Ethics

The maintenance and handling of all mice used in this study was performed in the Max Planck Institute for Multidisciplinary Sciences (MPI-NAT) animal facility according to international animal welfare rules (Federation for Laboratory Animal Science Associations guidelines and recommendations). Requirements of formal control of the German national authorities and funding organizations were satisfied, and the study received approval by the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (33.19-42502-04-18/2746).

Animals

Mouse (Mus musculus) strains used in this study were CAG-TAG (Srinivas lab, Oxford University; C57BL/6J background; 23–28 days of age), Oct4–GFP (Mann lab, Beckman Research Institute of the City of Hope, Duarte, CA; C57BL/6J background; 23–28 days of age) and C57BL/6J (23–28 days of age). All mice were female. Sex was determined by visual inspection of anal–genital distance of mice by a trained animal technician. All mice were kept in rooms with a constant temperature of 21 °C and humidity of 55%. The light–dark rhythm was 12:12 h, from 5:00 to 17:00. Health monitoring was carried out in accordance with Federation of European Laboratory Animal Science Associations recommendations with large annual examinations in January and smaller-scale examinations in May and September. For immunofluorescence and scRNA-seq experiments, 23–28-day-old C57BL/6J mice were superovulated by injection of 0.2 ml of 25 IU ml−1 pregnant mare serum gonadotropin (THP Medical Products, #hor-272-a) followed 48 h later by injection of 0.2 ml of 25 IU ml−1 hCG (Intervet, Ovogest 1000).

Medium preparation

To prepare the basic culture and ovulation medium, 0.5 g of powdered Alpha-MEM (Thermo Fisher 12000014) and 0.11 g of sodium bicarbonate (final concentration 2.2 g l−1; Sigma-Aldrich, S5761) were dissolved in 50 ml of embryo-tested water (Sigma-Aldrich, W1503) and filter-sterilized using a 0.22 µm filter (Krackeler Scientific, SE2M228I04).

Isolation and culture of mouse antral follicles

Our follicle isolation and culture protocol was adapted from previously established methods11,15. To obtain antral follicles, ovaries from 23–28-day-old mice were collected in Alpha-MEM + 25 mM HEPES (Thermo Fisher, 42360032) with 100 IU ml−1 penicillin–streptomycin (Gibco, 15140122), 5% foetal bovine serum (FBS; Gibco, 16000-044) and 30 ng ml−1 (1 nM) FSH (National Hormone and Peptide Programme, #NIDKK-oFSH-20 (ovine FSH)) (isolation medium). Ovarian antral follicles were isolated as previously described15 with minor modifications. Follicles of 300–500 µm diameter were dissected using 30 G needles in the collection medium detailed above, in 35 × 10 mm non-treated sterile culture dishes (CytoOne, CC7672-3340), then cultured in a 37 °C, 5% CO2, 20% O2 incubator for 24 h in media droplets on 0.4 µm polytetrafluoroethylene (PTFE) membrane cell culture inserts (Millicell, PICM03050) in six-well plates containing 1.6 ml Alpha-MEM (Thermo Fisher, 12000014) prepared with 2.2 g l−1 sodium bicarbonate (Sigma-Aldrich, S5761), supplemented with 100 IU ml−1 penicillin–streptomycin (Gibco, 15140122), 5% FBS (Gibco, 16000-044), 1× Insulin-Transferrin-Selenium (ITS-G; Gibco, 41400-045) and 30 ng ml−1 (1 nM) FSH (National Hormone and Peptide Programme, #NIDKK-oFSH-20 (ovine FSH)) (culture medium).

Live imaging of ovulation in cultured follicles by confocal and two-photon microscopy

