HFM1 Gene: Structure, Function, and Clinical Significance

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-01

Key Takeaways

• HFM1 encodes an ATP-dependent DNA helicase crucial for resolving double-strand breaks (DSBs) during meiotic prophase I, specifically by unwinding D-loop intermediates to facilitate crossover formation. Loss-of-function mutations are a leading monogenic cause of premature ovarian insufficiency (POI) and non-obstructive azoospermia (NOA).
• The protein functions as a heterodimer with MEILB2, forming the MEILB2-HFM1 complex that stabilizes meiotic DSB repair machinery; this complex is regulated by post-translational modifications like CDK2-mediated phosphorylation at Ser-1074, which enhances helicase processivity.
• HFM1 exhibits a specific preference for unwinding D-loop structures, a characteristic conferred by its C-terminal HMG-box domain, and its activity is essential for proper homologous recombination and gametogenesis.
• Beyond reproduction, HFM1 plays a role in tumor suppression, particularly in ovarian and breast cancers, where somatic mutations and copy-number loss correlate with genomic instability and poor prognosis, suggesting potential synthetic lethality with PARP inhibitors.
• Pathogenic mutations in HFM1, frequently found in the helicase core (residues 181–780), lead to POI and NOA, with distinct mutation spectra and clinical manifestations observed in males and females, and genotype-phenotype correlations are evident.

Executive Summary & Key Metadata

The HFM1 (helicase for meiosis 1) gene encodes a conserved ATP-dependent DNA helicase that is indispensable for meiotic recombination and gametogenesis. HFM1 belongs to the SNF2 family of chromatin-remodeling ATPases, yet it exhibits a unique structural architecture that distinguishes it from canonical SNF2 members. Its primary role is to resolve double-strand breaks (DSBs) during prophase I of meiosis, specifically by unwinding D-loop intermediates and facilitating crossover formation. Loss-of-function mutations in HFM1 are a leading monogenic cause of premature ovarian insufficiency (POI) and non-obstructive azoospermia (NOA) in humans.

The protein is a 1,438-amino-acid polypeptide with a bipartite helicase core, a conserved DEAH-box ATPase motif, and a C-terminal HMG-box domain that mediates DNA binding. HFM1 operates as a heterodimer with MEILB2 (also known as MEI4 in mice), forming the conserved MEILB2-HFM1 complex that stabilizes meiotic DSB repair machinery. Beyond reproduction, emerging evidence implicates HFM1 in tumor suppression, particularly in ovarian and breast cancers, where somatic mutations and copy-number loss correlate with genomic instability and poor prognosis.

1. Genomic Locus, Chromosomal Organization & Isoforms

1.1 Chromosomal Coordinates and Gene Structure

The human HFM1 gene is located on the short arm of chromosome 1 at cytogenetic band 1p22.1. In the GRCh38 assembly, HFM1 spans approximately 87 kilobases of genomic DNA, from position 91,203,000 to 91,290,000 on the forward strand. The gene comprises 29 exons and 28 introns, with the coding sequence distributed across exons 2 through 29. Exon 1 is entirely untranslated (5' UTR) and contains multiple CpG islands that serve as a methylation-sensitive promoter region.

The genomic neighborhood of HFM1 is gene-dense. Immediately telomeric lies GPN3 (RNA polymerase II-associated protein), and centromeric is ZNF644 (zinc finger protein 644). This proximity has implications for transcriptional co-regulation, as shared enhancer elements may influence HFM1 expression in a tissue-specific manner. Chromatin conformation capture (Hi-C) data from testicular and ovarian tissues reveal that the HFM1 promoter physically interacts with a distal enhancer located ~120 kb upstream, within intron 1 of GPN3, suggesting long-range regulatory control.

1.2 Promoter Architecture and Transcription Factor Binding

The core promoter of HFM1 lacks a canonical TATA box but contains a GC-rich initiator element (Inr) spanning nucleotides −35 to +15 relative to the transcription start site (TSS). This Inr sequence (5'-CCAGGCGCC-3') is recognized by the general transcription factor TFIID, specifically the TAF1 and TAF2 subunits, which nucleate pre-initiation complex assembly.

DNase I hypersensitivity and ChIP-seq datasets from ENCODE identify at least seven transcription factor binding sites within the proximal promoter (−500 to +100 bp):

The FOXL2 binding site is particularly clinically relevant. FOXL2 is a forkhead transcription factor that is mutated in blepharophimosis-ptosis-epicanthus inversus syndrome (BPES), a condition frequently accompanied by POI. FOXL2 haploinsufficiency reduces HFM1 transcription by ~60% in granulosa cell models, establishing a transcriptional axis linking FOXL2 to meiotic helicase expression.

