Article Text

Original article
Minichromosome maintenance complex component 8 (MCM8) gene mutations result in primary gonadal failure
  1. Yardena Tenenbaum-Rakover1,2,
  2. Ariella Weinberg-Shukron3,4,
  3. Paul Renbaum3,
  4. Orit Lobel3,4,
  5. Hasan Eideh5,
  6. Suleyman Gulsuner6,
  7. Dvir Dahary7,
  8. Amal Abu-Rayyan8,
  9. Moien Kanaan8,
  10. Ephrat Levy-Lahad3,4,
  11. Dani Bercovich9,
  12. David Zangen10
  1. 1Pediatric Endocrine Unit, Ha'Emek Medical Center, Afula, Israel
  2. 2The Rappaport Faculty of Medicine, Technion, Haifa, Israel
  3. 3Medical Genetics Institute, Shaare Zedek Medical Center, Jerusalem, Israel
  4. 4Hebrew University Medical School, Jerusalem, Israel
  5. 5Palestinian Medical Complex, Ramallah, USA
  6. 6Departments of Medicine (Medical Genetics) and Genome Sciences, University of Washington, Seattle, Washington, USA
  7. 7Toldot Genetics Ltd., Hod Hasharon, Israel
  8. 8Hereditary Research Laboratory, Bethlehem University, Bethlehem, Palestine
  9. 9Tel Hai College and GGA (Galilee Genetic Analysis lab), Tel Hai, Israel
  10. 10Division of Pediatric Endocrinology, Hadassah Hebrew University Medical Center, Jerusalem, Israel
  1. Correspondence to Dr Yardena Tenenbaum-Rakover, Pediatric Endocrine Unit, Ha'Emek Medical Center, Afula 18101, Israel; rakover_y{at}


Background Primary gonadal failure is characterised by primary amenorrhoea or early menopause in females, and oligospermia or azoospermia in males. Variants of the minichromosome maintenance complex component 8 gene (MCM8) have recently been shown to be significantly associated with women's menopausal age in genome-wide association studies. Furthermore, MCM8-knockout mice are sterile. The objective of this study was to elucidate the genetic aetiology of gonadal failure in two consanguineous families presenting as primary amenorrhoea in the females and as small testes and azoospermia in a male.

Methods and results Using whole exome sequencing, we identified two novel homozygous mutations in the MCM8 gene: a splice (c.1954-1G>A) and a frameshift (c.1469-1470insTA). In each consanguineous family the mutation segregated with the disease and both mutations were absent in 100 ethnically matched controls. The splice mutation led to lack of the wild-type transcript and three different aberrant transcripts predicted to result in either truncated or significantly shorter proteins. Quantitative analysis of the aberrantly spliced transcripts showed a significant decrease in total MCM8 message in affected homozygotes for the mutation, and an intermediate decrease in heterozygous family members. Chromosomal breakage following exposure to mitomcyin C was significantly increased in cells from homozygous individuals for c.1954-1G>A, as well as c.1469-1470insTA.

Conclusions MCM8, a component of the pre-replication complex, is crucial for gonadal development and maintenance in humans—both males and females. These findings provide new insights into the genetic disorders of infertility and premature menopause in women.

