TGFB2/IGFBP5 activated by transcription factors STAT2 and SMAD3 initiate granulosa cell degeneration and cause follicular atresia in chickens

Abstract

Follicular atresia is a crucial factor that affects the egg-laying performance of chickens, and granulosa cells (GCs) are the root cause of follicular atresia. Investigating the fate-determining genes of GCs can help improve egg production and laying period duration in chickens. Notably, transforming growth factor beta 2 (TGFB2) and insulin like growth factor binding protein 5 (IGFBP5) were previously identified as key genes responsible for GC degeneration. In this study, the promoters of TGFB2 and IGFBP5 were identified and subjected to DNA pull-down assays. The pulled-down proteins were then analyzed using mass spectrometry. Additionally, single-cell regulatory network inference and clustering (SCENIC) analysis results from previous single-cell RNA sequencing data were integrated to investigate the upstream transcriptional regulators of TGFB2 and IGFBP5. The results revealed that SMAD family member 3 (SMAD3) is a crucial transcription factor (TF) for IGFBP5, while signal transducer and activator of transcription 2 (STAT2) is a common TF for both TGFB2 and IGFBP5. The regulatory pathway of STAT2/TGFB2/SMAD3/IGFBP5 highlights the significant interaction between the janus kinase (JAK)/STAT and transforming growth factor-β (TGF-β)/SMAD signaling pathways during follicular atresia. Their synergistic regulation leads to GC degeneration, which causes follicular atresia. This study provides new evidence for identifying switch genes, further elucidates the regulatory mechanisms of follicular atresia, and offers new genetic targets for extending the laying period and improving the egg-laying performance.

Introduction

Increasing egg production and extending the laying period of chickens have long been scientific issues of concern. Egg-laying capacity is one of the most important economic traits of chickens. The number and proportion of atretic follicles (AFs) in chicken ovaries directly affect the laying rate and laying period duration. When hens begin to age, the number of AFs increases significantly, the number of follicles entering the hierarchical sequence decreases sharply, and the laying rate decreases rapidly. Therefore, delaying follicular atresia can effectively slow down the aging of hens and extend the laying period (Liu, et al., 2018; Pertynska-Marczewska and Diamanti-Kandarakis, 2017; Yao, et al., 2020). Granulosa cells (GCs) in the follicle have been widely recognized as the fundamental cause of follicular atresia. Apoptosis of GCs occurs earlier than that of oocytes and theca cells (TCs) (Yuan, et al., 2016). Further research has shown that abnormal autophagy in GCs precedes apoptosis. An abnormal increase in autophagy and imbalance in internal homeostasis can lead to GC apoptosis, ultimately resulting in follicular atresia (Zhao, et al., 2014; Zhou, et al., 2019).

The transforming growth factor-β (TGF-β) pathway plays a crucial role in numerous biological systems and is also extensively involved in the process of chicken follicle development (Li, et al., 2023; Nie, et al., 2022; Shen, et al., 2020; Yang, et al., 2022). TGF-β is a superfamily composed of structurally related multifunctional cytokines, including TGF-β proteins, activins, inhibins, bone morphogenetic proteins (BMPs), and growth differentiation factors (GDFs). In chickens, three types of TGF-β proteins are found, namely TGFB1, 2 and 3. TGFB1 regulates various ovarian functions under both normal physiological and pathological conditions. For example, it can increase egg weight (Kang, et al., 2002), maintain the germ cell pool (Zhou, et al., 2020), and stimulate GCs to secrete collagen (Zhou, et al., 2021). However, research on TGFB2/3 in chicken ovary is rare.

Insulin-like growth factor binding proteins (IGFBPs) are also major regulators of cell growth and proliferation and play a primary role in the “IGF transport and uptake by IGFBP” signaling pathway. In mammals, IGFBP family members are closely associated with follicular development (Ji, et al., 2022; Rubio, et al., 2021; Xiong, et al., 2017; Zhou, et al., 2003). In chickens, the members of this family include IGFBP1–5 and IGFBP7. This family has been screened out in many sequencing datasets related to follicular development (Ahmadi and Ohkubo, 2022; Hu, et al., 2024; Mohammadi and Ansari-Pirsaraie, 2016). Studies have shown that IGFBP2 can affect GC proliferation and progesterone (P4) synthesis (Hu, et al., 2024), promote primordial follicle activation (Ahmadi and Ohkubo, 2022), and positively affect the reproductive performance of aged laying hens (Mohammadi and Ansari-Pirsaraie, 2016). However, little research has been conducted on the specific functions and mechanisms of most of the other members.

In our previous single-cell RNA sequencing (scRNA-Seq) analysis of various follicular GCs, TGFB2 and IGFBP5 were screened as candidate switch genes and were verified to cause GC degeneration (Zhang, et al., 2024a). In this study, the promoter sequences of TGFB2 and IGFBP5 were used for DNA pull-down, and the pulled-down proteins were analyzed by mass spectrometry (MS). Combined with the results of single-cell regulatory network inference and clustering (SCENIC) analysis, the upstream transcriptional regulatory factors of TGFB2 and IGFBP5 were further explored. The results not only enrich the research on the regulatory mechanism of chicken follicular atresia but also provide a scientific basis for improving the egg-laying performance of chickens.

