (22.2)15—12593/12——3/12 (25)176670663/10; 1x CRC0/2 (Leu)—3/10 (30)21—27632/10—1/103/10 (30)962x 65iOnly 1 CRC sequenced for APC but negative for the c.835-8A>G variant.4653/5——3/5 (60)1225510552/4; 1x CRC—1/43/4 (75)1254x 552552/4; 2x CRC0/4 (C)—2/4 (50)128—10526/100/5 (2x C, Leu, Ur and BS)2/108/10 (80)152—28812/60/1 (C)—2/6 (33.3)156—36742/4——2/4 (50)168—17554/7——4/7 (57.1)174—10542/6——2/6 (33.3)194—22693/7——3/7 (42.9)19869gCRC not sequenced for APC.30692/4——2/4 (50)No mosaic412x 62hCRC sequenced for APC but negative for the c.835-8A>G variant.4621/6—1/62/6 (33.3)56—22751/3——1/3 (33.3)6572gCRC not sequenced for APC.24711/3——1/3 (33.3)88—21591/6——1/6 (16.7)120—30591/4——1/4 (25)1277330741/4; 1x CRC——1/4 (25)132—15651/8—1/82/8 (25)1492x 64, 708641/3; 1x CRC0/1 (C)—1/3 (33.3)15865hCRC sequenced for APC but negative for the c.835-8A>G variant.28651/3—1/32/3 (66.7)177—16651/4——1/4 (25)1782x 69hCRC sequenced for APC but negative for the c.835-8A>G variant.>20691/4——1/4 (25)182—11551/3——1/3 (33.3)184—10371/4——1/4 (25)188—14421/70/3 (Leu, Ur, and BS) (c.4348C>T as hybrid mutation)—1/7 (14.3)190—29831/4——1/4 (25)191—23671/4—1/42/4 (50)202—14631/6——1/6 (16.7)204—9591/3—1/32/3 (66.7)BS, buccal swab; C, normal colon mucosa; CRC, colorectal cancer; Leu, leukocyte; Ur, urine.a Cumulative number of adenomas.b Age at first adenoma.c Number of lesions with c.835-8A>G variant/total number of lesions tested; number of CRC with c.835-8A>G.d Number of normal tissue with c.835-8A>G variant/total number of normal tissue tested.e Number of lesions with 1 of the other 7 recurrent variants fulfilling the pks+ E coli mutational signature/total number of lesions tested (Supplementary Table 2).f Number of lesions with a pks+ E coli variant/total number of lesions tested.g CRC not sequenced for APC.h CRC sequenced for APC but negative for the c.835-8A>G variant.i Only 1 CRC sequenced for APC but negative for the c.835-8A>G variant. Open table in a new tab Supplementary Table 2Recurring APC Variants in Nonmosaic Colorectal LesionsVariant (NM_000038.5:)n%pks+ E coli mutational signature?c.835-8A>G617.0Yesc.4348C>T515.9Noc.2626C>T343.9Noc.646C>T222.5Noc.637C>T212.4Noc.694C>T192.2Noc.847C>T151.7Noc.1495C>T121.4Noc.4393_4394delAG121.4Noc.4099C>T111.3Noc.3964G>T91.0Noc.3927_3931delAAAGA80.9Noc.1690C>T80.9Noc.2413C>T80.9Noc.3340C>T80.9Noc.994C>T60.7Noc.4391_4394delAGAG60.7Noc.4630G>T60.7Noc.3493A>T60.7Yesc.4067C>G60.7Noc.1660C>T50.6Noc.2804dupA50.6Noc.1213C>T50.6Noc.2805C>A50.6Noc.4245delT50.6Noc.4285C>T50.6Noc.6363_6365dupTGC40.5Noc.4063delT40.5Yesc.1409-5A>G40.5Noc.1168A>G40.5Noc.1600A>T40.5Yesc.3862G>T40.5Noc.3991A>T40.5Noc.4108A>T40.5Noc.4501delT40.5Yesc.4747A>G40.5Noc.7490C>T40.5Noc.904C>T40.5Noc.4390G>T30.3Noc.645+1G>A30.3Noc.1312+1G>A30.3Noc.1548G>T30.3Noc.2008A>T30.3Yesc.2821G>T30.3Noc.3441C>G30.3Noc.3916G>T30.3Noc.3934G>T30.3Noc.4053dupT30.3Noc.4057G>T30.3Noc.4063dupT30.3Noc.4233delT30.3Noc.4463delT30.3Noc.4666dupA30.3Noc.646-1G>A20.2Noc.6934C>A20.2Noc.4067C>A20.2Noc.3982C>T20.2Noc.1307delA20.2Yesc.1548G>A20.2Noc.1968_1969delAA20.2Noc.423-6A>G20.2Yesc.4460_4464delCTTTA20.2Noc.1548+2T>C20.2Noc.1960C>T20.2Noc.3175G>T20.2Noc.4128T>A20.2Noc.4189G>T20.2Noc.4459dupA20.2Noc.4611_4612delAG20.2Noc.4634C>A20.2Noc.509_512delATAG20.2Noc.591_592delAG20.2Noc.6868T>C20.2Noc.1234C>T20.2Noc.1312+5G>A20.2Noc.1333C>T20.2Noc.1411G>A20.2Noc.1744-1G>A20.2Noc.1958+1_1958+2dupGT20.2Noc.2205delG20.2Noc.2222delA20.2Noc.2364_2365delGCinsAT20.2Noc.2741_2742delGTinsAG20.2Noc.2932C>T20.2Noc.3030delT20.2Noc.3193C>T20.2Noc.3289G>T20.2Noc.3471_3474delGAGA20.2Noc.3682C>T20.2Noc.3907C>T20.2Noc.3956delC20.2Noc.4135G>T20.2Noc.4216C>T20.2Noc.4260_4261delCA20.2Noc.4271delC20.2Noc.4300delA20.2Noc.4429C>T20.2Noc.4485delT20.2Noc.4549C>T20.2Noc.4612_4613delGA20.2Noc.4655_4656delAG20.2Noc.5626A>G20.2Noc.5937delC20.2Noc.70C>T20.2Noc.757_761dupGGCTC20.2Noc.7835G>A20.2Noc.8416C>G20.2Noc.933+1G>A20.2No Open table in a new tab BS, buccal swab; C, normal colon mucosa; CRC, colorectal cancer; Leu, leukocyte; Ur, urine.

Topics & Concepts