  • Chromosomal
  • Clinical genetics
  • Endocrinology
  • Genetics

Statistics from


Fetal sex differentiation occurs via multiple genes that affect the development of germ cells, their migration to the urogenital ridge, and the differentiation and formation of testis or ovary from the bipotential gonad.1–3 Studies in knockout mouse models have indicated that following the appropriate formation and migration of germ cells, several transcription factors, such as empty spiracles homeobox 2 (Emx2), Wilms tumour 1 (WT1), and Lim homeobox 9 (Lhx9), enable formation of the urogenital ridge from the intermediate mesoderm and, consequently, bipotential gonad development. The bipotenial gonad differentiates into testes in the presence of sex region Y (SRY). Differentiation into ovaries occurs when the SRY domain is lacking,2 but discovery of the roles of WNT4 (MIM 3603490), R-spondin 1 (RSPO1 (MIM 609595)), forkhead transcription factor L2 (FOXL2 (MIM 305597)), and β catenin has suggested that ovarian development is an active process rather than just a default pathway.2 ,4 Clinically, primary gonadal insufficiency is characterised by high serum gonadotropin concentrations and a lack of spontaneous pubertal development, associated in females with primary or secondary amenorrhoea, uterine hypoplasia, infertility and early menopause. In males, testicular insufficiency is associated with underdeveloped testes, oligospermia or azoospermia. Primary ovarian insufficiency (POI) results from either ovarian dysgenesis with depletion of the primordial follicular pool, or diminished size of the pool due to accelerated atresia. To date, a genetic cause has been determined in only a few cases of POI:5 the most common one involves X chromosome abnormalities as in Turner syndrome, or X-linked genes, in particular fragile X mental retardation 1 (FMR1 (MIM 309550)) and bone morphogenetic protein 15 (BMP15 (MIM 300247)).5 Additional genes have been identified in the last few years, including: FOXL2, the follicle stimulating hormone (FSH) receptor (FSHR (MIM 136435)), newborn ovary homeobox (NOBOX (MIM 610934)), nuclear receptor subfamily (NROB1/DAX1 (MIM 300473)),6 steroidogenic factor 1 (SF1/NR5A1 (MIM 184757)),7 PSMC3IP (MIM 608665),8 and very recently, the STAG3 gene,9 but the transcriptional cascade of ovarian development is still poorly understood. Genome-wide association studies (GWAS) aimed at identifying factors that determine the length of the reproductive lifespan and the aetiology of premature ovarian failure have revealed a few specific single nucleotide polymorphisms (SNPs) within genes (eg, SYCP2L, UIMCI, and MCM8)10 that are significantly associated with age of menopause. Interestingly, only a few families have been reported in which both genders are affected with primary gonadal failure, but their genetic aetiology remains unexplained in most cases.

In the present study, using whole exome sequencing (WES), we describe two novel mutations in the minichromosome maintenance complex component 8 gene (MCM8), leading to primary gonadal failure in humans, both males and females. Both the splice mutation, c.1954-1G>A, and a frameshift mutation, c.1469-1470insTA, compromise MCM8 function and lead to chromosomal instability.


Further information can be found in the online supplementary methods. This study was approved by Shaare Zedek and Ha'Emek Medical Centers Institutional Review Board and the National Helsinki Committee for Genetic Studies. Informed consent and blood samples were obtained from the study participants and 100 unrelated healthy controls of Arab origin.

Genetic studies

Genomic DNA was extracted from peripheral blood mononuclear cells. Chromosome microarray analysis (CMA) was performed on genomic DNA extracted from peripheral blood mononuclear cells in families A and B. Affymetrix cytogenetic 2.7 M arrays were used with the Human Mapping Cytogenetic 2.7 M Assay Kit according to the manufacturer's standard protocol. Results were analysed with the Affymetrix Chromosome Analysis Suite (genome build 37) V.31NA.

For homozygosity mapping, DNA samples were genotyped with the Affymetrix Gene Chip 250 K/750K Nsp SNP array. SNP data were analysed using KinSNP11 and examined for informative genomic regions >2 Mb that were homozygous and shared among affected individuals, but not by their unaffected siblings. WES in family A was used to search for candidate variants in the affected female proband (figure 1A, V-1). WES in family B was used to search for shared candidate variants in the three affected sisters (figure 1B, IV-2, IV-3 and IV-4). Genomic DNA libraries were created following standard protocols. Libraries were enriched using Nextera Exome enrichment kit (62 Mbp; Illumina) and sequenced on HiSeq2500 (2×170 bp PE run) at a 83× mean coverage. Reads were aligned to the human genome (hg19; NCBI build 37; February 2009). (Exome filtering criteria for both families are detailed in table 1, see online supplementary tables S1 and S2). In family A, Sanger sequencing was used to validate all coding and splicing variants that were homozygous, had a positive logarithm of odds (LOD) score, and were absent or found at very low frequencies in dbSNP, Exome variant server ( and 1000 Genomes (see online supplementary table S1). In family B, Sanger sequencing was performed for the MCM8 c.1469-1470insTA mutation in all available samples (figure 1B, D). PCR products were sequenced using BigDye Terminator V.3.1. Sequence analysis was performed on an Applied Biosystems 3130xl Genetic Analyzer. Both MCM8 c.1954-1G>A and MCM8 c.1469-1470insTA were analysed in 100 Arab controls by either PstI digestion or a Gene-Scan assay, respectively. The Gene-Scan PCR products were analysed on the 3130xl Genetic Analyzer with GeneScan 500 ROX size standards. Results were analysed using GeneScan analysis software.