For ex vivo ovulation induction, follicles were transferred into Alpha-MEM (Thermo Fisher, 12000014) prepared with 2.2 g l−1 sodium bicarbonate (Sigma-Aldrich, S5761), supplemented with 100 IU ml−1 penicillin–streptomycin, 5% FBS (Gibco, 16000-044), 1× Insulin-Transferrin-Selenium (ITS-G; Gibco, 41400-045), 30 ng ml−1 (1 nM) FSH (National Hormone and Peptide Programme, #NIDKK-oFSH-20 (ovine FSH)), 8 µg ml−1 (5 IU ml−1) hCG (MSD Animal Health Ovogest 1,000 IU ml−1) and 5 ng ml−1 epidermal growth factor (EGF; Roche, 11376454001) (ovulation medium). The live imaging setup was constructed as follows. Standing feet were removed from the bottom of the 0.4 µm PTFE membrane cell culture inserts (Millicell PICM01050, 12 mm diameter) using scissors and attached to the bottom of a 35 mm glass-bottom dish (Ibidi, 81158) or a glass-bottom two-well slide (Ibidi, 80287) using imaging spacers (Grace Bio-Labs SecureSeal Imaging Spacers, 654002). Then, 1.3 ml of imaging medium was carefully pipetted into the dish and underneath the culture insert. Follicles were transferred into the re-equilibrated dish using glass capillaries with an inner diameter of 0.6 mm (Hilgenberg, 1411012), keeping them in individual droplets on the membrane, and the dishes were transferred to the microscope to start imaging as quickly as possible. The imaging chamber was supplied with 5.5% CO2 and heated such that the imaging medium measured at 37.5 °C. Wet tissues ensured sufficient humidity to prevent evaporation.

All microscopy was performed using ZEN Blue 2.3 (Zeiss). Images were acquired with LSM800 or LSM980 confocal laser scanning microscopes (Zeiss) with an environmental incubator box and a LD LCI Plan-Apochromat 25×/0.8 Imm Korr DIC M27 on the water-immersion setting in combination with Immersol W2010 immersion oil (Zeiss). To generate our two-dimensional datasets, we used tiling to cover the entire follicles and imaged two planes 30 µm apart at 10 min time intervals. For two-photon microscopy to generate our 3D datasets, follicles were imaged using a Zeiss LSM980 confocal laser scanning microscope with a Plan-Apochromat 20×/0.8 M27-Air and a MaiTai AX EHPDS two-photon laser (Spectra-Physics). For image acquisition, Myr–TdTomato was excited with the two-photon laser tuned to 1,000 nm at 30% power and detected using the GaAsP-BiG-non-descanned detector and a BP 570-610 nm bandpass filter. We imaged 27 slices through a z-stack of 260 µm with 4 tiled regions. Control and drug-treated groups were imaged on the same microscope. Care was taken to avoid phototoxicity and photobleaching. The step-by-step protocol for isolation, culture and imaging of mouse ovarian follicles is available on protocols.io58.

Drug perturbation

Drugs used in this study include 4-MU (0.5 and 1 mM; Sigma-Aldrich, M1381), mifepristone (100 µM; Sigma-Aldrich, M8046), JKC-301 (10 µM; Sigma-Aldrich, SCP0141), FCCP (10 µM; Sigma-Aldrich, C2920), blebbistatin (−) and (+) (100 µM; Tocris, 1852 and 1853), Y-27632 (30 µM; Tocris, 1254) and SB-3CT (250 µM; Sigma-Aldrich, S1326). All drugs except mifepristone (resuspended in ethanol) were resuspended in DMSO (Sigma-Aldrich, D2650). Dextran for medium supplementation (25 and 50 mg ml−1; Sigma-Aldrich, D5376) was resuspended in the ovulation medium. For follicle expansion experiments (4-MU, dextran), follicles were incubated for 1.5 h in culture medium supplemented with the drug or DMSO, then transferred into the imaging setup. For contraction and rupture experiments (mifepristone, JKC-301, FCCP, blebbistatin, Y-27632 and SB-3CT), follicles were cultured in ovulation medium for 6 h before transfer into an equivalent medium containing the drug (or DMSO or ethanol) and live imaging was started immediately afterwards.

Follicle volume measurements

To calculate follicle volume, images from 27 z-slices through the height of the follicle were segmented into (A) ‘follicle’ or (B) ‘background’ on the basis of the Myr–TdTomato signal using the pixel segmentation feature in Ilastik59. The segmented images were imported into Fiji/ImageJ60, and a mask was created with all holes in the signal filled. The images were 3D-reconstructed in Imaris, and volume measurements were extracted for all timepoints. In all main figure volume traces, follicle volumes were normalized to 100% at 30 min post hCG addition to represent how follicles expanded and contracted regardless of differences in starting volume. Raw and normalized individual traces for all control and drug-treated follicles are shown in Extended Data Figs. 4–6, 8 and 10.

