Identification of cryosensitive niches and a targetable FOS/AP‑1 program in the human ovarian cortex by single‑cell and spatial transcriptomics

Abstract

Background

The ovary is a vital and dynamic reproductive organ. Ovarian tissue cryopreservation (OTC) plays a vital role in preserving female fertility. However, the cellular subtypes most susceptible to cryoinjury and the molecular mechanisms underlying cryopreservation-associated damage remain poorly understood. This study aimed to identify cell populations vulnerable to freezing-thawing and to elucidate the key transcriptomic alterations and signaling pathways associated with ovarian cryoinjury at the single-cell and spatial levels.

Methods

Ovarian cortical tissues from patients undergoing three gender reassignment surgery (GRS) were divided into fresh and vitrification-rapid warming groups. Following collagenase IV digestion, 10x Genomics single-cell RNA-seq was used for dissociated ovarian cell suspensions (27,185 fresh and 25,480 frozen-thawed cells). Eight major cell clusters were identified. Additionally, 110 oocytes (66 fresh, 44 vitrification-rapid warming) were isolated and analyzed using the Smart-seq2 platform. Spatial transcriptomics was performed via BGI Stereo-seq. Molecular validation was performed via β-galactosidase staining, immunofluorescence, and qRT-PCR.

Results

Cryopreservation significantly altered the activity of pathways related to focal adhesion, oxidative stress, and apoptosis, particularly in stromal and perivascular cells. The number of FOS-positive perivascular cells was notably increased after vitrification-rapid warming, whereas the number of PTGDS-positive stromal cells decreased. Oocyte analysis revealed that cryopreservation primarily disrupted pathways involved in the cell cycle and meiosis, although the damage was not irreversible, supporting the relative safety of long-term cryostorage. Spatial transcriptomics and functional validation further confirmed the rapid and robust activation of the FOS/AP-1 pathway after vitrification-rapid warming, particularly in perivascular and granulosa cells. Treatment with T-5224 (a FOS/AP-1 inhibitor) significantly rescued the morphology and function of cultured frozen-thawed ovaries.

Conclusions

Stromal and perivascular cells are the main cell types that are sensitive to ovarian cryopreservation. The FOS/AP-1 pathway is markedly activated after, suggesting the exacerbation of metabolic impairment. In oocytes within the ovarian cortex, the cell cycle and meiosis-related physiological processes were the primary processes affected.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12916-026-04757-4.

Background

Over the past two decades, advancements in cancer therapy have resulted in significant improvements in survival rates [1]. Advances in cancer detection and treatment have led to markedly improved survival rates, meaning that an increasing number of girls and women under 45 years of age now achieve long term remission [2]. However, treatments such as chemotherapy (especially involving alkylating agents), radiotherapy, and surgery can lead to premature ovarian insufficiency (POI), highlighting the urgent need to address quality of life after cancer remission [3–6]. Therefore, fertility preservation should be considered prior to any gonadotoxic treatment, particularly for adolescents and women of reproductive age [7]. Embryo and oocyte cryopreservation are now widely used fertility-preservation techniques [8]. Ovarian tissue cryopreservation (OTC), in many centers, was introduced even before routine oocyte vitrification or embryo cryopreservation [9]. It is therefore an established fertility-preservation option, particularly for patients who cannot undergo controlled ovarian stimulation, although its protocols and indications continue to evolve [6, 9–11].

OTC for fertility purposes has been performed for more than 25 years. Since the first report of a live birth after thawing and autotransplantation of ovarian cortex tissue in 2004, the procedure has become popular and is now accepted and performed worldwide [12]. As of 2022, more than 10,000 women worldwide have undergone OTC, and more than 200 live births have been achieved [13, 14]. More specifically, OTC and subsequent transplantation are the sole options that enable both the restoration of fertility and the resumption of ovarian endocrine function, avoiding morbidity associated with POI [4, 15–17]. Large cohort data show that most cryostored ovarian samples remain in storage for many years, and only a minority are ultimately transplanted or discarded [18]. In particular, children and adolescents undergoing gonadotoxic therapies often store ovarian tissue for 10–20 years before potential use [19]. Optimizing freezing protocols is therefore crucial for long-term follicle survival and endocrine function. Clinical meta-analyses have shown that more than 90% of women regain ovarian function after OTC-transplantation (OTT) [20]. However, the clinical outcomes of ovarian transplantation post-cryopreservation are significantly worse than those fresh ovary transplantation, indicating adverse effects associated with the cryopreservation process [21]. The live birth rate (LBR) is only 21% for in vitro fertilization (IVF) after tissue cryopreservation and transplantation but is 41% for IVF after embryo cryopreservation [20]. This discrepancy between high ovarian function recovery rates and low live birth rates warrants further detailed investigation. Therefore, to improve follicle development post-transplantation, elucidating the mechanisms through which temperature stress and cryoprotectant toxicity damage ovarian cells is crucial [22, 23].

There are two methods for OTC: slow freezing and vitrification. Slow freezing/thawing is the broadly approved protocol in routine practice. Vitrification is emerging as an efficient strategy for ovarian cryopreservation [24], which has been widely used for oocyte and embryo preservation [16, 25–27]. Vitrified ovarian tissue transplantation has so far resulted in several live births worldwide and is supported by three recent meta-analyses [28–30], as well as new European case series [31, 32]. The exploration of ovarian cryoinjury faces multiple challenges, mainly due to following issues. First, the human ovary is a complex structure consisting of numerous heterogeneous cell types, including stromal cells, granulosa cells, endothelial cells, immune cells, smooth muscle cells, and oocytes [33–37]. Recent studies have indicated that the ovary, owing to its diverse cellular composition, is highly responsive to external environmental factors, chemical drugs, and aging [38, 39]. Freezing-thawing inevitably leads to alterations in the metabolism of various cell types within the ovary. Preliminary investigations have been carried out to explore the impact of freezing on ovarian tissue via morphological staining and transmission electron microscopy [40, 41]. The effects of freezing-thawing on ovarian, especially at the single-cell transcriptome and spatiotemporal levels, remain unknown.

Enabled by advances in single-cell RNA sequencing techniques, the cellular changes in ovarian tissue under physiological or pathological conditions have been assessed [34–37, 39, 42, 43]. Smart-seq2-based single-cell sequencing enables the detection of abundant genes in each cell, especially for oocytes [44, 45]. Spatially enhanced resolution omics sequencing (Stereo-seq) is a novel and highly technical spatial RNA sequencing method with nanoscale sensitivity and a large field of view [46].

In this study, we aimed to explore the cellular spatiotemporal changes and key regulatory genes involved in human ovarian cortex cryoinjury. We present a spatiotemporal atlas that reflects the spatial archetypes and cellular heterogeneity after human ovarian tissue vitrification-rapid warming. Specifically, granulosa cells, smooth muscle/perivascular cells, and stromal cells were identified as cryoinjury-sensitive cell types. The in situ effects of cryopreservation on oocytes within the ovary were also explored via a Smart-seq2 strategy. Furthermore, the potentially crucial role of the FOS/AP-1 pathway in the process of ovarian tissue vitrification-rapid warming was tentatively explored. T-5224 is a selective small-molecule inhibitor of the FOS/AP-1 transcription factor complex that blocks c-Fos/c-Jun binding to AP-1 sites and downregulates pro-apoptotic and pro-inflammatory target genes. In this study, we used T-5224 as a pharmacological tool to test whether FOS/AP-1 activation contributes causally to the stress and injury responses of ovarian cells after cryopreservation. Overall, our data provide valuable insight into the mechanism of frozen-thawed stress and cryoprotectant toxicity-induced ovarian cryoinjury and potential therapeutic targets.

Methods

Human ovarian freezing and thawing

Human ovarian cortical tissue was obtained from three transmasculine individuals (25–35 years) undergoing planned gender-affirming surgery with bilateral oophorectomy at the International Peace Maternity and Child Health Hospital. Ovaries were removed under standard sterile conditions and transferred in aseptic containers for immediate processing. All samples were handled using the same workflow for transport, cortical trimming, vitrification/rewarming, and downstream analyses. Immediately after retrieval, ovarian tissue was placed in pre-cooled (≈4 °C) serum-free RPMI 1640 medium (Gibco) and transferred on ice to a clinical-grade clean laboratory. Throughout transport and trimming to remove excess fat, medulla, and connective tissue, samples were kept on ice (~0–4 °C) to minimize warm ischemia. Prepared cortical strips were then vitrified and stored long term in liquid nitrogen until use.

The KITAZATO ovarian tissue vitrification kit (82212, 82222, and 81213 KITAZATO) was used for human ovarian vitrification and thawing process. In brief, the ovarian cortex was cut into pieces (about 1 cm × 1 cm) and washed with Ova Rinse. Then, tissue pieces were equilibrated in gradually increasing concentrations of cryoprotective agents (CPA) [47]. After equilibration, excess medium was gently blotted on sterile gauze, and the tissue was positioned flat on an Ova Cryo Device with the cortical surface spread out to maximize exposure. Residual solution on the device was removed with blotting paper, after which the loaded devices were plunged into liquid nitrogen, transferred into pre-cooled cryovials, sealed, attached to canes, and stored in liquid-nitrogen tanks.

