Targeted Next-Generation Sequencing Identifies a Recurrent Mutation in MCPH1 Associating with Hereditary Breast Cancer Susceptibility

Breast cancer is strongly influenced by hereditary risk factors, a majority of which still remain unknown. Here, we performed a targeted next-generation sequencing of 796 genes implicated in DNA repair in 189 Finnish breast cancer cases with indication of hereditary disease susceptibility and focused the analysis on protein truncating mutations. A recurrent heterozygous mutation (c.904_916del, p.Arg304ValfsTer3) was identified in early DNA damage response gene, MCPH1, significantly associating with breast cancer susceptibility both in familial (5/145, 3.4%, P = 0.003, OR 8.3) and unselected cases (16/1150, 1.4%, P = 0.016, OR 3.3). A total of 21 mutation positive families were identified, of which one-third exhibited also brain tumors and/or sarcomas (P = 0.0007). Mutation carriers exhibited significant increase in genomic instability assessed by cytogenetic analysis for spontaneous chromosomal rearrangements in peripheral blood lymphocytes (P = 0.0007), suggesting an effect for MCPH1 haploinsufficiency on cancer susceptibility. Furthermore, 40% of the mutation carrier tumors exhibited loss of the wild-type allele. These findings collectively provide strong evidence for MCHP1 being a novel breast cancer susceptibility gene, which warrants further investigations in other populations.


Introduction
Breast cancer is the most common malignancy among women, and the contribution of hereditary susceptibility to its development has been well recognized. Numerous breast cancer susceptibility genes are involved in the DNA damage response, strongly indicating that certain pathways of DNA repair and checkpoint control are necessary for preventing malignancy, particularly in breast epithelial cells. These susceptibility genes, including the major ones BRCA1 and BRCA2 along with PALB2, are all characterized by several rare, loss-of-function mutations, usually protein truncations. These heterozygous germline mutations predispose carrier individuals to breast, but to some extent also to ovarian cancer [1]. However, as the currently known moderate-to-high risk genes explain only 30% of the familial and 5% of the total breast cancer incidence [1][2][3], the identification of new genetic susceptibility factors and understanding of their contribution to disease onset is imperative. For this purpose we have performed a targeted next-generation sequencing of altogether 796 genes involved in diverse DNA repair signaling pathways in individuals with indication of hereditary disease susceptibility, originating from the genetically isolated Northern Finnish population. Based on their strong prior evidence for breast cancer association, the analysis was focused on protein truncating mutations, which were evaluated for cancer association by case-control comparisons. This initial sequencing revealed a deletion in the MCPH1 gene, encoding an early DNA damage response protein.
We show here that this recurrent mutation significantly associates with breast cancer susceptibility, and that MCPH1 has an integral role in the maintenance of genomic instability and acts as a tumor suppressor in breast cancer.