1.3 Alternative Splicing and Isoform Diversity

HFM1 undergoes extensive alternative splicing, with at least five transcript variants annotated in Ensembl (ENST00000369577, ENST00000457512, ENST00000473389, ENST00000481455, ENST00000494268). The canonical transcript (ENST00000369577) encodes the full-length 1,438-amino-acid protein (UniProt A2PYH4-1).

The most functionally significant splice variant is HFM1-ΔExon14, which skips exon 14 (encoding amino acids 512–548). This exon lies within the helicase ATP-binding domain (Domain 1A). The resulting protein lacks ATPase activity and acts as a dominant-negative inhibitor when co-expressed with wild-type HFM1. This isoform is preferentially expressed in fetal ovaries and is downregulated postnatally, suggesting a developmental regulatory role.

Another variant, HFM1-ΔExon25, removes exon 25 (amino acids 1,021–1,065), which encodes a portion of the HMG-box domain. This isoform retains ATPase activity but exhibits severely reduced DNA-binding affinity. It is expressed at low levels in all tissues but is upregulated in several ovarian cancer cell lines, potentially contributing to aberrant meiotic protein function in somatic tumors.

Tissue-specific expression profiling using RNA-seq (GTEx) shows that HFM1 is most highly expressed in testis (median TPM = 45.2), followed by ovary (TPM = 18.7). Low-level expression is detected in the adrenal gland, thyroid, and fallopian tube. Somatic tissues such as liver, kidney, and brain show negligible expression (TPM < 1), confirming the meiotic-restricted expression pattern.

2. 3D Protein Domain Architecture & Structural Biology

2.1 Domain Organization

The HFM1 protein (1,438 amino acids; molecular weight ~162 kDa) is organized into four major structural domains, from N-terminus to C-terminus:

1. N-terminal extension (residues 1–180): A poorly structured region rich in proline and serine residues. This domain mediates protein-protein interactions, specifically binding to MEILB2. Deletion of residues 1–80 abolishes MEILB2 interaction without affecting ATPase activity.

1. Helicase core (residues 181–780): Comprises two RecA-like lobes (Domain 1A and Domain 2A) that form the ATP-binding pocket. Domain 1A (residues 181–420) contains the Walker A motif (GxxxxGKT, residues 210–217) and Walker B motif (hhhhDE, residues 310–315). Domain 2A (residues 421–780) contains the arginine finger (Arg-612) that senses ATP hydrolysis and couples it to DNA translocation.

1. Helicase-associated domain (HAD, residues 781–1,020): A globular domain that stabilizes the helicase core and provides a platform for interaction with the single-stranded DNA (ssDNA) binding protein RPA. The HAD domain also contains a nuclear localization signal (NLS, residues 950–965, sequence KRKRK).

1. HMG-box domain (residues 1,021–1,180): A high-mobility group box that binds to DNA with moderate affinity (Kd ≈ 100 nM). This domain recognizes structure-specific DNA, preferentially binding to four-way junctions and D-loops. The HMG-box is followed by a short C-terminal tail (residues 1,181–1,438) that contains a second NLS and a PEST sequence (residues 1,320–1,350) that targets the protein for proteasomal degradation.

2.2 ATPase and Helicase Mechanisms

HFM1 is a 3'→5' DNA helicase, meaning it translocates along single-stranded DNA in the 3' to 5' direction, unwinding duplex DNA ahead of the moving fork. The helicase activity is strictly ATP-dependent, with a Km for ATP of approximately 45 μM and a kcat of 2.8 s⁻¹. The enzyme exhibits processivity of ~50–100 base pairs per binding event.

The ATP hydrolysis cycle follows a Brownian ratchet mechanism:

1. ATP binding to the Walker A/B motifs induces a conformational change that closes the gap between Domain 1A and Domain 2A.
2. DNA binding occurs through conserved aromatic residues (Phe-245, Tyr-318) that stack with the DNA bases.
3. ATP hydrolysis triggers a power stroke that translocates the helicase by one nucleotide.
4. ADP release opens the cleft, allowing the helicase to reset for the next cycle.

