Transcriptomic regulation of sequential eggshell mineralization in laying hens: Molecular mechanism underlying eggshell quality and production performance implications Authors: Xiaofan Yu, Xinye Zhang, Xing Chen, Yalan Zhang, Dehe Wang, Mengyuan Wang, Yuhang Peng, Xiaolong Ma, Yuhui Xie, Zhonghua Ning, Lujiang Qu Published: 2026 ABSTRACT AbstractEggshell quality is a key economic trait in commercial layer production, which profoundly affects economic profits, flock elimination rate, and feeding management. Frequent occurrence of cracked, thin-shelled eggs has long plagued intensive laying hen farming. This study aimed to dissect the molecular mechanisms and regulatory networks underlying staged eggshell mineralization across the pre-mineralization, initial mineralization, and rapid mineralization phases in laying hens, and explore their links to eggshell quality and shell defect in production. By integrating RNA-sequencing, temporal expression profiling and WGCNA, we screened core functional genes and signaling pathways in uterus that governing the variation of eggshell thickness. In pre-mineralization stage, high expression of Ovalbumin (OVAL) may inhibit premature calcium carbonate crystallization and avoid abnormal shell development. During rapid mineralization, genes encoding eggshell matrix proteins, ATPase, TRP (transient receptor potential), SLC (Solute carrier) and collagen family members coordinately maintained uterine cellular calcium homeostasis, sustained calcium supply for eggshell deposition, and ultimately determined eggshell thickness and strength. Meanwhile, calcium concentration in uterine fluid increased gradually along with mineralization and remained stable throughout rapid mineralization phase, providing a steady calcium supply basis for preventing eggshell defects in commercial laying hen production. Findings of this study provided targets and theoretical support for the breeding of laying hen breeds, the regulation of nutritional management, and the optimization of farming environment management.Keywords: Calcification, Calcium transport, Eggshell matrix protein, Uterus, Chicken Introduction Reduced eggshell strength and frequent appearance of soft-shelled, sandy-shelled and shell-less eggs are prevalent issues in intensive cage rearing. These defective eggs result from arrested calcification at different developmental stages (Fig. 1). Unqualified eggshells lead failure to incubation, cause inconvenience to transportation and ultimately increase management costs. Currently, eggshell quality is primarily regulated through extensive management practices such as feed nutrition, lighting, and housing conditions. However, there is limited theoretical support at molecular level, and management measures lack precise targets. Previous studies on eggshell mineralization have primarily focused on basic molecular descriptions, evidence about the association of eggshell structure and thickness with hen’s production performance and uterine fluid environment at different stages of mineralization was insufficient. Overview of our research. EM = eggshell membrane stage, INI = initiation stage, RG = rapid growth stage, SEM = scanning electron microscope, TGEP = temporal gene expression profiling. Photographs showed laid eggs with eggshell defects collected in a farm. Avian eggshell is a key functional structure affecting laying hens’ egg quality and production performance, which not only protects embryos and enables gas exchange, but also determines the shelf life and commercial value of eggs in poultry industries. The avian eggshell assembles within the uterus over a period of about 19 hours(Gautron et al., 2021), which is divided into initiation, rapid growth and termination stages (Gautron et al., 1997). During initial phase, mammillary knobs grow on eggshell membrane and fuse together (Dennis et al., 1996). The rapid growth phase accounts for the largest portion of the entire deposition period, during which the palisade layer grows rapidly perpendicular to the mammillary layer. Mineralization ends in terminal phase when calcium carbonate deposition ceases, and the final organic matrix, pigments and the cuticle (Nys and Y., 2011) is deposited onto the shell surface prior to oviposition through vagina (Gautron et al., 2021). However, current research often overlooks the shell membrane formation stage, which occurs 4-5 hours after oviposition in isthmus before initiation phase. While residing in isthmus prior to uterine entry, the egg not only acquires the fibrous shell membranes but also initiates the formation of mammillary nucleation sites. Therefore, this stage is pivotal to