Follicle microinjection

Antral follicles were microinjected in isolation medium using a setup that has previously been described61,62. The protocol was adapted for injection of a high-molecular-weight (500,000 MW/500 kDa) fluorescently tagged dextran into the antrum of the follicle, instead of into the follicle-enclosed oocyte. The fluorescent dextran (Thermo Fisher, D7144) was labelled using the Alexa Fluor 647 Protein Labeling Kit (Thermo Fisher, A20173) according to the manufacturer’s instructions.

Oocyte tracking

To track the oocyte, the imaging files were opened in Imaris. A surface was generated on the oocyte signal, and the movement of the centre of mass was tracked through all timepoints. Oocyte speed and total distance moved were extracted from Imaris and plotted.

Tracking fluid influx with Alexa Fluor 647 Hydrazide dye

Antral follicles were isolated and cultured as described in the "Isolation and culture of mouse antral follicles" in Methods. Follicles were then transferred into imaging dishes containing (A) ovulation medium, (B) culture medium (without hCG) or (C) ovulation medium supplemented with 4-MU (1 mM; Sigma-Aldrich, M1381). All three medium types were additionally supplemented with Alexa Fluor 647 Hydrazide (10 µM; Thermo Fisher, A20502). Follicles were then transferred immediately to an LSM800 confocal laser scanning microscope (Zeiss), and images were acquired at 10 min intervals. Following acquisition, mean Alexa Fluor 647 Hydrazide fluorescence intensity measurements were made between mural granulosa cells, avoiding any antral regions, in Fiji/ImageJ.

Immunofluorescence of cryosectioned follicles and ovaries

The staining protocol was adapted from previously published protocols27,63. Fresh ovarian follicles or ovaries from superovulated C57BL/6J mice were placed into plastic moulds (Leica HistoMold 6 × 8 mm, 14702218311) containing optimal cutting temperature (OCT) compound mounting medium for cryotomy (VWR Chemicals, 361603E), immediately frozen on dry ice and stored at −80 °C until sectioning. The frozen blocks were sectioned into 12 µm cryosections using a Cryostat (Leica, CM3050S), mounted onto microscopy slides (Thermo Fisher Scientific Menzel-Gläser Superfrost Plus, J1800AMNZ) and stored at −80 °C. Sections were fixed using a fixative containing 100 mM HEPES (pH 7.0, titrated with KOH), 50 mM EGTA (pH 7.0, titrated with KOH), 10 mM MgSO4, 4% methanol-free formaldehyde and 0.5% Triton X-100 in ddH2O at room temperature (RT) for 12 min, followed by three phosphate-buffered saline (PBS) washes for 5 min. Follicles were permeabilized in pre-chilled acetone (−20 °C) for 7 min, washed three times in PBS for 5 min each and blocked with 5% bovine serum albumin (BSA; Fisher Scientific, BP1605-100) in 1× PBT (PBS + 0.1% Tween) for 2 h at RT. For hyaluronan binding protein (HABP) staining, follicles were washed once with PBS + 5% BSA, stained using 1:50 dilution biotinylated HABP (b-HABP from Amsbio, AMS.HKD-BC41, 0.25 mg ml−1 stock concentration) in PBS with 5% BSA at 4 °C overnight. Sections were washed three times with PBS for 5 min, then incubated in PBS + 5% BSA with 1:500 Alexa Fluor 488-conjugated streptavidin (Thermo Fisher, S11223) for 2 h at RT. Sections were washed three times for 5 min in PBS + 5% BSA, stained with 100 µM Hoechst in PBS + 5% BSA for 5 min at RT, followed by three 5 min washes in PBS. For antibody incubations, blocked sections were stained with anti-smooth muscle actin (SMA) (rabbit polyclonal; Proteintech 55135-1-AP; 1:400; final concentration 1.5 μg ml−1) and anti-phospho-myosin light chain 2 (Ser19; mouse monoclonal; Cell Signaling #3675; 1:400) in blocking buffer (5% BSA in 1× PBT (PBS + 0.1% Tween)) overnight, washed using PBT, incubated with secondary antibodies (Alexa Fluor 647 chicken anti-mouse, Invitrogen A21200 or Alexa Fluor 488 donkey anti-rabbit, Invitrogen A31573, both at 1:500) for 1 h at RT, then washed for 1 h with PBT. All steps were carried out in a humid chamber. Samples were mounted using Prolong Glass Antifade Mounting Solution (Invitrogen ProLongGlass Antifade Mountant, P36984) and cover glasses (22 × 50 mm, 0.13–0.17 mm thick), and imaged within a week of staining using an LSM980 or LSM800 confocal laser scanning microscope (Zeiss) equipped with a multi-immersion LD LCI Plan-Apochromat 25×/0.8 Imm Korr DIC M27 (420852-9871-000) on the water-immersion setting or a C-Apochromat 40×/1.2 W Corr M27 (421767-9971-711) in combination with Immersol W2010 immersion oil (Zeiss).