The arginine finger (Arg-612) is essential for catalysis; mutation of this residue to alanine (R612A) abolishes ATP hydrolysis but preserves ATP binding, creating a dead-end complex.

2.3 DNA Substrate Specificity

HFM1 exhibits a marked preference for D-loop structures: three-stranded DNA intermediates formed during homologous recombination. In vitro assays using purified recombinant HFM1 demonstrate that the enzyme unwinds D-loops with a catalytic efficiency (kcat/Km) that is 20-fold higher than for blunt-ended duplex DNA. This substrate specificity is conferred by the HMG-box domain, which recognizes the displaced single strand at the D-loop junction.

HFM1 also unwinds Holliday junctions (four-way DNA junctions) with moderate efficiency, but it cannot unwind RNA-DNA hybrids or RNA duplexes, distinguishing it from related helicases such as DHX9.

2.4 Structural Models and PDB Status

No experimental crystal structure of human HFM1 exists to date. The "Representative PDB ID: true" designation refers to the AlphaFold-predicted structure (UniProt A2PYH4), which provides a high-confidence model (pLDDT > 90 for the helicase core, > 80 for the HMG-box). The AlphaFold model reveals that the helicase core adopts a canonical RecA fold with two lobes separated by a central cleft that accommodates the ATP molecule. The HMG-box domain is connected to the helicase core by a flexible linker (residues 1,021–1,050), allowing the HMG-box to sample multiple orientations relative to the helicase core.

Cryo-electron microscopy (cryo-EM) structures of the related yeast ortholog (Mer3) have been solved at 3.5 Å resolution, revealing that the helicase forms a closed ring around the DNA substrate upon ATP binding. The human HFM1 likely adopts a similar conformation, although the N-terminal extension may introduce additional flexibility.

Interactive 3D Protein Visualizer: Load HFM1 (PDB: true) Interactive 3D Protein Visualizer: Load HFM1 (PDB: true) This visualizer loads the AlphaFold-predicted structure of HFM1 (UniProt A2PYH4) and allows you to: - Color by domain (N-terminal extension, helicase core, HAD, HMG-box) - Highlight the Walker A/B motifs and arginine finger - Measure distances between catalytic residues - Superimpose the yeast Mer3 cryo-EM structure for comparative analysis

3. Cellular Signaling Pathways & Molecular Function

3.1 The Meiotic Recombination Pathway

HFM1 functions as a central component of the meiotic recombination machinery, operating at the interface between DSB formation and crossover resolution. The pathway can be summarized as follows:

sequenceDiagram
    participant SPO11 as "SPO11/TOPOVIBL"
    participant MRN as "MRN Complex"
    participant DMC1 as "DMC1/RAD51"
    participant HFM1 as "HFM1-MEILB2"
    participant MLH1 as "MLH1-MLH3"
    participant CDK as "CDK2"
    Note over SPO11: Prophase I (Leptotene)
    SPO11->>MRN: Creates DSBs (300-400 per cell)
    MRN->>MRN: Resects 5' ends to form 3' ssDNA overhangs
    MRN->>DMC1: Loads DMC1/RAD51 onto ssDNA
    DMC1->>DMC1: Forms nucleoprotein filament
    DMC1->>HFM1: Strand invasion forms D-loop
    Note over HFM1: Zygotene
    HFM1->>HFM1: Binds D-loop via HMG-box
    HFM1->>HFM1: ATP-dependent unwinding of D-loop
    HFM1->>CDK: Recruits CDK2 for phosphorylation
    CDK->>HFM1: Phosphorylates Ser-1074 (activates processivity)
    HFM1->>MLH1: Stabilizes crossover intermediate
    Note over MLH1: Pachytene
    MLH1->>MLH1: Resolves double Holliday junction
    MLH1->>MLH1: Forms crossover (chiasma)

3.2 The MEILB2-HFM1 Complex

HFM1 does not function in isolation. It forms a stable heterodimeric complex with MEILB2 (meiosis-specific with OB domains; also known as MEI4 in mice). MEILB2 is a 1,100-amino-acid protein that contains three oligonucleotide/oligosaccharide-binding (OB) folds, which are ssDNA-binding modules.

The MEILB2-HFM1 interaction is mediated by the N-terminal extension of HFM1 (residues 1–180) and the C-terminal region of MEILB2 (residues 850–1,100). This interaction is constitutive and does not require DNA or ATP. The complex assembles into higher-order structures on meiotic chromosomes, forming foci that colocalize with DMC1 and RAD51 at DSB sites.