Materials and Methods

Animals and cells

The chickens used in the experiment were yellow-bearded chickens (Zheng, et al., 2023), which were provided by the experimental chicken farm at China Agricultural University. All chickens were provided ad libitum access to food and water. To control for variables, all chickens were sacrificed at approximately 3:00 p.m. The experimental process complied with the rules of the Animal Welfare Committee of the State Key Laboratory of Agricultural Biotechnology at China Agricultural University (License No.: XK257).

The chicken fibroblast cell line (DF-1) has been preserved in our laboratory for a long time. GCs are primary cells isolated from chicken follicles.

Construction of the pySCENIC reference database

Files for the areas 5 kb upstream and downstream of the gene were prepared (http://hgdownload.soe.ucsc.edu/goldenPath/galGal6/bigZips/ (upstream5000.fa)). Then, the motif file was prepared (https://resources.aertslab.org/cistarget/motif_collections/v10nr_clust_public/singletons/ (motifs.tbl)). The Motif list was then prepared and stored in a file according to the motif file name motifs.lst. The reference command was as follows: create_cistarget_motif_databases.py -f upstream5000.fa -M singletons/ -m motifs.lst -o gallus -t 20. The constructed reference data were as follows: gallus.regions_vs_motifs.rankings.feather; gallus.regions_vs_motifs.scores.feather; gallus.motifs_vs_regions.scores.feather.

Analysis process of cellular transcription factor activity

We then prepared the transcription factor (TF) file (https://resources.aertslab.org/cistarget/motif2tf/motifs-v10nr_clust-nr.chicken-m0.001-o0.0.tbl) and motif annotation file (https://resources.aertslab.org/cistarget/motif2tf/motifs-v10nr_clust-nr.chicken-m0.001-o0.0.tbl). The gene expression matrix file was then generated as follows: 1) extract the gene expression matrix from the Seurat file object into a file: seurat_rnacount.csv; 2) use Python to convert the file from Step 1) to rnacount.loom.

The gene regulatory network (GRN) was then constructed. Using the grn subcommand of pySCENIC, the expression matrix and list of TFs were input, and based on co-expression, a regulatory network between TFs and potential target genes was constructed. The following reference command was used: pyscenic grn –num_workers 40 –output adj.sample.tsv –method grnboost2 rnacount.loom TFs.txt

Gene motifs were then validated. Regulatory network modules (regulons) were validated through motif discovery. This step was completed using the ctx subcommand. The reference command was as follows: pyscenic ctx adj.sample.tsv gallus.regions_vs_motifs.rankings.feather –annotations_fname motifs.tbl –expression_mtx_fname rnacount.loom –mode "dask_multiprocessing" –output reg.csv –num_workers 40 –mask_dropouts.

Cell activity was then calculated. The Area Under the Curve for Cells (AUCell) subcommand was used to calculate the activity of regulons. This step involved mapping the activity of TFs to each cell type to analyze the TF activity of different cell types. The reference command was as follows: pyscenic aucell rnacount.loom reg.csv –output aucell.csv –num_workers 40.

Vector construction

Follicles from yellow-bearded chickens were collected, and DNA was extracted using a TIANamp Genomic DNA Kit (TIANGEN, Beijing, China) according to the manufacturer’s instructions. The predicted promoter regions were amplified from the five flanking regions of TGFB2(NC_052534.1) and IGFBP5(NC_052538.1) according to the homologous recombination principle. The primer sequences are shown in Table 1, and the promoter sequences of TGFB2 and IGFBP5 are provided in Supplementary Files S1 and S2, respectively. The promoter region of TGFB2 spans from -2025 to +258, and that of IGFBP5 spans from -2048 to +43. The amplified fragments were inserted into the pEGFP-N1 vector digested with Ase I/Xho I (Takara Bio, Beijing, China). The constructed vectors were verified using sequencing (Beijing TsingKe Biotech Co., Ltd., Beijing, China) and named pTGFB2-EGFP and pIGFBP5–EGFP.

Table 1.

Information on amplification primers.

Gene	Primer name	Sequence	Product length
TGFB2	F-TGFB2	GCCATGCATTAGTTATTAATGTGGAGCAGTTCAGGCATGT	2323bp
	R-TGFB2	ATTCGAAGCTTGAGCTCGAGAGCCTAGTGAGAGTCGGGG
IGFBP5	F-IGFBP5	GCCATGCATTAGTTATTAATGAAAGCTCCCGGACAGACTA	2329bp
	R-IGFBP5	TCGAAGCTTGAGCTCGAGGATCTCCCAACACTTTTCCCCTG

Note: The gray areas are vector homologous sequences.