Results
The targeted next-generation sequencing revealed a recurrent deletion in the MCPH1 gene (also known as BRIT1, NM_024596.3: c.904_916del, rs747489687) in 3/189 of the analyzed patients, these carriers being negative for any other known breast cancer associated gene mutations. All observed PTVs and their corresponding frequency in ExAC database (http://exac. broadinstitute.org/) are summarized in S1 Table. The MCPH1 c.904_916del mutation results in a frameshift and premature translation stop (p.Arg304ValfsTer3). In total, the MCPH1 c.904_916del allele was genotyped in 1370 breast cancer cases (145 familial cases, 75 young cases diagnosed below the age of 40 years, and 1150 cases unselected for a family history of cancer or age at disease onset) and 1160 healthy geographically matched controls ( Table 1). The highest prevalence for MCPH1 c.904_916del was observed among the familial cases (5/145, 3.4%), whereas only 5 of the 1160 healthy controls (0.4%) carried the mutation (P = 0.003, OR 8.3, 95% CI 2.4-28.9). The association with breast cancer was replicated with the unselected breast cancer cohort, where 16 additional MCPH1 c.904_916del carriers were identified (16/ 1150, 1.4%, P = 0.016, OR 3.3, 95% CI 1.2-8.9). The average age at disease onset for all mutation carriers was 59 years (variation 27-90 years), which did not differ from the mean in the unselected cohort (58 years, variation 28-93 years). In line with this observation, no mutation carriers were identified in the cohort of young breast cancer cases (0/75).
MCPH1 c.904_916del resides within a common haplotype, suggesting that it is a founder mutation (S2 Table). It has also been reposited in ExAC, where its frequency in Finnish population (not stratified for any cancer phenotypes) equals to that observed in our control cohort (16/3305, 0.48% and 5/1160, 0.43%, respectively). When comparing MCPH1 c.904_916del allele's frequency in the currently analyzed breast cancer cohort to that reported in ExAC Finns, the statistical evidence for cancer association is even more significant: for familial cohort P = 0.001 (OR 7.3, 95% CI 2.7-20.3), for unselected cohort P = 0.003 (OR 2.9, 95% CI 1. 4-5.8), and for all breast cancer cases P = 0.0004 (OR 3.2 and 95% CI 1.7-6.2). Curiously, MCPH1 c.904_916del is also reported in non-Finnish cohorts in ExAC, although at much lower frequency (3/33 277 in non-Finnish Europeans, and 1/446 in "other" cohort of unknown origin).
In order to exclude the presence of any other known cancer predisposing gene mutations in MCPH1 c.904_916del carriers, all 21 index cases were analyzed using TruSight One Sequencing panel (Illumina). In 20 of the analyzed patients, MCPH1 c.904_916del was the only deleterious mutation identified, whereas one unselected case (BR-0336, diagnosed with breast cancer at the age of 79 and melanoma at the age of 82, Table 2) carried also pathogenic mutation (rs80357571) in BRCA1. In this one case, the disease predisposition can be caused by either MCPH1 or BRCA1 mutation, or by their combined effect. However, for the majority (20/21) of MCPH1 c.904_916del carriers the possibility that other known cancer predisposition mutations could explain their cancer phenotype was excluded.
MCPH1 encodes an important early DNA damage response protein. Following DNA damage, MCPH1 is rapidly recruited to nuclear foci along with MDC1, 53BP1 and γH2AX. It is responsible for the recruitment of several other integral DNA repair proteins, including ATM, NBS1, BRCA2 and RAD51 to the site of damage [4,5]. The multifunctionality of this protein is further emphasized by reports providing a role for MCPH1 as an important link between chromatin remodeling and DNA damage response. The loss of MCPH1 causes impaired chromatin relaxation, resulting from decreased association of the ATP-dependent chromatin remodeling complex SWI-SNF with chromatin. This might provide the explanation for its crucial role in the recruitment of repair proteins to DNA lesion [6]. Reflecting these functions, biallelic MCPH1 mutations result in microcephaly and premature chromosome condensation syndrome (MIM #251200). At cellular level, these patients show genomic instability and defective DNA repair [4,7,8]. Analogous to many other proteins crucial for the DNA damage response, MCPH1 contains BRCA1 COOH terminal (BRCT) [9] domains: the N-terminal BRCT is essential for the rescuing of premature chromosome condensation in MCPH1-deficient cells [10], whereas the two C-terminal BRCTs are necessary for ionizing radiation induced foci formation and DNA damage response [4,11]. The currently observed deletion abolishes both C-  terminal BRCT domains (Fig 1A), and is predicted to cause defective function in DNA damage response as modeled previously [4]. As expected, the amount of full-length protein was reduced to 52% in the studied patient-derived heterozygous MCPH1 c.904_916del lymphoblastoid cell lines ( Fig 1C). Curiously, the mutant transcript was, to some extent, able to escape nonsense-mediated decay as demonstrated by cDNA specific sequencing ( Fig 1B). In addition, a faint band marking the truncated protein of the expected 35 kDa in size was observed in immunoblotting ( Fig 1C), indicating