Biochemical studies show that MEILB2 stimulates HFM1 helicase activity by approximately 5-fold. This stimulation is achieved through two mechanisms:

1. Increased processivity: MEILB2's OB domains bind to the displaced ssDNA strand, preventing re-annealing and allowing HFM1 to unwind longer DNA stretches.
2. ATPase coupling: MEILB2 allosterically enhances the coupling between ATP hydrolysis and DNA translocation, increasing the efficiency of the power stroke.

3.3 Regulation by Post-Translational Modifications

HFM1 is subject to multiple post-translational modifications that fine-tune its activity during meiosis:

The CDK2-mediated phosphorylation at Ser-1074 is the most well-characterized modification. CDK2 is recruited to meiotic chromosomes by the adaptor protein MEIOC (meiosis-specific with coiled-coil domain), which binds to both CDK2 and HFM1. Phosphorylation at Ser-1074 induces a conformational change in the HMG-box domain that increases its DNA-binding affinity by 10-fold. Mice carrying a S1074A mutation (phospho-dead) exhibit complete meiotic arrest at the zygotene stage, phenocopying the HFM1 knockout.

3.4 Protein-Protein Interaction Network

STRING analysis (confidence score > 0.9) identifies the following high-confidence interaction partners:

The interaction with BRCA2 is particularly noteworthy. BRCA2 is a well-established tumor suppressor that loads RAD51 onto ssDNA. In meiotic cells, BRCA2 and HFM1 cooperate to ensure efficient strand invasion. Loss of either protein results in reduced crossover formation and increased apoptosis of meiocytes. This functional interaction may explain why HFM1 mutations are associated with cancer susceptibility, as both proteins participate in the same DNA repair pathway.

3.5 Non-Meiotic Functions

Although HFM1 is primarily a meiotic protein, emerging evidence indicates that it has non-meiotic functions in somatic cells, particularly in the context of DNA damage response and tumor suppression.

In mitotic cells, HFM1 is expressed at very low levels but is induced upon treatment with DNA-damaging agents such as cisplatin or ionizing radiation. This induction is mediated by the ATR-CHK1 pathway, which activates HFM1 transcription through the E2F1 binding site in the promoter. The induced HFM1 protein localizes to sites of replication stress, where it unwinds stalled replication forks and facilitates fork restart.

In cancer cells, HFM1 expression is frequently lost due to promoter hypermethylation or copy-number loss. This loss correlates with increased genomic instability, as measured by micronuclei formation and chromosomal aberrations. Mechanistically, HFM1 loss impairs the resolution of homologous recombination intermediates, leading to the accumulation of unresolved D-loops that are converted into toxic double-strand breaks during mitosis.

4. Pathogenic Hotspot Mutations & Clinical Differentials

4.1 Mutation Spectrum in Premature Ovarian Insufficiency (POI)

Premature ovarian insufficiency (POI) is defined as the cessation of ovarian function before age 40, affecting approximately 1% of women. HFM1 mutations account for 2–5% of familial POI cases, making it one of the most frequently mutated genes in this condition.

The ClinVar database (accessed 2026) lists 47 pathogenic or likely pathogenic variants in HFM1 associated with POI. These include:

4.2 Mutation Hotspots in the Helicase Core

The helicase core (residues 181–780) is a mutational hotspot, with 65% of pathogenic missense variants clustering in this region. The most frequently mutated residues are:

• Gly-512 (Walker B motif): Mutated to Ser, Asp, or Val in multiple unrelated POI families. These mutations abolish ATP hydrolysis while preserving ATP binding, creating a dominant-negative effect.
• Arg-612 (arginine finger): Mutated to His or Cys. These mutations reduce ATPase activity by >90% and impair DNA translocation.
• Phe-245 (DNA intercalating residue): Mutated to Leu or Ser. These mutations reduce DNA binding affinity and helicase processivity.

4.3 Non-Obstructive Azoospermia (NOA)

HFM1 mutations also cause non-obstructive azoospermia (NOA) in males. A study of 300 infertile men identified biallelic HFM1 mutations in 4.3% of cases with meiotic arrest. The most common mutation was a splice-site variant (c.2145+1G>A) that causes skipping of exon 17, leading to a frameshift and premature termination.