Cell culture and transfection

DF-1 was cultured at 37°C in 5% CO₂ using complete medium (Dulbecco's Modified Eagle Medium [DMEM] [Gibco, Waltham, USA] + 10% Fetal Bovine Serum [FBS] [Gibco, Waltham, USA] + 1% Penicillin-Streptomycin Solution [PS] [Gibco, Waltham, USA]). When the cells reached approximately 60% confluence, the plasmids were transfected. The following groups were established: pEGFP-N1 (positive control), pTGFB2-EGFP, pIGFBP5-EGFP, and pLinker-EGFP (negative control).

Quantitative reverse transcription PCR (qRT-PCR)

The GCs of small white follicle (SWF), atretic small white follicle (ASWF), small yellow follicle (SYF), atretic small yellow follicle (ASYF) and follicle 6 (F6) were collected. All GC layers were peeled off under a stereomicroscope to ensure purity. The details of GC separation can be found in the Materials and Methods section and the operation video of our previously published articles (Zhang, et al., 2024a). For total RNA extraction, TRIzol reagent manufactured by TIANGEN (Beijing, China) was employed. Reverse transcription was conducted utilizing the FastKing gDNA Dispelling RT SuperMix Kit, also from TIANGEN. qRT-PCR was performed with the 2 × Universal SYBR Green qPCR Premix supplied by Accurate Biotechnology Co. Ltd. (Changsha, China). GAPDH served as the reference gene for normalization purposes. The primer sequences utilized are detailed in Table S1. Data analysis was carried out employing the 2^‐ΔΔCt methodology.

Protein extraction

AFs were collected and washed with PBS. AF pieces were placed into pre-chilled tissue lysis buffer (containing protease inhibitors) (Beyotime Biotechnology, Nanjing, China), homogenized on ice using a tissue homogenizer or ultrasonic disruptor until no visible tissue chunks remained, and incubated on ice for 10–15 min to ensure complete cell lysis. The lysate was centrifuged at low speed (500–1000 × g, 4°C, 5–10 min) to remove unhomogenized tissue and debris, and then the supernatant was collected, transferred to a new centrifuge tube, and centrifuged at high speed (3000–5000 × g, 4°C, 10–15 min) to pellet the nuclei. The supernatant was then discarded, and the nuclear pellet was retained, resuspended in nuclear protein extraction buffer (containing protease inhibitors) (Beyotime Biotechnology, Nanjing, China), and incubated on ice for 30 min with intermittent shaking or vortexing to fully release nuclear proteins. Then, the solution was centrifuged at high speed (12000–15000 × g, 4°C, 15 min), and the supernatant (containing nuclear proteins) was collected. The protein concentration was determined using the Bicinchoninic Acid Assay (BCA) (Beyotime Biotechnology, Nanjing, China), and protein samples were aliquoted and stored at -80°C for future use.

Western blot

Nucleoprotein (34.2 μg) was loaded and separated using FuturePAGETM 4–20% gel (ACE Biotechnology, Changzhou, China) and transferred onto a polyvinylidene fluoride membrane (Merck Millipore, Billerica, USA). The membrane was blocked with 5% skimmed milk for 1 h and then incubated overnight at 4°C with corresponding primary antibodies Histone H3 rabbit pAb (1:5000) (Proteintech, Wuhan, China), followed by incubation with goat anti-rabbit secondary antibodies (Solarbio, Beijing, China) for 1 h at room temperature (RT).

Biotin-streptavidin pull-down assay

Biotin-labeled primers were designed and synthesized based on the promoter sequences of TGFB2 and IGFBP5. The plasmids constructed in section “Vector construction” were used as templates for PCR amplification to obtain crude biotin-labeled DNA products. A DNA Gel extraction kit was then used to recover the DNA from the gel. Nucleic acid-compatible streptavidin magnetic beads were prepared and washed. After setting up control and experimental tubes, 450 μL extracted protein was added to each tube, with protein dilution buffer added to a volume of 1 mL. The tubes were incubated overnight (approximately 16 h) at 4°C on a silent mixer. One hundred microliters of lysis buffer was reserved as the input group. The tubes were placed on a magnetic rack for 1 min, and the supernatant was discarded. Then, 1 mL of protein dilution buffer was added to resuspend the magnetic beads; the tubes were placed on a magnetic rack again for 1 min, and the supernatant was discarded. This washing step was repeated five times. Subsequently, 100 μL elution buffer was added to both the control and experimental tubes. After mixing, the tubes were incubated in boiling water for 8–10 min. The tubes were then placed on a magnetic rack for 2 min. The supernatant was transferred into new eppendorf tubes, which were the pull-down products, and marked as the control and experimental groups. Then, 20 μL 6 × loading buffer was added to each of the two tubes and incubated in boiling water for 8–10 min, and 20 μL 6 × loading buffer was added to the reserved 100 μL lysis buffer for the input group and incubated in boiling water for 8–10 min. The samples were then stored at -20°C. Silver staining and MS identification were then performed. The protein-DNA-streptavidin-agarose complex was analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE).

