Clinical syndromes linked to biallelic germline variants in MCM8 and MCM9

Estimation of population allele and biallelic carrier frequencies in gnomAD v.2.1.1

The gnomAD (v.2.1.1, containing 141,456 samples; https://gnomad.broadinstitute.org/) database was accessed in May 2023 to estimate the allele frequencies of MCM8/MCM9 variants across multiple populations. Predicted loss of function variants (pLoF; including splice acceptor, splice donor, frameshift, and stop gained variants) and missense variants, as classified based on the Ensembl Variant Effect Predictor (VEP) consequence⁴¹, were analyzed separately for both genes (Figure 1). Allele frequencies are presented as # of cases per 100,000 persons, unless otherwise specified.

• Download figure
• Open in new tab

Figure 1. Flow chart of study approach.

Pathogenicity-based filtering of (A) population-based cohorts, (B) our case series and cancer-specific cohorts, and (C) the TCGA Pan-Cancer atlas dataset. ^afCRCX cohort comprised 24 CRC-affected members of 16 Amsterdam-positive non-polyposis CRC families; ^bSPS cohort comprised 44 unrelated serrated polyposis families; ^cHMF cohort comprised 632 metastasized CRCs and 25 metastasized ECs. Tumors from the TCGA Pan-Cancer Atlas were selected based on the presence of somatic MCM8/MCM9 variants and are not related to germline variant carriers. ACMG/AMP, American College of Medical Genetics and Genomics; AF, allele frequency; CADD, Combined Annotation-Dependent Depletion; CRC, colorectal cancer; EC, endometrial cancer; pLoF, predicted loss of function; VUS, variant of uncertain significance.

The gnomAD (v.2.1.1, exomes only, containing 125,748 samples; https://gnomad.broadinstitute.org/) database was assessed in May 2023 to estimate the biallelic carrier frequencies of MCM8/MCM9 variants. These carriers could be either homozygous or compound heterozygous for the MCM8/MCM9 variant(s). For compound heterozygous variant carriers, we included individuals who carried two heterozygous variants in trans (on different copies of the genes), while individuals with two heterozygous variants in cis (on the same copy of the gene) were excluded from the analysis. Phasing of the variants was predicted by the gnomAD variant co-occurrence tool (gnomad.broadinstitute.org/variant-cooccurrence).

Identification of carriers and variant enrichment analysis in 200K UK Biobank and 100K Genomes Project datasets

Germline variants in MCM8 and MCM9 were identified from the 100,000 Genomes Project (project code 1142, version 17) and the 200K exomes release of the UK Biobank (project code 86977, released on November 17, 2021). Variants were annotated using Variant Effect Predictor (VEP) v.107.⁴¹ We retained missense variants with a Combined Annotation-Dependent Depletion (CADD)⁴² score ≥20 and a deleterious Condel score⁴³, as well as pLoF variants (including splice acceptor, splice donor, frameshift, and stop gained variants), provided their allele frequency was <1% in gnomAD v2.1.1 (Figure 1). The impact of variants on the canonical transcripts was reported for MCM8 (ENST00000610722.4) and MCM9 (ENST00000619706.5).

The International Classification of Diseases 10th Revision (ICD10) codes from participants’ diagnosis information (Participant Explorer in 100K Genomes Project, field ID 41270 in 200K UK Biobank), along with the International Classification of Diseases for Oncology (ICD-O) codes obtained from cancer histology and behavior fields (field ID 40011 and 40012 in 200K UK Biobank), as shown in Supplementary Table 1, were searched to identify participants with phenotypes associated with MCM8/MCM9 variants, as selected based on literature⁴⁴ as well as our case series.

We conducted case-control tests to assess whether potentially pathogenic biallelic (homozygous or compound heterozygous) MCM8/MCM9 variants were enriched in participants with the phenotypes of interest compared to a control cohort. A total of 15,091 controls were identified from the 100K Genomes Project dataset and 90,897 from the 200K UK Biobank dataset. Controls were selected based on the absence of personal or family history of common cancers and any phenotypes listed in Supplementary Table 1. To account for differences in age and ethnicity between cases and controls, association testing was performed using PLINK v1.9, adjusting for both variables. Sex was included as a covariate in all analyses, except for breast cancer, endometrial cancer, and female infertility, where only female controls were considered.

