Germline mutations in the spindle assembly checkpoint genes BUB1 and BUB3 are infrequent in familial colorectal cancer and polyposis

Much of the heritability associated with colorectal cancer (CRC) cannot be explained by the currently known genetic risk factors for CRC. Genome-wide genetic and genomic screenings have tried to identify novel high penetrance genes for CRC. Despite the identification of novel candidate genes for CRC predisposition, overall, they account for a very low number of familial cases, and for many of the suggested genes, identification of additional mutated families is essential to conclusively define their actual implication in CRC predisposition [1].

Based on genome-wide copy number profiling and exome sequencing in early-onset and familial CRC, De Voer et al. identified germline mutations in BUB1 and BUB3, components of the spindle assembly checkpoint (SAC) and thus controllers of correct chromosome segregation, as cause for CRC predisposition [2, 3]. Six germline mutations affecting BUB1 and BUB3 were identified in 6 independent families. Recently, Broderick et al. assessed the presence of germline mutations in proposed genes for CRC predisposition, including BUB1 and BUB3, finding no increased frequency of mutations in cases compared to controls in either gene [4]. In view of the controversial results, the purpose of the present study is to evaluate the impact, supported by functional studies, of BUB1 and BUB3 germline variants in the predisposition to CRC and/or polyposis.

Using a strategy that combines pooled DNA amplification and massively parallel sequencing, we sequenced BUB1 and BUB3 in 456 familial colorectal cancer cases (60 Amsterdam-positive families), and in 88 polyposis cases, without identified mutations in known high-penetrance genes. Considering novel and rare (population minor allele frequency (MAF) < 1%) non-synonymous genetic changes, a total of 4 variants, BUB1 c.1965-1G>A, BUB1 c.2296G>A (p.E766K), BUB1 c.2473C>T (p.P825S) and BUB3 c.77C>T (p.T26I), were detected in 4 families (Table 1, Fig. 1). Only BUB1 c.2473C>T was reported in public databases. Loss of heterozygosity (LOH) causing the elimination of the wild-type allele or promoter methylation was not detected in any of the tumors studied (Table 1, Additional file 1: Table S1), as previously observed [2].

BUB1 c.1965-1G>A, identified in a male patient diagnosed with CRC and 25 adenomatous polyps at age 40 with no relevant family history of cancer (Fig. 1), causes an out-of-frame deletion of 11 bases, produced by the disruption of the canonical acceptor site of exon 18 and usage of the next AG as novel splicing acceptor (r.1965_1975del; p.S655Rfs*32) (Additional file 1: Figure S1). The EBV-transformed lymphoblastoid cell line obtained from this carrier showed reduced BUB1 mRNA expression (p = 0.0020) and increased chromosome segregation errors (p = 0.0050) compared to control lymphoblastoid cell lines (Fig. 2a-b). The c.1965-1G>A cells showed reduced BUB1 levels at the kinetochores, where the SAC is active, compared to controls (30% reduction; p < 0.0001) (Fig. 2c-d).

BUB1 c.2296G>A (p.E766K), located in a β-strand outside the protein kinase domain, was identified in three CRC-affected members of an Amsterdam-positive CRC family (Fig. 1). Computational analyses predicted a destabilizing effect on the protein (Table 1). However, the EBV-transformed lymphoblastoid cell line obtained from the proband did not show reduced BUB1 expression (Fig. 2a) or reduced BUB1 levels at the kinetochores (Fig. 2c-d) when compared to controls. Assessment of chromosome segregation errors could not be performed.

BUB1 c.2473C>T (p.P825S) (rs748392521, MAF_ExAC: 0.004%), which affects a highly conserved amino acid (considering 13 species) and is located in the catalytic kinase domain of the protein, was identified in an individual diagnosed with CRC at age 44, with no immediate relatives affected with cancer but with a cousin diagnosed with CRC at age 44 whose mutation status could not be assessed (Fig. 1). The variant was predicted to affect the function and structure of the protein (Table 1, Fig. 2e and f). Sample unavailability prevented us from carrying out the specific functional studies in the patient-derived lymphoblastoid cell line.

