Generation of ER+pcMSCs

In the present study, human placentas were donated by women who received cesarean sections in Taipei Medical University Hospital with procedures approved by the Institutional Review Board at Taipei Medical University (IRB Approval No. TMU-JIRB 201501063 and TMU-JIRB N202304143). Written informed consent was obtained from all donors and experiments were performed in accordance with relevant guidelines and regulations.

The pcMSCs from human placental tissues (provided by Prof. Thai-Yen Ling’s Lab in National Taiwan University) were isolated and characterized as described in our previous study [36]. In brief, the placental tissues were carefully dissected, and the harvested tissue pieces were washed several times with phosphate-buffered saline (PBS); minced; and enzymatically digested using digestion buffer containing DNase I, protease, and collagenase B in minimum essential medium (MEM, Thermo Fisher Scientific, NY, USA). The digested tissues were neutralized with 10% fetal bovine serum (FBS) in MCDB201 medium (Sigma-Aldrich, MO, USA) and filtered twice through a nylon membrane (pore size: 100 μm) and, subsequently a 100 μm cell strainer (BD Bioscience, NJ, USA) to remove undigested pieces. The mixture was centrifuged at 300 × g for 20 min, after which the supernatant was discarded. The cell pellets were resuspended and cultured in ITS and 10 mg/ml EGF in MCDB201 at 37°C and 5% CO2. The pcMSCs from passage 6 were subjected to phenotypic marker identification through flow cytometry to identify their MSC characteristics [36].

Generation of CM and E2-CM derived from ER+pcMSCs

To investigate the differences in protein concentrations between CM and E2-CM, pcMSCs were cultured in serum-free MCDB201 medium; for this procedure the seeding number (5 × 105 cells/10 cm dish) and harvest duration was set at 48 h. At approximately 50% confluence, the medium was replaced with fresh medium containing E2 (100 nM) for 48 h. The culture dishes were washed three times with PBS, and serum-free medium was added. The cells were cultured for another 48 h to obtain E2-CM. Normal CM was also harvested by using this procedure without the E2 priming. CM and E2-CM were then subsequently concentrated 10- or 50-fold by using an ultraconcentrator (Vivaspin 20 with 3000 Dalton molecular weight cutoff filters, Cytiva, USA) per the manufacturer’s instructions and were stored at − 80°C for future use.

Human granulosa-like cell line (KGN)

Cells from the human granulosa-like cell line (KGN, RCB1154), a steroidogenic human granulosa cell line, were purchased from RIKEN Bioresource Centre (Tsukuba, Japan). The physiological characteristics of granulosa cells are maintained in KGN cells [37]. These characteristics include the expression of functional FSH receptors and steroidogenic activity, such as estradiol production in response to FSH stimulation. The purchased KGN cells were cultured in Dulbecco’s Modified Eagle Medium: Nutrient Mixture F-12 (DMEM/F-12) supplemented with 10% fetal bovine serum (Corning, AZ, USA) and 1% antibiotic–antifungal 100 × antibiotic antimycotic solution (Life Technologies, USA) in 5% CO2 at 37°C. The cells were seeded at density of 15,000 cells/well in 96-well plates; after incubation for 24 h, the wells were treated with various concentrations (0–12 μg/ml) of 4-hydroperoxycyclophosphamide (4-OOH-CP), which is the active metabolite of CTX (39800-16-3, Cayman Chemical Company, USA).

Human umbilical vein endothelial cells

Human umbilical vein endothelial cells (HUVECs) were maintained in 10-cm fibronectin-coated dishes containing endothelial cell medium (ScienCell Research Laboratories, USA) supplemented with 5% fetal bovine serum, 50 μg/ml endothelial cell growth supplement, and 1% penicillin–streptomycin solution at 37°C and 5% CO2. All experiments were performed using HUVECs at passages 3–5.