a partial stability of the truncation product. All available information and additional DNA samples from the c.904_916del mutation carrier families were used to study the potential segregation of the mutation with cancer phenotype (Fig 2 and Table 2). In support of inherited susceptibility to the disease, 44% (7/16) of the 16 MCPH1 mutation carriers from the unselected cohort also had at least one breast and/or ovarian cancer case among their 1st and 2nd degree relatives. Besides initially studied index cases, four relatives with breast cancer were available for mutation testing. Of these three were positive for the MCPH1 c.904_916del mutation. The relatives of MCPH1 c.904_916del carriers were also reported to have several other types of malignancies, the most common being lung cancer, which occurred altogether in 38% (8/21) of all carrier families. Notably, some of these lung cancers were diagnosed at very young age (25 years) and in confirmed non-smokers. Also brain tumors and/or sarcomas were overrepresented in the carrier families (28.6%, Table 2); the difference being highly significant when compared to the incidence of these cancers in the analyzed cohorts (5.1%, P = 0.0007, OR 7.4 95% CI 2.7-19.9). One of the initially analyzed breast cancer cases was even diagnosed with two brain tumors, meningiomas. The association between breast cancer and meningiomas has previously been reported [12], suggesting shared genetic and/or environmental risk factors. The tumor spectrum of breast cancer, brain tumor, sarcoma and to some extent also lung cancer is reminiscent to that of Li-Fraumeni syndrome (LFS), caused by germline mutations in TP53, although meningiomas are only rarely reported in LFS patients [13]. The similar tumor spectrum might be explained by a recent report providing a role for MCPH1 as a regulator of p53 stability by blocking MDM2-mediated p53 ubiquitination [14]. Given the numerous functions of MCPH1 in DNA damage response, we determined the impact of c.904_916del to genomic stability assessed by chromosomal analysis [15] in nontransformed peripheral blood lymphocytes of 7 mutation carriers negative for other known breast cancer associated mutations. Samples from nine non-carrier individuals were used as controls. Based on epidemiological studies people with elevated frequency of chromosomal aberrations in their peripheral blood lymphocytes have a significantly elevated risk of developing cancer, reflecting either early biological effects of genotoxic carcinogens or individual's inherited cancer susceptibility [16]. The MCPH1 c.904_916del carriers exhibited a significantly increased frequency of spontaneous chromosomal rearrangements (P = 0.0007) when compared to controls ( Table 3 and S1 Fig). All observed chromosomal aberrations were considered random, as no preference for specific break site or evidence for clonality was observed. This suggests an effect for MCPH1 c.904_916del mutation already in the heterozygous state. The prevalence of increased incidence of chromosomal rearrangements in MCPH1 carriers is consistent with the increased genome instability caused by defective DNA damage response genes [15,17,18]. Total depletion of MCPH1 has also been reported to lead to chromosomal aberrations in mammary epithelial cells [4], and experiments in mice show that MCPH1 deficiency results in increased chromosomal aberrations and promotes long-latency tumor formation [19].
Loss of heterozygosity (LOH) analysis at the MCPH1 locus demonstrated that the wild-type allele was lost in 40% (8/20) of the studied c.904_916del carrier breast tumors (S3  [20]. Although MCPH1 resides in chromosomal region 8p23.1, frequently found aberrated in different malignancies including lung, ovarian and breast cancer [21][22][23], only one of the studied tumors had lost the mutant allele. This provides evidence for the observed LOH being directional rather than a random event. In line with this, 2/4 studied non-breast tumors (lung carcinoma and liver metastasis from gastric adenocarcinoma) of MCPH1 c.904_916del carriers had lost the wild-type allele. The MCHP1 mutation carrier breast tumors (n = 22, S3 Table) also showed a trend towards triple-negativity (i.e. ER-, PR-, HER2-; 28% of the tumors, p = 0.07), but larger cohorts would be needed to confirm any significant associations.
Based on the observation of LOH in the tumors and the role of MCPH1 in DNA DSB repair, we tested whether malignant (MCF7) mammary epithelial cells depleted in MCPH1 by Table 3. Occurrence of chromosomal aberrations in blood lymphocyte metaphases of heterozygous MCPH1 c.904_916del mutation carriers and healthy controls. transient RNA-interference are sensitive to poly (ADP-ribose) polymerase (PARP) inhibitors, similarly to BRCA1 and BRCA2 depletion [24]. Our results show only a modest sensitization to PARP inhibition and these results were replicated using MCF10A cells (S3 Fig). This indicates that the role of MCPH1 in DNA damage response differs somewhat from those directly involved in DSB repair via homologous recombination. Thus, MCPH1 mutation carriers might benefit from treatment with PARPi only when combined with some other treatments, which should be tested further.