HABP staining intensity measurements

For HABP intensity quantifications, all samples were handled at the same time and using the same treatment throughout (two biological replicates and one technical replicate per biological replicate). All images were acquired with the same imaging settings on the same microscope. All measurements were done in Fiji/ImageJ60. The signal was only compared for the granulosa cell compartment by drawing custom regions of interest, excluding HABP signal in the outer layers of the follicle.

Phosphorylated myosin light chain 2 occupancy analysis

For analysis of phosphorylated myosin light chain 2 occupancy on SMA, a mask was created using the SMA signal, and the intensity of both SMA and phosphorylated myosin light chain 2 signal was measured in the masked region. We then calculated the ratio between the intensities of SMA and phosphorylated myosin light chain 2 in the masked region.

Generation of single-cell suspensions and scRNA-seq

For the in vivo dataset, follicles were mechanically isolated from ovaries of superovulated 23–28-day-old C57BL/6J mice at 0, 3, 6, 9 and 12 h post-hCG administration. For the ex vivo dataset, follicles from C57BL/6J mice were collected from the ex vivo culture system at 0, 3, 6, 9 and 12 h post-hCG addition. In both cases, three to five follicles were collected in low-binding 1.5 ml tubes (Eppendorf, VB-0285). To initiate the cell dissociation process, 1 ml of Collagenase IV (Gibco, 17104019) at 10 mg ml−1 was added to the samples. Samples were incubated for 30 min at 37 °C at 500 rpm, inverted every 5 min and pipetted with a wide-orifice low-retention tip (Mettler Toledo) every 10 min. After incubation, samples were pipetted ten times with a wide-orifice low-retention tip, pelleted at 400g for 5 min and incubated with Accumax (PAN-Biotech, P10-21200) for 3 min at RT with constant tube inversion, followed by termination of the reaction with 500 µl PBS supplemented with 10% FBS. The sample was further dissociated by pipetting using a wide-orifice low-retention tip and a normal-bore low-retention tip. The sample was strained twice through a 35 µm filter into a 1.5 ml low-bind tube, pelleted at 400g for 5 min at 4 °C, washed once with 500 µl PBS without Mg2+/Ca2+, resuspended in 100 µl PBS with 0.4% BSA using a wide-orifice low-retention tip and strained using Flowmi (Merck, 136800040). Viability and concentration were assessed using Trypan Blue with Countess II (Thermo Fisher Scientific, AMQAX1000).

The single-cell suspensions were processed using the 10x Genomics Chromium Single Cell System and the Chromium Single Cell 3′ v3 Reagent Kits following the manufacturer’s instructions. Around 12,000 cells per sample were loaded into the reaction well to recover around 7,000 cells per library. Following partitioning into gel bead-in-emulsion in the Chromium controller and downstream library preparation, samples were processed for paired-end sequencing in a NovaSeq 6000 by the Sequencing Core Facility, Max Planck Institute for Molecular Genetics, Berlin, Germany.

Initial data processing and cluster annotation

Count matrices were converted to FASTQ files using the Cell Ranger software (version 3.0.2, 10x Genomics). Merging and quality check of the number of genes per cell, per cent mitochondrial content and number of total transcripts per cell were performed in Seurat v4.3.0.164. Based on the simultaneous inspection of all parameters, the following quality check cut-offs were established: per cent mitochondrial genes ≤12.5%, per cent ribosomal genes ≤30%, number of unique genes detected >500, and total number of molecules >1,000 and <150,000. Doublet detection and removal were run on each sample separately using DoubletFinder v2.0.365 with an assumed doublet formation rate of 5.6%. Overrepresented genes, such as Malat1, Gm42418 and AY036118, were filtered out to prevent technical bias. Normalization was performed with Scran v1.20.166, and Harmony v1.2.067 was used to integrate the datasets using the library preparation batch as the grouping variable.