Tube formation assay (thin layer method)

Geltrex LDEV-Free reduced growth factor Basement membrane matrix (Invitrogen, CA, USA) was thawed at 2–8°C overnight. Pre-chilled 24-well plates were coated with the matrix (50 µL/cm2 onto the growth surface of each well and incubated for 30 min at 37°C. The cells (2 × 104/cm2) were placed in 500 µL DMEM/F12 with various supplements: 10% FBS and endothelial cell growth supplement (positive control group), 20% CM in basal DMEM/F12, and 20% E2-CM in basal DMEM/F12. Anti-angiogenin antibody at 1/100 dilution (sc-74528, Santa Cruz, USA) was preincubated with CM (anti-ANG/CM group) or E2-CM (anti-ANG/E2-CM group) 1 h before tube formation assays were performed. Nonsupplement medium was used for the negative control group. After an additional 8 h of incubation, the endothelial cell tubular structure was formed. Tubes were labeled using 4-µg/ml (approximately 4 µM) Calcein AM Fluorescent Dye (354,217, Corning, USA) in 0.5-ml basal medium for another hour at 37°C. The tube areas from five random fields per well were photographed using a fluorescence microscope (Olympus BX51, Japan). The total tube length and total branching points were measured using the online software WimTube (Onimagin, Spain). This experiment was repeated three times in triplicate. Data are expressed as means ± standard deviations (SDs).

CTX-induced KGN granulosa cell injury model

To establish an in vitro granulosa cell injury model, the KGN cells were seeded and cultivated upon reaching 50–60% confluence, after which they were treated with 4-OOH-CP (8 µg/ml) for 12 h to induce cell apoptosis and senescence. The 4-OOH-CP-treated KGN cells were cultured with basic medium (DMEM/F12 mixed with 20% MCDB201 medium) and divided into three groups, namely the Mock group (supplemented with basic medium), CM group (supplemented with CM), and E2-CM group (supplemented with E2-CM). After 48 h of incubation, the cells were collected to enable analysis the markers of apoptosis and proliferation. Untreated KGN cells maintained in regular medium served as the control group.

CTX-induced C57BL/6J POI mouse model

The CTX-induced mouse POI model was established per the protocol described previously [38]. In brief, seven-week-old adult female C57BL/6J mice were purchased from the National Laboratory Animal Center, Taipei, Taiwan. The mice were fed a standard laboratory diet, provided with ad libitum access to water, and subjected to a 12:12-h light:dark cycle. To evaluate the effects of the pcMSC secretome on chemotherapy-induced POI, mice were randomly divided into 4 groups, namely the control, POI, CM, and E2-CM groups. The mice in the POI, CM, and E2-CM groups were intraperitoneally injected with CTX (Sigma-Aldrich, USA) resuspended in normal saline (50 mg/kg/day) for 14 consecutive days to establish a POI model [38], and the mice in the control group were injected with saline. To study the mRNA expression of clock genes in the ovaries and SCN, ovarian and SCN tissue sampling was performed every 4 h for 24 h, starting from Zeitgeber time 0 (ZT0). In each group, we conducted an average of three independent biological replicates per Zeitgeber time (ZT) point (six time points in total). The data for the 0-h and 24-h time points were identical [39]. All animal experiments conducted in the present study were approved by the Institutional Animal Care and Use Committee of Taipei Medical University (approval numbers: LAC-2021-0049 and LAC-2022-0171) and complied per the Guide for the Care and Use of Laboratory Animals.