Pathogenicity-based filtering and classification of the identified MCM8/MCM9 variant carriers

The MCM8/MCM9 variants identified in the case series and cancer-specific cohorts (see next section) were filtered based on their predicted pathogenicity (Figure 1). Initially, we annotated the MCM8/MCM9 variants using the guidelines from the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) for variant interpretation^{53, 54}, along with the CADD scoring⁴² and gnomAD v.2.1.1 allele frequency, accessed through Franklin.⁵⁵ We excluded from the analysis (i) carriers of benign or likely benign MCM8/MCM9 variants and (ii) variants of uncertain significance (VUS) with a CADD score lower than 20 or a gnomAD AF higher than 1%.

Individuals with homozygous or compound heterozygous variants that met the pathogenicity-based filtering criteria were considered as biallelic carriers. Conversely, compound heterozygous carriers with one variant meeting the criteria and one that did not were categorized as monoallelic carriers. Additionally, compound heterozygous carriers with a pathogenic or likely pathogenic variant and a VUS that met the pathogenicity-based filtering criteria were included in the pathogenic or likely pathogenic group, i.e., as biallelic carriers.

DNA sequencing and bioinformatic analysis of tumors from the case series

Bioinformatic analysis of publicly available whole genome sequencing data

To further evaluate potential SBS mutational signatures associated with MCM8/MCM9 deficiency, we analyzed tumor WGS data from two publicly available sources. Firstly, we examined tumor data from cases with germline monoallelic MCM8/MCM9 variants from the HMF cancer-specific cohort, as described earlier. Additionally, we evaluated tumor data from the TCGA Pan-Cancer Atlas, accessed through cBioPortal for Cancer Genomics (https://www.cbioportal.org/) between February and April of 2023.

For the TCGA Pan-Cancer Atlas samples, tumors from any cancer type were selected based on the presence of somatic MCM8/MCM9 variant(s) that met the following criteria: (i) a tumor variant allele frequency of ≥ 20%; (ii) classified as pathogenic or likely pathogenic or as a VUS according to the ACMG/AMP recommendations for variant interpretation.^{53, 54} Additionally, variants were excluded in case they were assessed as tolerated by the Sorting Intolerant from Tolerant score⁶⁹ and deemed benign by the PolyPhen score (Figure 1).⁷⁰

In both the HMF and TCGA WGS datasets, SBS mutational signatures were identified by fitting the counts of single nucleotide variants (SNVs) per 96 tri-nucleotide context to the COSMIC v3.3 reference mutational signatures⁷¹, using the MutationalPatterns tool.⁷²

Individuals with (biallelic) germline MCM8/MCM9 variants are rare in gnomAD v.2.1.1

The occurrence of pLoF variants of MCM8 in gnomAD (v.2.1.1) was 1.4 individuals per 100,000 persons across all populations, with the highest prevalence (5.5 per 100,000 persons) in the African/African American population (Table 1). Regarding MCM9, the prevalence of a pLoF variant was 2.5 individuals per 100,000 persons across all populations, with the highest prevalence (5.7 per 100,000 persons) found in the European Finnish population. The prevalence of missense MCM8 and MCM9 variants was 462.4 and 1173.3 individuals per 100,000 persons, respectively. Twenty-three individuals (0.02%) were identified as biallelic carriers of missense variants or more severe mutations of the MCM8 gene (Table 1). With respect to the MCM9 gene, 22 (0.02%) individuals were predicted to be biallelic carriers, including 21 carriers of missense variants or worse and one carrier of a homozygous pLoF variant.

Biallelic MCM9 variant carriers in the 100K Genomes Project have an increased risk of polyposis and gastric cancer, while no enrichment was observed for biallelic MCM8 variants or in the 200K UK Biobank dataset