SDS-PAGE and silver staining

After performing SDS-PAGE, the gels were rinsed thoroughly with clean water. Deionized (DI) water was added to cover the gel, which was then shaken on a decolorizing shaker at RT for 5 min. Then, the DI water was discarded. Fixative solution was added to cover the gel, which was fixed at RT for 30 min. The fixative solution was discarded, and the cells were washed with DI water at RT for 5 min. This washing step was repeated once. Sensitizing solution was then added, and the mixture was shaken at RT for 30 min. The sensitizing solution was discarded, and the cells were washed with DI water at RT for 2 min. This washing step was repeated once. The staining solution was added and incubated at RT for 20 min. The staining solution was discarded; DI water was added, and the washing step was repeated twice. Chromogenic solution was added and incubated at RT for approximately 2 min. The solution was discarded when it became turbid; new chromogenic solution was added, and the process was continued until the target bands were clear.

Sample preparation for MS

The reaction solution (2.5% SDC/100 mM Tris-HCl, pH 8.5) was added to the sample and incubated at 95°C for 10 min. The sample was centrifuged, and the protein in the supernatant was precipitated based on a trichloroacetic acid precipitation assay. The precipitation was resuspended in redissolved solution (1% SDC/100 mM Tris-HCl, pH = 8.5) and incubated at 60°C for 30 min to complete reduction and alkylation. The resulting mixture was centrifuged, and the supernatant was diluted with an equal volume of ddH₂O. Trypsin was added at a ratio of 1:50 (enzyme: protein, w/w) for overnight digestion at 37°C. The next day, trifluoroacetic acid was used to reduce the pH to 6.0 and end the digestion. After centrifugation (12000 g, 15 min), the supernatant was purified using an in-house sulfonated styrene-divinylbenzene copolymer reversed phase desalting column. The peptide eluate was vacuum dried and stored at -20°C for later use. This work was supported by GeneCreate (WuHan GeneCreate Biological Engineering Co., Ltd.).

MS analysis

All specimens underwent analysis utilizing an UltiMate 3000 RSLCnano system, which was interfaced online with a Q Exactive HF MS equipped with a Nanospray Flex ion source, all sourced from Thermo Fisher Scientific. The peptide samples were introduced into a C18 trap column measuring 75 µm × 2 cm, featuring a 3 µm particle size and 100 Å pore size, also from Thermo. Subsequent separation occurred within a reversed-phase C18 analytical column, custom-packed in-house using ReproSil-Pur C18-AQ resin with dimensions of 75 µm × 25 cm, possessing a 1.9 µm particle size and 100 Å pore size. The mobile phases employed were A (comprising 0.1% formic acid, 3% dimethyl sulfoxide [DMSO], and 97% H₂O) and B (consisting of 0.1% formic acid, 3% DMSO, and 97% acetonitrile), utilized to establish a separation gradient at a flow rate of 300 nL/min. The MS operation was conducted in DDA top20 mode, encompassing a full scan range from 350 to 1500 m/z. For the full MS scan, the AGC target value was set at 3E6 charges, with a maximum injection time of 30 ms and a resolution of 60,000 at m/z 200. The precursor ion selection window was maintained at 1.4 m/z, and fragmentation was executed via higher-energy collisional dissociation at a normalized collision energy of 28. Fragment ion scans were documented at a resolution of 15,000, with an automatic gain control of 1E5 and a maximum fill time of 50 ms. Dynamic exclusion was activated and configured for a duration of 30 s. This research was facilitated by GeneCreate, specifically WuHan GeneCreate Biological Engineering Co., Ltd.

Database search

Raw MS data were analyzed with MaxQuant (version 2.2.0.0) using the Andromeda database search algorithm. Spectra files were searched against the Gallus_gallus_Chicken protein sequence database downloaded from Uniprot (20241213) using the following parameters: variable modifications were set as carbamidomethyl (C)-57.021464, oxidation (M)-15.994915, acetyl (protein N-term)-42.010565, and deamidation (NQ)-0.984016; fixed modification was set as carbamidomethyl (C)-57.021464; enzyme digestion specificity was set to trypsin/P with a maximum of two missed cleavages; peptide mass tolerance in the first search and main search was set to 20 and 4.5 ppm, respectively; and fragment match tolerance was set to 20 ppm. Unique peptides incorporated into one proteome based on MaxQuant cannot be used to distinguish proteins. The search results were filtered with a 1% false discovery rate (FDR) at both the peptide and protein levels.

Statistical analysis

Proteins denoted as reverse or potential contaminants or only identified by site were removed. Further bioinformatics analysis was conducted using the R statistical programming environment. Functional annotations were performed using the Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), EggNOG, Pfam, and UniProt subcellular localization databases.