Table 1

Next generation whole exome data filtering criteria

Figure 1

Family pedigrees and MCM8 mutation sequencing. (A) Pedigree of family A. Affected siblings V-1 and V-2 are homozygous for the MCM8 c.1954-1G>A mutation (V- variant, N-normal). The parents (IV-1 and IV-2), and the healthy brother V-3 are heterozygous carriers. (B) Pedigree of family B. Affected siblings IV-2, 3 and 4 and affected cousins IV-6 and 7 are homozygous for the MCM8 c.1469-1470insTA mutation. The parents (III-1 and III-2), healthy aunt III-3 and a healthy brother (IV-5) are heterozygous. An older healthy sister (IV-1) is wild type. (C) Sanger sequencing in family A of the genomic MCM8 c.1954-1G>A mutation at chr20:5966317, showing the mutation at the acceptor splice site between intron 14 and exon 15. First row: sequence from an unrelated wild-type (WT) individual, with the WT G allele at the splice site (surrounded by a black square). Second row: sequence from a carrier parent, harbouring both the WT G allele and the mutant A allele (R, surrounded by a black square). Third row: sequence from an affected individual, harbouring the mutated base A on both alleles at the splice site (surrounded by a black square). Intron–exon junction is marked by a black vertical line at the right of the mutated base. (D) Sanger sequencing in family B of the genomic MCM8 c.1469-1470insTA mutation at chr20: 5958595, showing the 2 bp insertion and the frameshift it causes. First row: sequence from an unrelated WT individual. WT amino acid codons are marked above the WT sequence. Second row: sequence from a carrier parent, harbouring both the WT allele and the mutant 2 bp insertion allele. Both the WT amino acid codons and the mutated codons that are generated by the insertion allele are marked above the sequence. Third row: sequence from an affected individual, harbouring the mutated 2 bp insertion on both alleles (the inserted 2 bp are surrounded by a black square). The new amino acid codons which are generated by the insertion allele are marked above the sequence.

RNA and cDNA studies

RNA was extracted from peripheral blood leucocytes drawn into Tempus Blood RNA Tubes using the Applied Biosystems Tempus Blood RNA kit. RNA was reverse-transcribed using random hexamers in the presence of RNase inhibitor (rRNasin, Promega). In family A, cDNA was amplified by PCR with primers surrounding the MCM8 mutation located in intron 14 at c.1954-1 (see online supplementary table S3). PCR products were sequenced using BigDye Terminator V.3.1. Quantitative real-time PCR analysis (qRT-PCR) was performed using primers spanning exon 7 and exon 8 of MCM8 (see online supplementary table S3). Reactions were performed using Power SYBR master mix (Applied Biosystems) on the ABI PRISM 7000/7900 Sequence Detector (Applied Biosystems). MCM8 threshold cycle (Ct) values were normalised to the Ct values of the housekeeping gene GAPDH in each relevant sample. Relative mRNA levels were quantified using the comparative method (ABI PRISM 7700 Sequence Detection System 1997) and calculated as 2−ΔCt. Experiments were performed at least three times on all family members and healthy controls. MCM8 expression in ovarian tissue was assessed by performing PCR reactions surrounding exons 12 and 15 of MCM8 in an IGROV-1 ovarian carcinoma cell line.

Chromosomal instability studies

We assayed DNA repair capabilities of lymphocytes cultured in peripheral blood karyotyping medium with phytohaemagglutinin (cat# 01-201-1, Biological Industries Ltd, Israel) in the presence of mitomycin C (MMC) from Streptomyces caespitosus (cat# m0503-2 mg, Sigma) at increasing concentrations: 0, 150 and 300 nM.12 Cells were harvested after 72 h of incubation at 37°C and dropped onto microscope slides. At least 10 metaphase spreads per sample were evaluated under the microscope for aberrations. Two-tailed t test assuming equal variance was used to compare the two cell lines at each concentration.


