A homozygous variant in HFM1 causes preimplantation embryo developmental arrest by disrupting zygotic genome activation

Abstract

STUDY QUESTION

Does a homozygous HFM1 mutation cause human embryonic developmental arrest by disrupting zygotic genome activation?

SUMMARY ANSWER

A pathogenic homozygous HFM1 mutation causes aberrant mRNA splicing and produces a protein that fails to localize to the nucleus, leading to widespread transcriptional dysregulation, failure of zygotic genome activation, and consequent embryonic arrest.

WHAT IS KNOWN ALREADY

HFM1 (Helicase 1) is a germ cell-specific gene that plays a pivotal role in meiotic recombination and DNA damage repair, and its mutations are linked to premature ovarian insufficiency. While HFM1 knockout mice exhibit fertility defects, the mechanism by which HFM1 mutations cause preimplantation embryonic arrest in humans, particularly its role in zygotic genome activation, remains unclear.

STUDY DESIGN, SIZE, DURATION

This was a case-based experimental study conducted from June to November 2024, involving a single infertile patient carrying a homozygous HFM1 mutation and experiencing recurrent embryonic arrest. Analyses included molecular characterization of patient embryos and functional validation in a mouse model.

PARTICIPANTS/MATERIALS, SETTING, METHODS

The patient was recruited from the Reproductive Medicine Centre of the Affiliated Guangdong Second Provincial General Hospital of Jinan University. Whole-exome sequencing identified a homozygous HFM1 mutation. Minigene assays, RNA-seq, immunofluorescence, and confocal imaging were used to characterize the mutation’s impact on splicing, protein localization, and transcriptomic and epigenetic states. Functional rescue experiments were performed in mouse embryos.

MAIN RESULTS AND THE ROLE OF CHANCE

Functional analysis confirmed that the HFM1 mutation disrupts normal mRNA splicing, leading to the production of a protein variant that is excluded from the nucleus. Transcriptomic and epigenetic profiling of arrested human embryos linked the abnormal localization of this protein to a failure in zygotic genome activation and aberrant retention of H3K27me3. The essential role of HFM1 was further verified in a mouse model, where embryonic defects induced by HFM1 knockdown were specifically rescued by wild-type HFM1 mRNA, but not by the mutant version.

LIMITATIONS, REASONS FOR CAUTION

The findings are based on a single clinical case and a limited number of embryos. Further studies with larger cohort studies are needed to validate the prevalence and pathogenicity of such mutations. Further mechanistic studies are also required to fully elucidate how HFM1 regulates gene expression and epigenetic remodeling.

WIDER IMPLICATIONS OF THE FINDINGS

This study establishes that nuclear localization of HFM1 may be essential for ZGA and early embryogenesis in humans. It provides a mechanistic link between noncoding HFM1 variants, transcriptional dysregulation, epigenetic dysregulation, and embryonic arrest, expanding the genetic understanding of female infertility and informing future diagnostic approaches.

STUDY FUNDING/COMPETING INTEREST(S)

This study was supported by National Key R&D Program of China (2022YFC2702200), the National Natural Science Foundation of China (82271728), and the Key Basic and Applied Research Project of Guangdong Province (2023B1515120027). None of the authors have any competing interests.

TRIAL REGISTRATION NUMBER

n/a.

Introduction

Infertility is a multifactorial reproductive disorder characterized by diverse etiologies and reflects abnormalities in the reproductive system. While the widespread application of ART, particularly IVF and ICSI, has enabled many couples to achieve successful pregnancies, a significant number of individuals continue to encounter obstacles such as oocyte maturation defects, fertilization failures, or pre-implantation embryonic arrest (Chen and Heilbronn, 2017; Chambers et al., 2021). Successful progression of early embryonic development is a precisely regulated process that critically depends on zygotic genome activation (ZGA). ZGA, which occurs during the maternal-to-zygotic transition, is indispensable for subsequent embryogenesis. Disruption or failure of ZGA frequently leads to preimplantation developmental arrest and consequently compromises the success of ART (Jukam et al., 2017; Ji et al., 2023).

HFM1 (Helicase for Meiosis 1), also known as Mer3 in Saccharomyces cerevisiae, is a germ cell-specific gene comprising 39 exons and encoding a DNA helicase protein (Tanaka et al., 2006; Altmannova et al., 2023). Studies have established HFM1 as a critical regulator of meiotic recombination and DNA repair (Guiraldelli et al., 2013). HFM1 co-localizes with GM130 at the spindle pole, participating in spindle assembly and cell division. In mouse oocytes, HFM1 silencing disrupts Golgi-associated spindle formation during both Meiosis I and II through the MAPK signaling pathway, ultimately resulting in meiotic defects (Wang et al., 2020). Additionally, HFM1 interacts with the FUS protein to facilitate the repair of DNA double-strand breaks. It also regulates the expression of BRCA1 and key synaptonemal complex components—including SYCP1, SYCE1, REC8, and STAG3—thereby ensuring accurate chromosome synapsis and segregation (Zhong et al., 2024). HFM1 deficiency disrupts meiotic progression, compromises oocyte quality, and has been associated with premature ovarian failure (POF) in knockout mouse models, which exhibit reduced follicle numbers and increased oocyte apoptosis (Guiraldelli et al., 2013; Wang et al., 2020). Consistent with these findings, HFM1 mutations are associated with various reproductive disorders in humans, including POF, non-obstructive azoospermia, and meiotic arrest (Wang et al., 2014; Zhe et al., 2019; Tang et al., 2023; Yao et al., 2023).

Despite these advances, the mechanisms by which HFM1 deficiency leads to preimplantation embryonic arrest—particularly its role in ZGA—remain poorly understood. In this study, we identify a novel homozygous HFM1 mutation in a Chinese patient with recurrent infertility due to preimplantation embryo developmental arrest. Through integrated genomic, transcriptomic, and functional analyses, we investigate how this mutation disrupts embryonic development and ZGA. Our findings aim to elucidate the pathogenic impact of the HFM1 mutation and provide deeper insight into the molecular mechanisms underlying HFM1-associated infertility.

Materials and methods

Nomenclature

For clarity, the gene name HFM1 is used throughout this manuscript to refer to both the human (HFM1) and mouse (Hfm1) orthologs, unless otherwise specified.

Study participants

The family underwent genetic evaluation at the Second People’s Hospital of Guangdong Province. The patient had been married for 10 years but was unable to conceive. She had undergone two unsuccessful IVF attempts. In 2018, the patient underwent several surgical procedures, including laparoscopic pelvic adhesiolysis, right salpingostomy, removal of a right mesosalpinx cyst, hysteroscopy, hydrotubation, and curettage. None of the patient’s parents, sister, or brother reported a history of infertility. The patient was diagnosed with infertility based on the criteria recommended by the ACOG Committee Opinion.