After 2 weeks of storage, the Ova Cryo Devices carrying vitrified ovarian tissue were retrieved under liquid nitrogen and immediately immersed in thaw 1 solution at 37 °C. After 30 s of warming, tissues were sequentially exposed for 10 min each to solutions with decreasing DMSO/EG concentrations (20%, 10%, 5%, and 0%). This was followed by incubation in solutions containing 0.4-, 0.2-, 0.1-, and 0-M sucrose to allow osmotic CPA removal. Tissues were then washed in L-15 medium and processed for digestion and fixation.

Ovarian single-cell suspension preparation

For the 10 × Genomics scRNA-seq experiments, fresh and vitrified-thawed human ovarian cortex was finely minced with sharp scalpels and enzymatically dissociated in collagenase I (2 mg/mL in PBS) for 45 min at 37 °C. Before flow cytometry, erythrocytes were removed using a red blood cell lysis buffer (BD Pharmingen). The cell suspension was then stained with 4′,6-diamidino-2-phenylindole (DAPI; Sigma, D9452; 1:1000), and DAPI-negative (viable) cells were sorted on a MoFlo Astrios EQ (Beckman Coulter) directly into PBS containing 0.04% bovine serum albumin (BSA).

For Smartseq2 strategy, fresh and vitrification-thawing processed human ovarian cortex were cut into pieces with very sharp knives, followed by dissociating in collagenase I solution (2 mg/ml) for 45 min at 37 °C. RPMI 1640 medium containing 10% FBS was used to stop digestion. The resulting cell suspension was then transferred into a 3.5-cm dish. Then, oocytes were manually picked under a dissection microscope and transferred to a PBS drop (containing 0.1% BSA) by mouth pipetting. Finally, the oocytes were placed separately into an individual PCR tube with lysis buffer and stored at −80 °C for subsequent experiments.

Single-cell RNA-seq library preparation and sequencing

10x Genomics scRNA-seq

FACS-enriched live ovarian cells were loaded onto the 10x chromium controller (single cell 3′ v2); libraries were prepared per manufacturer’s protocol and sequenced on an Illumina NovaSeq 6000 (PE150).

Smart-seq2 scRNA-seq

Libraries were generated following the published protocol (with minor modifications). Each oocyte library was pre-screened by qPCR for DDX4 (oocyte) and NR5A2 (granulosa) to exclude multi-cell captures and then sequenced on an Illumina NovaSeq 6000 (150-bp paired end; Novogene).

Tissue processing for spatial transcriptomic experiment

To collect tissues for Stereo-seq analysis, ovaries were dissected from GRS patients. The fresh tissues were rinsed in pre-chilled PBS twice to eliminate surface impurities, and any remaining liquid or blood was gently wiped dry. Next, one piece of ovaries cortex had been treated with standard human ovarian vitrification and thawing procedure described above. The other one was washed with pre-chilled tissue freezing medium OCT (Leica, Germany) and embedded together in a new OCT. The ovaries were oriented with a blunt metal needle to ensure proper positioning. The entire OCT block was snap-frozen in liquid nitrogen that had been pre-chilled with isopentane and then transferred to a −80 °C freezer for storage before cryosection. Two weeks later, the treatment and OCT embedding methods adopted by the vitrification group were the same as those of the fresh group.

To minimize RNA degradation, the fresh and cryo ones were embedded within 30 min, and the entire dissection procedure was performed in a low-temperature environment. The OCT block was sliced transversely at a thickness of 10 μm using Leica CM1950 cryostat (Leica). Total RNA was extracted from the sections using the RNeasy Mini Kit (Qiagen, USA) in accordance with the manufacturer’s protocol. A sample with an RNA integrity number (RIN) of 7–10, as measured by the 2100 Bioanalyzer (Agilent, USA), was used for the Stereo-seq. The targeted section was directly adhered to the surface of the Stereoseq chip (BGI, Qingdao, China), which had capture probes that contained a 25-bp coordinate identity (CID) barcode, a 10-bp molecular identifiers (MID), and a 22-bp polyT for in situ mRNA hybridization. The adjacent section was stained with H&E for tissue histology examination later. The section on the chip was incubated at 37 °C for 3 min, fixed in pre-cooled methanol for 30 min at −20 °C, and then stained with nucleic acid dye (Thermo Fisher Scientific) for ssDNA visualization. The Ti-7 Nikon Eclipse microscope (Nikon, Japan) was used for ssDNA and histological imaging. Library construction and sequencing were completed with the help of OE Biotech, China.

Single-cell RNA-seq data analysis

The Cell Ranger “count” pipeline (version 3.1.0) was applied with the FASTQ data produced to map the human reference genome (version hg19, GRCh38). The data matrixes in “outs” files were then loaded in R (version 4.3.0) using the Seurat package (version 4.3.0.1) [48].

After aggregation of the sequencing data from fresh and vitrification-thawed human ovarian cortex, R package Seurat 4.3.0 was used for cell filter, data normalization, variable gene selection, unsupervised clustering, and uniform manifold approximation and projection (UMAP) according to their recommended steps [49]. Briefly, Seurat objects were created from the aggregated library as matrix containing gene-by-cell expression data. Cells with less than 200 genes expression or a percentage of more than 0.05 mitochondrial genes were filtered out. Then, data were log-normalized and scaled for subsequent analysis. Variable genes were found and used for principal component analysis (PCA), which was performed for dimension reduction. ElbowPlot function was used for the determination of the numbers of principal components, followed by unsupervised clustering and UMAP. FindAllMarkers function was used to identify the genes exclusively expressed in each cluster. Visualization of total profiles of each cluster was generated with Seurat function DimPlot. Visualization of gene expression with feature plot, dot plot, and heatmap was generated with Seurat function FeaturePlot, DotPlot, and DoHeatmap, respectively. Differentially expressed genes (P < 0.01) between two identities were found with FindMarkers function. Gene Ontology (GO) biological function analysis was performed with marker genes of each cluster found by FindMarkers function with average adjusted p < 0.05 on ToppGene website (https://toppgene.cchmc.org/) and then plotted with R package GOplot.

Unsupervised clustering and bin clusters annotation of Stereo-seq data

The downloaded GEM file was converted to the Seurat object and then processed with Seurat v4.4.4. The data were normalized and scaled, multiple samples were integrated, and unsupervised clustering was performed using bins (resolution = 0.4). Marker genes for each cell type were defined using the DEGs that were identified through the “FindAllMarkers” function. The cell identities of the clusters were then annotated based on the marker genes and histological morphology of the H&E images. To further confirm the relationship between cell types, the Pearson correlation was calculated across the matrix, and hierarchical clustering of the Stereo-seq clusters was performed using ComplexHeatmap package in R (v2.9.1).

Integrated mapping of cell types in Stereo-seq spots with scRNA-seq data

To combine scRNA-seq and Stereo-seq data, the “FindTransferAnchors” and “TransferData” functions of Seurat were employed to determine the possibility of anchors in single spots. The cell type associated with the highest probability among all cell types was subsequently identified as the cell type of the spot. The integrated mapping of cell types was then visualized using the “SpatialDimPlot” and “SpatialFeaturePlot” functions. Additionally, the gene and pathway score had been visualized by “AddModuleScore” and “ggplot2.”

RNA in situhybridization

Human ovarian cortex sections from fresh and vitrification-thawed were fixed in 4% PFA overnight and embedded in paraffin wax. An 8-µm-thick sections were mounted on poly-L-lysine-coated slides (Thermo Scientific). For RNAscope® in situ hybridization (ISH) analysis, Hs-PTGDS (431471), with the target region 15–808 in human PTGDS, probes (Advanced Cell Diagnostics) were used according to manufacturer’s instructions. Quantification of PTGDS positive signals had been performed by ImageJ via using equally sized region of interest (ROI) for the ovarian cortex zone and incorporating all cortex layers.

Transmission electron microscopy

For transmission electron ultrastructural analysis, fresh and vitrification-thawed tissues were fixed in PBS (phosphate-buffered saline), which is supplemented with 2% glutaraldehyde and 2% paraformaldehyde. After tissue processing, semithin (1 mm) and ultrathin (700 nm) sections were cut; semi-thin sections were evaluated under a brightfield microscope (DMi8 Microscope; Leica), whereas ultra-thin sections were observed by transmission electron microscopy (TEM; LEO 906 E TEM 60 kV; Zeiss, Jena, Germany).