Values of qRT-PCR are expressed as the mean ± standard error of the mean. One-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test was used to compare statistical data between groups that conformed to a Gaussian distribution and had equal standard deviations (SDs). Non-parametric tests followed by Dunn’s multiple comparison test were applied if the data did not conform to a Gaussian distribution. If the data conformed to a Gaussian distribution but the SDs were not equal, Brown–Forsythe and Welch ANOVA tests followed by Dunnett’s T3 multiple comparisons were used. GraphPad Prism 9.5.1 software (La Jolla, CA, USA) was used to plot the statistical data. Statistical significance was set at P < 0.05 (ns = not significant; *P < 0.05; **P < 0.01; ***P < 0.001)

Results

TF characteristics of GCs during follicular atresia analyzed using SCENIC

Based on the previous research results (Zhang, et al., 2024a), SCENIC was used to analyze TFs in the following five samples: SWF, ASWF, SYF, ASYF, and F6. The regulon (regulatory network that includes a set of TFs and their target genes) score trends of all TFs classified by sample or cell subtype were consistent with the expression trends of differentially expressed genes (DEGs) from previous studies (Figure S1a, b). Specific regulons classified by sample type and cell subtype were similar (Figure S1c, d). The top five regulons (according to the AUCell ranking) in each group displayed a partial overlap under the two classification methods. For example, the top five regulons of SWF combined the top five regulons of GCs-Ⅰ (Type 1 GCs) and GCs-Ⅱ (Type 2 GCs) (Figure 1a, b, f). The top five regulons of SYF were combined with those of GCs-Ⅱ and GCs-Ⅲ (Type 3 GCs) (Figure 1b, c, h). F6 completely overlapped with GCs-Ⅲ (Figure 1c and j). Significant overlap was observed between ASYF and GCs-Ⅳ (Type 4 GCs) (Figure 1d, i), and a large overlap was observed between ASWF and dGCs (degenerated GCs) (Figure 1e, g). These results confirmed the accuracy of the cell classification and were consistent with the biological background. Among the subtypes of GCs, GCs-Ⅳ (a subtype before degeneration) has received significant attention because it might be the key subtype triggering follicular atresia. Therefore, specific TFs and their target genes in GCs-Ⅳ were analyzed. In the UMAP plots of the cell profiles of all samples, the top 5 TFs of GCs-Ⅳ are shown as follows: SMAD3 (Figure 2a), FOSL2 (Figure 2b), ETV4 (Figure 2c), MAF (Figure 2d), and JUN (Figure 2e). The regulon scores of these five TFs were consistent with their mRNA expression levels, and the results indicated that their AUCell scores or mRNA expression levels were higher at GCs-Ⅳ and dGCs locations. The analysis of cell profiles, cell subtype identification, cell clustering, cell annotation, and pseudotime trajectory can be found in our previous reference (Zhang, et al., 2024a). The target genes were mainly enriched in signaling pathways such as MAPK, TGF-β, GnRH, WNT, tight junction, and Notch, and their functions were primarily associated with cell differentiation and migration, intercellular adhesion, cell junctions, and extracellular matrix.

Figure 1.

Top five regulons of each group under the classification of samples and cell subtypes. a∼e, RSS of different regulons in different GC subtypes. X-axis represents the different regulons, and Y-axis represents the RSS. A higher RSS value indicates more specific expression of this regulon in this subtype. Top five TFs are shown. f∼j, RSS of different regulons in different samples.

RSS: regulon specificity score, GCs: granulosa cells, GCs-Ⅰ: proliferative GCs subtype, GCs-Ⅱ: transitional subtype between GCs-Ⅰ and GCs-Ⅲ, GCs-Ⅲ: mature GCs subtype, GCs-Ⅳ: the subtype before degeneration, dGCs: degenerated GCs, TF: transcription factor.

Figure 2.

Top five regulons of GCs-Ⅳ. a∼e, the first column shows the UMAP visualizations of the AUCell scores. A higher AUCell score corresponds to stronger target gene set activity in the cell. The second column shows the gene expression of the top five regulons in GCs-Ⅳ; the violin plot shows the distribution of TF expression in different clusters, and the last column lists the possible motif sequences.

UMAP: Uniform Manifold Approximation and Projection; AUCell: Area Under the Curve; GCs-Ⅳ: subtype before degeneration; TF: transcription factor.

Promoter region determination of TGFB2 and IGFBP5

In a previous study, TGFB2 and IGFBP5 were screened as candidate switch genes for follicular atresia, and they have been shown to cause GC degeneration. To further explore the upstream regulatory factors, we investigated the proteins binding to the promoter regions of TGFB2 and IGFBP5. The amplified promoter sequences of TGFB2 and IGFBP5 were used to replace the CMV promoter in the vector pEGFP-N1 (Figure S2a∼e). The plasmids pTGFB2-EGFP, pIGFBP5-EGFP, pEGFP-N1 (positive control), and pLinker-EGFP (negative control) were transfected into DF-1 cells. At 24 h after transfection, EGFP expression was detected in the pTGFB2-EGFP and pIGFBP5-EGFP groups but not in the negative control group pLinker-EGFP, which lacked a promoter (Figure S2f). This indicates that the cloned 5′-flanking regions of the TGFB2 and IGFBP5 genes can initiate the expression of EGFP and possess promoter activity.