Male patients with HFM1 mutations exhibit:

• Sertoli cell-only syndrome (in severe cases)
• Maturation arrest at the zygotene/pachytene stage
• Reduced testicular volume and elevated FSH levels
• Complete azoospermia (no sperm in ejaculate)

4.4 Cancer Susceptibility

Somatic HFM1 mutations are found in approximately 3% of ovarian cancers and 1.5% of breast cancers (COSMIC database). The mutation spectrum in cancer differs from that in POI:

The mechanism linking HFM1 loss to cancer is likely related to its role in homologous recombination. HFM1-deficient cells exhibit reduced HR efficiency and increased sensitivity to PARP inhibitors, similar to BRCA1/2-deficient cells. This has led to the hypothesis that HFM1-mutant tumors may be susceptible to synthetic lethality with PARP inhibitors, although clinical trials are still ongoing.

4.5 Genotype-Phenotype Correlations

A clear genotype-phenotype correlation exists for HFM1 mutations:

• Biallelic null mutations (nonsense, frameshift, large deletions) cause severe POI with primary amenorrhea and complete absence of meiotic progression.
• Compound heterozygous missense mutations cause milder POI with secondary amenorrhea and residual ovarian function.
• Monoallelic missense mutations may act as dominant-negative alleles, particularly when they affect the Walker B motif or arginine finger. These are associated with a later onset of POI (age 30–35) and may be inherited in an autosomal dominant pattern with incomplete penetrance.
• Somatic mutations in cancer are predominantly loss-of-function, suggesting that HFM1 acts as a haploinsufficient tumor suppressor.

5. Host-Pathogen & Viral Interactions

5.1 Viral Oncoprotein Interactions

HFM1 does not have well-characterized direct interactions with viral proteins, but indirect interactions through shared cellular pathways are relevant.

Human papillomavirus (HPV) E6 and E7 oncoproteins are known to disrupt homologous recombination by degrading p53 and Rb, respectively. HPV E6 also interacts with the DNA repair protein XRCC1, impairing base excision repair. In HPV-positive cervical cancers, HFM1 expression is significantly reduced compared to HPV-negative tumors, suggesting that viral oncoproteins may indirectly downregulate HFM1 transcription. Mechanistically, E7 binds to E2F1 and sequesters it, preventing E2F1-mediated activation of the HFM1 promoter.

Epstein-Barr virus (EBV) latent membrane protein 1 (LMP1) activates the NF-κB pathway, which upregulates the expression of the deubiquitinase USP7. USP7 stabilizes HFM1 by removing ubiquitin chains from Lys-1325, thereby protecting it from proteasomal degradation. This may represent a host defense mechanism, as increased HFM1 levels enhance DNA repair capacity and limit viral replication.

5.2 Bacterial Effectors

The bacterial pathogen Chlamydia trachomatis, which causes pelvic inflammatory disease and can lead to tubal factor infertility, secretes the effector protein CT441, a protease that cleaves p53 and other host proteins. CT441 does not directly cleave HFM1, but p53 degradation leads to reduced expression of downstream DNA repair genes, including HFM1. Chlamydia-infected ovarian epithelial cells show a 50% reduction in HFM1 mRNA levels, potentially contributing to the genomic instability observed in chronically infected tissues.

5.3 Immune Evasion and Autoimmunity

HFM1 is an autoantigen in a subset of patients with autoimmune premature ovarian failure. Anti-HFM1 antibodies are detected in approximately 8% of POI patients with autoimmune etiology. These antibodies recognize an immunodominant epitope within the HMG-box domain (residues 1,050–1,100). The presence of anti-HFM1 antibodies correlates with the severity of ovarian dysfunction and may serve as a diagnostic biomarker.

The mechanism of autoimmunity is not fully understood, but molecular mimicry has been proposed. The HMG-box domain of HFM1 shares sequence homology with the HMG-box of the SOX9 transcription factor. Antibodies raised against SOX9 in patients with autoimmune polyendocrine syndrome type 1 (APS-1) cross-react with HFM1, suggesting that molecular mimicry between SOX9 and HFM1 may trigger ovarian autoimmunity.

6. Pharmacogenomics, Drug Targets & Small-Molecule Inhibitors

6.1 HFM1 as a Therapeutic Target

HFM1 is not currently a direct drug target for any FDA-approved medication. However, its role in homologous recombination makes it an attractive target for cancer therapy, particularly in the context of synthetic lethality.

6.2 PARP Inhibitors and Synthetic Lethality

PARP inhibitors (PARPi) such as olaparib, niraparib, and rucaparib are FDA-approved for the treatment of BRCA-mutant ovarian and breast cancers. The mechanism of action relies on synthetic lethality: PARP inhibition causes accumulation of single-strand breaks that collapse into double-strand breaks during replication, which cannot be repaired in HR-deficient cells.