Ethical approval

This study was approved by the Ethics Committee of the Second People’s Hospital of Guangdong Province. All experimental procedures and sample collections were conducted in accordance with the ethical principles of the Declaration of Helsinki, the Ethical Review Measures for Biomedical Research Involving Humans, the Measures for the Administration of Human Assisted Reproductive Technology of the People’s Republic of China, and other applicable national regulations and institutional guidelines. All oocytes and embryos used in this study were obtained through voluntary donation, with written informed consent provided by both partners prior to participation. The consent process and documentation were completed under the oversight of the hospital’s ethics committee.

DNA extraction and whole-exome sequencing

Genomic DNA was extracted from the patient’s peripheral blood, and whole-exome sequencing (WES) analysis was performed using next-generation sequencing technology. The sequencing generated 16.352 Gb of raw data, achieving a targeted depth coverage of 99.69% and a coverage of 99.53% at ≥30× depth. After quality control and filtering of low-quality reads, the clean data were aligned to the GRCh37/hg19 reference genome for variant calling and annotation. The analysis primarily targeted exonic regions and adjacent splice junctions of genes known to be highly associated with female infertility. Additionally, the analysis encompassed known pathogenic genes associated with the clinical data of the subjects, such as primary infertility. Variants with a minor allele frequency >0.01 were excluded based on data from the gnomAD, ExAC, 1000 Genomes, and ESP databases.

Online prediction software predicts mutation hazard

The Mutation Taster (https://www.mutationtaster.org/) online prediction software was employed to ascertain the impact of mutations on gene expression, while the NNSPLICE version 0.9 (https://fruitfly.org/seq_tools/splice.html) was utilized to evaluate deleterious splicing variants. To predict nuclear localization signals (NLS) and subcellular localization of the HFM1 protein, cNLS Mapper (https://nls-mapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_form.cgi) and DeepLoc-2.1 (https://services.healthtech.dtu.dk/services/DeepLoc-2.1/) were used, respectively.

Sanger sequencing

Genomic DNA was extracted from the peripheral blood of the patient’s immediate family members. Following the instructions provided by the manufacturer, the HFM1 mutation site was amplified by PCR using the 2 × Phanta Flash Master Mix reagent (Vazyme, Nanjing, China). The PCR products were purified and analyzed by Sanger sequencing to determine the parental origin of the pathogenic variant. The primer sequences used for amplification are listed in Supplementary Table S1.

Construction of plasmids

To investigate the molecular consequences of the identified HFM1 mutation, two expression systems were established using the pcDNA3.1 vector. A minigene construct, containing partial genomic sequences of HFM1, was designed to evaluate mutation-induced splicing alterations. In parallel, a full-length HFM1 expression construct was generated to determine the subcellular distribution of the wild-type (WT) and mutant proteins in oocytes and embryos.

For minigene assays, the minigene DNA fragment containing Exons 23–26 and Introns 23–25 of HFM1 was amplified from genomic DNA by PCR and cloned into the PcDNA3.1 expression vector using the ClonExpress^® Ultra One Step Cloning Kit (Vazyme). The mutant minigene, containing the c.2680 + 3 _ 2680 + 4del deletion mutation, was generated using the Mut Express^® II Fast Mutagenesis Kit V2 (Vazyme). For the localization experiments, the WT HFM1 coding sequence was cloned into pcDNA3.1 with an enhanced green fluorescent protein tag, while the mutant construct, harboring the same c.2680 + 3_2680 + 4del variant, was cloned into pcDNA3.1 with an RFP (red fluorescent protein) tag. These fluorescently labeled plasmids were microinjected into mouse oocytes and zygotes, enabling direct visualization of HFM1 subcellular localization under confocal fluorescence microscopy. All primer sequences used for plasmid construction are listed in Supplementary Table S1.

Plasmid transfection

293T and HeLa cells were cultured in DMEM high glucose medium containing 10% (v/v) fetal bovine serum at 37°C, in a 5% CO₂ incubator. When cell confluence reached ∼80%, cells were transfected with either WT or mutant HFM1 minigene plasmids using Lipofectamine 6000 (Beyotime, Shanghai, China) according to the manufacturer’s instructions. After 48 h of transfection, cells were collected, and total RNA was isolated using the RNA/DNA Extraction Kit (Beyotime) for subsequent splicing analysis.

Minigene assays

Total RNA was reverse transcribed into complementary DNA (cDNA) using the HiScript III RT SuperMix for qPCR (plus gDNA wiper) (Vazyme) according to the manufacturer’s protocol. The cDNA was amplified by PCR, and the PCR products were analyzed by agarose gel electrophoresis. The DNA fragments were excised and purified using the DNA gel recovery kit (Tiangen, Beijing, China), and Sanger sequencing was performed to confirm the HFM1 mRNA splicing products. The sequence of the primers is provided in Supplementary Table S1.

mRNA synthesis and microinjection in oocytes and embryos

The target gene fragments were amplified by PCR, and in vitro transcription of capped mRNA was performed using the HiScribe^® T7 ARCA mRNA Kit (NEB, USA). A poly(A) tail was subsequently added with the HyperScribe™ Poly(A) Tailing Kit (APExBIO, USA). The synthesized mRNA was purified using VAHTS RNA Clean Beads (Vazyme) according to the manufacturer’s instructions. Purified WT and mutant mRNAs were mixed to a final concentration of 250 ng/µl each. Microinjection was performed under a micromanipulator, delivering 5–10 pl of mRNA mixture into oocytes and zygotes.

For embryo injection, 30 zygotes collected at 18 h post-hCG injection were injected and subsequently cultured in KSOM + aa medium (10 embryos per drop). Embryonic development was monitored at 34, 42, 52, and 60 h post-hCG. For oocyte injection, 30 germinal vesicle (GV)-stage oocytes were injected and cultured in M2 medium (10 oocytes per drop). To maintain GV arrest, oocytes were cultured in M2 medium supplemented with Milrinone (MCE, China) for 8 h. For meiotic progression analysis, oocytes were cultured for 8 h (MI stage) and further cultured for 12 h (MII stage) in standard M2 medium without Milrinone. All imaging and fluorescence analyses were performed using an inverted confocal laser scanning microscope.

Collection and culture of mouse oocytes and embryos

We used 35 female C57BL/6 mice aged 8 weeks for the experiments. To collect oocytes, each female C57BL/6 mouse was injected i.p. with 5 IU of PMSG (NSHF, Ningbo, China). After 46–48 h, ovaries were excised, and oocytes were isolated, washed, and placed in M2 medium covered with mineral oil (Sigma-Aldrich, Livonia, MI, USA). To collect embryos, each C57BL/6 mouse was administered 5 IU of PMSG (NSHF) via injection to obtain mouse embryos. After 46–48 h, 5 IU of hCG (Bio-ss, Ningbo, China) was injected into the mice and placed in cages with male mice. After 18 h, the mice were euthanized by cervical dislocation. The fallopian tubes were excised, and the fertilized eggs were harvested in the M2 culture medium. The fertilized eggs were placed in KSOM + aa culture medium, covered with mineral oil (Sigma-Aldrich, Livonia, MI, USA), Each culture drop was 20 µl in volume, with 10 oocytes or embryos cultured per drop. All cultures were maintained in an incubator at 37°C and 5% CO₂ for the experiment.