Finally, BUB3 c.77C>T (p.T26I) was identified in patient diagnosed with prostate cancer at age 70 and two synchronous CRCs and 22 adenomatous polyps at age 73, belonging to a family fulfilling the Amsterdam I criteria. The variant, located in the WD40 repeat 1 of a seven-bladed beta-propeller fold, was predicted to be functionally deleterious and destabilizing (Table 1). The location of the mutated residue together with the previously reported BUB3 CRC-predisposing mutations is shown in Fig. 2e-f. Co-segregation results in other cancer-affected family members did not support a causal role of the variant in the aggregation of cancer in the family (Fig. 1). The in vitro functional studies could not be performed due to the unsuccessful transformation of the patient’s lymphocytes.

Cytogenetic analysis of non-transformed lymphocytes obtained from BUB1 c.1965-1G>A, BUB1 c.2296G>A (p.E766K), BUB3 c.77C>T (p.T26I) mutation carriers and from unrelated controls, did not show differences in the number of metaphases with chromosome number alterations. Notably, complex chromosomal translocations were identified in one of 41 metaphases of the BUB1 c.1965-1G>A carrier, while no chromosomal translocations were detected in any of the 91 control metaphases analyzed (Additional file 1: Table S2).

No reminiscent traits of the mosaic variegated aneuploidy syndrome or variegated aneuploidy in lymphocytes were found in any of the studied carriers, not even in the carrier of BUB1 c.1965-1G>A (p.S655Rfs*32). However, the EBV-transformed lymphocytes from the latter revealed chromosome segregation errors, which may lead to aneuploidy in the next cell cycle. It is unclear what the cell fate of the missegregated cells is, as the subsequent chromosomal instability might be detrimental. De Voer et al. identified mosaic aneuploidy in all three BUB1/BUB3 mutation carriers they studied, while only two of the three carriers showed dysmorphic features [2]. On the other hand, no aneuploidies were observed in a 54 year-old Dupuytren patient without family history of CRC and a 1.7 Mb deletion of chromosome 2q13, which includes BUB1 [5]. These findings suggest that BUB1/3 monoallelic mutations may or not cause mosaic aneuploidy and/or phenotypic affectation.

The role of BUB1 and BUB3 in familiar cancer has been topic of debate in the last years. Recently, Broderick et al. scrutinized the exomes of 863 familial/early-onset CRC cases and 1604 cancer-free controls in order to validate the proposed hereditary CRC genes, including BUB1 and BUB3 [4]. Neither the herein identified variants nor the ones identified by De Voer et al. [2] were detected in the exome study. While only one novel/rare missense variant predicted to be deleterious (none stop-gain, frameshift or splice-site variants) was identified in BUB1 in the 863 cases (0.11%), and none in BUB3, several (4 in BUB1 (0.25%) and 2 in BUB3 (0.12%)) novel/rare disruptive variants were identified in controls. Of note, the frequency of disruptive (stop-gain, frameshift and splice acceptor/donor) mutations in controls identified by Broderick et al. is remarkably higher than the frequency of disruptive mutations annotated in large population-based browsers (BUB1: ExAC, 0.065% (40 carriers in 60,703 individuals); GnomAD, 0.063% (88/138,044). BUB3: ExAC, 0.006% (4/59,569); GnomAD, 0.008% (11/123,084)) [4]. Taking into account the two largest series analyzed ([4] and present study), 0.14% (2/1392; 1/863 and 1/529, respectively) of familial and/or early-onset CRC cases would be explained by functionally relevant germline mutations in BUB1 (0% in BUB3), compared to a population frequency of 0.063–0.065%. Nevertheless, despite the demonstrated functional effect of several BUB1/BUB3 variants, their causal implication in colorectal carcinogenesis, either alone or in combination with other mutations/variants in other genes, is yet to be proven.

To date, 11 CRC families and 1 pancreatic ductal adenocarcinoma family with different novel/rare germline mutations in BUB1 or BUB3 have been reported (Additional file 1: Table S3; Fig. 2e-f) [2, 4, 6]. Of the 12 variants, 7 were proven deleterious based on functional experimental evidence and/or on their splice-site or frameshift nature, one did not show any effect, and the remaining 4 missense variants were not subjected to functional studies. Six of the 8 mutation carriers (7 families) in whom functionally relevant mutations have been identified were diagnosed with CRC at young ages (range: 29–40 years), one of them together with the presence of 25 adenomas. The increased risk to other tumor types, such as lung cancer, remains uncertain until additional functionally relevant mutation carriers with the disease are identified.

Based on the evidence gathered to date, we conclude that the paucity of functionally-relevant cancer-predisposing mutations in BUB1 and BUB3 do not support the need for germline genetic testing of these genes in familial CRC for diagnostic purposes.