Discussion
Current study provides strong genetic evidence for the association of MCPH1 c.904_916del heterozygosity with inherited breast cancer predisposition, and adds yet another link to human disease for the multifunctional MCPH1 protein. In the currently studied cohorts the MCPH1 c.904_916del mutation explains 3.4% of the familial and 1.4% of the unselected cases, these frequencies being comparable to that observed for Finnish PALB2 c.1592delT founder mutation [3]. Although c.904_916del allele's frequency in familial cohort is lower than the frequency of BRCA1 and BRCA2 mutations, which collectively explain 20% of the Finnish breast and ovarian cancer families [25,26], in unselected cases the combined frequency of different BRCA1/2 mutations (1.8%) [27] is roughly the same as observed for this single MCPH1 mutation. Based on case-control comparisons, the MCPH1 c.904_916del carriers have a comparable breast cancer risk to that observed for the Finnish PALB2 c.1592delT carriers, evaluated by using both unselected breast cancer (3.3-fold risk for MCPH1 and 3.9-fold risk for PALB2) and familial breast cancer cohort (8.3-fold risk for MCPH1 and 11.3-fold risk for PALB2) [3], the risk being at least moderate and potentially high. PALB2 has recently been reported to confer a breast cancer risk comparable to that of BRCA2 mutations, indicating the need to include it in the clinical diagnostics along with BRCA1 and BRCA2 mutation testing [28]. As MCPH1 is also important for brain development, demonstrated by the fact that inherited biallelic MCPH1 mutations result in microcephaly, it is of note that many of the currently identified MCPH1 c.904_916del carrier families also exhibited brain tumors. The importance of MCPH1 for prevention of malignancy, particularly in breast epithelial cells, has been indicated by previous studies reporting somatic MCPH1 downregulation in multiple breast cancer cell lines and breast cancer specimens [4,29]. MCPH1 alterations have also been reported in TCGA datasets retrieved from cBioPortal Cancer Genomics database [30], not only in breast tumors (about 7%) but also in other malignancy types observed in the currently identified carrier families (sarcomas 3.5%, brain tumors 1% and lung cancer 8%). The current genetic data brings now firm evidence for the role of MCPH1 in breast cancer prevention. The effect of MCPH1 c.904_916del mutation on the maintenance of genomic integrity and cancer predisposition is also supported by the significant increase in chromosomal rearrangements observed in the untransformed cells of the mutation carriers, and the tumor suppressor role of MCPH1 is reinforced by loss of wild-type allele in 40% of the MCPH1 c.904_916del mutation carrier breast tumors. Altogether, this data indicates that common causal molecular events can occur both in inherited forms and in non-inherited forms of breast cancer. In order to reveal the molecular details linked to this mutation and its role in cancer predisposition more detailed functional characterization using genome editing of breast epithelial cells is required. These subsequent studies on MCPH1 c.904_916del carriers could serve as a model for somatically MCPH1 deficient cancers as well.
The identification of MCPH1 as a novel breast cancer susceptibility gene further reinforces the essential involvement of compromised DNA DSB response pathway in malignancy development. Thus, targeted high depths of coverage next-generation sequencing of the genes from DSB pathway provides an excellent approach for the identification of novel rare germline mutations predisposing to this common disease. The use of founder populations provides further advantage for this: gene's contribution to hereditary cancer susceptibility is easier to prove, if many families harboring the same predisposing mutation can be identified. Even though the currently identified MCPH1 c.904_916del is evidently Finnish founder mutation, ExAC database reports it in other populations as well, although at much lower frequency, thus warranting further investigations. Given the mutation spectrum of MCPH1 gene reported in ExAC, other population specific breast cancer associated MCPH1 mutations are also likely to exist. Altogether, the current results collectively provide strong evidence for MCHP1 being a novel breast cancer susceptibility gene, which will provide further tools for the clinical risk assessment of individuals with family burden of this disease.