Microinjection and siRNA synthesis

Following the manufacturer’s instructions, siRNA targeting mouse HFM1 and a non-targeting control siRNA (siNC), which does not share sequence homology with any known mouse genes, were synthesized using the T7 RNAi Transcription Kit (Vazyme) and microinjected into mouse fertilized eggs (primer sequence detailed in Supplementary Table S1). All microinjections were performed using an ECLIPSE Ti2 microinjection platform. The siRNA concentration was adjusted to 600 ng/µl, and microinjection was performed with a volume of 5–10 pl per fertilized egg at 18 h post-hCG. For each experimental group, 30–40 embryos were microinjected. In all embryo microinjection experiments, siNC (a non-targeting siRNA sequence) was used as a negative control to account for any non-specific effects of the injection procedure. The fertilized eggs were washed with KSOM + aa and cultured in KSOM + aa medium at 37°C and 5% CO₂ to facilitate development after microinjection.

Real-time fluorescent quantitative PCR

Following the manufacturer’s instructions, a cDNA library was constructed using a Single Cell Sequence Specific Amplification Kit (Vazyme), and real-time fluorescent quantitative PCR (qRT-PCR) analysis was conducted using AceQ qPCR SYBR Green Master Mix (Vazyme) and a LightCycler 480 system. The mRNA levels of the target gene in each group were compared by normalizing the endogenous GAPDH mRNA levels. The primer sequence is presented in Supplementary Table S1.

EU dyeing

The embryos were collected at 44 h post-hCG injection. Following the manufacturer’s instructions, the cells were treated with the Cell-Light EU Apollo567 RNA Imaging Kit (Ribobio, Guangzhou, China). Subsequently, the treated cells were imaged using an immunofluorescence microscope, and fluorescence intensity was quantified using ImageJ software. Regions of interest (ROIs) were manually selected around the nuclei of individual oocytes or embryos, and the mean fluorescence intensity within each ROI was measured. Background signal was subtracted using adjacent non-nuclear regions as controls. This method allowed for a semi-quantitative comparison of transcriptional activity based on EU incorporation.

Immunofluorescence

The embryos were fixed in 4% paraformaldehyde (Beyotime) at room temperature for 20 min, followed by washing three times with a washing solution containing 0.1% Tween-20 and 0.01% Triton X-100, each wash lasting 5 min. The samples were blocked for 1 h using an immunostaining blocking solution (Beyotime) and then incubated overnight at 4°C primary antibodies (diluted at 1:100 to 1:200). After washing four times with washing solution, each for 10 min, the samples were incubated at room temperature for 1 h with Alexa Fluor 488-labeled goat anti-rabbit IgG (H + L) secondary antibodies (1:300 dilution, Beyotime). Following four additional 10 min washes with washing solution, the samples were mounted with an anti-fluorescence quenching mounting medium containing DAPI (Beyotime), and the confocal images were obtained using an inverted confocal microscope. For each experimental condition, both experimental and control groups were analyzed. Immunofluorescence staining was quantified using ImageJ software. ROIs were manually defined based on DAPI staining to delineate nuclear or cytoplasmic areas, depending on the localization of the target protein. The mean fluorescence intensity within each ROI was measured, and background fluorescence was subtracted using adjacent regions lacking specific signal. All quantification was performed under identical imaging and exposure settings to ensure consistency and comparability across samples and groups. At least three biological replicates were included for each group to ensure statistical reliability.

Human oocyte and embryo collection

All oocytes and embryos were obtained under the stipulations of informed consent, as evidenced by the signed document from the donor couple. Embryos in the experimental group were collected at the eight-cell stage on Day 4 or Day 5 post-fertilization. Those who failed to progress to the compaction stage and remained morphologically stable, without signs of degeneration, were classified as developmentally arrested. The control group’s GV oocytes, zygotes, and eight-cell stage embryos were obtained from discarded frozen oocytes and embryos of patients undergoing ART treatment due to tubal factor infertility. Only cells that exhibited good developmental morphology after thawing were included. Four eight-cell stage embryos were collected from the patient group, while five GV oocytes, six zygotes, and four eight-cell embryos were collected for the control group.

The construction of single cell RNA-seq library

The embryos were digested in Tyrode’s solution (pH 7.4, Solarbio, Beijing, China) for 2–3 s to remove the zona pellucida. They were then washed three times with 1× PBS buffer containing 0.2% bovine serum albumin. The oocytes and embryos were subsequently placed into 200 µl PCR centrifuge tubes containing 4 µl of cell lysate (1.86 µl Nuclease-Free Water, 1 µl 10 μM Oligo-dT primer (Thermofisher, Waltham, MA, USA), 1 µl 10 mM dNTP (Thermofisher), 0.04 µl 10% Triton X-100 (Sigma-Aldrich), 0.1 µl RNase Inhibitor (Transgen, Beijing, China)), and then 12,000g for 10 s. After 3 min of incubation at 72°C, the samples are immediately placed on ice. The first strand of cDNA is synthesized by adding 6 µl of reverse transcription premix (2 µl 5× first strand buffer (Thermofisher), 0.5 µl SuperscriptII (Thermofisher), 0.25 µl DTT (100 mM) (Thermofisher), the 2 µl 5M Betaine (Sigma-Aldrich), 0.9 µl MgCl2 (100 mM) (Beyotime), 0.1 µl TSO (100 µM) (Thermofisher), 0.25 µl RNase Inhibitor (Vazyme)). In comparison, the second strand of cDNA is synthesized by adding 15 µl of premixed PCR reaction solution and, meanwhile, carries out 14 cycles of pre-amplification. The amplified cDNA products were purified using VAHTS DNA Clean Beads (Vazyme) according to the reagent vendor’s instructions. The purified cDNA was then fragmented, and junction ligated using the TransNGS Tn5 DNA Library Prep Kit for Illumina (Transgen). TransNGS Tn5 Index Kit for Illumina (Transgen) was used to amplify the library and insert the index primer. The final product was purified by VAHTS DNA Clean Beads (Vazyme) to obtain the sequencing library. The constructed sequencing library was sent to AnoYuda for sequencing.

Analysis of RNA-seq data

The data from Illumina sequencing were transformed into raw sequencing sequences by bcl2fastq software, screened, and filtered to remove low-quality fragments. The human reference genome was downloaded from the Ensembl database. The screened data reads were mapped to the GRCh38 human reference genome using HISAT2 software. Following the quantification of gene-level data, raw counts were obtained. Fragments Per Kilobase of the exon model per Million mapped fragments (FPKM) were calculated to represent transcript levels. The R package was used to count differentially expressed genes (DEGs). Principal component analysis (PCA) clustering, Gene Ontology (GO) and KEGG enrichment analysis, and K-means clustering analysis were conducted using the R package.