In the 100K Genomes Project, we identified 51 biallelic carriers (21 homozygous and 30 compound heterozygous) and 2,782 monoallelic carriers of pLoF or predicted deleterious missense variants in the MCM8 gene. Moreover, we found 64 biallelic carriers (21 homozygous and 43 compound heterozygous) and 3,166 monoallelic carriers of pLoF or predicted deleterious missense variants in the MCM9 gene. Among the 51 biallelic MCM8 variant carriers in the 100K Genomes Project, 2 individuals (3.9%) had colorectal cancer (CRC), 3 (5.9%) had colonic polyps, 3 (5.9%) had colonic adenomas, 3 (5.9%) had rectal polyps, 2 (3.9%) had hypothyroidism, and 5 (9.8%) had breast cancer. Additionally, 1 individual (2.0%) had epilepsy, 1 had endometrial cancer, 1 had short stature, and 1 experienced delayed puberty. Among the 64 biallelic MCM9 variant carriers in the 100K Genomes Project, 3 individuals (4.7%) had colorectal cancer (CRC), 2 (3.1%) had colonic polyps, 2 (3.1%) had colonic adenomas, 2 (3.1%) had rectal polyps, 3 (4.7%) had hypothyroidism, 5 (7.8%) had breast cancer, and 2 (3.1%) had epilepsy. Additionally, 1 individual (1.6%) had melanoma, 1 had gastric cancer, and 1 had endometrial cancer. While no significant enrichment of biallelic MCM8 pLoF or predicted deleterious missense variants was observed for any of these phenotypes compared to controls, we did observe significant associations between biallelic MCM9 pLoF or predicted deleterious missense variants and colonic polyps (odds ratio (OR) 6.51, 95% confidence interval (CI) 1.24-34.11; P = 0.03), rectal polyps (OR 8.40, 95% CI 1.28-55.35; P = 0.03), and gastric cancer (OR 27.03, 95% CI 2.93-248.5; P = 0.004) (Table 2).

In the 200K exomes release of the UK Biobank, we identified 110 biallelic carriers (47 homozygous and 63 compound heterozygous) and 8,453 monoallelic carriers of pLoF or predicted deleterious missense variants in the MCM8 gene. Additionally, we found 74 biallelic carriers (15 homozygous and 59 compound heterozygous) and 4,991 monoallelic carriers of pLoF or predicted deleterious missense variants in the MCM9 gene. Among the 110 biallelic MCM8 variant carriers in the 200K UK Biobank, 2 individuals (1.8%) were registered with CRC, 3 (2.7%) with colonic polyps, 4 (3.6%) with adenomas, 1 (0.9%) with female infertility, and 6 (5.5%) with hypothyroidism. Among the 74 biallelic MCM9 variant carriers in the 200K UK Biobank, 1 individual (1.4%) was registered with colorectal cancer (CRC), 3 (4%) with colonic polyps, 6 (8%) with adenomas, 1 (1.4%) with rectal polyps, and 2 (2.7%) with hypothyroidism. However, no significant enrichment of biallelic MCM8/MCM9 pLoF or predicted deleterious missense variants was observed for any of these phenotypes compared to controls in the 200K UK Biobank (Table 3).

None of the other phenotypes investigated (see Supplementary Table 1) were registered among the biallelic MCM8/MCM9 variant carriers, based on ICD10/ICD-O registrations.

Phenotype of biallelic germline MCM8/MCM9 variants carriers

In our case series, we identified 26 biallelic MCM8 variant carriers (including 15 with pathogenic or likely pathogenic variants and 11 with a VUS) and 28 biallelic MCM9 variant carriers (including 22 with pathogenic or likely pathogenic variants and 6 with a VUS) that met the pathogenicity-based filtering criteria. This group included 3 biallelic MCM8 and 4 biallelic MCM9 variant carriers who had not been previously described (Figure 1). An overview of all identified MCM8/MCM9 variant carriers, including their sources, is presented in Supplementary Table 2. A detailed description of all newly identified MCM8/MCM9 variant carriers and previously documented carriers for whom we obtained updated clinical information (individuals meeting the pathogenicity-based filtering criteria only), including pedigrees, is available from the corresponding author upon reasonable request.

Biallelic MCM8/MCM9 variant carriers often present with hypogonadism linked to impaired gonadal development

The majority of individuals with biallelic MCM8 (23 out of 26, 88%) or MCM9 (26 out of 28, 93%) variants from our case series experienced hypogonadism (HP:0000815) (Figure 2). Apart from five males (three with biallelic MCM8 variants and two with biallelic MCM9 variants) with azoospermia (no sperm in the semen; HP:0000027), these issues involved women affected by POI. Fourteen out of twenty (70%) individuals affected by POI and carrying biallelic MCM8 variants had undetectable or small ovaries coupled with an infantile or absent uterus upon ultrasound in thirteen (65%) of the cases. Among the biallelic MCM9 variant carriers affected by POI, 14 out of 23 (61%) exhibited invisible or small ovaries, and 12 out of 23 (52%) had infantile or absent uteri. Furthermore, osteoporosis or delayed bone age (HP:0000939) was reported in seven individuals with biallelic MCM9 variants and one individual with biallelic MCM8 variants, which were all affected by hypogonadism. In both the MCM8 and MCM9 groups, hypogonadism manifested at a relatively young age, typically between 10 and 30 years (Figure 3). Many of these patients were part of earlier studies, with no updated clinical data available upon request, so most were lost to follow-up post-publication.