Clinical presentations

Family A

Two siblings from a highly consanguineous Arab family presented with gonadal failure. The oldest sister (V-1; figure 1A) was completely prepubertal with primary amenorrhoea at 15 years of age. Her karyotype was 46,XX. Her gonadotropin values were elevated, consistent with primary ovarian failure (table 2). At her last visit, at 22 years of age, her pubertal development was at Tanner stage B2 and P5. Her 21-year-old brother (V-2; figure 1A), born small for gestational age (weight 1900 g), had neonatal hypocalcaemic seizures (treated with calcium and vitamin D), mild ventricular septal defect (VSD), right aortic arch and T cell derived immune deficiency, and was diagnosed with DiGeorge syndrome (22q11 microdeletion) by fluorescence in situ hybridisation (FISH). His karyotype was 46,XY. At the age of 17 years and at his last visit at 21 years of age, his pubertal development was at Tanner stage 5 for pubic hair and penile development but his testicular volume was only 3 mL, as confirmed by ultrasound. High basal and GnRH-stimulated gonadotropins and azoospermia consistent with primary testicular failure (table 2). The father underwent delayed puberty, and, in addition, at the age of 49 years, his serum FSH concentrations were high at 21.2 mIU/L (normal male range 1.5–12.4 mIU/L), with normal luteinising hormone (LH) and testosterone concentrations. The mother had delayed menarche at the age of 15 years but ever since then has had regular menses and a normal hormonal profile (age 40 years). Both parents were negative for the 22q11 microdeletion. There were two additional siblings: a 10-year-old brother with a small VSD but otherwise no somatic phenotype and a normal prepubertal gonadotropin response to GnRH stimulation; and a 14-year-old brother (V-3, figure 1A) with a normal age related hormonal profile. No known history of malignancies was reported in the extended family.

Table 2

Summary of clinical, hormonal and imaging results

Family B

Three sisters from a highly consanguineous Arab family (figure 1B, not related to family A) presented consecutively at the age of 14.5–15 years with delayed puberty, primary amenorrhoea, hypergonadotrophic hypogonadism, absence of ovaries, and a small uterus at both ultrasound and MRI imaging (table 2). Apart from delayed pubertal signs, the medical history and physical examination were unremarkable. The karyotype was 46,XX and SRY testing was negative. They responded well to oestrogen and progesterone replacement therapy, achieved normal height and pubertal development (Tanner stage 5 for breast and pubic hair), and experienced regular menstrual cycles. The parents and an older healthy sister had normal pubertal development and a normal hormonal profile. Following the diagnosis of POI in the three sisters, two paternal female cousins aged 30 and 28 years were diagnosed with primary hypergonadotrophic hypogonadism. No known history of malignancies was reported in the extended family.

Copy number variation analysis

We first performed CMA analysis on both affected siblings from family A and on the three affected sisters from family B (figure 1A, cases V-1 and V-2, and figure 1B, cases IV-2, IV-3 and IV-4); as expected, in patient V-2 from family A, who was previously diagnosed with DiGeorge syndrome, we identified a 2.724 Mb deletion in 22q11.21 (chr22:19 040 100–21 457 100, genes: DGCR11, SERPIND1, ARVCF, COMT, PI4KAP1, UFD1L, FLJ39582, RIMBP3B, ZNF74, and SLC7A4). He also had a 427 kb duplication in chromosome 6q27 (chr6:170 300 320–170 827 320, genes: FAM120B, TBP, PDCD2, DLL1, LOC154449, and PSMB1). His non-gonadal phenotype is consistent with the 22q deletion, whereas genes in the duplicated 6q27 region have not been reported to cause similar features or gonadal effects. Furthermore, these genomic rearrangements were not found in his sister, V-1, whose CMA revealed no significant duplications or deletions. No significant copy number variations (CNVs) were observed in family B.

Homozygosity mapping

Homozygosity mapping in family B identified approximately 32 Mb of homozygous regions, including two large regions on chromosomes 17 and 18, but no candidate genes in these regions were related to gonadal dysgenesis.

Whole exome sequencing

Family A

WES was performed on genomic DNA from the female proband in family A (figure 1A, V-1). Applying the filters indicated above, 13 homozygous coding variants with a quality score >10, LOD score >0, and frequency ≤1% were identified (table 1). Sanger sequencing confirmed that the proband (V-1) was indeed homozygous for 12/13 variants, but her affected brother (V-2) was either heterozygous or wild type for 12 of these variants. Only a novel homozygous MCM8 IVS14-1G>A (c.1954-1, G>A) variant at the canonical acceptor splice site of intron 14 (genomic position chr20:5966567 G>A dbSNP ID rs138761187, NM_001281520.1, NP_001268449) segregated with the disease in family A (see online supplementary table S1). Both affected siblings were homozygous for this mutation, whereas the parents (IV-1 and IV-2) and the healthy brother (V-3) were heterozygous (figure 1A, C). The MCM8 c.1954-1 G>A splice site mutation (rs138761187) has been reported, in the heterozygote state, in 1/8599 (0.012%) European Americans and in 0/4066 African Americans, in the Exome Variant Server (National Heart, Lung, and Blood Institute (NHLBI) Exome Sequencing Project ( and in 1/4545 (0.022%) in the 1000 Genomes project. No homozygotes are reported in any database, and it was not identified in 100 ethnically matched healthy control subjects.