Acquisition of the ZGA genes

The human ZGA genes were obtained from a previously published dataset (Zou et al., 2022), and only those mRNAs with reliable sequence annotations were retained for further analysis. Genes were defined as human ZGA genes if the FPKM (eight cell/ one cell) ≥2.5 and FPKM (FGO + MII) <25. This criterion was used to screen ZGA genes from published datasets (GSE197265) for subsequent analysis.

Statistical analyses

Data are presented as the mean ± SD of three biological replicates, using GraphPad Prism (GraphPad Software, San Diego, CA, USA) for statistical analysis. For RNA-seq analysis, one biological replicate was defined as a single transcriptome library, and one embryo was used per group. In immunofluorescence quantification experiments, each biological replicate typically included 20–30 embryos per group. For quantitative RT-PCR, one qPCR run was considered one biological replicate, with 10 embryos collected per group for each run. When comparing the experimental group, the difference was analyzed by a two-tailed unpaired t-test, and P < 0.05 was considered significant.

Results

A homozygous HFM1 mutation leads to infertility in patients

Proband II-3 (Fig. 1A), a 34-year-old woman from a Chinese family, presented with infertility. She experienced menarche at the age of 15 and has maintained a regular menstrual cycle of 28–32 days. Despite over 10 years of regular unprotected intercourse, she failed to conceive. The patient was diagnosed with infertility based on the ACOG Committee Opinion diagnostic criteria. Both the patient’s and her husband’s chromosomal karyotypes were normal. Ovarian ultrasound revealed normal uterine and ovarian sizes (the size of the left ovary was 29 × 15 mm, the size of the right ovary was 25 × 18 mm, and the size of the uterus was 56 × 47 × 47 mm), and the endometrial thickness was within the normal range. Her blood anti-Mullerian hormone level was 3.2 ng/ml, and ultrasound showed that each ovary contained 7 to 8 antral follicles, indicating that the ovarian reserve of the patient is normal (Table 1). Additionally, the patient’s mother denied a history of POF.

Figure 1.

Identification of a novel HFM1 mutation. (A) A patient with an HFM1 gene mutation was identified in a Chinese family. The female patient (II-3) of the family was diagnosed with infertility. The double line represents infertility; the square represents a male member; the circle represents a female member; and the all-black solid symbol represents an infertile patient with a homozygous mutation. The semi-black solid symbol expresses the carrier of the gene mutation, and the hollow indicates that the gene mutation is not carried. WT, wild type; nM _ 001017975.6: c.2680 + 3 _ 2680 + 4del, indicating the mutation position of the gene. (B) Sanger sequencing peak map of the patient’s family. The black arrow represents the base before the mutant base, and the red box represents the mutant base and the affected base sequence behind it. (C) This figure represents the structural diagram of HFM1 at the gene, mRNA, and protein levels. The red arrow indicates the mutation site of HFM1 in the patient. The schematic gene structure and mRNA structure diagram data were obtained from the NCBI database, and the protein structure diagram data were obtained from the UniProt database.

Table 1.

Clinical features and genetic information of HFM1 variations in the proband II-3.

Clinical characters	II-3
Diagnosis	Unexplained infertility
Age	35
Somatic karyotype	46, XX
Secondary sexual characteristics	Normal
Hormone levels
LH (mIU/ml)	3.58
FSH (mIU/ml)	3.67
PRL (ng/ml)	18.71
E2 (pg/ml)	53.99
Testosterone (ng/ml)	0.15
Anti-Müllerian Hormone (ng/ml)	3.2
Ovary ultrasound
Ovarian size (left/right, mm)	29 × 15/25 × 18
Antral follicle count (both ovaries combined)	15-16
Information of HFM1 mutation
Mutation site	NM_001017975.6: c.2680 + 3_2680 + 4
Mutation type	Deletion

To investigate potential genetic factors contributing to infertility, we performed WES. After screening the data, we identified a two-base pair deletion mutation (NM_001017975.6: c.2680 + 3_2680 + 4del) located in the 24th intron of HFM1 (Fig. 1C). This mutation was present in a homozygous form and is classified as a rare variant, with a frequency of 0.00030 in the East Asian population, according to the gnomAD database. Previous research has shown that deletion or mutation of HFM1 can lead to POF and disrupt the nuclear maturation of oocytes, resulting in meiotic arrest during the first meiotic division (Wang et al., 2014, 2020; Zhong et al., 2024). However, based on the patient’s medical history and ovarian reserve evaluations, we determined that her ovarian reserve was normal, and she did not meet the diagnostic criteria for POF. Subsequent Sanger sequencing of her immediate family members revealed that both her parents were heterozygous carriers of the mutation. At the same time, her sister and brother did not carry the variant (Fig. 1B). These findings suggest that the mutation follows an autosomal recessive inheritance pattern.

Recurrent embryo developmental arrest and ART failure

The female patient (II-3) underwent three oocyte retrieval cycles and two embryo transfer cycles using ART between 2017 and 2024 (Table 2). The first retrieval and two transfer cycles were conducted at other hospitals, with two eight-cell embryos transferred in each cycle. However, both attempts failed, and the reasons for this require further investigation. The following two oocyte retrievals were carried out in the Reproductive Medicine Centre of the Affiliated Guangdong Second Provincial General Hospital of Jinan University. In the second cycle, eight oocytes were retrieved, and seven cleavage embryos were obtained, including four high-quality embryos. Unfortunately, these embryos experienced developmental arrest at the eight-cell (D3) and compaction (D4) stages, leading to the decision not to proceed with cryopreservation or embryo transfer. In the third cycle, 12 oocytes were retrieved. Two were donated for research, so only 10 were used for IVF, producing nine eight-cell embryos. Two eight-cell embryos were cryopreserved; the others continued to be cultured further in vitro. Regrettably, the remaining embryos are also arrested at the eight-cell and compaction stages (Fig. 2A), failing to meet the criteria for embryo transfer, which prevented embryo transfer once again.

Table 2.

Information about ART attempts by proband II-3.

IVF attempts	IVF-1	IVF-2	IVF-3
Number of oocytes retrieved	12	8	12
Number of MII oocytes	N/A	8	12
Fertilization rate (Fraction (%))	N/A	6/8（75）	9/10 (90)
Cleavage rate (Fraction (%))	N/A	6/6（100）	9 (100)
Number of good quality cleavage embryos	N/A	4	6
Compaction rate (Fraction (%))	N/A	6/6(100)	4/9 (44)
Blastocyst rate (Fraction (%))	N/A	0	1/9（11）
Number of embryos transferred	4	0	0
Clinical pregnancy	NO	NO	NO
Live birth	NO	NO	NO

IVF-1 was done in other medical facilities. It included one oocyte retrieval but two embryo transfers each of two embryo but complete information is currently unavailable. In IVF-3, two MII oocytes were donated for scientific research purposes, while the remaining 10 were used for IVF. In IVF-3 stage, one embryo developed to the Stage 2 blastocyst stage but did not progress to the Stage 4 blastocyst stage (transferable blastocyst stage).