Immunohistochemistry and H&E staining in human ovarian tissue

For immunohistochemistry, human ovarian tissues were fixed in 4% paraformaldehyde (in PBS) overnight, embedded in paraffin, and sectioned at 5 µm. Sections were deparaffinized in xylene, rehydrated through graded ethanol, and rinsed in PBS before heat-mediated antigen retrieval in sodium citrate buffer (YEASEN, 36319ES60). After blocking with 5% BSA in PBSTx, sections were incubated with primary antibodies at 4 °C overnight and subsequently developed using the UltraSensitive™ SP IHC Kit (MXB, KIT-9710) with DAB (MXB, MAX-001). Nuclei were counterstained with Mayer’s hematoxylin (Beyotime), and sections were dehydrated, cleared in xylene, and mounted with a neutral resin mounting medium. Negative controls were processed in parallel by omitting the primary antibody.

For H&E staining, paraffin sections were deparaffinized, rehydrated to water, stained with hematoxylin, rinsed, and counterstained with eosin. After dehydration and clearing in xylene, slides were mounted with the same neutral resin mounting medium for microscopic evaluation. And the histological analyses were performed on 5 ovarian cortical fragments per donor, for a total of 15 fragments across the 3 donors.

Immunofluorescence staining

The cryosections (7-μm thickness) were dried at room temperature for 1 h, fixed by 4% PFA for 10 min, and then rinsed with 1× PBS for three times. After being permeabilized with 0.1% Triton X-100 (T8787, Sigma) for 10 min at room temperature, the cryosections were blocked in PBSTx containing 5% BSA for 1 h at room temperature and then incubated with primary antibodies at 4 °C overnight in a humidified chamber. The primary antibodies utilized for these experiments were as follows: PTGDS (1:200, ab182141, Abcam), EGR1 (10 μg/ml; 55117-1-AP, Proteintech), FOSB (2 μg/ml, 2251 Cell Signal Technology), SOHLH2 (1:100, bs-12279R Biossusa), αSMA (1:500; A5228 Sigma), and DDX4 (2 μg/ml; ab180642 Abcam). Afterward, slides were rinsed with 1× PBSTx three times for 15 min, followed by the incubation with Alexa fluorescence-conjugated secondary antibodies (Invitrogen, Carlsbad, CA, USA). Then, the slides were rinsed with 1× PBSTx again three times and finally incubated with DAPI (Beyotime, Shanghai, China) at room temperature for 15 min. Images were visualized using a fluorescence microscope (Leica, Germany).

TUNEL assay

TUNEL staining was used to assess DNA fragmentation in frozen ovarian sections. Cryosections were fixed in 4% paraformaldehyde, permeabilized in 0.1% Triton X-100, and processed with the In Situ Cell Death Detection Kit (CST, No. 25879) according to the manufacturer’s instructions. After incubation with the TUNEL reaction mixture at 37 °C, nuclei were counterstained with DAPI, and samples were imaged on a Zeiss Axio fluorescence microscope.

Murine ovarian tissue 3D culture

The murine ovaries at 7 days postpartum (dpp) were subjected to organ culture as described below. The whole ovarian tissues were directly placed onto a 12-well culture plate (LABSELECT, 12 mm, 0.4 μm). Each well contained 1.2 mL of Dulbecco’s Modified Eagle’s Medium/Ham’s F12 nutrient mixture (Gibco, USA) supplemented with 5% fetal bovine serum, 1% insulin–transferrin–selenium (ITS), and 100 UI/mL penicillin–streptomycin. We tested four different culture conditions: Fresh (the Control group), standards vitrification-thawing treatment (the Cryo group), and with T-5224 post-vitrification thawing (the Cryo + T-5224 group, and during culture, T-5224 were added to the medium with the final concentration which is 80 μM per well). Ovarian tissues were cultured under each the four conditions for 2–4 days at 37 °C, 5% CO2. Half of the culture media was replaced every other day.

T-5224 preparation and application

T-5224 was provided by MedChemexpress (MCE) (New Jersey, USAs). T-5224 was dissolved in dimethyl sulfoxide (DMSO) and diluted in CPA and ovarian culture medium to the target concentration for each experiment.

KGN cell culture and treatments

The human granulosa-like tumor cell line KGN was cultured in DMEM/F12 medium supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin at 37 °C in a humidified 5% CO₂ incubator. For stimulation experiments, cells were seeded at an appropriate density in 6- or 96-well plates and allowed to adhere overnight. KGN cells were then exposed to the AP-1 agonist phorbol 12-myristate 13-acetate (PMA) with or without the FOS/AP-1 inhibitor T-5224; vehicle controls received the corresponding volume of DMSO. Cell viability and proliferation were assessed using the CCK-8 assay, apoptosis was evaluated by Annexin V/propidium iodide staining followed by flow cytometry, and protein expression was analyzed by Western blotting and co-immunoprecipitation as described in the corresponding sections.

Statistical analysis

For experimental studies, all quantitative data were evaluated whether they followed the normal distribution by the Shapiro–Wilk test and equal variance by F-test. For data passed both tests, data are expressed as means ± SEM. Student’s t-test was used for the comparison between two groups, and for comparison among multiple groups, the data were analyzed by one-way ANOVA with Tukey’s post hoc test. For the data that were not normally distributed, nonparametric test (Mann–Whitney U-test) was performed and presented as median ± SD. All P-values are two-sided, and P < 0.05 was considered a statistically significant difference.

Results

scRNA-seq profiling and spatial localization of human ovarian cortex cells

An overview of the study used to generate a spatiotemporal atlas of human ovarian tissue cryoinjury is shown in Fig. 1A. The ovarian cortex tissues of three patients who underwent gender reassignment surgery were divided into fresh and cryopreserved (cryo) groups. The ages of the three patients were as follows: 29, 34, and 35 years. Following collagenase IV digestion, 27,185 fresh ovarian cells were subjected to single-cell RNA sequencing (scRNA-seq). After standard ovarian tissue vitrification-rapid warming, 25,480 dissociated cells from the cryopreserved group were subjected to scRNA-seq. Tissues from each donor were processed separately for dissociation and library preparation. And fresh and cryopreserved datasets were integrated together for comparative analysis. Since large cells cannot be enclosed in microdroplets (diameters > 40 µm), Smart-seq2 was used for sequencing of oocytes in the ovarian cortex. Cells for which mitochondrial genes accounted for more than 25% of the total unique molecular identifiers (UMIs) were excluded (Additional file 1: Fig. S1B). The total cellular RNA content and number of expressed genes did not significantly differ among the three groups (Additional file 1: Fig. S1A and C). More detailed information can be found in the Additional file 1: Table S1.

Fig. 1.

scRNA-seq profiling and spatial location of human ovarian cortex cells. A The flow chart of single-cell RNA sequencing strategy carried on human fresh and frozen-thawed ovarian (including 10x Genomics strategy, Smart-seq2 strategy, and BGI’s Stereo-seq). Collagenase I was used to dissect ovarian cortex tissues. Fluorescence-activated cell sorting (FACS) was used to isolate live cells (DAPI^-). Seurat 4.3 was used for single-cell analysis. Scale bar = 200 μm. scRNA-seq: single-cell RNA sequencing. B Human ovarian tissues were classified into eight populations. Colors denoted different populations. C Featureplots characterized representative marker genes for different cell types and were used to define clusters of human ovarian. D Left: Heatmap showing expression signatures of top 50 specifically expressed genes in each cell type; the value for each gene is row-scaled Z score. Right: Representative GO terms. E The spatial mapping of scRNA-seq cell clusters in representative slides of fresh and frozen-Thawed human cortex tissue. F Cell clusters identified by Stereo-seq (bin = 50, resolution = 0.4). G The cell type mapping resulted from E were shown separately. H Heatmap showing marker genes expressed in the same cell types in both Stereo-seq and scRNA-seq

Using the UMAP algorithm for nonlinear dimensionality reduction analysis, we identified eight distinct cell types on the basis of the expression of specific markers (Fig. 1B and Additional file 1: Fig. S1D). The cell types included granulosa cells (GCs; AMH+), oocytes (DDX4+), stromal cells (SCs; STAR+), smooth muscle/perivascular cells (PCs; MACM+), blood endothelial cells (BECs; VWF+), lymphatic endothelial cells (LECs; PROX1+), macrophages (M; CD68+) and natural killer T cells (NKTs; CD3D+). The expression levels of markers of the different cell types in the ovarian cortex were visualized using a feature plot (Fig. 1C). Additionally, the biological function of each cell cluster was analyzed via GO analysis of DEGs (Fig. 1D), revealing unique characteristics of these ovarian subclusters. For example, GO terms specific to oocytes, such as “meiotic cell cycle” and “gamete generation,” were identified. The DEGs in GCs were enriched in GO terms including “regulation of follicle-stimulating hormone secretion” and “reproductive system development.” Moreover, the DEGs in SCs were enriched in “cytoplasmic translation” and “extracellular matrix organization,” indicating the role of these cells in providing nutritional and structural support to the ovary.