Circadian locomotor activity measurement and analysis

We performed actimetry according to established procedures [40, 41]. In brief, the mice were individually housed in specially designed light-sealed boxes equipped with computer-controlled light-emitting-diode lighting and a ventilating fan, as described in the previous subsection [40]. The mice had ad libitum access to food and water, and the light inside the boxes was computer-controlled to match the conditions in the breeding room. Outside the enclosures, the animal room was illuminated solely with safety lights, with computer monitors shielded by safety films. Inside the box, circadian locomotor activity of each mouse was continuously monitored using a passive infrared (PIR) motion sensor and recorded at 1-min intervals. Scheduled light control and data acquisition were managed through an Arduino Mega 2560 microcontroller and custom software [42]. The mice underwent a habituation phase for one week (Week 1) in the boxes under a 12:12-h light:dark (LD) cycle; the lights were turned on at 7:00 a.m. and turned off at 7:00 p.m. Subsequently, daily intraperitoneal (IP) injections of CTX (50 mg/kg) were administered to induce the POI model during Week 2 under LD and Week 3 (constant darkness, DD) of the experiment. Subsequently, in Weeks 4 and 5, the mice in the CM and E2-CM groups received CM and E2-CM treatments every two days under DD. Intensive observation under DD was conducted in Week 6, and observation under an LD cycle was conducted in Week 7. On the final day, the entrainment reverted to DD, and the mice were sacrificed at circadian timepoints for the collection of ovaries and SCN tissues, as well as blood samples. IP injections (Weeks 3–5) and sample collections were conducted under dim safety lighting. Circadian locomotor activities were visualized side-by-side as double plots and spectrograms. Time-dependent period and rhythmicity were assessed using a sliding window fast Fourier transform to generate a circadian heatmap [41]. The resulting data were presented as spectrograms, showing the power spectrum over a range of periods for each time point. During the observation phase of Week 6 (DD), the circadian free-running period was estimated from the peak power in the FFT. The circadian power ratio (CPR) was calculated by dividing the area spanning 20 h and 28 h by the total area of the spectral power in the frequency domain.

Histology

The collected ovaries were fixed in 10% neutral buffered formalin at 4°C for 24 h. The specimens were then dehydrated, cleared in xylene, embedded in paraffin wax, serially sectioned into 5-μm-thick slices, and mounted on glass slides. Hematoxylin–eosin staining was performed on every tenth sample from three consecutive slides, sampled from a total of 20 ovaries across 4 experimental groups (n = 5), followed by blinded follicle counting. Whole-slide scanning at 200 × magnification was carried out using TissueFaxs (TissueGnostics GmbH, 1020 Vienna, Austria) for analysis. Follicles with a visible nucleus were counted and classified as primordial, primary, secondary, antral, or atretic. A primordial follicle was identified as an oocyte surrounded by a single layer of flat pregranulosa cells. Primary follicles had a single layer of cuboidal granulosa cells, whereas secondary follicles had multiple layers of cuboidal granulosa cells. In antral follicles, an antrum was present in the granulosa cell layers. Atretic follicles were identified by eosinophilia of the ooplasm, nuclear pyknosis of granulosa cells, cytoplasmic contraction, cytoplasmic vacuoles, and the dissociation of granulosa cells from the basal membrane [43]. To evaluate ovarian fibrosis, modified Masson’s staining was performed using the Modified Masson’s Trichrome Stain Kit (ScyTek Laboratories, USA). Specifically, tissue slides were cut from paraffin-embedded blocks, randomly selected (n = 4), then subjected to deparaffinization, rehydration, and staining according to the manufacturer's protocol. After mounting with resin, slides were photographed at 200 × magnification. ImageJ software was used to quantify the area of fibrosis indicated by collagen deposition (stained blue).

RNA isolation and real-time quantitative polymerase chain reaction

Ovarian and SCN tissues, pcMSCs, and KGN cells were collected and homogenized, and their total RNA was isolated using an EasyPure Total RNA Spin Kit (Bioman Scientific, Taiwan) per the manufacturer’s instructions. RNA quantity and quality were assessed using a NanoDrop spectrophotometer (Thermo Scientific, USA).

We synthesized cDNA using Moloney murine leukemia virus reverse transcriptase (M-MLV RT, M1705, Promega, USA). Ovary RNA (2000 ng) and SCN RNA (400 ng) were mixed with a reverse transcriptase master mix in a final volume of 25 μL. The reaction mixtures were incubated at 25°C for 5 min, 50°C for 60 min, and 70°C for 10 min.