N/A, not available; MII, metaphase II.

Fertilization rate = (number of fertilized egg/total number of MII eggs) × 100%.

Cleavage rate = (number of fertilized and cleaved embryos/number of fertilized eggs) × 100%.

Compaction rate = (number of compacted embryos/(number of fertilized eggs) × 100%.

Blastocyst formation rate = (number of blastocysts formed/number of embryos cultured) × 100%.

Figure 2.

Minigene assays and pathogenicity prediction to assess the pathogenicity of the homozygous variants in HFM1. (A) Representative images showing the morphology of MII oocytes and embryo development on Days 3 and 4 after IVF from the control and proband II-3, who carries the homozygous HFM1 variant. In two independent ART attempts, all arrested embryos displayed similar morphology. (B, C) Prediction results from Mutation Taster and splice site prediction tools. In (B), higher ‘Model’ scores indicate greater prediction reliability. In (C), the ‘-’ symbol denotes the loss of splice sites caused by the HFM1 mutation. (D) Overview of sample processing and transcriptomic analysis in human eight-cell embryos. (E) RNA-seq analysis of HFM1 transcripts in control and mutant embryos, revealing altered mRNA splicing patterns in HFM1 mutant embryos. (F) Schematic diagram of plasmid transfection into HEK293T and HeLa cells, sample collection, mRNA extraction, reverse transcription, and agarose gel electrophoresis. (G) Minigene assays and corresponding HFM1 mRNA bands in HEK293T and HeLa cells. Red arrows indicate mRNA bands corresponding to wild-type and mutant constructs. Images are representative of three independent experiments. (H) Schematic diagram illustrating aberrant mRNA splicing caused by the HFM1 mutation. The wild-type plasmid produces a single 364 bp mRNA band, whereas the mutant plasmid generates three bands: 447 bp, 266 bp, and 136 bp, corresponding to Intron 24 retention, Exon 24 skipping, and Exons 24 and 25 skipping, respectively.

Pathogenic HFM1 mutation causes aberrant splicing as validated by predictions and minigene assays

The mutation identified in this patient is located at the exon–intron boundary of HFM1 and is likely to disrupt normal mRNA splicing. To evaluate its potential impact, we used MutationTaster (https://www.mutationtaster.org/) to predict effects on both protein structure and mRNA splicing. The results indicated a high probability that the mutation alters both protein structure and mRNA splicing (Fig. 2B). In addition, a splice site prediction tool (https://fruitfly.org/seq_tools/splice.html) confirmed that the mutation causes loss of the splice site between Exon 24 and Intron 24 of HFM1 (Fig. 2C). Together, these predictions indicate that the HFM1 mutation may disrupt canonical splicing signals, potentially leading to exon skipping or cryptic splice site activation.

To experimentally validate these predictions, we performed RNA-seq and functional assays to further evaluate its effect on splicing patterns. We collected developmentally arrested eight-cell embryos from proband II-3 as the experimental group (H8cell-del) and cryopreserved embryos without developmental arrest, voluntarily provided by other patients as controls (H8cell-C). All embryos were collected only after obtaining informed consent from both spouses, with signed consent forms completed before the procedures. RNA-seq analysis was performed on samples from all groups to explore gene expression profiles and potential mechanisms (Fig. 2D).

We first analyzed HFM1 alternative splicing events using RNA-seq data. The results showed that the HFM1 mutation led to exon skipping in H8cell-del embryos, disrupting normal HFM1 mRNA splicing (Fig. 2E), which was consistent with the predicted splice site loss (Fig. 2C). This exon skipping is expected to cause a frameshift, resulting in premature termination codons and subsequent mRNA degradation or non-functional protein production. However, due to the limited sensitivity of RNA-seq in detecting low-abundance isoforms, we were unable to definitively identify minor aberrant transcripts. To fully characterize the impact of the mutation on splicing and confirm its pathogenicity, we conducted minigene assays by constructing WT and mutant (Mut) plasmids containing the HFM1 genomic region spanning Exons 23–26 and Introns 23–25. These plasmids were transfected into 293 T and HeLa cells. After 24 h, RNA was extracted, reverse transcribed, and analyzed by PCR and agarose gel electrophoresis (Fig. 2F). Results revealed that the WT plasmid displayed a dominant 364 bp splicing product, representing normal mRNA. However, the Mut plasmid generated multiple aberrant isoforms, including products that band at ∼266 bp and 447 bp and a weaker band at 136 bp. In comparison, the standard 364 bp band was absent (Fig. 2G). Collectively, these results validate our predictions and demonstrate the mutation’s detrimental functional impact. Specifically, it significantly disrupts the normal splicing of HFM1 mRNA and generates aberrant isoforms, thereby impairing the production of functional mRNA. This splicing dysregulation likely contributes to preimplantation embryonic arrest.

To definitively ascertain the architecture of these aberrant splicing events, we isolated individual DNA fragments from the gel electrophoresis of the mutant group and proceeded with Sanger sequencing. The results demonstrated that the products of the aberrant splicing pattern resulting from the mutation were the isoform that retains Intron 24, the isoform that excises Exon 24, and the isoform that excises both Exons 24 and 25, respectively (Fig. 2H). Collectively, these findings delineate a pathogenic splicing mechanism whereby the HFM1 mutation generates aberrant mRNA isoforms through exon skipping and intron retention.

The HFM1 mutation disrupts nuclear localization and protein stability

Given that the HFM1 mutation results in aberrant mRNA splicing, we next asked whether these aberrant transcripts could be translated into protein and, if so, whether the resulting mutant proteins exhibited altered functionality.

HFM1 is known to function as a nuclear-localized protein. Consistent with this notion, our immunofluorescence analyses confirmed its nuclear localization in mouse oocytes and embryos (Fig. 3A) and in human embryos (Fig. 3B). We therefore specifically examined whether the mutation might disrupt this localization.

Figure 3.