Stereo-seq-based single-cell spatial transcriptome sequencing (ST-seq) of tissues in the fresh and cryopreserved groups was subsequently conducted. Specifically, sections were obtained from the Fresh-1 and Cryo-2 groups. The sequencing depth for ST-seq was 1000 genes/dot (fresh: 934 genes/pot, vitrification-rapid warming: 841 genes/pot) when bin = 50, indicating high data quality (Additional file 1: Fig. S1E and F). Utilizing the scRNA-seq atlas, factor analysis was conducted to infer the probable single-cell composition of each spot, effectively mapping all the scRNA-seq clusters. We identified eight cell types, namely GCs, two subtypes of SCs (SCs1 marked by COL1A2+ and SCs2 marked by GPR78+), PCs, smooth muscle cells (SMCs), ECs, theca-interstitial cells, oocytes, and red blood cells (RBCs) (Fig. 1E, F, G). SpatialFeaturePlot was used to visualize the in situ spatial expression patterns of subgroup marker genes in the ovarian cortex (Additional file 1: Fig. S1G). Additionally, a heatmap was generated to display the top cluster markers, revealing significant overlap between cell cluster markers identified by both scRNA-seq and Stereo-seq (Fig. 1H). Overall, spatial cell mapping elucidated the distribution of clusters predicted by scRNA-seq within the human ovarian cortex.

Cell-specific changes in transcriptional programs following ovarian tissue freezing-thawing

A comparative analysis of gene expression patterns across various ovarian cell types in the fresh and frozen-thawed (F&T) groups was conducted (Fig. 2A). A UMAP plot revealed seven ovarian cell clusters in each sample (Fig. 2B). Quantitative analysis revealed that the proportion of PCs significantly increased in all three replicate samples, whereas the opposite trend was detected for SCs (Fig. 2C and D). To evaluate the reliability of the data shown in Fig. 2C, we integrated the publicly available single-cell dataset GSE255690 (GEO No.) [39] with our dataset. As expected, most ovarian subclusters from the same tissue sources were clustered together (Additional file 1: Fig. S2A and B). Integration and cell proportion analysis also revealed significant decreases in the number of stromal cells in different subclusters after vitrification-rapid warming (Additional file 1: Fig. S2C). Notably, the apparent “increase” in certain cell populations in the vitrified group reflects changes in their relative abundance among captured cells rather than true expansion in situ. Given the short interval between tissue retrieval, cryoprocessing and dissociation, and the absence of any ex vivo expansion step, these shifts are most likely driven by differential survival, loss, or dissociation efficiency after cryopreservation, with more resilient populations (e.g., perivascular cells) becoming relatively enriched and more vulnerable stromal subsets under-represented. Additionally, the existence of the eight cell types identified from the integrated single-cell data was further validated using specific cell markers (Additional file 1: Fig. S2D). Collectively, the apparent “increase” in some cell populations in the vitrified group reflects shifts in their relative abundance among captured cells rather than true in situ expansion. Given the short interval between cortical trimming, freezing-thawing, and dissociation, these changes are most likely driven by differential survival after cryopreservation, with more resilient populations (e.g., perivascular cells) becoming relatively enriched and more vulnerable stromal subsets under-represented.

Fig. 2.

Cell type-specific spatiotemporal changes in transcriptional regulatory programs throughout ovarian vitrification-thawing’s programs. A Integration analysis of fresh and vitrification-rapid warming human ovarian scRNA-seq data (only 10× Genomics strategy data). B UMAP plots showing batch-dependent cell distribution. C Quantitative analysis of the proportions of seven ovarian cell types changed during vitrification-thawing treatment. D Statistical difference analysis of PCs and SCs after freezing and thawing in three repeats. E Heatmap displaying top 50 differentially expressed genes between fresh (left) and vitrified-thaw processed (right) samples with 3 individual repeats. F KEGG pathway analysis of total DEGs between fresh and vitrification-rapid warming-treated cells within human ovarian cortex. G and H Gene set score analysis of focal adhesin, apoptosis, and AP-1 pathways in various ovarian cell types of different groups. J Electron microscopy of fresh and vitrification-rapid warming human ovarian tissue. Abbreviations: GC, granulosa cells; O, oocyte; M, mitochondria; N, nucleus; SC, stromal cells. Scale bar in upper = 10 μm. Scale bar in zoomed screen = 2 μm. I Masson staining evaluates ovarian tissue fibrosis after vitrification-rapid warming in human ovary. K TUNEL analysis of human ovarian cortex section after freezing-thawing treatment

Differentially expressed genes (DEGs; top 50 genes with |avg_logFC|> 0.25 and P adj < 0.05) between the fresh and vitrification-rapid warming groups in at least one cell type were identified and are shown in a heatmap (Fig. 2E). Pathway analysis demonstrated that the DEGs were involved mainly in “focal adhesion” and “apoptosis.” In addition to apoptosis- and adhesion-related pathways, we also noted an increase in PI3K/Akt signaling scores in vitrified ovarian cortex (Fig. 2F). PI3K/Akt–FOXO3 is a key regulator of primordial follicle activation, and its hyperactivation can drive excessive recruitment of dormant follicles and premature depletion of the ovarian reserve [50, 51]. The PI3K/Akt increase we observe after vitrification may support short-term survival but could also predispose to accelerated posttransplant follicle activation and loss.

CellPhoneDB is a bioinformatics toolkit designed to infer cell–cell communication via analysis of the combined expression of multisubunit ligand-receptor complexes [52]. After the vitrification-rapid warming process, there was a significant reduction in ligand-receptor interactions between ovarian cells, notably between granulosa cells and other cell types (Additional file 1: Fig. S3A). Further analysis of individual subpopulations revealed that certain DEGs, such as complement 3 (C3), were unique to specific subpopulations, being predominantly expressed in stromal and granulosa cells (Additional file 1: Fig. S3B). These findings indicate that different cell types respond heterogeneously to the vitrification-rapid warming process (Additional file 1: Fig. S3C, D, E, F).

On the basis of the above results, apoptosis and extracellular matrix organization may play important roles in human ovarian tissue cryoinjury. After vitrification-rapid warming, the signaling pathway score for focal adhesion decreased, whereas that for apoptosis increased in the majority of ovarian cells (Fig. 2G). Pathway analysis of spatial gene expression data further confirmed increased apoptosis and decreased activity of the focal adhesion pathway in ovarian cortex tissue in the vitrification-rapid warming group (Fig. 2H). Masson’s trichrome staining revealed that compared with tissue that underwent vitrification-rapid warming, fresh ovarian tissue had a higher collagen content (Fig. 2I and Additional file 1: Fig. S3G). Transmission electron microscopy analysis revealed noticeable shrinkage of the oocytes and surrounding cells, confirming a reduction in intercellular communication (Fig. 2J). Additionally, TdT-mediated dUTP nick-end labeling (TUNEL assay) further demonstrated a significant increase in the number of cells undergoing increased DNA fragmentation following vitrification-rapid warming (Fig. 2K).

Comparing slow freezing—a method clinically approved and tested over decades for ovarian tissue cryopreservation—and vitrification is crucial. Researchers have conducted single-cell sequencing of slow-frozen human ovarian cortex tissue (M. Wagner et al., Nat Commun (2020), EMBL-EBI database, accession codes: E-MTAb − 8381) [35], allowing direct integration and comparison with our data. UMAP analysis results indicate that the eight major subgroups were consistently present across all three groups without notable absences (Additional file 1: Fig. S4A). In-depth analysis of the differentially expressed genes revealed distinct variations in gene expression among the groups (Additional file 1: Fig. S4B). Specifically, comparisons between samples in the slow-freezing and vitrification groups highlighted significant differences in the expression of genes associated with focal adhesion, the p53 signaling pathway, and the MAPK pathway (Additional file 1: Fig. S4C). DNA damage and apoptosis analyses indicated that compared with vitrification, slow freezing resulted in slightly less negative outcomes (Additional file 1: Fig. S4D and E). Furthermore, cellular crosstalk analyses conducted via CellPhoneDB and CellChat revealed that compared with vitrification, slow freezing significantly reduced intercellular interactions (Additional file 1: Fig. S4G and H). We further explored the changes in WNT and NOTCH signaling via ligand-receptor interactions between oocytes and other cells in the slow freezing, vitrification, and fresh groups. A dot plot indicated that WNT-FZD and JAG-NOTCH signaling was nearly absent in the slow-freezing group (Additional file 1: Fig. S4I). These pathways are closely related to cell proliferation, which may also explain the more severe apoptosis observed in the slow-freezing group. It should be noted that our vitrification protocol and the slow-freezing protocol of Wagner et al. differ in key parameters, including cryoprotectant composition, equilibration times, and cooling/warming conditions; therefore, the comparisons below are exploratory and not a strict head-to-head evaluation.

Overall, our findings suggest that mesenchymal disruption and increased apoptosis are predominant features of cryoinjury in the human ovarian cortex. However, specific cell types within the ovarian cortex may exhibit distinct changes in response to cryoinjury.