Family B

WES was performed on genomic DNA from three affected females (figure 1B, IV-2, 3 and 4), one healthy sister (IV-1), and the parents (III-1 and III-2). Only a novel homozygous MCM8 c.1469-1470insTA 2-bp frameshift mutation (genomic position chr20:5958595 insTA, NM_001281520.1, NP_001268449) segregated with the disease in this family (table 1). All affected individuals were homozygous for the MCM8 c.1469-1470insTA mutation (IV-2, 3, 4, 6 and 7), whereas the parents (III-1 and III-2), a healthy unaffected aunt (III-3), and the unaffected brother (IV-5) were heterozygous. The unaffected sister (IV-1) was wild type (figure 1B, D). Although the probable inheritance model in this family is homozygosity for an autosomal recessive allele, we searched for both compound heterozygous and de novo heterozygous mutations in the affected sisters and no mutations were found to fit these models (see online supplementary table S2). The frameshift c.1469-1470insTA mutation is predicted to generate a premature stop codon 88 amino acids downstream, p.Leu491Ilefs*88. This variant has not been reported in any database and was absent in 100 ethnically matched healthy control subjects.

MCM8 c.1954-1 G>A cDNA analysis

To confirm that the MCM8 c.1954-1 G>A mutation affects splicing, MCM8 cDNA was amplified and sequenced in all sampled individuals from family A (figure 2A). Whereas in normal controls there was a single transcript of the expected length, in the affected homozygotes (family A, V-1 and V-2) there were three different MCM8 transcripts: one of seemingly similar size to the wild type, and two shorter transcripts which were not present in normal controls (figure 2B). Sequencing revealed that all three MCM8 transcripts produced from the mutant allele differ from the wild type, either by usage of alternative splice sites in exon 15 or by skipping exon 15 entirely (figure 2C): (1) the largest transcript (figure 2A, transcript b) is only 1 bp shorter than the wild type, based on an alternative splice site 1 nucleotide into exon 15—this results in a frameshift leading to a premature stop codon after six alternative codons, p.Val652Trpfs*6; (2) the second transcript (figure 2A, transcript c) is the product of an alternative splice site 39 nucleotides into exon 15, predicted to delete the first 13 amino acids in exon 15 (in frame), leaving the rest of the protein intact, p.(Val652_Gln664del); (3) the smallest transcript (figure 2A, transcript d) is produced by complete skipping of exon 15, predicted to delete (in frame) 70 amino acids from the protein, p.(Val652_Glu721del).

Figure 2

cDNA analysis of family 1. (A) MCM8 gDNA: Schematic representation of wild-type (WT) and c.1954-1G>A-associated transcripts in the MCM8 gene. (A) Normally spliced WT exon 15, resulting in a 358 bp PCR product. Transcripts observed with the c.1954-1G>A mutation; (B) alternative exon 15 acceptor splice site at +1 bp into exon 15; (C) alternative exon 15 acceptor splice site at +39 bp into exon 15; (D) skipping of exon 15. MCM8 protein: Schematic representation of the MCM8 protein and its known domains. Marked by a purple oval is the MCM domain, typical of all proteins in the MCM family, which resides between amino acids 402 and 609. Exons 14 to 16 are marked by blue squares and the dashed lines mark the regions in the protein that are translated from these exons. (B) Alternative splice products observed using exon 14/16 PCR primers (as described in A) viewed on a 3% agarose gel. Lane 1, wild type; lane 2, heterozygous c.1954-1G>G/A; lane 3, homozygous c.1954-1G>A. Transcripts a (wild type) and b (del 1 bp) are indistinguishable on this gel. (C) Sequencing of alternative splice transcripts purified from gel shown in panel B. The expected amino acid sequence of the aberrant splicing in the mutant transcripts is shown above the chromatograms. Transcript b: premature protein truncated after six alternative codons; transcript c: deletion of 13 amino acids; transcript d: deletion of 70 amino acids as a result of exon 15 skipping. (D) Semi-quantitative analysis of WT and mutant MCM8 transcripts. Shown are the cDNA of the c.1954-1G>A homozygote (Hom), the heterozygous (Het) parents, and the WT individuals, amplified with exon 14/16 primers. The homozygous mutant sample does not contain any WT transcript (358 bp). Percentage of each mutant transcript in the homozygous sample is indicated. (E) Quantitative real-time PCR analysis of MCM8 mRNA in family A. Homozygous MCM8 c.1954-1G>A samples contain threefold less mRNA than the WT samples, including no WT allele. Heterozygous values are intermediate. The decrease in mRNA quantity in homozygous MCM8 c.1954-1G>A samples is significant compared with both WT and heterozygous values. aIndicates significant p value in comparison between WT and heterozygous levels; bindicates significant p value in comparison between heterozygous and homozygous levels; and cindicates significant p value in comparison between WT and homozygous levels. SD is represented as error bars. SEM for homozygous and heterozygous values is 0.05 and 0.1, respectively. Experiments were performed at least three times, each time on two homozygous individuals, three heterozygous family members (two parents and one healthy brother), and three unrelated WT samples from the same ethnic background.