The HFM1 mutation disrupts nuclear localization and protein stability. (A) Immunofluorescence images showing nuclear localization of HFM1 protein in mouse embryos. Images are representative of three independent experiments. (B) Immunofluorescence images showing nuclear localization of HFM1 protein in human embryos. Images are representative of three independent experiments. (C) Predicted nuclear localization signal (NLS) of HFM1 protein using cNLS Mapper. The red-highlighted sequence indicates the predicted NLS motif, with higher scores reflecting stronger predicted NLS activity. (D) Prediction of HFM1 protein NLS using DeepLoc-2.1. The red dashed box marks the predicted NLS region. Images are representative of three independent experiments. (E) Representative immunofluorescence images showing subcellular localization of WT and mutant HFM1 protein upon overexpression in human early embryos. Images are representative of three independent experiments. (F, G) Representative immunofluorescence images (E) showing subcellular localization of WT and mutant HFM1 protein upon overexpression in mouse oocytes, and corresponding line plots (F) illustrating the relative stability of WT versus mutant HFM1 protein in oocytes. Images are representative of three independent experiments. (H, I) Representative immunofluorescence images (D) showing subcellular localization of WT and mutant HFM1 protein upon overexpression in early mouse embryos, and line plots (F) depicting relative stability of WT versus mutant HFM1 protein in early embryos. Data are shown as mean ± SD from three biological replicates (unless otherwise indicated).

To address this question, we used two online prediction tools—cNLS Mapper (https://nls-mapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_form.cgi) and DeepLoc-2.1 (https://services.healthtech.dtu.dk/services/DeepLoc-2.1/)—to assess the presence and possible position of NLS within the HFM1 protein sequence. Our results showed that both tools consistently predicted a strong NLS located near the C-terminal region of the HFM1 protein, approximately at amino acid residue 1190 (Fig. 3C and D). As the mutation causes premature translational termination, the resulting protein truncation is predicted to eliminate this NLS, thereby compromising nuclear import. Such nuclear exclusion is likely to disrupt the critical nuclear functions of HFM1 during early embryonic development.

To experimentally test this hypothesis, we constructed plasmids encoding WT and Mut forms of HFM1 and overexpressed the corresponding mRNAs in human and mouse oocytes and early embryos. We first examined human GV oocytes and early embryos (zygote, two-cell, and four-cell stages; sample availability being limited by ethical and practical constraints). We found that the Mut protein displayed aberrant localization and a degradation pattern, while the WT protein remained relatively stable in the nucleus (Fig. 3E). This core finding—that the mutation disrupts HFM1 protein localization—was corroborated in mouse oocytes and embryos, where live-cell imaging definitively showed the Mut protein’s nuclear exclusion and accelerated degradation (Fig. 3F–H). It is noteworthy that under the same conditions, the mouse WT protein, despite undergoing degradation, maintained relatively stable expression in two-cell embryos, which contrasts with its pronounced stability in human embryos (Fig. 3I). This difference in stability suggests that while the nuclear localization of HFM1 is conserved between human and mouse oocytes and embryos, its protein stability may be species-specific. The sustained stability of HFM1 in early human embryos may thus reflect a continued functional requirement during these stages, underscoring the importance of nuclear localization in early development.

HFM1 mutation is associated with H3K27me3 reprogramming failure and disrupted ZGA

These findings indicate that HFM1 protein is retained in the nucleus during early embryogenesis and may support transcriptional activation. Building on the observation that the HFM1 mutation leads to defective nuclear localization, we next asked whether this impairment disrupts transcriptional activation during embryogenesis. Given the critical role of ZGA at the four- to eight-cell stage in human embryos, we performed RNA-seq analysis to comprehensively assess global gene expression changes in mutant embryos (H8cell-del) compared to controls (H8cell-C) (see Material and methods and Fig. 2D). RNA-seq analysis revealed a total of 12,110 DEGs between the diseased and control groups (hereafter referred to as DC-DEGs), accounting for 41.7% of the total detected genes (Fig. 4A). Among these, 6043 DC-DEGs genes were upregulated, and 6067 DC-DEGs genes were downregulated (Fig. 4B). GO enrichment analysis revealed that the majority of DC-DEGs were associated with gene expression-related processes (Fig. 4C), suggesting widespread transcriptional dysregulation in the mutant embryos.

Figure 4.

HFM1 mutation is associated with H3K27me3 reprogramming failure and disrupted zygotic genome activation. (A) Pie chart showing the proportion of differentially expressed genes (DEGs) among all detected genes. A total of 12,110 DEGs were detected, accounting for 41.7% of all detected-genes. (B) Volcano plot showing the distribution of upregulated (n = 6043) and downregulated (n = 6067) DEGs as determined by DESeq2. Yellow dots represent upregulated genes (Fold Change > 2) with P < 0.05, and blue dots represent downregulated genes (Fold Change < 0.5) with P < 0.05. (C) GO enrichment analysis of DEGs, showing significant enrichment in gene expression-related processes (P < 0.05). (D) Volcano plot of upregulated and downregulated zygote genome activation (ZGA) genes. Red dots represent upregulated ZGA genes (Fold Change > 2) with P < 0.05, blue dots represent downregulated ZGA genes (Fold Change > 2) with P < 0.05. (E) Pie chart showing the proportion of ZGA-related DEGs within the current transcriptomic dataset relative to all known ZGA genes. (F) Pie chart showing the proportion of upregulated and downregulated ZGA-related DEGs among all detected ZGA-related DEGs. (G) Bar graph showing the relative expression of key transcription factors (TPRXL, LEUTX, and POU5F1) in H8cell-del embryos; ** represent significant differentiation of the gene; Fold Change < 0.5 or Fold Change > 2.0 to be a significant differentiation gene. (H) Heatmap showing transcriptomic profiles of genes associated with H3K27me3 modification. (I) RNA-seq analysis showing downregulation of KDM6B expression (Fold Change < 0.5) in H8cell-del embryos compared to controls. (J) Immunofluorescence staining of H3K27me3 in HFM1 mutant embryos. (K) Heatmap showing transcriptomic profiles of H3K27me3 target genes obtained from the Cistrome DB database. (L) Bar graph showing H3K27me3 target genes; ** represents significant differential expression (Fold Change < 0.5 or Fold Change > 2.0). Data are shown as mean±SD from three biological replicates. Statistical significance was assessed by two-tailed unpaired t-test: **P < 0.01, ***P < 0.001 (unless otherwise indicated).

ZGA is a critical process during the maternal–zygotic transition, and its failure often leads to developmental arrest (Zou et al., 2022; Yuan et al., 2023). Therefore, we investigated whether the HFM1 mutation causes ZGA failure. Based on the published definition of human ZGA genes (including minor ZGA and major ZGA) and RNA-seq data (Zou et al., 2022), we compiled a list of ZGA genes using FPKM for quantification (Supplementary Table S2). We performed a joint analysis of these ZGA genes with our RNA-seq data. Our analysis identified 1041 ZGA-related genes within the DC-DEGs, representing 40.98% of all ZGA genes (Fig. 4D and E). Among these, 52.35% ZGA genes exhibited significant downregulation (Fig. 4F), supporting the view that HFM1 deficiency broadly disrupts ZGA activation.