STAT2 is a crucial upstream TF of TGFB2

A biotin-labeled probe was designed for the TGFB2 promoter as the experimental group. Meanwhile, control (unlabeled probe) and input groups were set up (Figure S3a). Nuclear proteins from AFs were extracted for the DNA pull-down assay (Figure S3c, d, e), and the pull-down proteins were subjected to MS. A total of 2,065 proteins were identified. The number of proteins and peptides identified in the experimental and control groups is shown in Figure 3a. These proteins were mainly enriched in GO/KEGG/COG terms such as cellular aromatic compound metabolic process, metabolic pathways, focal adhesion, tight junctions, apoptosis, autophagy, lipid transport, and metabolism (Figure 3b, c, d), which aligns with previous scRNA-Seq analysis results. Of these, 255 proteins were specifically pulled-down in the experimental group (Figure 3e). A joint analysis of these specific proteins and the TFs identified in a previous SCENIC study revealed that STAT2 is not only a differentially regulated TF in AF-GCs (Figure S1a) but also binds to the TGFB2 promoter region. The regulon activity of STAT2 and gene expression patterns of STAT2 and TGFB2 were similar (Figure 3f, g, h). The qRT-PCR results further confirmed that STAT2 and TGFB2 were highly differentially expressed in the GCs of AFs (Figure 3i, j), suggesting that STAT2 is an important upstream TF of TGFB2.

Figure 3.

STAT2 is a crucial upstream TF of TGFB2. a, Numbers of identified proteins and peptides in the MS results of TGFB2-pulldown. Ctl: control group; Exp: experimental group. b, GO annotation of all proteins. c, COG homologous relationship annotation of all proteins. d, KEGG annotation of all proteins. e, Venn diagram of all proteins in the control and experimental group. f, UMAP visualizations of the STAT2 AUCell scores. g, UMAP visualizations of the STAT2 gene expression. g, UMAP visualizations of the TGFB2 gene expression. i, qRT-PCR results of STAT2 in GC layers at different stages. j, qRT-PCR results of TGFB2 in GC layers at different stages. (ns = no significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

UMAP: Uniform Manifold Approximation and Projection, AUCell: Area Under the Curve, GO: Gene Ontology, KEGG: Kyoto Encyclopedia of Genes and Genomes, COG: Clusters of Orthologous Groups of Proteins.

SMAD3 and STAT2 are crucial upstream TFs of IGFBP5

The same DNA pull-down assay was performed for the IGFBP5 promoter (Figure S3b, c, d, f), and the pull-down proteins were subjected to MS. In total, 2,438 proteins were identified. The number of proteins and peptides identified in the experimental and control groups are shown in Figure 4a. These proteins were mainly enriched in GO/KEGG/COG terms such as metabolic pathways, focal adhesion, tight junctions, phagosomes, various disease processes, lipid transport and metabolism, and defense mechanisms (Figure 4b, c, d), which aligns with the previous scRNA-Seq analysis results. Among these, 185 proteins were specifically pulled-down in the experimental group (Figure 4e). Joint analysis of these proteins and the TFs identified in a previous SCENIC study revealed that STAT2 and SMAD3 are not only differentially regulated TFs in AF-GCs (Figure S1a) but also bind to the IGFBP5 promoter region. The regulon activity of SMAD3 and gene expression patterns of SMAD3 and IGFBP5 were similar (Figure 4f, g, h). SMAD3 had the highest regulon score among all TFs in the AFs and GC-IV subtypes (Figure 1d, i). qRT-PCR results further confirmed that SMAD3 and IGFBP5 were highly differentially expressed in the GCs of AFs (Figure 4i, j). These findings suggest that SMAD3 and STAT2 are crucial TFs during follicular atresia and are important upstream TFs of IGFBP5.

Figure 4.

SMAD3 and STAT2 are crucial upstream TFs of IGFBP5. a, Numbers of identified proteins and peptides in the MS result of IGFBP5-pulldown. Ctl: control group; Exp: experimental group. b, GO annotation of all proteins. c, COG homologous relationship annotation of all proteins. d, KEGG annotation of all proteins. e, Venn diagram of all proteins in the control and experimental group. f, UMAP visualizations of the SMAD3 AUCell scores. g, UMAP visualizations of the SMAD3 gene expression. g, UMAP visualizations of the IGFBP5 gene expression. i, qRT-PCR results of SMAD3 in GC layers at different stages. j, qRT-PCR results of IGFBP5 in GC layers at different stages. (ns = no significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

UMAP: Uniform Manifold Approximation and Projection, AUCell: Area Under the Curve, GO: Gene Ontology, KEGG: Kyoto Encyclopedia of Genes and Genomes, COG: Clusters of Orthologous Groups of Proteins.