establishes both the physical scaffold and the biochemical cues that direct the entire subsequent mineralization process, and this is the first step to determine eggshell quality. Eggshell is composed of approximately 95% calcium carbonate and 3.5% organic matrix (Moreau et al., 2022), wherein calcium ions serve as the primary inorganic constituent of the eggshell (Dominguez, 2004), and its concentration in uterine fluid is closely associated with the efficiency of shell deposition. Calcium homeostasis and eggshell matrix protein modification are key regulatory factors influencing eggshell formation and stability of laying hens’ production. Calcium required for eggshell deposition derives from plasma, and enters uterus through several channels including Voltage-Operated Ca2+ Channels (VOCC), Receptor-Operated Ca2+ Channels (ROCC), and Store-Operated Ca2+ Channels (SOCC) (Schwaller, 2012). Intracellular calcium homeostasis is maintained by endoplasmic reticulum (ER) transport system and calcium-binding buffers such as calbindin 1 (CALB1) (Schwaller, 2010). CALB1, a vitamin D-dependent calcium-binding protein first identified in chicken duodenal mucosa (Taylor and Wasserman, 1967), and it possesses high affinity to calcium (Cedervall et al., 2005; Schwaller, 2009; Gloux et al., 2019). Recent research demonstrated that CALB1 regulated intracellular concentration of calcium ion in uterine tubular gland cells of chicken (Cui et al., 2025). Proteomic analyses of eggshell have revealed that many key matrix proteins are highly glycosylated and implicated in both biomineralization and immune functions (Xie et al., 2023). For instance, ovocleidin 17 (OC-17) and extracellular matrix phosphoglycoprotein, (MEPE, also called OC-116) possess unique glycosylation sites (Hincke et al., 1999; Reyes-Grajeda et al., 2004), glycosylation provides recognition and binding sites for matrix proteins and forms anionic sites for ion-mineral binding (Evans, 2020). For instance, N-glycosylation influences calcium ion transport and formation of eggshell mineralization layer (Zeng et al., 2023) to further modulate shell ultrastructure. The focus on the pre-mineralization stage was driven by two factors. Firstly, since calcification has not yet initiated at this stage, it can serve as a “shell-less” trait for comparison with samples from the subsequent shell-forming stage. Secondly, although the eggshell has not yet formed during pre-mineralization, uterine cells are likely already making certain preparations. This study aims to identify candidate genes and pathways associated with calcium carbonate deposition in eggshells by comparing transcriptomic profiles and uterine fluid calcium concentrations during different stages. In practical layer management, exploring the regulatory mechanisms of calcium metabolism and protein glycosylation in laying hens can provide a theoretical basis for optimizing nutritional supplementation, improving eggshell quality, reducing breakage rate, and maintaining stable production performance in laying hens. Materials and methods Ethics Statement All experimental protocols were performed according to the Ministry of Science and Technology (Beijing, China). We declare that experiments in this study were conducted approved by the Ethics Committee of China Agricultural University (permit number: AW026022021-1). Animals selection and sample collection A total of six hundred 300-day-old Rhode Island Red hens that laid normal eggs were used in this study. The oviposition time of each hen was manually recorded at 20-minute intervals for five consecutive days. Approximately 100 hens with similar body weight, good health status, and minimal variation in oviposition intervals were selected for euthanasia and subsequent procedures. During dissection, the precise position of the egg within the oviduct was documented. The entire oviduct was then excised, and the egg was carefully retrieved. Eggshell with membranes and uterine tissue samples were separately collected. All tissue samples were immediately preserved in RNA stabilization solution (RNAwait, purchased from Beijing Bio-Top Biotechnology Co., Ltd.) and stored at −20°C. Observation of eggshell ultrastructure Eggshell samples were cleaned and trimmed into 1 mm × 1 mm pieces using surgical scissors and forceps. All fragments were collected from the equatorial region of the eggshell, ensuring flat fracture surfaces and intact structural integrity. The samples were air-dried in a 100% desiccant, mounted on specimen stubs with conductive adhesive, and sputter-coated with a thin layer of gold using an ion sputter coater to enhance surface conductivity. The