Since ZGA depends on key transcriptional regulators, we examined three essential ZGA activators—TPRXL, LEUTX, and POU5F1—in human embryos (Taubenschmid-Stowers et al., 2022; Zou et al., 2022; Gawriyski et al., 2023; Sinha et al., 2023). We found that all three were significantly reduced in H8cell-del embryos (Fig. 4G). For TPRXL and LEUTX, however, predicted downstream targets showed no significant alterations (Supplementary Fig. S1A and B), suggesting limited downstream effects under these conditions. By contrast, POU5F1 was markedly downregulated, and many of its established targets were suppressed (Supplementary Fig. S1C). Integration of hTFtarget-predicted targets with RNA-seq data revealed 1003 significantly downregulated genes (Supplementary Fig. S1D), representing 16.5% of all downregulated DC-DEGs, of which 24.4% were ZGA-related (Supplementary Fig. S1E). GO enrichment analysis further showed that these POU5F1 targets were enriched in essential processes, including transcriptional regulation and mitochondrial function (Supplementary Fig. S1F). Moreover, multi-database cross-referencing identified 24 genes of interest, including 14 DEGs, 13 of which were downregulated in H8cell-del embryos (Supplementary Fig. S1G and H).

Notably, analysis of our RNA-seq data revealed significant downregulation of KDM6B (Fig. 4H and I), the principal demethylase responsible for removing H3K27me3 marks (Supplementary Fig. S2A). This finding prompted us to directly examine whether H3K27me3 levels were altered in HFM1 mutant embryos. Immunofluorescence staining confirmed a significant increase in H3K27me3 levels in patient-derived eight-cell embryos compared to controls (Fig. 4J). Given the essential role of H3K27me3 removal in enabling ZGA (Liu et al., 2016; Fukushima et al., 2019; Xia et al., 2019), it is plausible that the impaired erasure of this repressive histone mark contributes to the observed ZGA failure.

To investigate whether impaired H3K27me3 erasure could constitute a potential mechanism linking HFM1 nuclear mislocalization to embryonic developmental arrest, we analyzed genes associated with H3K27me3 modification from the Cistrome DB database. Integration with our RNA-seq data showed a general downregulation of these genes in H8cell-del embryos (Fig. 4K and L). To strengthen this observation across species, we incorporated well-characterized mouse embryonic H3K27me3 data (Xia et al., 2019). After filtering for evolutionarily conserved targets (Supplementary Fig. S2B), we identified 248 overlapping genes in our dataset, of which 148 were differentially expressed in mutant embryos. Strikingly, 63.6% of these conserved H3K27me3 target genes were significantly downregulated (Supplementary Fig. S2C and D).

Collectively, our findings indicate that the HFM1 mutation is associated with defective ZGA, likely through two interconnected pathways: direct dysregulation of key transcription factors and impaired epigenetic reprogramming marked by abnormal H3K27me3 persistence. While the mechanistic link between HFM1 nuclear dysfunction and H3K27me3 dynamics remains to be fully established, the observed disruptions in both transcriptional and epigenetic landscapes may collectively contribute to the failure of embryonic genome activation and early developmental arrest.

Activation of stress-response pathways in the HFM1 mutant embryos

In addition to defects in repressive histone modifications, aberrant HFM1 function may perturb genome integrity and cellular homeostasis during early embryogenesis. Given the established role of HFM1 in maintaining genome stability during meiosis, we asked whether its dysfunction might similarly influence stress-response pathways in early embryos. Consistent with this possibility, RNA-seq analysis revealed significant upregulation of TP53 (Fig. 5A), the key effector of the p53 signaling pathway, which serves as a central mediator of cellular stress and apoptotic regulation (Jackson and Bartek, 2009; Ciccia and Elledge, 2010). KEGG analysis of DEGs revealed significant enrichment of stress-response and genome-stability-related pathways among upregulated genes, including the p53, Fanconi anemia, and Hippo signaling pathways (Fig. 5B). These pathways are known to participate in cell-cycle regulation and cellular stress responses. The Fanconi anemia pathway has an established role in maintaining genome stability and coordinating cellular stress responses (Jones and Huang, 2012; Engel et al., 2024). The p53 signaling pathway, as a core mediator of cellular stress, regulates both cell-cycle arrest and cell fate. When stress is detected, p53-mediated arrest provides time for recovery, whereas excessive stress can induce apoptosis (Engeland, 2022; Heltberg et al., 2022; Liu et al., 2024; Rodencal et al., 2024). Together, these findings indicate that the HFM1 mutation is associated with activation of multiple stress-response pathways, consistent with the developmental arrest phenotype observed in mutant embryos.

Figure 5.

Activation of stress-response pathways in the HFM1 mutant embryos. (A) Bar graph showing TP53 expression levels. Data are shown as mean±SD from three biological replicates. Statistical significance was assessed by two-tailed unpaired t-test: **P < 0.01. (B) KEGG pathway bar chart illustrating enriched pathways of differentially upregulated genes in H8cell-del embryos.

Loss of HFM1 function alters early embryonic gene expression and impairs embryonic development in mice

Given the ethical constraints on human embryo research and the limited clinical sample availability, we used a murine model to further assess the role of HFM1 in early embryonic development. Since human ZGA occurs at the eight-cell stage, corresponding to the late two-cell stage in mice (Tadros and Lipshitz, 2009), we microinjected siRNA targeting HFM1 (siHFM1) or a negative control siRNA (siNC) into mouse zygotes (Fig. 6A). RT-QPCR and fluorescence immunoassay confirmed efficient knockdown of HFM1 transcripts (Fig. 6B–D). Developmental monitoring revealed that HFM1 knockdown significantly impaired embryonic development, with most embryos arresting at the two-cell stage (Fig. 6E and F), phenocopying the developmental arrest observed in human embryos carrying the homozygous HFM1 mutation.

Figure 6.

Loss of HFM1 function alters early embryonic gene expression and impairs embryonic development in mice. (A) The schematic diagram and grouping pattern of microinjection. (B) Bar graphs shows the relative expression levels of HFM1 in two-cell embryos after microinjection of siHFM1 and siNC. (C) Immunofluorescence images showing the assessment of HFM1 knockdown efficiency. (D) Quantification of HFM1 Fluorescence Intensity. (E) Embryonic development following HFM1 knockdown and rescue. Shown are representative images from the control, HFM1-KD, HFM1-KD+WT rescue, and HFM1-KD+Mut rescue groups across different time points. (F) Line graph comparing embryonic development rates across groups at various stages post-hCG. (G) Confocal microscopy showing decreased transcriptional activity in late two-cell embryos after siHFM1 microinjection. EU (red) labels nascent RNA and DAPI (blue) stains nuclei. Images are representative of three independent experiments. (H) Quantification of the mean fluorescence intensity of EU-labeled RNA. (I) Immunofluorescence images showing KDM6B expression in late two-cell embryos across experimental groups. Images are representative of three independent experiments. (J) Quantification of the mean fluorescence intensity of KDM6B. (K) Immunofluorescence images showing H3K27me3 expression in late two-cell embryos across experimental groups. Images are representative of three independent experiments. (L) Quantification of the mean fluorescence intensity of H3K27me3. Data are shown as mean±SD from three biological replicates. Statistical significance was assessed by two-tailed unpaired t-test: *P < 0.05, **P < 0.01, ***P < 0.001 (unless otherwise indicated).