Real-time quantitative polymerase chain reaction (PCR) was performed in 96-well plates on a LightCycler 96 (Roche Diagnostics, Switzerland) by using the Fast SYBR Green Master Mix (15,350,929, Applied Biosystems, MA, USA). PCR reactions were conducted using a mixture comprising 5 μL of SYBR Green I Master, 2.5 μL of RNAse-free water, 2 μL of 30 nM primer mix, and 0.5 μL of cDNA for a total volume of 10 μL. Three technical replicates were performed for each sample. The cycling conditions were as follows: an initial cycle was performed for 20 s at 95°C, followed by 40 cycles for 10 s at 95°C, 20 s at 60°C, and 20 s at 72°C. Melting curve analyses for determining the dissociation of PCR products were performed at between 65 and 95°C. The PCR efficiencies of the primers was evaluated by examining the samples through the use of six standards in triplicate (diluted in fivefold series). Two reference genes (Actb and Gapdh) were used for normalization when the expression of the clock genes was analyzed. Cq values were exported using the software LightCycler 96 (Roche Diagnostics, Switzerland) and analyzed using Microsoft Excel. The PCR efficiency factor (1 denoting 100% efficiency) and Cq values were quantified without any weighting. A PCR efficiency of 90–110% was regarded as acceptable [39]. Table S1 lists the primer sequences for the targeted genes in the real-time quantitative PCR.

Western blot analysis

The collected cells and ovaries were lysed with RIPA buffer (Energenesis Biomedical, Taiwan) along with a protease inhibitor cocktail (Roche Diagnostics, Switzerland) and phosphatase inhibitor cocktail (Roche Diagnostics). Total protein solutions were obtained by centrifuging the lysed samples at 14,000 rpm for 25 min at 4°C. The protein concentration was detected using the Pierce BCA assay kit (Thermo Fisher Scientific, USA) per the manufacturer’s instructions. Equal amounts of protein (30 µg) in sodium dodecyl sulfate (SDS) buffer were separated by applying electrophoresis to 10% SDS–polyacrylamide gel, after which the separated proteins were transferred to pure polyvinylidene fluoride membranes (Immobilon, Merck Millipore, Germany). The membranes were placed in 5% skimmed milk for 1 h to block nonspecific binding and were then incubated with corresponding primary antibodies (Table S2). They were then washed three times with tris-buffered saline (TBS) buffer (Bioman, Taiwan) containing 0.2% Tween 20 (Bioshop, Canada). After undergoing rinsing, the membranes were incubated with anti-rabbit or anti-mouse IgG secondary antibody (diluted 1:3000). Finally, they were visualized using Immobilon Western (Millipore, Germany) and imaged using the ImageQuant LAS 4000 mini system (GE Healthcare, USA). Band density was quantified using ImageJ software 1.53u (NIH Image, USA), β-actin was used as an internal control.

Immunohistochemical and immunocytochemical staining

Slides were incubated in a heating chamber at 60°C for 30 min, immersed in xylene, and rehydrated through a graded alcohol series. Deparaffinized sections were treated with 10 mM citrate buffer (pH 6.0) for 20 min at 98°C for antigen retrieval and were subsequently washed with TBS and permeabilization with TBST (0.2% Tween 20 in TBS solution). These sections were then blocked with 5% (v/v) normal goat serum (Vector Laboratories, S-1000, USA) for 1 h. After blocking, the slides were incubated overnight with the following primary antibodies: CYP19A1 (Abcam, ab18995, UK), PER2 (Abcam, ab227727, UK), RORA (Proteintech, 10616-1-AP, USA), and Rev-Erbα (NR1D1, Proteintech, 13906-1-AP, USA). Subsequently, they were incubated with goat anti-rabbit horseradish peroxidase–conjugated secondary antibody (Vector Laboratories, USA) for 1 h. Finally, the sections were stained with 3,3’-diaminobenzidine substrates (Vector Laboratories, USA).