Freezing and thawing altered the gene expression profiles of ovarian oocytes in situ

Dormant primordial follicles, situated within the approximately 1-mm-thick ovarian cortex (Additional file 1: Fig. S1E), are the primary structures affected by freezing in ovarian tissue. This has prompted significant interest in investigating the impact of cryoinjury on the physiological metabolism of oocytes within the ovarian cortex. We performed sequencing of 115 oocytes, and the sequencing data from 5 oocytes were excluded because of abnormally high expression of mitochondrial genes (> 30%). Finally, single-cell sequencing data from 110 oocytes at various stages were analyzed (66 oocytes from the freezing group and 44 from the fresh group). During oocyte isolation from the dissociated ovarian tissue, we could not identify the oocyte stage under light microscopy accurately. Fortunately, the increasing availability of transcriptomic data enables the identification of genes specifically expressed in unique stages of the oocyte cycle [53, 54]. Vlnplot analysis revealed high expression levels of LMOD3, RPS4X, and FIGLA in these oocytes, suggesting that they were predominantly at the primordial or primary stage (Additional file 1: Fig. S5A). AddModuleScore and Doheatmap analyses revealed that genes specific to primary oocytes at the primordial stage were enriched in Cluster 1, whereas those at later stages were upregulated in Cluster 2 (Additional file 1: Fig. S5B and C). These findings indicate that the majority of the oocytes used for Smart-seq were derived from primordial follicles (~75 oocytes).

Integrated analysis demonstrated substantial overlap in the expression profiles of oocytes from fresh and vitrified ovarian cortex tissues, indicating that they were generally identical (Fig. 3A). Owing to the extensive sequencing depth of Smart-seq2, we identified numerous differentially expressed genes, as shown in a heatmap (Fig. 3C). Notably, the expression of genes related to meiosis, namely ZP family members (including ZP2, ZP3, and ZP4) [55], OOEP and OOSP4B, was considerably downregulated (Fig. 3B and C). In addition, the expressions of cell cycle-related genes, including CKS1B and TUBB8  [56, 57], was significantly downregulated in ovarian oocytes that had undergone freezing and thawing, whereas the expression of CDCA7L and RAB18 was upregulated (Fig. 3C). Intriguingly, genes related to the heat response, namely DNAJA2 (HSP40), were also significantly upregulated in vitrified-thawed ovarian oocytes (Fig. 3B). To validate the results of the bioinformatics analysis, multiplex immunofluorescence staining was employed. The results revealed an increased proportion of oocytes coexpressing SOHLH2 and DDX4 (Fig. 3F and H), which is consistent with the results of FeaturePlot analysis (Fig. 3G). The upregulated gene SOHLH2 has been reported as a candidate gene for premature ovarian insufficiency [58]. This prompted us to further explore the impact of changes in the expression levels of these genes on later embryonic development and the physiological metabolism of offspring.

Fig. 3.

Vitrification-thawing procedures altered the gene expression profiles of ovarian oocytes in situ. A The UMAP plot showing two ovarian oocyte types. B The VlnPlot showed the marker genes which are expressed predominantly in fresh and vitrification-rapid warming processed oocytes separately. C The heatmap showed the top 20 different expressed genes between fresh and frozen-thawed treatment oocyte groups within ovary. D and E Enriched GO terms and KEGG pathways of the up- and downregulated genes in cryopreserved oocytes in situ. F and H Dual immunofluorescence staining of DDX4 and SOHLH2 in both fresh and vitrification-rapid warming-treated murine ovarian section. Scale bar = 200 μm. And quantification analysis of SOHLH2&DDX4-positive oocytes in F. N = 5. G Featureplot showing the expression profile changes of SOHLH2 in fresh and frozen-thawed human ovarian cortex. I Circle plot showing the intercellular communication between oocyte and major other cell types in fresh and vitrification-rapid warming subjects, respectively. J and K CellphoneDB analysis of Notch, TGFβ, chemokines, and oocyte-rich interactions between oocyte and surrounding cell clusters in fresh vs vitrification-rapid warming-treated human ovarian cortex. Depicted are −log10 p-values (circle size) and log2 means (circle color) for the interacting pairs for selected pairwise cluster combinations

GO analysis revealed that genes upregulated after vitrification-rapid warming were primarily associated with cell cycle-related pathways (Fig. 3D and E). Intriguingly, our observations also indicated the involvement of epigenetic pathways, such as “histone H2A acetylation” and “DNA methylation involved in gamete generation” (Fig. 3D upper). These findings suggest that epigenetic regulatory mechanisms following ovarian tissue vitrification-rapid warming merit further exploration. In addition, the downregulated genes were enriched in some pathways related to apoptosis and embryonic development (Fig. 3E upper). Furthermore, KEGG pathway analysis revealed that the downregulated genes were enriched in several key pathways in vitrified ovarian oocytes, including “carbon metabolism,” “the citrate cycle,” “RNA degradation,” and “endocytosis” (Fig. 3D and E bottom). Additionally, we compared the list of genes involved in DNA damage and repair and generated scores during ovarian freezing and thawing. The DNA damage score decreased, and the DNA repair score increased in ovarian oocytes that had undergone vitrification-rapid warming (Additional file 1: Fig. S5D).

Cell–cell communication analysis via CellPhoneDB revealed that after vitrification-rapid warming, oocytes displayed diminished interactions with granulosa, perivascular, and stromal cells in situ (Fig. 3I). In the cell–cell communication analysis, TGFβ family signaling was increased in vitrified cortex (Fig. 3I, J, K). TGFB ligands were mainly produced by perivascular and immune populations, whereas stromal and endothelial cells expressed the corresponding receptors. TGFβ is a key regulator of immune modulation, fibrosis, and tissue remodeling [59–61]. Thus, the enhanced TGFβ signaling is interpreted as part of a broader stress and remodeling response that links sterile inflammation to matrix remodeling, rather than as a purely pro-inflammatory signal. However, Notch signaling from oocytes to peripheral cells (i.e., NOTCH1_DLK1, NOTCH1_DLL3&4, and NOTCH1_JAG1) decreased (Fig. 3J and K). Intriguingly, the interaction between the AMH receptor (MIS receptor) and AMH from peripheral cells to oocytes was diminished, yet the reverse interaction remained unaffected (Fig. 3J and K). The underlying molecular mechanism deserves further exploration.

PTGDS-positive stromal cells (PTGDS⁺ SCs) sharply decreased after ovarian tissue vitrification-rapid warming

Stromal cell subpopulations were now a focal point of our investigation (Fig. 4A). Quantitative analysis revealed a significant decrease in the number of cells in Subcluster 1 after the freezing‒thawing process (Fig. 4B). A heatmap was used to display the top genes predominantly expressed in Subcluster 1, including PTGDS, C3, SFRP4, and IGFBP6 (Fig. 4C). GO analysis of the top differentially expressed genes in stromal cells following the vitrification-rapid warming process was conducted. These genes were significantly enriched in pathways related to “reactive oxygen response” and stress-inflammatory response-related pathways, such as the “MAPK signaling pathway” (Fig. 4D). Here, we defined Subcluster 1 as the PTGDS-positive subpopulation (PTGDS⁺ SCs) (Fig. 4E). Additionally, RNA from both fresh and frozen-thawed ovarian cortex samples was isolated and analyzed via real-time PCR, and the results confirmed the bioinformatics data (Fig. 4F). ST data revealed decreased expression of C3, SFRP4, and PTGDS, particularly in COL1A2+ stromal cells located in the cortex (Fig. 4G).

Fig. 4.

PTGDS-positive stromal cells (PTGDS⁺ SCs) were sharply decreased post-ovarian vitrification-thawing progress. A Seven subclusters clustered from integrated stromal cells subset. B The fraction of each subcluster calculated separately from fresh and cryo SCs. C Heatmap showing expression signatures of top 50 specifically expressed genes in each subclusters. D Pathway analysis of significant different expressed genes between fresh and vitrification-rapid warming processed stromal cells within ovary. E Expression profiles of marker genes (PTGDS, C3, SFRP4, and IGFBP6) in subcluster 1 shown by FeaturePlot. F Real-time PCR analysis of top genes’ mRNA expression profile changes post-vitrification-rapid warming process in human ovarian cortex. N = 3. G Gene set score analysis of C3, SFRP4, PTGDS, and PTGDS+ cells’ top 10 expressing genes in ST-seq data. H In situ hybridization (ISH) was employed to detect the expression site of PTGDS in sections of human ovary tissue, both fresh and following vitrification-rapid warming treatment, using the RNAscope strategy. Scale bar = 200 μm. I Western blot analysis was conducted to evaluate the changes in PTGDS protein expression levels in murine ovaries before and after vitrification-rapid warming

To validate the results of the bioinformatics analysis, RNAscope® ISH was carried out on paraffin human ovary tissue sections. As expected, PTGDS-positive cells were distributed mainly in the extracellular matrix. In the frozen-thawed ovarian cortex sections, the expression of PTGDS was downregulated compared with that in the fresh control tissue (Fig. 4H), in agreement with the IHC results (Additional file 1: Fig. S6F). We then performed cryopreservation of murine ovaries. The changes in the mRNA and protein expression levels of PTGDS were similar to those observed in human tissue samples (Fig. 4I and Additional file 6: Fig. S6G). These findings raise an intriguing question about the role of PTGDS-positive stromal cells and the PTGDS gene in ovarian tissue cryopreservation. The underlying pathways and metabolic processes involved warrant further exploration.