Highly variable genes were identified using a mean variability plot with the following parameters: ‘mean.cutoff = (0.0125,3)’ and ‘dispersion.cutoff = (0.5,Inf)’. The best principal component analysis (PCA) dimension was chosen on the basis of either reaching over 90% cumulative variance with less than 5% individual variance, or if the difference in variance between successive components was greater than 0.1, whichever was lower. Uniform manifold approximation and projection (UMAP)68 was used on the Harmony embeddings for the downstream nonlinear dimensionality reduction. Clustering was performed using K-nearest neighbour graph69 and the Louvain algorithm70. The ideal cluster resolution was identified using Clustree v0.5.171.

Cluster annotation was performed by inspection of the expression of known marker genes, temporal distribution of the cell types and examination of upregulated GO terms, as well as by performing direct comparisons of clusters to differentiate between closely related cell types.

Differential gene expression and GO enrichment analyses

All differential gene expression analysis was performed using DESeq2 v1.32.072. Pseudobulk counts were generated with Libra v1.0.073. The following design was used for the direct comparison of timepoints: ‘~LPbatch + folliclecount + type + time’, where ‘LPbatch’ corresponds to the library preparation batch, ‘folliclecount’ to the number of follicles used in the sample, ‘type’ to the setup from which the follicles were isolated (either ex vivo or in vivo) and ‘time’ to the number of hours after hCG addition. Differential gene expression analysis was performed with the LRT test with ‘~LPbatch + folliclecount + type’ as the reduced variables. A gene was considered to be differentially expressed if the average absolute log2 fold-change of its expression (|avglog2FC|) was ≥0.25 and adjusted P value <0.05, and if it was expressed in at least 10% of the cells in the cluster. GO enrichment analyses were performed with the ‘enrichGO’ function from clusterProfiler v4.0.574 and using genes with at least one unique molecular identifier count as the background.

For global comparison of the two setups (ex vivo versus in vivo) at each timepoint, the Wald test in DESeq2 v1.32.071 with ‘~type’ as the design was used. Only genes that did not yield ‘NA’ P values (that is, genes with more than zero counts that are not outliers and with high mean normalized counts) in the differential gene expression analysis were used for the analysis. A gene was considered to be upregulated in a setup if its |avglog2FC| ≥0.5 and adjusted P value <0.01. Area-proportional Venn diagrams were generated with VennDiagram v1.7.375.

Figure preparation

Microscopy images shown in figures were processed using a Gaussian filter with a sigma of 1.0 in Fiji/ImageJ. Plots were generated using OriginPro 2022 (64-bit) SR1 9.9.0.225 and Graphpad Prism 9.3.1. All schematic diagrams were created with BioRender.com. Figures were assembled using Adobe Illustrator 27.1.1.

Statistics and reproducibility

Sample size was determined by the maximum number of follicles that fit on the imaging membrane. This was typically between 5 and 12 follicles each for controls and drug inhibitions, and 15–25 in total. No statistical methods were used to pre-determine sample sizes. Follicles were not used for live imaging if they were deemed unhealthy on the basis of both follicle and oocyte morphology. Each experimental condition reported in this study includes at least two experimental replicates. Some conditions consisted of three or more experimental replicates. Biological replicates are indicated in the manuscript alongside the data figures, in the figure legends, and/or in Methods. Where applicable, each experiment in this study contained internal controls to ensure that variability between experimental replicates would not bias the outcome of experiments. Most experiments used follicles collected from multiple mice pooled together before random assignment to control and treatment groups. Control (DMSO-treated) follicles were always cultured and handled alongside drug-treated follicles. Follicles belonging to each treatment condition were processed and analysed blindly to avoid bias. While researchers knew the treatments involved in experiments, analysis of acquired data was performed blindly to limit bias. Automated software was used in analysis where possible. For statistical analyses, P values derived from two-sided unpaired t-tests were calculated using OriginPro 2022 (64-bit) SR1 9.9.0.225; P values derived from chi-squared test (two-sided) with Yates’s correction were calculated manually. Data distribution was assumed to be normal, but this was not formally tested. All data points are shown on graphs in which statistical testing has been done.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.