• Download figure
• Open in new tab

Figure 2. Phenotype of biallelic MCM8/MCM9 variant carriers.

The phenotype is presented for all (A) biallelic MCM8 and (B) biallelic MCM9 variant carriers from our case series. Each column represents an individual, while each row corresponds to one of the four primary observed phenotypes: CRC, other type(s) of cancer, hypogonadism, and polyposis. Person IDs are provided below each column. CRC, colorectal cancer; VUS, variant of uncertain significance.

• Download figure
• Open in new tab

Figure 3. Disease onset in biallelic MCM8/MCM9 variant carriers.

The onset of the four primary observed phenotypes (CRC, other type(s) of cancer, hypogonadism, and polyposis) is displayed for each biallelic MCM8/MCM9 variant carrier with available age details in our case series. Those without age details were excluded from the analysis. Individuals are ordered by ACMG/AMP classification (pathogenic or likely pathogenic, VUS)^{53, 54} and current age or age at the time of death/lost to follow-up. ACMG, American College of Medical Genetics and Genomics; AMP Association for Molecular Pathology; CRC, colorectal cancer; VUS, variant of uncertain significance.

Biallelic MCM9 variant carriers may face polyposis, gastric cancer and early-onset colorectal cancer, while both biallelic MCM8/MCM9 carriers may face female germ cell tumors

Polyposis (typically > 20 polyps, including hyperplastic, adenomatous, and serrated types) was reported in 6 out of 28 (21%) biallelic MCM9 variant carriers from our case series (Figure 2). Similarly, CRC was observed in 6 of 28 (21%) biallelic MCM9 variant carriers in our case series. This includes three carriers of likely pathogenic variant(s) who developed CRC between the ages of 30 and 40, and three carriers with a variant of uncertain significance (VUS) diagnosed between 40 and 60 years (Figure 3). No CRC or polyp diagnoses were reported among the biallelic MCM8 variant carriers. Three female carriers—two with biallelic MCM8 variants and one with a biallelic MCM9 variant—were diagnosed with germ cell tumors (HP:0100728) between the ages of 11 and 15 years. These included two endodermal sinus tumors originating from dysgerminomas, which themselves arose from gonadoblastomas, and one germ cell tumor of unspecified origin. Single biallelic MCM9 variant carriers were diagnosed with gastric cancer (HP:0012126), an HPV-unrelated clear cell carcinoma of the cervix (HP:0031522), and melanoma (HP:0012056), whereas a biallelic MCM8 variant carrier was diagnosed with breast cancer (HP:0003002).

Monoallelic MCM8/MCM9 variants may experience hypogonadism

During the pathogenicity-based filtering process of our case series, we filtered 49 mono-allelic MCM8 variant carriers and 45 monoallelic MCM9 variant carriers. Out of these 49 monoallelic MCM8 variant carriers, hypogonadism was noted in 14 (29%) individuals, with two having a likely pathogenic variant and 12 carrying a VUS (Supplementary Figure 1-2). Two monoallelic MCM8 variant carriers were diagnosed with CRC, another two with polyposis, and two individuals with a monoallelic MCM8 variant were diagnosed with breast cancer.

Among the 45 monoallelic MCM9 variant carriers from our case series, ten (22%) were known to have hypogonadism, including one individual who was also diagnosed with CRC and polyposis (Supplementary Figure 1-2). CRC and polyps were additionally reported in five and six other monoallelic MCM9 variant carriers, respectively. No other types of cancer were reported in the monoallelic MCM9 group.

Genotype-phenotype correlations reveal potential hotspot sites

Mapping of variants onto the MCM8 and MCM9 protein domains revealed that the variants in our case series clustered in two key regions: the N-terminal DNA binding domain, which is crucial for protein-DNA binding (6 of 11 MCM8 variants in biallelic carriers, 55%; 4 of 20 MCM9 variants in biallelic carriers, 20%), and the AAA+ core domain, essential for DNA helicase activity (5 of 11 MCM8 variants in biallelic carriers, 45%; 12 of 20 MCM9 variants in biallelic carriers, 60%) (Figure 4, Supplementary Figure 3).⁴⁴ Additionally, several variants were found to be shared among multiple families with hypogonadism from our case series. For instance, the c.482A>C [p.(His161Pro)] VUS in the MCM8 gene, previously linked to hypogonadism^{26, 39, 46, 49, 51}, was shared by six biallelic carriers across two families. Similarly, the pathogenic c.394C>T [p.(Arg132*)] variant in the MCM9 gene, also associated with hypogonadism^{39, 49, 51}, was shared by seven biallelic carriers from four unrelated families.