To determine whether this developmental arrest was associated with defective ZGA, we assessed global transcriptional activity in late two-cell embryos using EU staining. The results showed that the transcriptional activity of the embryos was significantly decreased in siHFM1 embryos (Fig. 6G and H), indicating decreased nascent RNA synthesis during the ZGA stage. This finding is consistent with the RNA-seq results of the human HFM1 mutant embryos, supporting the conserved role of HFM1 in zygotic genes.

To validate phenotype specificity, rescue experiments were performed via microinjection of in vitro-transcribed mRNA encoding either WT or mutant HFM1. Microinjection of WT HFM1 mRNA partially rescued developmental progression beyond the two-cell stage, whereas mutant HFM1 failed to rescue the arrest (Fig. 6E and F). Moreover, immunofluorescence analysis showed reduced nuclear signals of KDM6B in siHFM1 embryos (Fig. 6I and J), accompanied by persistent H3K27me3 marks (Fig. 6K and L), consistent with the defects observed in human mutant embryos. These consistent observations in both human and mouse embryos point to an association between HFM1 deficiency and impaired epigenetic remodeling during early development.

Together, these mouse model data, combined with human embryonic RNA-seq profiles, suggest that HFM1 plays an evolutionarily conserved role in regulating embryonic gene expression and epigenetic remodeling. Loss of HFM1 function disrupts transcriptional activation and histone modification resetting during early embryogenesis, likely contributing to ZGA failure and developmental arrest across species.

Discussion

This study identifies a novel intronic mutation in HFM1 associated with human embryonic developmental arrest. The mutation likely disrupts RNA splicing, providing a mechanistic link between aberrant RNA processing and infertility. While previously reported exonic HFM1 mutations primarily impair oocyte meiosis (Xie et al., 2022), our findings reveal the possibility that intronic mutations can also affect early embryogenesis, possibly through reduced nuclear localization and subsequent alterations in gene regulation. This observation may expand the known phenotypic spectrum of HFM1 deficiency.

Many genetic mutations have been associated with infertility; most of these mutations disrupt the developmental progression of oocytes or zygotes, leading to arrest (Sha et al., 2020; Zhang et al., 2024; Li et al., 2025). Notably, in the present case, the patient had undergone multiple failed IVF attempts with recurrent embryonic arrest—a costly and emotionally taxing journey that long eluded etiological diagnosis. The normal maturation of her oocytes, contrasted with consistent failure to form blastocysts, pointed to a post-fertilization defect. Our data show that the intronic mutation does not block translation but destroys nuclear localization, leaving the protein restricted to the cytoplasm while preserving its functional domains. Previous studies have shown that cytoplasmic HFM1 during metaphase I contributes to the maturation of the oocyte (Wang et al., 2020). Based on these observations, we speculate that HFM1 may have distinct functions at different stages: In oocytes, the cytoplasmic function of HFM1 may be sufficient to support oocyte maturation, whereas after fertilization, when chromatin remodeling and transcriptional regulation become essential, loss of nuclear HFM1 could compromise these processes and lead to early developmental arrest. Furthermore, as a DNA helicase, nuclear HFM1 may normally facilitate chromatin relaxation. And its absence during early embryogenesis likely restricts transcriptional access, creating a non-permissive environment for embryonic genome activation. This notion is supported by the persistence of repressive H3K27me3 marks and the downregulation of genes essential for ZGA in HFM1 mutant embryos. These findings provide a mechanistic framework linking nuclear HFM1 deficiency to early developmental arrest.

Mechanistically, HFM1 has been reported to participate in DNA replication and chromatin organization (Guiraldelli et al., 2013; Zhong et al., 2024). Our results support a mechanism in which nuclear HFM1 may be important for proper transcriptional activation at the onset of embryogenesis. Sustained repressive histone modifications coupled with chronic stress pathway activation in mutant embryos indicate a transcriptionally repressed chromatin state, characterized by persistent compaction and dysregulated stress sensitivity. Although these pathways may temporarily alleviate genomic instability, such activation is frequently associated with developmental delays or cell cycle arrest (Zhao et al. 2021; Engeland, 2022; Rodencal et al., 2024). These alterations are likely downstream consequences of HFM1 nuclear mislocalization and are consistent with the notion that loss of nuclear HFM1 disrupts early embryonic gene expression.

Functional evidence from mouse experiments reinforces this interpretation. HFM1 knockdown impaired embryo development and reduced nascent RNA synthesis. In addition, immunofluorescence analysis revealed reduced nuclear KDM6B signals and exhibited persistently high levels of H3K27me3 in siHFM1 embryos, a finding consistent with the persistence of H3K27me3 observed in human mutant embryos. These results suggest that HFM1 deficiency may impair not only transcriptional activation but also aspects of epigenetic remodeling during early development. Importantly, injection of WT HFM1 but not mutant HFM1 partially restored progression. These data indicate that HFM1 may be critical for supporting transcriptional activation and epigenetic remodeling during early embryo development.

In conclusion, our study demonstrates that an intronic HFM1 mutation impairs nuclear localization, leading to transcriptional dysregulation and defects in epigenetic remodeling, which collectively contribute to early embryonic arrest. These findings establish nuclear HFM1 as a critical regulator of early human embryogenesis. For the patient in this study, this genetic diagnosis finally resolved the long-standing mystery behind her recurrent IVF failure, ending a protracted and costly diagnostic odyssey. Guided by these findings, she underwent donor egg IVF, which successfully resulted in a pregnancy, highlighting the direct clinical impact of mechanistic embryogenesis research.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Authors’ roles

Y.-W.Z. and X.-G.Z. contributed equally to this work. Y.-W.Z., X.-G.Z., S.L., X.-H.O., and J.C. conceived and designed the study. S.L., X.-H.O., J.C., and Z.Y. were responsible for sample collection. Y.-W.Z., P.-Y.L., F.-F.X., and M.-Y.L. conducted the experiments, analyzed the data, and contributed to the interpretation of results. T.-G.M., H.-J.Z., L.Z., L.-N.C., J.L., and Z.Y. assisted with patient information collection and provided technical support. Y.-W.Z., X.-G.Z., S.L., X.-H.O., and J.C. authored the manuscript. S.L., X.-H.O., J..C., Y.-W.Z., T.-G.M., S.-M.L., and Q.-Y.S. contributed to the manuscript revision and additional experimental validation. All authors reviewed and approved the manuscript.

Funding

National Key R&D Program of China (2022YFC2702200), the National Natural Science Foundation of China (82271728), and Key Basic and Applied Research Project of Guangdong Province (2023B1515120027).

Conflict of interest

The authors declare no conflict of interest.