Combined effect of STAT2/TGFB2/SMAD3/IGFBP5 causes GC degeneration

SMAD3, as a key downstream effector in the TGF-β signaling pathway, is regulated by TGFB2. Moreover, SMAD3 is a transcriptional regulator of IGFBP5, indicating that TGFB2 regulates IGFBP5 through SMAD3. Previous functional validation studies have shown that overexpression of either TGFB2 or IGFBP5 leads to consistent changes in GC functions. Therefore, the TGFB2/SMAD3/IGFBP5 regulatory pathway reveals that the combined action of the TGF-β and “IGF transport and uptake by IGFBP” pathways contribute to GC degeneration. Additionally, STAT2, a key downstream effector in the JAK/STAT pathway, is an upstream TF of both TGFB2 and IGFBP5. The mRNA expression trend of STAT2 in the scRNA-Seq results was consistent with that of TGFB2 and IGFBP5 (Figure 3g, 3h, 4h), which was further verified using qRT-PCR (Figure 3i, 3j, 4j). This suggests that STAT2 can regulate GC degeneration not only through the TGFB2/SMAD3/IGFBP5 pathway but also by directly regulating IGFBP5. These results demonstrate that the complex synergistic interactions among STAT2, TGFB2, SMAD3, and IGFBP5 contribute to GC degeneration (Figure 5).

Figure 5.

Pathway of GC degeneration (by Figdraw, an online drawing tool).

GCs: granulosa cells; GCs-Ⅰ: proliferative GCs subtype; GCs-Ⅱ: transitional subtype between GCs-Ⅰ and GCs-Ⅲ; GCs-Ⅲ: mature GCs subtype; GCs-Ⅳ: subtype before degeneration; dGCs: degenerated GCs; FC: free cholesterol; SYF: small yellow follicle; OE: overexpression.

Discussion

Exploring the regulatory mechanisms of follicular atresia hinges on elucidating the degeneration program of GCs, which is of great significance for improving the egg-laying performance of chickens and extending the laying period. In a previous study, the scRNA-Seq results of GCs from chicken follicles were used to identify TGFB2 and IGFBP5 as candidate switch genes for follicular atresia. In vitro experiments demonstrated that both can induce GC degeneration and present consistent effects on multiple functions of GCs (Zhang, et al., 2024a). In this study, the scRNA-Seq data were further analyzed using SCENIC. Combined with the MS results from the DNA pulldown of the promoter regions of TGFB2 and IGFBP5, STAT2 was revealed as an important upstream TF of TGFB2 and SMAD3 and STAT2 were identified as important upstream TFs of IGFBP5.

SMAD is a downstream effector TF in the TGF-β signaling pathway. As a key node gene in this pathway, it is commonly used to detect pathway activity. TGF-β is transcribed and translated within cells and then released into the extracellular space, where it becomes activated. The activated TGF-β proteins are bound by their membrane receptors. The phosphorylated TGF-β receptors recruit and activate downstream SMAD proteins, and phosphorylated SMAD proteins then accumulate in the nucleus and function as TFs to regulate gene expression (Akhurst and Hata, 2012; Wang, et al., 2022b). In chickens, SMAD include SMAD1–10. BMP15 participates in the activation of primordial follicles in chicks through the SMAD1, SMAD5, and SMAD4 signaling pathways (Zhao, et al., 2022). In addition, corticosterone can inhibit the proliferation of chicken ovarian prehierarchical GCs and induce apoptosis by blocking SMAD1/ID3 signal transduction (Yang, et al., 2022). BMP4 is highly expressed in goose GCs and phosphorylates SMAD8, and activated SMAD8 binds to SMAD4 and translocates to the nucleus to regulate the expression of BAMBI and promote lipid synthesis (Wei, et al., 2023). Few studies have reported the specific functions of other SMAD family members in poultry follicles. Most studies have assessed the activity of the TGF-β pathway by detecting whether the SMAD signal is activated (Fu, et al., 2025; Qin, et al., 2015). The importance of the TGF-β pathway in follicle development is widely recognized. Among the members of the TGF-β superfamily, many genes are known to affect follicle development. For example, AMH is considered a marker of follicle selection (Huang, et al., 2021b). In addition, TGF-β has gradually become the focus of biomedical research as a target for tumors and fibrosis. It can promote fibrosis through intracellular signal transduction pathways (including the SMAD-dependent pathway) (Huang, et al., 2021a; Pawlak and Blobe, 2022) and plays a crucial role in regulating fibrosis in organs such as the lungs (Wang, et al., 2024b), liver (Zhang, et al., 2024b), myocardium (Vistnes, 2024), kidneys (Wang, et al., 2024a), and pancreas (Kweon, et al., 2023). Many DEGs during follicular atresia have been shown to be enriched in the TGF-β pathway, which is one of the reasons for our focus on TGFB2. GC degeneration is a fibrosis-like process, with many fibrosis marker genes showing differential expression, including genes in the collagen family. This study found that SMAD3 is the most potent effector TF during follicular atresia and a TF of IGFBP. It is regulated by TGFB2, indicating that the TGFB2/SMAD3/IGFBP5 signaling pathway likely causes GC degeneration through a fibrosis-like process.