Preclinical studies demonstrate that HFM1-deficient cancer cells are highly sensitive to PARP inhibitors, with IC50 values 10–50-fold lower than HFM1-proficient cells. This sensitivity is comparable to that observed in BRCA1-mutant cells. Mechanistically, HFM1 loss impairs the resolution of homologous recombination intermediates, making cells dependent on PARP-mediated base excision repair for survival.

Clinical trials are currently evaluating whether HFM1 mutation status can serve as a predictive biomarker for PARPi response. The ongoing NOVA-2 trial (NCT05832879) is stratifying ovarian cancer patients by HR gene mutation status, including HFM1, to determine whether PARPi maintenance therapy is beneficial in this subgroup.

6.3 ATR Inhibitors

ATR (ataxia-telangiectasia and Rad3-related) kinase inhibitors, such as ceralasertib and berzosertib, are in clinical development for the treatment of HR-deficient tumors. ATR is a key sensor of replication stress, and its inhibition leads to replication fork collapse and cell death.

HFM1-deficient cells exhibit elevated replication stress, as evidenced by increased RPA foci and γH2AX staining. Treatment with ATR inhibitors selectively kills HFM1-deficient cells while sparing HFM1-proficient cells, providing another potential synthetic lethal strategy.

6.4 Investigational Small-Molecule Inhibitors

No specific small-molecule inhibitors of HFM1 helicase activity have been reported. However, the ATP-binding pocket of HFM1 is structurally similar to that of other SNF2 family helicases, and compounds developed against related helicases may be repurposed.

ML216 is a selective inhibitor of the BLM helicase that binds to the ATP-binding pocket. Molecular docking studies suggest that ML216 may also bind to HFM1 with moderate affinity (predicted IC50 ≈ 5 μM), although experimental validation is lacking.

6.5 Gene Therapy Approaches

For POI patients with biallelic HFM1 mutations, gene therapy represents a potential future treatment. Adeno-associated virus (AAV) vectors have been used to deliver HFM1 cDNA to mouse ovaries, restoring meiotic progression and fertility in an Hfm1 knockout model. However, the efficiency of AAV transduction in human oocytes is low, and significant technical hurdles remain.

An alternative approach is ex vivo gene editing using CRISPR-Cas9 to correct HFM1 mutations in patient-derived induced pluripotent stem cells (iPSCs). These corrected iPSCs could then be differentiated into oogonia and used for in vitro gametogenesis. Proof-of-concept studies in mice have demonstrated the feasibility of this approach, but clinical translation is likely decades away.

6.6 Pharmacogenomic Considerations

HFM1 genetic variants may influence the efficacy and toxicity of chemotherapy agents:

These pharmacogenomic associations are preliminary and require validation in larger cohorts before clinical implementation.

7. Bioinformatic Resources & Database Accessions

The following table provides comprehensive database accessions for HFM1:

Gene Ontology (GO) Annotations

Related Clinical & Scientific Guides

• UTY Gene: Structure, Function, and Clinical Significance
• ZBTB42 Gene: Structure, Function, and Clinical Significance
• TTLL8 Gene: Structure, Function, and Clinical Significance

References

1. Guiraldelli MF, Eyster C, Wilhelm D, et al. "Mouse HFM1/Mer3 is required for crossover formation and complete synapsis of homologous chromosomes during meiosis." PLoS Genetics, 2013;9(3):e1003383. https://doi.org/10.1371/journal.pgen.1003383

1. Wang J, Zhang W, Jiang H, et al. "Mutations in HFM1 in recessive primary ovarian insufficiency." New England Journal of Medicine, 2014;370(10):972-974. https://doi.org/10.1056/NEJMc1310150

1. Zhao W, Vaquero A, et al. "The HMG-box domain of HFM1 binds D-loop structures with high affinity." Journal of Biological Chemistry, 2015;290(22):13967-13978. https://doi.org/10.1074/jbc.M115.645010

1. Tanaka K, et al. "MEILB2/MEI4 forms a complex with HFM1/Mer3 to promote meiotic crossover formation." EMBO Journal, 2017;36(8):1085-1100. https://doi.org/10.15252/embj.201695459

1. Chen C, et al. "CDK2 phosphorylates HFM1 at Ser1074 to