For confocal fluorescence detection, the slides were subjected to incubation overnight with mouse anti-PCNA (Chemicon, CBL407, Germany) or double staining with cleaved caspase-3 (Cell Signaling, #9961, USA) by using the In Situ Cell Death Detection Kit, POD (TUNEL assay, Roche, Cat. No.11684817910, Switzerland). Double immunofluorescence staining was also performed using antibodies against CD31 (550274, BD Pharmingen, USA) and VEGF-A (ab183100, Abcam, UK). On the next day, after being washed with TBST, the sections were incubated for 1 h at room temperature with Alexa Fluor-594–labeled goat anti-mouse IgG/goat anti-rabbit IgG (Life Technologies, USA) and then counterstained with DAPI (4',6-diamidino-2-phenylindole). Fluorescence signals were detected using a Stellaris 8 confocal microscope (Leica, Germany).

KGN cell cultures grown on glass coverslips were fixed with 4% newly prepared paraformaldehyde for 10 min at room temperature, permeabilized with cold TBST for 10 min, and incubated with a blocking solution containing 5% normal goat serum in TBST for 1 h. The coverslips were incubated overnight at 4°C with primary antibodies for the proteins: Ki67 (Abcam, ab15580, UK) and cleaved caspase-3 (Cell Signaling, #9961, USA). Subsequently, the coverslips were incubated with goat anti-rabbit Alexa Fluor 488-conjugated IgG (Life Technologies, USA), and F-actin was labeled with Alexa Fluor 594 phalloidin (Invitrogen, A12381, USA). Fluorescence signals were detected using a confocal Stellaris 8 microscope (Leica, Germany). Table S2 presents the experimental conditions for the antibodies used for Western blotting or immunostaining.

Enzyme-linked immunosorbent assay

To perform enzyme-linked immunosorbent assays (ELISA), serum samples were obtained from the mice in each experimental group to evaluate the levels of AMH (MBS2507173, MyBioSource, CA, USA), E2 (MBS261250, MyBioSource, CA, USA), and FSH (MBS2507988, MyBioSource, CA, USA). Fresh CM and E2-CM from four consecutive passages were used to quantify angiogenin (ELH-ANG, Raybiotech, GA, USA). All assays were conducted using ELISA kits according to the manufacturers’ instructions.

Flow cytometric analysis

KGN cells were treated with Allophycocyanin (APC) Annexin V in a staining buffer containing propidium iodide (BioLegend, USA) per the manufacturer’s recommendations. The samples were analyzed using a BD FACSVerse system and the software BD FACSuite (CA, USA).

Cytokine array assay

The cytokines in CM/E2-CM were detected using antibody array-based technology (Human Cytokine Arrays C5, RayBiotech, USA). Each array was incubated with CM/E2-CM (700 µg/ml) at 4°C overnight in accordance with the manufacturer’s instructions. The signals on the membranes were imaged using the ImageQuant LAS 4000 mini system (GE Healthcare, USA). Integrated densities were measured using ImageJ 1.53u. The samples were normalized on the basis of six spots of positive controls and the background of the surrounding area.

Exosome isolation and characterization

To obtain the exosomes in CM and E2-CM, the ER+pcMSCs were cultured under serum-free conditions with or without E2 priming until they reached 80% confluence, after which they were maintained in new serum-free medium for 48 h. Cell debris was removed by centrifuging the samples at 300×g for 10 min at 4°C. The supernatant was filtered through membranes (pore size: 0.22 μm). The exosomes were isolated through the following steps. The medium was ultracentrifuged at 100,000×g at 4°C for 90 min (Optima L-90 K Ultracentrifuge, Beckman Coulter, USA), the supernatant was carefully collected into another tube, and the pellet was resuspended in 10 ml iced PBS. The suspension was ultracentrifuged again at 100,000×g and 4°C for 90 min, and the supernatant was carefully removed. The final pellets were resuspended in 250 μL iced PBS as crude exosomes. The exosomes present in fresh conditioned media (CM and E2-CM) were quantified by employing digital exosome counting technology (JVC Exocounter, Japan), in which anti-CD9 and anti-CD63 antibodies are used for exosome labeling. The system excludes larger particles, such as microvesicles, through a size-selective nano-structure on a disc with a 260-nm width. Each exosome is labeled with a single nanobead and detected through optical pickup, which was developed on the basis of Blu-ray technology.