Materials and Methods
Targeted next-generation sequencing Agilent HaloPlex 5 Mb custom target enrichment system was used for the initial screening of germline mutations in 189 breast cancer patients with indication of hereditary disease susceptibility [n = 62 cases with young disease onset (<40y), and n = 127 cases with family history of breast or breast and ovarian cancer]. 16/189 of the sequenced patients carried known mutations in BRCA1/2, PALB2 or TP53 [3,26,31]. They were included for the validation of the mutation detection sensitivity, and also to identify potential risk modifiers. Target genes (n = 796) for the sequencing were selected among those encoding 1) proteins identified as being part of DNA repair processes using the GeneOntology database searches and STRING v.9.0 (n = 612) [32,33] and 2) novel BRCA1 and PALB2 interacting proteins identified in protein complex purification assays performed with epitope tagged versions of the proteins in HeLaS3 cells (n = 184). Gene list, genomic interval and percentage of gene coverage are presented in S1 Table. Sequencing of the enriched target regions (coding regions, splice sites and 5' and 3' UTR regions) of selected genes was performed with HiSeq2500 platform (Illumina). The sequencing resulted in mean read depth of 174x per sample for the captured region (total bait of 8.38 Mb). The minimum read depth of 7 was used as a threshold level for the variant calling procedure in the bioinformatics pipeline developed by the Finnish Institute of Molecular Medicine [34]. In total, mean of 90.5% of the captured region was covered at least by 7 reads for the analyzed samples. Annotation and characterization of variants was performed with wANNOVAR [35] and SureCall v.1.0 (Agilent), with the focus on protein truncating mutations. The manual examination and visualization of the sequence data was done using the Integrative Genomics Viewer v.2.3 [36]. Mutations were confirmed with Sanger sequencing (ABI3500xL Genetic Analyzer, Applied Biosystems).
Mutation genotyping MCPH1 c.904_916del was genotyped with High-resolution Melt (HRM) analysis (CFX96, Bio-Rad) in familial, young and unselected breast cancer cases, and in geographically matched population controls. Positive control for MCPH1 c.904_916del was included in all analyses and samples with differing melting curves were confirmed by Sanger sequencing. The familial cases were affected index cases of 145 Northern Finnish breast, or breast and ovarian cancer families, all negative for known pathogenic BRCA1/BRCA2/PALB2/TP53 mutations [3,26,31]. Of these, 98 were considered as high risk families with 1) three or more breast and/or ovarian cancers in 1st or 2nd degree relatives, or 2) two cases of breast, or breast and ovarian cancer in 1st or 2nd degree relatives, of which at least one with early disease onset (<35 years), bilateral breast cancer, or multiple primary tumors including breast or ovarian cancer in the same individual. The remaining 47 families were indicative of moderate disease susceptibility, and had two cases of breast cancer in 1st or 2nd degree relatives, of which at least the other was diagnosed below the age of 50. Young breast cancer cohort consisted of 75 Northern Finnish patients, diagnosed with breast cancer at or below the age of 40 (median 38, variation 25-40 years. These patients were unselected for a family history of the disease, and negative for known pathogenic BRCA1/ BRCA2/PALB2 mutations [37]. The unselected cohort consisted of 1150 consecutive breast cancer cases operated at the Oulu University Hospital during 2000-2013. They were unselected for a family history of cancer or age at disease onset. 1160 healthy geographically matched anonymous Northern Finnish Red Cross blood donors (n = 704 females and n = 456 males) were used as population controls. Only the health status at the time of blood donation was known in addition to birth year and gender. All the analyzed patients and healthy controls were geographically matched, and have given their informed consent. Permission to use the above mentioned patient and control materials for the study has been obtained from the Finnish Ministry of Social Affairs and Health (Dnr 46/07/98), and the Ethical Committee of the Northern-Ostrobothnia Health Care District (Dnr 88/2000+amendment).