The graphs revealed that after vitrification-rapid warming, there was not only a decrease in the number of PTGDS-positive cells but also an increase in the number of cells subcluster 2 (Fig. 4B), indicating a rapid and varied response of stromal cells to freezing. Subsequent detailed analysis of cluster 2 revealed a subgroup with increased abnormalities post-vitrification-rapid warming (Additional file 1: Fig. S6A and B, Subgroup 4). FeaturePlot and temporal-based expression profiling demonstrated that after freezing and thawing, genes from the heat-shock protein family, namely DNAJB1, HSP90A1, and HSP90B1, were notably upregulated and expressed in the newly emerged Subgroup 4 (Additional file 1: Fig. S6C and E). Additionally, pseudotemporal analysis revealed that certain PTGDS-positive stromal cells tended to transform into cells with increased expression of temperature-sensitive genes (Additional file 1: Fig. S6D and E). These findings collectively suggest the presence of heightened sensitivity and unique cryoinjury-related changes in SCs after ovarian tissue vitrification-rapid warming.

The FOS/AP-1 signaling pathway is activated in smooth muscle/perivascular cells and granulosa cells following the vitrification-rapid warming process.

Particular attention should be given to the perivascular subtissue structures surrounding blood vessels, especially smooth muscle cells, as they participate in multicellular stratified systems [62]. As shown in Fig. 1D, the number of cells in the smooth muscle/perivascular subcluster clearly increased after vitrification-rapid warming. Reclustering analysis was performed to divide smooth muscle/perivascular cells into six clusters, with cluster 3 (FOS+) exhibiting a significant increase in number after vitrification-rapid warming (Fig. 5A). The specifically highly expressed genes in cluster 3 predominantly included FOS, FOSB, JUN, JUNB, and JUND, as evidenced by the FeaturePlot and ViolinPlot results (Fig. 5G and Additional file 1: Fig. S7A). Moreover, heatmaps revealed consistent activation of AP-1 protein family members, including FOS, FOSB, and JUN, across the three sets of replicate experiments (Fig. 5C). Furthermore, pathway analysis indicated that the DEGs in the ovarian tissue freezing group were predominantly related to “amyotrophic lateral sclerosis,” “Huntington disease,” etc., which are closely linked to muscle dysfunction (Fig. 5D). Granulosa cells play important roles in follicle development and oocyte maturation [63]. Reclustering analysis revealed that the granulosa cells could be divided into three subgroups. Moreover, the proportion of cells in subcluster 3 (FOS/AP-1 active GCs) significantly increased after vitrification-rapid warming (Fig. 5B). Intriguingly, the genes with the greatest expression in cluster 3 included FOS, FOSB, JUN, JUNB, and JUND, as shown by FeaturePlot analysis (Fig. 5H). SpatialFeaturePlot analysis indicated that the expression of FOS, FOSB, JUN, and JUNB significantly increased in situ following vitrification-rapid warming, corroborating the results of the scRNA analysis (Additional file 1: Fig. S8A). Multi-immunofluorescence labeling revealed a significant increase in FOSB protein expression in smooth muscle and perivascular cells following vitrification-rapid warming (Additional file 1: Fig. S8B).

Fig. 5.

The FOS/AP-1 signaling pathway is activated in perivascular and granulosa cells following the frozen-thawed procedure. A Six subclusters clustered from integrated smooth muscle/perivascular cells (SM/PCs) subset. B Three clusters subclustered from integrated granulosa cells (GCs) subset. C Heatmap displaying differentially expressed genes between perivascular and granulosa clusters in fresh (left) and frozen-thaw processed (right) samples. D Pathway analysis of significate different expressed genes between fresh and vitrification-rapid warming processed perivascular cells within ovary. E and F Gene set score analysis of stress-related expressing genes in scRNA-seq data and ST-seq data. G Expression profiles of marker genes (FOS, FOSB, JUN, JUNB, and JUND) in subcluster 3 (FOS⁺ PCs) shown by FeaturePlot. H Expression profiles of marker genes (FOS, FOSB, JUN, JUNB, and JUND) in FOS/AP-1 active subsets shown by FeaturePlot. I RNAscope strategy-based in situ hybridization (ISH) was employed to detect the expression site of FOS in sections of human ovary tissue from fresh and following vitrification-rapid warming treatment. Scale bar = 200 μm. J Immunofluorescence staining was employed to detect the expression site of FOSB in sections of human ovary tissue, both fresh and following vitrification-rapid warming treatment. Scale bar = 200 μm. K Real-time PCR analysis of genes in AP-1 complex pathway expression profile changes post-vitrification-rapid warming process in human ovarian tissues. N = 3

These observations prompted a more detailed examination of the FOS/AP-1 pathway. When the gene scoring algorithm was used, both the scRNA-seq and the ST-seq data confirmed that FOS/AP-1 pathway activation was increased in PCs compared to other subgroups (Fig. 5E and F). Additionally, we focused on DNA damage repair, ROS-related genes, the NF-κB pathway, and the AP-1 complex, but the effects of cryoinjury were not obvious (Additional file 1: Fig. S9).

To validate the increase in the number of FOS⁺ perivascular cells, RNAscope-based ISH was conducted. As expected, the expression of FOS clearly increased in vitrified human ovarian cortex samples compared with fresh ovarian cortex samples (Fig. 5I), which was consistent with the immunochemical staining results for FOSB (Fig. 5J and Additional file 1: Fig. S7B). Furthermore, mRNA from both fresh and frozen-thawed ovarian cortex samples was isolated and analyzed using real-time PCR, and the results corroborated those of the bioinformatics analysis (Fig. 5K). FOS family members (including FOS, FOSB, FOSL1, and FOSL2) can dimerize with proteins of the JUN family, forming the transcription factor complex AP-1 [64, 65]. FOS proteins are regulators of cell proliferation, differentiation, and transformation. Together, these results suggest that the FOS/AP-1 pathway is activated immediately following temperature stress and cryoprotectant toxicity, providing insight into the mechanism underlying the role of FOS in preventing cellular cryoinjury in ovarian tissue.

FOS/AP-1 is rapidly activated in the early stage of the cryopreservation process, and its inhibition maintains the viability of frozen-thawed ovaries in vitro

Previous studies have shown that the FOS/AP-1 pathway rapidly responds to external stimuli and stress, primarily facilitating the translation and accumulation of stress-related inflammation [66–68]. AP-1 activity is induced by numerous extracellular matrix and genotoxic agents, suggesting its involvement in programmed cell death [69]. Cell–cell communication analysis revealed a significant reduction in ligand-receptor interactions between FOS/AP-1-activated granulosa/perivascular cells and other types of cells, especially oocytes (Fig. 6A and Additional file 1: Fig. S10A). Furthermore, we focused on certain ligand-receptor sets involved in cell proliferation, differentiation, and inflammatory responses. As expected, FOS/AP-1-activated PCs and GCs exhibited a decrease in ligand-receptor interactions involved in NOTCH signaling, which promotes proliferating (Fig. 6B). Unexpectedly, although the TGFβ pathway is involved in the activation of FOS/AP-1, there was a significant decrease in ligand-receptor interactions in this pathway after freezing–thawing (Fig. 6B). These findings suggest that FOS/AP-1 activation is primarily driven by direct stress and chemical toxicity at this stage.

Fig. 6.