ultrastructural features of the eggshells were examined and imaged using a scanning electron microscope (Prisma E SEM, Thermo Fisher Scientific Ltd., USA) at magnifications of 500× and 1000×. Eggshell thickness was measured in Adobe Photoshop 2020 based on the scale bar provided in the electron micrographs, using the built-in ruler tool for precise quantification. RNA extraction and transcriptomic sequencing Total RNA was extracted from uterine tissue using the Trizol reagent. RNA quality was comprehensively assessed by measuring concentration and determining A260/A280 and A260/230 ratios with a Nanodrop spectrophotometer, while RNA integrity and fragment size distribution were evaluated via 1% agarose gel electrophoresis (200 V, 10 min). cDNA libraries were constructed using the V8 RNA-seq Library Prep Kit for Illumina and subsequently sequenced on an Illumina NovaSeq 6000 high-throughput sequencing platform. The raw image data files generated by the sequencer were processed through base calling to generate raw sequencing reads (Raw Data), which were stored in FASTQ format. Each FASTQ file contains both nucleotide sequence information and corresponding base quality scores for all sequenced reads. Data quality control Raw reads were subjected to a stringent filtering process to obtain high-quality clean reads. Specifically, paired reads containing adapter sequences were removed. Paired reads were also discarded if either end contained over 10% of undetermined bases (N) or if more than 50% of the bases in either read were of low quality (Phred score ≤ 5). The resulting clean reads were further processed using fastp (v0.23.4), where sequences shorter than 30 nucleotides were filtered out, and low-quality bases at both ends of the reads were trimmed. Subsequently, the HISAT2 software (v2.2.1) was employed to align the quality-controlled reads to the chicken reference genome (Galgal7), following the construction of the corresponding genome index. The alignment results were saved in SAM format and converted into sorted BAM files using Samtools (v1.19.2). Finally, gene expression levels were quantified as raw counts from the sorted BAM files using featureCounts (v2.0.8). Bioinformatic analysis Identification of Differentially Expressed Genes and Functional Enrichment. Differentially expressed genes (DEGs) were identified using the DESeq2 (v1.42.1), with thresholds set at false discovery rate (FDR) < 0.05 and |log₂FoldChange| > 1. Functional enrichment analyses of KEGG pathways and GO terms were performed using clusterProfiler (v4.10.1) with the org.Gg.eg.db annotation database (v3.18.0), considering terms with p < 0.05 as statistically significant. Resulting plots were generated using the ggplot2 (v4.0.1). Temporal Expression Profiling. From the gene expression matrix comprising 23 individuals across five developmental stages, normalized cpm values were averaged per group. The top 5,000 genes exhibiting the highest median absolute deviation (MAD) were selected for temporal expression analysis. A heatmap was constructed using the online platform (https://www.bioinformatics.com.cn) for data visualization. Weighted Gene Co-expression Network Analysis (WGCNA). Co-expression networks were constructed using the WGCNA package (v1.72-5). Genes with low expression or minimal variation across samples were filtered based on MAD. Normalized cpm values were log₂-transformed as log₂ (cpm + 1). A sample clustering tree was generated, and the optimal soft-thresholding power was selected to achieve a scale-free topology fit index R² > 0.8 while maintaining a low mean connectivity (<100). Then, a weighted co-expression network is constructed to identify gene modules through two main steps: first, calculating the proximity value, which is the expression correlation coefficient (Pearson correlation rho) between two genes across different samples; second, calculating the topology overlap similarity (TOM), which represents the similarity between two genes in terms of network structure. Modules were identified using hierarchical clustering with a minimum module size of 30 and a dissimilarity threshold (1-TOM). Modules with similarity above 75% (height cut = 0.25) were merged. Module-trait associations were evaluated by correlating module eigengenes with phenotypic data. Module membership (gene-module correlation) and gene significance (gene-trait correlation) were calculated and visualized in scatter plots using absolute values. KEGG and GO analyses for genes in the green and salmon modules were performed following the procedure as described above. Network Visualization and Integration. Gene-pathway interaction networks were generated using ClueGO and CluePedia plugins in Cytoscape (v3.10.3). Protein-protein interaction (PPI) networks were obtained from the STRING database, imported into Cytoscape, and analyzed with the CytoHubba and MCODE plugins to identify hub genes and functionally significant subnetworks. Schematic diagrams of molecular mechanisms were created with BioGDP.com.