Semi-quantitative analysis of c.1954-1 G>A cDNA transcripts

To determine the relative amounts of the alternative transcripts derived from the mutant MCM8 c.1954-1 G>A allele, we performed semi-quantitative analysis of cDNA from homozygous compared to heterozygous and wild-type (WT) samples in family A (figure 2D). The main alternative transcript in homozygous individuals was the one lacking exon 15 due to complete skipping of this exon (p.(Val652_Glu721del), transcript d, 55%). The two other transcripts—alternative splice site 1 bp into exon 15 (p.Val652Trpfs*6, transcript b) and alternative splice site 39 bp into exon 1 (p.(Val652_Gln664del), transcript c)—accounted for 35% and 10% of the mutant MCM8 c.1954-1,G>A alleles, respectively. No WT cDNA was identified in c.1954-1 G>A homozygotes.

qRT-PCR analysis of c.1954-1 G>A mRNA

To determine whether the MCM8 c.1954-1 G>A splice mutation affects the total amount of MCM8 mRNA and quantify this effect, we performed qRT-PCR using primers spanning exons 7 and 8 of MCM8 (see online supplementary table S3), which are present in all alternatively spliced MCM8 transcripts and are located upstream of the splice mutation we identified. Experiments were performed at least three times, each time on both homozygous individuals from family A, three heterozygous family members (two parents and one healthy brother), and three unrelated wild-type samples from the same ethnic background. We found a significant (p=0.002) threefold decrease in the quantity of MCM8 mRNA in mutant homozygous individuals from family A compared to WT controls. Heterozygote values were intermediate (figure 2E). The MCM8 gene normally undergoes alternative splicing, resulting in several WT transcripts, but according to published expression data, exons 12–16 are present in all of them.

MCM8 expression in ovarian tissue

To assess directly the expression of exons 12 and 15 of MCM8 in ovarian tissue, we prepared cDNA from an IGROV-1 ovarian carcinoma cell line and amplified the segment flanking exon 12 and exon 15, confirming that these exons are indeed expressed in ovarian tissue. Exons 12–16 were expressed in all of the alternative splicing transcripts of the gene.

Chromosomal instability studies

Given MCM8's crucial role in homologous recombination mediated DNA repair, we assessed chromosomal breakage repair of both MCM8 c.1954-1G>A and MCM8 c.1469-1470insTA mutations. DNA repair capabilities were assessed in peripheral lymphocytes exposed to MMC. Chromosomes derived from WT controls showed a few chromosomal breaks at 150 and 300 nM MMC (figure 3A–C), whereas chromosomes derived from homozygous individuals of both families showed a significantly increased number of breaks per cell at both 150 and 300 nM MMC (Family A 150 nM: Homozygous affected V-1: 3.8±3.3; V-2: 2.4±0.9 vs WT control: 0.25±0.44, p=1.17e−05, 300 nM: Homozygous affected V-1: 11.6±3.5; V-2: 11.8±3.3 vs WT control: 1.5±1.3, p=5.19e−15. Family B 150 nM: Homozygous affected IV-3: 4.4±3.4 vs WT control: 0.25±0.44, p=9.2e−06, 300 nM: Homozygous affected IV-3: 12.9±5.9 vs WT control: 1.5±1.3, p=6.18e−09) (figure 3A–C, E). Heterozygous carriers showed an intermediate number of breaks per cell at both 150 and 300 nM MMC (figure 3A–E). These results indicated impaired repair of chromosome breaks in homozygotes for both MCM8 c.1954-1G>A and MCM8 c.1469-1470insTA mutations.