TruSight One Sequencing panel
All identified MCPH1 mutation carrier index cases were analyzed for the prevalence of additional deleterious germline mutations using TruSight One Sequencing panel. Sequencing of the 4813 genes included in the panel (including 94 genes from TruSight Cancer suspected to play a role in cancer predisposition, but also other important genes such as RAD50 and currently studied MCPH1) was performed with NextSeq550 platform (Illumina). Sequencing resulted in mean read depth of 183x per sample for the captured region (12 Mb). In total, mean of 98.5% of the captured region was covered at least by 10 reads and 92.1% at least by 50 reads for the analyzed samples. Within BaseSpace Onsite Genomics computing environment (Illumina), BWA Enrichment (BWA Genome Aligner Software and the GATK Variant Caller) was used for sequence alignment and variant calling, Illumina VariantStudio for annotation, filtering and classification of the variants, and Intergrative Genomics Viewer for data visualization.

Loss-of-heterozygosity analysis of carrier tumors
Genomic DNA was extracted from formalin-fixed, paraffin-embedded tumors of MCPH1 c.904_916del carriers. Loss-of-heterozygosity (LOH) was studied by sequencing of 198 bp segment flanking the mutation site. The comparison of allelic ratios in tumor and corresponding normal DNA was based on the exact peak height values from the sequence chromatograms (ab1Peak Reporter Tool, Applied Biosystems). Allelic imbalance (AI) value was calculated for each sample pair (tumor vs. normal), and those showing AI (values >1.67 or <0.60) were considered as having LOH [38].

Chromosomal analysis
Chromosomal analysis of 7 MCPH1 c.904_916del carriers and 9 wild-type controls was carried out on metaphases obtained from regular short-term 3-day cultures of peripheral blood T-lymphocytes [15]. The blood samples of the patients selected for chromosomal analysis were collected at least 5 years after the initial breast cancer diagnosis and received treatment. The controls used were healthy, age-matched female individuals. A minimum of 50 Giemsa-banded metaphases for each sample were evaluated by light microscopy and photographed with an automatic chromosome analyzer (CytoVision version 7.2, Applied Imaging). Chromosomal aberrations were divided into five classes: 1) telomeric associations, 2) chromatid/chromosome breaks and deletions, 3) simple chromosomal rearrangements, 4) complex chromosomal rearrangements, and 5) total rearrangements.

Sequencing of cDNA
Total mRNA was isolated from the LCLs using RNeasy Mini Kit (Qiagen) and reverse transcribed with iScript cDNA synthesis Kit (Bio-Rad). The expression of MCHP1 wild-type and mutant allele was studied the by cDNA specific sequencing.

Immunoblotting analysis
For immunoblotting of MCPH1, protein lysates from LCLs of 3 mutation carriers and 3 noncarrier controls were extracted in NETN300 buffer, analyzed by SDS-PAGE using 7.5% mini-PROTEAN TGX gel (Bio-Rad) and transferred onto Immobilon-P PVDF membrane (Millipore). The primary MCPH1 antibody (R&D systems, AF3998) was a goat polyclonal antibody raised against the amino acids 1-250 mapping to the N-terminus of MCPH1 of human origin. Horseradish peroxidase-conjugated antibody (Jackson ImmunoResearch Laboratories, 305-035-045) was used as a secondary antibody. Visualization was done using SuperSignal West Femto chemiluminescent substrate (Thermo Scientific). Results were repeated in three independent experiments with variable loading order of the samples.

Statistical analyses
Carrier frequencies between cases and controls were compared using Fisher´s exact or Pearson's Chi-Square test. Mann-Whitney U-test was used to determine the difference in the number of different chromosomal aberrations between MCPH1 c.904_916del carriers and controls (IBM SPSS Statistics 20.0 for Windows, SPSS Inc.). All P-values were two-sided.