(Jiang et al., 2025) Calcium concentration assay of uterine fluid Using sterile cotton thread, ligate the isthmus and vaginal portion of the oviduct separately. Carefully dissect the egg from the uterus, then aspirate uterine fluid using a 2 ml sterile pipette. Store the fluid at −20°C for later use. Prior to measurement, preheat the spectrophotometer for at least 30 minutes, adjust the wavelength to 520 nm, and zero the instrument using distilled water. Proceed according to the protocol provided in the calcium concentration detection kit (YX-W-B801, Beijing Sierai Bio-Tech Co., Ltd., China). Statistical analysis Statistical significance was defined as p < 0.05. For RNA-seq differential expression analysis, FDR was controlled using the Benjamini-Hochberg procedure, with p-value < 0.05 considered significant. All statistical analyses were performed in R (v4.3.3). Visualization of calcium concentration assay was completed by Prism 10 (v10.3.0). T-test was used to analyze the difference of calcium concentration across 5 stages. Results Ultrastructure of eggshells at distinct mineralization stages Four to five hours after ovulation, the egg contents reside within white isthmus (Fig. 1), where the eggshell membranes and initial mammillary precursors are assembled, precursors appeared to be small spherical granules, which were deposited on surface of membranes (Fig. 2A). This stage is designated as the Eggshell Membrane phase (EM). As egg entered red isthmus which was adjacent to uterus (Fig. 1), the mammillary bodies underwent further growth and coalescence (Fig. 2B-C), defining the Initiation stage (INI) of mineralization occurring 5-10 hours post-ovulation. This is followed by the Rapid Growth phase (RG), a period of sustained calcium carbonate deposition lasting approximately 12 hours, which was further subdivided into RG1, RG2, and RG3 substages based on incremental shell thickness (Fig. 2D-F, Table 1). Fig. 2G illustrated the ultrastructure of eggshell from a laid egg, comprising mammillary layer, palisade layer, and vertical crystal layer. Thickness of eggshell at each stage is indicated in figs. Fig. 2H showed morphology of eggs in chronological progression during shell calcification. Laid eggs that were cracked or incompletely mineralized can be matched to the corresponding stage based on eggshell thickness, to determine at which specific phase the mineralization ceased. Table 1 showed details of traits used to define phenotypes. Ultrastructure observation of eggshell samples at 5 different stages. (A) Eggshell collected during EM phase (magnification 1000×), red and yellow arrows represent eggshell membranes and mammillary bodies respectively. (B-C) Eggshell collected during INI phase (magnification 500×, magnification 1000×), blue arrow indicates fusion of mammillary bodies. (D-F) Eggshell collected during RG1/2/3 phase (magnification 1000×), ML = mammillary layer, PL = palisade layer. (G) Eggshell collected during terminal phase (magnification 500×). VCL = vertical crystal layer, green arrow indicates cuticle. (H) Morphology of eggs and corresponding stages. Table 1List of traits used to define the phenotypes of different stages during eggshell mineralization in chicken.Table 1 dummy alt textStagesTime (p.o4)Position of eggsEggshell thickness (µm)Number of individualsEM14-5 hIsthmus/5INI25-10 hUterus90-1003RG1310-22 hUterus125-1504RG210-22 hUterus160-1806RG310-22 hUterus200-25051EM = eggshell membrane.2INI = initiation.3RG = rapid growth.4p.o = post ovulation. List of traits used to define the phenotypes of different stages during eggshell mineralization in chicken. EM = eggshell membrane. INI = initiation. RG = rapid growth. p.o = post ovulation. Uterine transcriptomic profiling across mineralization stages Principal component analysis among 23 individuals revealed a clear separation between EM stage and all subsequent mineralization stages. While the samples of INI stage showed proximity to those of EM phase, there is some overlap among three RG substages (Fig. 3A). EM was compared with INI, RG1, RG2, and RG3 stages, yielding 403, 1,095, 1,316, and 1,184 differentially expressed genes (DEGs), respectively (Fig. 3B, Supplementary Table S2). Among these four sets of DEGs, 129 shared genes were down-regulated, and 