Exosomal miRNA extraction, library preparation, sequencing and qPCR

Total RNA was extracted using TRIzol LS reagent (10296010, Invitrogen, CA, USA) per the manufacturer’s instructions. The RNA concentration was determined by using a spectrophotometer (ND-1000, NanoDrop Technology, USA) to measure absorbance at 260 nm. The quantity of RNA was evaluated by using a LabChip RNA 6000 kit (Agilent Technologies, USA) and the Bioanalyzer 2100 (Agilent Technology). Sample libraries were prepared using the QIAseq miRNA Library Kit (Qiagen, Germany) per the manufacturer’s guidelines, and subsequently they were sequenced using an Illumina instrument (75-cycle single-end read, 75SE). The miRNA data obtained in the present study have been deposited in the Gene Expression Omnibus database under the accession number GSE247568.

The collected sequencing data were analyzed using the Illumina BCL2FASTQ v2.20 software (Illumina, USA) for demultiplexing. We used Trimmomatic to isolate high-quality reads, discarding those shorter than 18 nucleotides [44]. The processed data were then analyzed using miRDeep2 [45]; during this process, the reads were aligned with the GRCh38 reference genome obtained from the University of California, Santa Cruz [46]. Given that human miRNAs can only align with a limited number of genomic locations, we only considered reads that matched the genome up to five times for miRNA identification [47]. We employed reads per million mapped reads to measure normalized miRNA expression. This model was derived by dividing the signal values of each miRNA by the total number of mapped reads. We used the MirTarget V4 tool [48] of miRDB version 6 [49] to predict miRNA targets. For this process, we only included functional human miRNAs from the FuncMir collection (http://www.mirdb.org/FuncMir.html). In addition, when searching multiple miRNAs or genes for target mining, we excluded gene targets with fewer than 60 target prediction scores and miRNAs with more than 2000 predicted targets in the genome. TargetScan (v8.0; targetscan.org [50, 51]) was employed to screen for miRNAs targeting genes of interest, with selection based on a TargetScan context +  + score ≤ − 0.2. For the differentially expressed miRNA analysis, we estimated the read counts of mature miRNA obtained from miRDeep2 Quantifier module. Then the read counts of each miRNA were normalized to total number of miRNAs as RPM (Reads per millions mapped reads) to compare the abundance between samples.

Total exosomal RNA extracted above, including miRNA, was used to conduct quantitative analysis of miRNA by qPCR. cDNA was synthesized using the miRCURY LNA RT Kit (Cat#339340, Qiagen, Hilden, Germany) according to the manufacturer’s recommendation. Reverse transcription reactions were performed using the miScript HiSpec Buffer in the kit, allowing selective conversion of mature miRNAs but not precursors. Mature miRNAs were polyadenylated by poly(A) polymerase and reverse transcribed into cDNA using oligo-dT primers. Polyadenylation and reverse transcription were performed in parallel in the same tube. The expression of the miRNAs of interest was assessed via qPCR using the miRCURY LNA SYBR Green PCR kit (Cat#339345 Qiagen, Hilden, Germany). U6 (RNU6-1) snRNA was chosen as the internal control for microRNAs. Primers were designed based on a miRNA-specific stem loop-RT assay [52]. Each experiment was performed at least in triplicate. Primer sequences for miRNAs are shown in the supplementary table (Table S3). Relative gene expression was calculated using the 2−ΔΔCq method.

Statistical analysis

All experiments were performed at least three times. The collected data were analyzed using GraphPad Prism (version 9.3, USA) and were presented as means ± SDs. An unpaired Student’s t test was conducted to perform between-group comparisons. One-way analysis of variance (ANOVA) and two-way ANOVA with Tukey’s test were performed to compare the results of multiple groups. A p-value < 0.05 was considered to indicate significance.