• Download figure
• Open in new tab

Figure 4. Biallelic MCM8/MCM9 variants mapped onto the respective protein domains.

(A) MCM8 and (B) MCM9 variants from all biallelic variant carriers in our case series are mapped onto the domains of the MCM8 and MCM9 proteins, respectively. Each homozygote variant carrier corresponds to one diamond symbol, whereas for compound heterozygous variant carriers, both variants are separately plotted. The fill and color of the diamond symbols correspond to the phenotype of the individual (CRC, other type(s) of cancer, hypogonadism, polyposis) and the ACMG/AMP classification of the variant (pathogenic or likely pathogenic, VUS)^{53, 54}, respectively. ACMG, American College of Medical Genetics and Genomics; AMP, Association for Molecular Pathology; CRC, colorectal cancer; RF, arginine finger; VUS, variant of uncertain significance; WA, Walker A; WB, Walker B; WH, winged-helix; ZF, zinc-finger.

Clock-like mutational processes dominate in tumors from the case series and HMF cancer-specific cohort

SBS mutational signatures SBS1 and SBS5, which represent clock-like mutational processes⁷³, were present across all tumors from our case series with matched germline sequencing data available (5 out of 16 sequenced tumors, 31.3%, involving three CRCs and two breast cancers), and were similarly present in a CRC and two polyps from a wildtype control (Figure 5A-B). With respect to MMR deficiency-associated signatures, only signature SBS26 was detected and only in one tumor (P2_2T, breast cancer), where it contributed to a minority of the mutations (269 out of 2026, 13%) (Figure 5B).^64–68 Sequenced tumors without matched germline sequencing data available were dominated by sequencing artefact signatures (SBS45, SBS47, SBS50, SBS51, SBS54, SBS56, SBS58, SBS95)^64–68, precluding us from the possibility of comparing these cases to those with matched controls or previously published cases.⁴ Clock-like signatures SBS1 and SBS5, along with signatures SBS93 and SBS40 of unknown etiology, were the most prevalent signatures in metastasized CRCs from seven individuals carrying monoallelic MCM8/MCM9 variants in the HMF cancer-specific cohort (Supplementary Figure 4).

• Download figure
• Open in new tab

Figure 5. Mutational signatures analysis of tumors from MCM8/MCM9 variant carriers from our case series.

(A) Sample overview of the tumors that were available from our case series for mutational signature analysis, including the corresponding genotype, next generation sequencing (NGS) approach, the availability of normal control tissue, and the tumor type. Control tissue originated from an individual tested negative for germline MCM8/MCM9 variants. (B) The number of mutations in each signature is presented for every tumor. Mutational signature assignment was performed using SigProfilerAssignment (v0.0.32)⁶³ based on the COSMICv3.3 single base substitution (SBS) reference signatures. SBS1 and SBS5 were classified as clock-like mutational signatures, while SBS3 was considered to be caused by defective homologous recombination (HR) repair. SBS26 was linked to defective MMR, and SBS30 and SBS36 were associated with defective base excision repair (BER). SBS88 was attributed to colibactin exposure, and SBS92 to tobacco smoking. SBS37, SBS40, and SBS94 were considered to be of unknown etiology, while SBS40, SBS45, SBS50, SBS51, SBS54, SBS56, SBS58, and SBS95 were considered possible sequencing artifacts. BER, base excision repair; CRC, colorectal cancer; HR, homologous repair; MMR, mismatch repair; NGS, next generation sequencing; SBS, single base substitution; WES, whole exome sequencing; WGS, whole genome sequencing.

Somatic MCM8/MCM9 mutations may occur as a result of other DNA repair deficiencies and mutational processes, potentially involving copy number variations

In the TCGA Pan-Cancer Atlas dataset, insights into the somatic mutational behavior of MCM8 and MCM9 were gained through the observation of copy number alterations in both genes. Furthermore, unsupervised hierarchical clustering of SBS mutational signature profiles revealed clusters characterized by signatures such as SBS7a/b (UV damage), SBS2 and SBS13 (APOBEC activity), SBS6, SBS14, SBS15, SBS20, and SBS21 (MMR deficiency), and SBS10a/b (POLE deficiency)^64–68, which suggest that somatic MCM8/MCM9 variants may be secondary to other DNA repair deficiencies and mutational processes (Supplementary Figure 5).