FOS/AP-1 is rapidly activated in the early flow of vitrification, and its inhibition maintains the viability of frozen-thawed ovaries in vitro. A The circle plots demonstrated the cross talk between PCs (left)/GCs (right) and other cell types in fresh and vitrification-thawed subjects respectively. B Dot plot of selected ligand/receptor (left) and receptor/ligand pair(right) interactions between FOS⁺ PCs or FOS/AP-1 active GCs and other cell components, analyzed separately for fresh and vitrification-rapid warming groups. C Regulon specificity scores (RSSs) identified top specific regulons in fresh and vitrification-rapid warming GCs. The top 6 regulons from each group are marked in red and annotated. D Circle plot shows gene regulatory network of the TFs FOS and JUND and their target genes. E Immunofluorescence staining was employed to detect the expression changes of EGR1 in human ovary tissue following vitrification-rapid warming treatment. Scale bar = 200 μm. F Schematic diagram illustrating the freezing and thawing process of mouse ovarian tissue, with corresponding protein collection points for G. G Western blotting of AP-1 complex family (FOS, FOSB, and JUN) and the upper stream protein (p-AKT and AKT) and down streams (EGR1) expression changes during the whole ovarian vitrification and thawing procedure. The protein collection time point is as follows: A Fresh ovary (FO), B vitrification dehydration (VD), C liquid nitrogen freezing (F), D thawing (T). H Whole-mount immunofluorescence was used to assess vascular (CD31) and extracellular matrix (αSMA) maintenance in in vitro cultured mouse ovarian tissues across different treatment groups. I TUNEL assay of ovaries from fresh and frozen-thawed and T-5224-treated group. Data are presented as the mean ± SEM. n = 5 for each group (unpaired two-tailed t-test). J Schematic illustration showing that activation of FOS/AP-1 in PCs and GCs cells contributes to ovarian cryoinjury-induced apoptosis

The maintenance of cell identity involves the coordinated action of many regulators, among which transcription factors have long been recognized to play a central role [70]. SCENIC was used to identify transcription factors in granulosa cells to predict essential regulators involved in cryoinjury. Regulon specificity scores (RSSs) revealed the top 3 specific regulons, including FOSB (Fig. 6C). Here, we further identified EGR1 as a co-target of FOS and JUN (Fig. 6D). Moreover, the results of co-immunofluorescence and histochemical staining indicated that EGR1 expression was significantly upregulated in smooth muscle/perivascular cells following vitrification-rapid warming (Fig. 6E and Additional file 1: Fig. S10B). The key pathways we focused on in our research, as well as the related molecules (FOS, FOSB, EGR1), are highly conserved in mice (Additional file 10: Fig. S10C and D). Additionally, we confirmed through co-immunoprecipitation experiments that there is no direct interaction between FOS and EGR1 (Additional file 1: Fig. S11D). FOS is a transcriptional regulatory factor [71] that predominantly enters the cell nucleus after phosphorylation to regulate the upstream promoter of EGR1.

These intriguing results prompted us to systematically explore the specific mechanisms and key points where and when the FOS/AP-1 pathway changes (Fig. 6F). We divided the vitrification process into four key steps: obtaining fresh ovarian tissue cortex (FO), vitrification dehydration (VD), liquid nitrogen freezing (F), and thawing (T). Interestingly, Western blotting revealed that FOS/AP-1 activation occurred at the onset of the vitrification dehydration phase and persisted after thawing (Fig. 6G). Additionally, we conducted a preliminary exploration of the function of FOS/AP-1 in the ovary. FOS/AP-1 was activated by the agonist phorbol 12-myristate 13-acetate (PMA) (Additional file 1: Fig. 11SA), which has been reported to be an AP-1 agonist. Following this activation, the proliferation of a human granulosa-like cell line (KGNs) was inhibited (Additional file 1: Fig. S11B). Activation of the FOS/AP-1 pathway increased the percentage of apoptotic KGN cells. Concurrently, T-5224 administration mitigated apoptosis induced by PMA treatment (Additional file 1: Fig. S11C). T-5224 is a FOS/AP-1 inhibitor with anti-inflammatory effects that specifically inhibits the DNA-binding activity of FOS/AP-1 without affecting other transcription factors [72]. Hence, T-5224 was added to the cryoprotectant during vitrification-rapid warming of murine ovaries (Fig. 6H upper). Whole-mount immunofluorescence staining demonstrated that the vascular network density in vitrified-thawed ovaries was well maintained following the addition of T-5224 in vitro (Fig. 6H bottom). Additionally, TUNEL assay confirmed that T-5224 treatment reduced DNA fragmentation in vitrified-thawed ovarian cells (Fig. 6I & J). In summary, these findings suggest that inhibiting FOS/AP-1 via T-5224 improves ovarian function after vitrification-rapid warming.

Discussion

In this study, a comprehensive analysis of single-cell and spatial transcriptomic changes in human ovaries was conducted throughout the cryopreservation procedure, illuminating spatial and temporal variations in gene expression during freezing and thawing. Previous studies have assessed ovarian freeze-thaw injury using histology, follicle counts, and bulk assays. However, these approaches generally lack single-cell and spatial resolution, making it difficult to pinpoint the most vulnerable cell types, niches, and early stress programs in situ [23, 28, 73]. As a result, cell-type-specific cryoinjury responses in the human ovarian cortex remain incompletely defined.

The noteworthy contributions of our work are as follows: Firstly, we delineated gene expression signatures and spatial localization for eight types of human ovarian cells, pinpointing cell type-specific DEGs during ovarian cortex tissue vitrification-rapid warming (Figs. 1 and 3). In both the frozen-thawed and fresh human ovarian cortex, stromal cells, granulosa cells, oocytes, endothelial cells, and smooth muscle/perivascular cells were classified according to their structures (Fig. 1). After vitrification-rapid warming, significant alterations were observed in pathways related to apoptosis, extracellular matrix synthesis, and fibrosis. Notably, genes such as FOS, C3, JUN, and JUNB were upregulated across all cell types (Fig. 2). Secondly, cryoinjury-associated changes in gene expression identified by Smart-seq2 highlighted the cell cycle as a biological pathway involved in oocyte cryopreservation in situ (Fig. 3). Thirdly, subclustering analysis revealed PTGDS-positive stromal cells as one of the primary cell types sensitive to vitrification-rapid warming, showing changes in transcriptomic features during cryopreservation (Fig. 4). Fourthly, integrating the scRNA-seq data with the ST-seq data revealed FOS/AP-1 as a potential key transcription factor for ovarian cell cryoinjury, as it accelerates the transcription of EGR1 (Fig. 5). In addition, inhibiting FOS with T-5224 improved the maintenance of ovarian function in vitro after freezing and thawing (Fig. 6). These findings provide new insights into human ovarian cortex cryopreservation and present potential targets for the treatment of ovarian tissue cryoinjury.

Previous studies on ovarian cryoinjury have primarily focused on two aspects: (1) Direct cellular damage, such as DNA breaks and apoptosis, and (2) disruption of the intercellular microenvironment, including degradation of matrix collagens. We further elucidate the underlying molecular and cellular mechanisms driving these changes. Our analyses revealed elevated DNA repair activity and activation of NF-κB and AP-1 signaling, accompanied by altered apoptosis-related gene expression. A marked loss of stromal cells indicated extracellular matrix disruption, while rapid FOS/AP-1 activation in perivascular and granulosa cells suggested heightened apoptotic and inflammatory responses after vitrification-rapid warming.

Changes in the expression of genes in oocytes within the ovarian cortex after vitrification-rapid warming merit detailed examination. Oocyte-corona-cumulus complex (OCCC) development has been investigated systematically at the single-cell level [74]. A recent study revealed that the effects of vitrification on the transcriptomes of mature human oocytes in vitro (metaphase II oocytes) are induced by the procedure itself rather than by the storage time [75]. However, the cryoinjury on oocytes in situ is unclear. Here, cell cycle and meiosis-related physiological processes were identified as the primary pathways altered in oocytes within the ovary after vitrification-rapid warming. Moreover, the DEGs were enriched in some epigenetics-related terms, such as histone H2A acetylation and DNA methylation, according to GO analysis. Additionally, the ligand-receptor interactions between oocytes and surrounding cells were diminished, especially those involved in the NOTCH and AMH signaling, which is related to development. While our approach has certain limitations, importantly, owing to variability in stages, we could not entirely exclude heterogeneity within the fresh or frozen-thawed oocyte groups in vivo.

Mesenchymal cells of the ovarian stroma (herein called stromal cells), which are similar in morphology to fibroblasts, make up the connective tissue throughout the ovary and surround follicles [76]. The ovarian stroma comprises mostly incompletely characterized stromal cells (e.g., fibroblast-like, spindle-shaped, and stromal cells) [77]. Recent studies have revisited the role of ovarian stromal cells, highlighting their significant contributions to folliculogenesis, particularly in the activation of primordial follicles and the differentiation of theca cells [78, 79]. The plasticity of ovarian stromal cells is a key point of research, and a recent study reported that inhibition of ovarian fibrosis, in part through regulation of stromal cell differentiation via the TGF-β1/Smad3 signaling pathway, could restore ovarian function in POI model rats [80]. Here, we identified a new cryoinjury-sensitive stromal cell cluster, called PTGDS+ stromal cells, whose biological function deserves further study.

Prostaglandin D2 synthase (PTGDS) was purified and first identified by Y. Urade et al [81]. Previous studies have reported that PTGDS catalyzes the conversion of PGH2 to PGD2, a prostaglandin involved in smooth muscle contraction/relaxation and a potent inhibitor of platelet aggregation [82]. This enzyme (PTGDS) is a dual-function protein; it acts as a PGD2-producing enzyme and as a lipophilic ligand-binding protein [83, 84]. PTGDS-knockout (KO) mice are new models of aging, as they exhibit progressive age-related cartilage degradation [85] and adenomyosis [86]. The function of PTGDS has been studied in the male reproductive system, such as the testis and epididymis [87–89]. Little is known about the function of PTGDS in the human ovary. We demonstrated that PTGDS⁺ SCs are among the most sensitive cell populations to cryopreservation. The upstream and downstream signaling pathways of PTGDS involved in ovarian function will be explored in the future.