Figure 3

Chromosomal breakage analysis in chromosomes derived from peripheral lymphocytes of individuals in families A and B. (A) Number of breaks per cell in homozygous and heterozygous individuals from family A, compared to a wild-type (WT) control. (B) Number of breaks per cell in homozygous and heterozygous individuals from family B, compared to a WT control. aIndicates significant p value in comparison between WT and heterozygous; bindicates significant p value in comparison between heterozygous and homozygous; and cindicates significant p value in comparison between WT and homozygous. At least 10 metaphases were counted per individual. (C) WT chromosomes exposed to 300 nM mitomycin C (MMC). (D) Heterozygous chromosomes. (E) Homozygous chromosomes. Arrows indicate the observed chromosomal breaks.


The reported novel homozygous splicing and frameshift mutations in MCM8 implicate the MCM family of proteins in gonadal failure in humans, both male and female. The MCM8 c.1954-1G>A mutation identified in family A results in lack of normal MCM8 transcript and leads to production of three aberrant MCM8 transcripts with significant reduction in the total amount of MCM8 mRNA. This might be due to nonsense mediated decay in the frame-shifted transcripts, and/or reduced stability of the in-frame deleted transcripts. The MCM8 c.1469-1470insTA mutation identified in family B results in a frameshift and premature stop codon 88 amino acids downstream.

MCM8 and MCM9 were the last genes to be discovered in the evolutionarily conserved family of eight proteins (MCM2–9) characterised by Walker A and B motifs for ATP hydrolysis and zinc- and arginine-finger motifs.13–15 Proteins MCM2–7 are related to each other and form a family of DNA helicases implicated in the initiation of DNA synthesis.

The functions of MCM8 and MCM9 are only partially known, as they have only been studied to a limited extent in cell lines and animal models. MCM8 is conserved in most eukaryotic species; it is located at chromosome band 20p12.3–13, and consists of 19 exons.15 Studies of MCM8 from Xenopus,16 ,17 Drosophila,18 chicken DT40 cells,19 and mice20 have demonstrated its crucial role in homologous recombination mediated DNA repair during gametogenesis. Furthermore, MCM8- and MCM9-deficient mice had chromosome damage and were hypersensitive to replication stress, hallmarks of unrepaired double-strand breaks.20

While the current study was under review, an MCM8 missense mutation was reported in three sisters of a consanguineous family affected with POI21 and MCM9 mutations were identified in two families with POI.22 Interestingly, the novel MCM8 splice mutation we report is the only mutation described to cause gonadal dysgenesis in a male subject as well, whereas the mutations described in MCM9 have only been described in females. This is compatible with the mouse knockout model in which only MCM8-deficient male mice were sterile, whereas MCM9-deficient males were fertile.20 Although the two proteins could act independently, it has been shown that together they form a stable complex that is required for DNA repair.18–20

Male MCM8-knockout mice have 50% lower testicular volume and arrested spermatogenesis in meiotic prophase I. The ovaries of newborn female MCM8-knockout mice contain fewer small-size oocytes with condensed nuclei, indicating that apoptosis is already occurring at birth. Later in life, MCM8-knockout females have arrested primary follicles, and by adulthood they have no observable follicles at all. Furthermore, 100% of the females develop ovarian adenomas.20

Interestingly, the late testicular apoptosis observed in MCM8-mutated mice20 suggests that the gonadal failure is an evolving process. Indeed, the homozygous affected male in family A (V-2) had ‘normal’ pubertal development until 17 years of age, when small testes and primary testicular failure became evident, suggesting that the gonadal failure developed gradually over a period of years. Moreover, the normal male serum testosterone concentrations indicated preserved Leydig cell function.

GWAS aimed at identifying genes associated with early menarche or menopause have revealed SNPs in MCM8 associated with either late age of menopause and higher follicle counts, or increased probability (by up to 85%) of early menopause.10 ,23–28 Interestingly mutations in STAG3, also encoding a protein that is important in meiosis, namely a meiosis-specific subunit of the cohesion ring, was recently reported in a large consanguineous family of Arab-Palestinian origin with inherited POI.9 That report, together with our findings, suggest that additional unexplained cases of ovarian dysgenesis, as well as premature menopause, might be attributed to defects in the meiotic process.