STAT is a downstream effector of TF in the JAK/STAT signaling pathway. Cytokines bind to receptors on the cell membrane and activate receptor-associated JAK kinases that undergo phosphorylation. The JAK kinases then further phosphorylate STAT proteins, which subsequently translocate into the nucleus to regulate the transcription of target genes (Mahjoor, et al., 2023; Xin, et al., 2020). As a stress-induced inflammatory signaling pathway, the JAK/STAT pathway is a key component of many signaling pathways that regulate cell growth, differentiation, survival, and pathogen resistance, and it is closely related to various diseases, inflammation, and immune responses in chickens (Janesick, et al., 2022; Lu, et al., 2023; Miao, et al., 2022). However, reports on chicken follicle development are limited. In chickens, STAT proteins include STAT1–6. Studies on STAT2 that are mainly related to virus or immune research in chickens are relatively few (Nissly, et al., 2024; Vu, et al., 2023). Similar to SMAD, activation of the STAT signal was detected to determine the activity of the JAK/STAT pathway. In this study, STAT2 was a common TF for both TGFB2 and IGFBP5, indicating that STAT2 can not only directly regulate IGFBP5 to affect GC degeneration but also indirectly regulate IGFBP5 through the TGFB2/SMAD3 pathway.

Notably, the synergistic pathway between JAK/STAT and TGF-β/SMAD has shown great significance and relevance in the physiological and pathological fibrosis processes of numerous tissues, such as the liver (Tang, et al., 2017), lung (Gu, et al., 2023; Wang, et al., 2022a), heart (Eid, et al., 2019), and peritoneum (Mo, et al., 2023). Therapeutic drugs targeting these two pathways are subjects of ongoing research (Huang, et al., 2021a; Kapoor, et al., 2024; Stanilov, et al., 2024). However, the upstream–downstream relationship between these two pathways remains controversial. Some studies suggest that JAK-STAT is a downstream signaling pathway regulated by TGF-β. Inhibiting TGF-β1 can attenuate the fibroblast activation and epithelial cell damage induced by the JAK/STAT pathway (Gu, et al., 2023). Nevertheless, other research has found that inhibiting the JAK/STAT pathway reduces the TGF-β level, thereby alleviating skin fibrosis (Karatas, et al., 2022). Additionally, JAK1/STAT3 acts as a direct regulator of TGF-β signaling in lung fibroblasts (Wang, et al., 2022a). The synergistic regulation of these two pathways is complex and involves interactions, binding, or feedback regulation between their effector factors. Follicular atresia is a physiological fibrotic process that resembles tissue aging. Our results indicate that the synergistic pathway of JAK/STAT2 and TGF2/SMAD3 may also serve as a novel and effective target for studying follicular atresia.

However, this study still has some limitations. For instance, the comprehensiveness of SCENIC analysis needs further improvement. Currently, it can only analyze chicken data by converting it to human or mouse homologs, which means the quantity and accuracy of identified TFs require enhancement. There is still much work to be done in developing and refining a TF database specifically for chickens. Furthermore, the functions of STAT2 and SMAD3 require further investigation. Our ongoing research project focuses on the family members of these two genes, and some of the findings have already demonstrated their significance in follicle development (Guo, et al., 2025).

Conclusion

In conclusion, STAT2 binds directly to the promoter region of IGFBP5 and influences its transcription. It can also act as a TF for TGFB2 to regulate its expression, and activated TGFB2 further influences the binding of SMAD3 to the IGFBP5 promoter region, thereby indirectly regulating the transcription of IGFBP5. The STAT2/TGFB2/SMAD3/IGFBP5 regulatory pathway revealed an important interaction mechanism between the JAK/STAT and TGF/SMAD signaling pathways during follicular atresia. The complex synergistic regulation between them initiates the cellular fibrosis program, leading to GC degeneration and ultimately resulting in follicular atresia. This study provides new evidence for identifying the switch genes of follicular atresia, contributes to improving our understanding of the regulatory mechanism of follicular atresia, and offers new targets for extending the egg-laying period and enhancing the egg-laying performance.

CRediT authorship contribution statement

Wenhui Zhang: Writing – original draft, Visualization, Validation, Methodology, Investigation, Conceptualization. Siyu Huang: Validation, Investigation, Formal analysis. Axiu Guo: Validation, Investigation. Zongyi Zhao: Software, Investigation. Bo Zhang: Supervision, Conceptualization. Fuwei Li: Writing – review & editing, Resources, Methodology. Haigang Bao: Writing – review & editing, Conceptualization. Hao Zhang: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Data curation, Conceptualization.

Disclosures

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Hao Zhang reports financial support was provided by National Key Research and Development Program of China. Hao Zhang reports financial support was provided by Key Research and Development Program of Shandong. Hao Zhang reports financial support was provided by China Agricultural Research System. Bo Zhang reports financial support was provided by Natural Science Foundation of Shandong Province. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.