The molecular network/key factors governing the cryoinjury of ovarian cells are poorly understood. In this study, our analyses revealed FOS/AP-1 as a crucial TF that regulates cellular cryoinjury in the ovarian cortex. During vitrification-rapid warming, we noted an increase in FOS expression in ovarian PCs and GCs. Stereo-seq analysis allowed us to discern whether FOS/AP-1 pathway activation was due to digestion to prepare single-cell suspensions or the cryopreservation process itself. As expected, FOS/AP-1 was upregulated in situ during the vitrification-rapid warming process. Here, we propose that FOS/AP-1 FOS/AP-1 activation results from the combined effects of the following: (i) temperature stress and repeated cooling/rewarming, (ii) cryoprotectant toxicity and osmotic stress during equilibration and dilution, and (iii) secondary mitochondrial dysfunction with ROS accumulation.

FOS is part of the AP-1 complex family, which includes FOS, FOSB, FOSL1, and FOSL2. These proteins dimerize with members of the JUN family to form the transcription factor complex AP-1 [90, 91]. Previous research has reported that FOS/AP-1 regulates metabolic changes and cholesterol synthesis in human periovulatory granulosa cells [92]. Additionally, AP-1 can have oncogenic or antioncogenic effects by regulating genes involved in cell proliferation, differentiation, apoptosis, angiogenesis, and tumor invasion [93]. Furthermore, previous studies have reported the ability of FOS/AP-1 to bind to the FOXP3 gene locus and promote the expression of this master regulator of Treg identity, which leads to the accumulation of inflammation [94, 95]. Consistently, we observed significant increase in the activity of inflammation-accumulation pathways, such as the NF-κB pathway. Previous studies have shown that antioxidants, such as N-acetylcysteine and melatonin, can attenuate ROS-driven damage after cryopreservation and grafting [96, 97]. Metabolic or mitochondrial support has similar protective effects on follicle survival [98]. Pro-angiogenic strategies, including VEGF or bFGF delivery and biomaterial scaffolds that accelerate revascularization, also reduce ischemia–reperfusion injury and maintain the follicle pool [99, 100]. Within this context, the FOS/AP-1 program likely participates mainly via inflammatory responses that shape subsequent vascular remodeling and functional recovery after transplantation. Notably, FOS/AP-1 appears to respond rapidly to temperature stress and cryoprotectant-related toxicity in human ovarian tissue, making it a potential pharmacological target (e.g., T-5224) to complement existing antioxidant and pro-angiogenic approaches. Although our data strongly implicate FOS/AP-1 as a nodal stress-response pathway, further work is required to dissect the relative contributions of temperature versus CPA toxicity using dedicated experimental designs.

The novel benzophenone derivative T-5224 was rationally designed to inhibit transcription regulated by AP-1 [101, 102]. In this study, we discovered its efficacy in improving ovarian function in vitro after freezing and thawing. Notably, orally administered T-5224 is safe for humans [103, 104]. In fact, the current data are based on short term, in vitro rescue of frozen-thawed ovarian tissues, and that longer-term functional studies (including hormone production, follicle growth, and in vivo transplantation models) will be needed to fully establish the translational potential of FOS inhibition. We further clarify that T-5224 should at this stage be considered a tool compound that demonstrates the druggability of the FOS/AP-1 axis in the context of ovarian cryoinjury.

There are several limitations that need to be acknowledged in our study. Firstly, when comparing slow freezing and vitrification, it is important to acknowledge that the underlying protocols differ in cryoprotectant composition, exposure times, and cooling/warming rates, all of which can affect the pattern and extent of cryoinjury. Our analysis therefore provides a contextual comparison rather than a head-to-head test under identical conditions, and these findings should be interpreted as hypothesis-generating rather than definitive evidence for the superiority of one method over the other. Several studies comparing ovarian tissue vitrification and slow freezing have reported lower rates of follicular or oocyte apoptosis after vitrification, with similar final densities of viable follicles and survival in culture under certain conditions [105, 106]. Our findings are consistent with this trend, showing slightly less apoptosis after vitrification than after slow freezing, but we emphasize that these comparisons remain protocol dependent and should be interpreted as hypothesis-generating rather than definitive evidence that one method is universally superior. Secondly, the reliance of our primary analyses on RNA-based measurements (scRNA-seq and spatial transcriptomics) is a limitation, as these data reflect transcriptional states but do not always translate linearly into protein abundance or functional outcomes. Thirdly, this study is based on ovarian cortical tissue from three donors, which limits the generalizability of our findings. The single-cell and spatial patterns described here should therefore be interpreted as exploratory and hypothesis-generating. Larger, independent cohorts will be needed to validate these signatures and assess inter-individual variability.

Conclusions

We mapped the spatiotemporal single-cell transcriptomic landscape of human ovarian cryoinjury and revealed that FOS/AP-1 is activated by vitrification-rapid warming and modulates ovarian function recovery. The pharmacological inhibition of FOS by T-5224 is a promising therapeutic strategy for mitigating ovarian damage post-vitrification-rapid warming. Our study deepens the understanding of human ovarian cryoinjury during freezing–thawing, providing a valuable resource for investigating potential therapeutic interventions. In the future, we aim to explore the use of FOS/AP-1 as a potential therapeutic target for restoring ovarian function after frozen-thawed transplantation.

Abbreviations

F&T

Frozen-thawed

SF&T

Slow freezing and thawing

V&RM

Vitrification-rapid warming

OTC

Ovarian tissue cryopreservation

scRNA-seq

Single-cell RNA sequencing

Stereo-seq

Spatial enhanced resolution omics sequencing

RNAscope® ISH

RNAscope-based in situ hybridization

DAPI

4′,6-Diamidino-2-phenylindole

IVF

In vitro fertilization

PCs

Perivascular cells

SCs

Stromal cells

GCs

Granulosa cells

BECs

Blood vascular endothelial cells

LECs

Lymphatic endothelial cells

SMCs

Smooth muscle cells

GRS

Gender reassignment surgery

DEGs

Differentially expressed genes

CP

Cryoprotective agents

Authors’ contributions

F.G., D.S., H.D., Y.L. and B.Y. contributed equally to this work. W.L., Q.Z. and M.Z. conceived the study. F.G., D.S., Y.L., B.Y., and H.D. performed the experiments. F.G., D.S., Y.L., and M.G. analyzed and visualized the data. M.Z., Y.M., R.Q., S.L., and L.Z. provided the ideas and suggestions. M. G. and Q. Z. helped to collect clinical samples. F. G. and W. L. wrote and revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Key Research and Development Project of China (2022YFC2703002), the National Natural Science Foundation of China (82371726, 82571873, 82200541), the Joint Funds of the National Natural Science Foundation of China (U24A20658), the Natural Science Foundation of Shanghai (No. 21ZR1428600), the Innovative Research Team of High-Level Local Universities in Shanghai (SHSMU-ZDCX20212200), the Shanghai Hospital Development Center Foundation (SHDC22022303), and the Key Project of Medical and Industrial Intersection of Shanghai Jiao Tong University (YG2023ZD27). The funders played no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The RNA-sequencing data have been deposited in the Gene Expression Omnibus (GEO) under the accession number GSE285362. The publicly available software applied in this study is listed and described in the Methods section. All the codes are available from the corresponding authors upon reasonable request (GSE285362:Guo F, Sun D, Li W, et al. Identification of Cryosensitive Niches and a Targetable FOS/AP‑1 Program in the Human Ovarian Cortex by Single‑Cell and Spatial Transcriptomics https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?&acc=GSE285362 (2025)).

Declarations

Ethics approval and consent to participate

All patients were informed and signed informed consent for this study, which was approved by the Ethics Committee of the International Peace Maternity and Child Health Hospital, School of Medicine, Shanghai Jiao Tong University (ethics approval no. B2022269P). Written informed consent was obtained from all participants. Donors were informed that their ovarian tissue would be used for biomedical research, including vitrification and storage in liquid nitrogen for the duration of this study. Remaining tissue was retained in a clinical biobank for potential future-related research under appropriate ethics approval and governance. All procedures were approved by the institutional ethics committee and conducted in accordance with relevant institutional and national guidelines. All experimental procedures complied with institutional guidelines for the care and use of laboratory animals in China and were approved by the Animal Ethics Committee of Shanghai Jiao Tong University School of Medicine (Shanghai, China).

Consent for publication

All authors give their consent for publication of this manuscript.

Competing interests

The authors declare that they have no competing interests.