MCM8 has been shown to be involved in homologous recombination mediated DNA repair during gametogenesis and in somatic cell DNA repair, functions that are associated with genomic instability and higher tumour susceptibility,16 as female MCM8-knockout mice all developed adenomas, and 50% of them developed sex cord stromal tumours later in life.20 Our results, which showed increased chromosomal breaks per cell in homozygote chromosomes of both mutations, along with similar results from the other very recently reported MCM8 and MCM9 mutations,21 ,22 and the development of ovarian adenomas and carcinomas in MCM8-deficient female mice, strongly indicate close monitoring and possibly even removal of non-functioning gonads in cases with MCM8 mutations. Additionally, since chromosomes derived from heterozygous individuals (in our two families) also displayed some increased number of chromosomal breaks per cell compared to controls, it is possible that delayed phenotypes might also manifest later in life in heterozygous carriers—as observed in the delayed puberty and the elevation of gonadotropins at 49 years of age in the father IV-2 in family A.

In family A, POI was associated with additional clinical features, including hearing loss, agenesis of the left kidney and temporal epilepsy (in case V-1), and mild mental retardation in both siblings. Whether these clinical findings can be attributed to the identified MCM8 gene mutation or are related to the morbidity associated with the familial consanguinity has yet to be defined. Interestingly, the affected male (V-2) had 22q11del which, in addition to gonadal failure, presented with DiGeorge syndrome phenotype. We speculate that this gene deletion might be caused by the MCM8 gene mutation, resulting in genetic instability due to meiotic DNA repair defect.

The MCM8 gene, transcribed into several alternatively spliced transcripts, is widely expressed in human tissues, including the ovaries and testes. Most transcripts include all 19 exons, but may differ in the length of their untranslated regions or in the inclusion of extra exons between exons 8 and 12. The splice mutation reported here, at chr20:5966317, is at the acceptor splice site before exon 15, which is present in all transcripts of the MCM8 gene in the different tissues. It was expressed in RNA extracted from peripheral blood leucocytes as well as from an ovarian cell line. Nevertheless, the exact function of the amino acid sequence encoded by exon 15—that is, amino acids 652–722 of the MCM protein—is unknown. Although the splice mutation, which eliminates exon 15 from the protein or generates a premature stop codon in this exon after six amino acids (from position 652 to 658), is not thought to affect the known functional domain of the protein (amino acids 402 to 609), we assume that it must have an important role given its ubiquitous tissue expression, its evolutionary sequence conservation across species, and the chromosomal instability observed in homozygous mutant cells. The frameshift mutation reported here occurs in exon 12 of the MCM8 gene and disrupts the MCM8 protein in the middle of its MCM domain (amino acids 402–609), which contains Walker A- and Walker B-type nucleotide binding motifs. This domain is conserved throughout all MCM proteins (MCM2–9) and is crucial for their functionality.

In summary, we identified novel splice and frameshift mutations in the MCM8 gene underlying primary gonadal failure in humans of both sexes. In view of the major role of MCM8 in the development and maintenance of gonads in mice, our finding of mutations in this gene leading to a human clinical phenotype might serve as a basis for elucidation of the genetic causes of other cases of POI and male infertility. Together with GWAS implicating an association between MCM8 SNPs and age of menopause, our findings provide important evidence that MCM8 is essential for human gonadal development and maintenance, and suggest its relevance in the larger context for reproductive lifespan in humans.


We thank Professor Serge Amselem for his constructive comments and suggestions, Camille Vainstein for professional language editing, and the families for participating in and providing samples for this study.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • YT-R, AW-S, DB and DZ contributed equally.

  • Contributors All authors contributed to this manuscript. Planning: YT-R, DZ, AW-S, EL-L and DB. Conducting: YT-R, AW-S, DB, DD, OL, SG, AA-R, MK, HE, PR and DZ. Reporting: YT-R, DZ, AW-S and EL-L.

  • Funding This study was supported by the Legacy Heritage Biomedical Program of the Israel Science Foundation (grant 1531/2009 to DZ), by a grant from the US Agency for International Development (USAID) program for Middle East Regional Cooperation (TA-MOU-10-M30-021 to EL-L and MK), and the Academic committe of Ha'Emek Medical Center (to YT-R).

  • Competing interests None declared.

  • Ethics approval Shaare Zedek and Ha'Emek Medical Center Institutional Review Board and the National Genetic Committee.

  • Patient consent Obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.