94 shared genes were up-regulated (Fig. 3C-D). These overlapped DEGs displayed a clear and progressive expression pattern throughout mineralization (Supplementary Fig. S1). The 129 downregulated DEGs included ATPase plasma membrane Ca2+ transporting 2 (ATP2B1), ATPase sarcoplasmic/endoplasmic reticulum Ca2+ transporting 3 (ATP2A3), transient receptor potential cation channel subfamily C member 7 (TRPC7), ATPase Na+/K+ Transporting subunit alpha 1 (ATP1A1), and CALB1. Transcriptomic analysis of uterine tissues at different stages during eggshell calcification. (A) PCA plot. (B) Volcano plot illustrating the DEGs of four comparisons between EM and other stages (log2|fold change| > 1.0 at a false discovery rate (FDR) < 0.05), gray dots indicate all DEGs and top DEGs were plotted in color. (C-D) Venn diagrams identifying overlapping DEGs across all comparisons. (E-F) KEGG pathway analysis and GO enrichment analysis of DEGs across all comparisons (p < 0.05). KEGG pathway analysis also revealed some shared pathways, including ligand-receptor interaction, protein processing in endoplasmic reticulum, calcium signaling pathway, cytokine-cytokine receptor interaction, and steroid hormone biosynthesis. Notably, EM vs RG1 comparison showed particularly significant enrichment in various metabolic pathways (Fig. 3E), indicating enhanced metabolism when rapid growth began. Shared enrichment of GO terms included extracellular region, extracellular space, endoplasmic reticulum lumen, and tissue development. Furthermore, the EM vs INI group was specifically enriched in bone mineralization and biomineralized tissue development, while EM vs RG3 comparison was additionally enriched in signaling receptor activity, binding, multicellular organism development, and cell migration (Fig. 3F). Temporal gene expression profiling Gene expression varied across five developmental stages, and they were classified into 12 distinct clusters (Fig. 4). Expression of genes in clusters 2, 4, and 9 was low from the beginning but rose dramatically upon the onset of mineralization. Cluster 2 was functionally enriched for electron transport-coupled ATP synthesis, mitochondrial complex I assembly, and NADH dehydrogenase activity. Cluster 4 was notably enriched in ion transport, calcium binding, genes encoding eggshell matrix proteins, ER protein processing, ER stress response, protein export, and osteoblast differentiation. Similarly, cluster 9 was enriched for ER-related functions and N-glycan biosynthesis. Information about critical genes in these clusters were shown in Supplementary Table S1. Genes in cluster 8 exhibit the highest expression during RG3 stage, and they were primarily involved in cell adhesion, extracellular region, and calcium channel complexes. Temporal expression patterns and functional enrichment of genes from EM to RG3. In contrast, clusters 5, 6, 7, and 10 displayed peak expression levels during the early mineralization stages (EM and INI), followed by a decline. These included OVAL, Ovodefensin A2 (OvoDA1-2), Ovodefensin A3 (OvoDA3), Ovodefensin B1 (OVODB1) and Ovotransferrin (TF). These clusters were significantly enriched in pathways such as cytokine-cytokine receptor interaction, ribosome biogenesis, G protein-coupled receptor activity, oxidative phosphorylation, the polycomb repressive complex, and inositol phosphate metabolism. Results above indicated that, during early stages of mineralization (EM, INI), biosynthetic and metabolic activities increased in preparation for subsequent shell deposition, and calcium ion transport and eggshell matrix proteins may be the core regulatory factors to support the rapid growth of eggshell. Stage‑specific modules identified by WGCNA WGCNA was conducted using transcriptomic data from 23 individuals. The sample dendrogram and corresponding developmental stages are shown in Fig. 5A. A soft-thresholding power of 8 was chosen to construct a scale-free network, with the scale-free topology fit index exceeding 0.8 and adequate mean connectivity preserved (Supplementary Fig. S2). We identified five co-expression modules (excluding the grey module of unassigned genes; Fig. 5B). Module-stage association analysis revealed significant relationships (Fig. 5C). The green module was strongly positively correlated with the early mineralization (EM) stage (p = 2e-09) and negatively correlated with the RG2 stage (p = 0.02). The salmon module showed a marked positive correlation with the RG3 stage (p = 4e-04) and negative correlations with the EM (p = 0.01) and INI (p = 0.01) stages. The topological overlap matrix (TOM) heatmap clearly separated regions with high gene co-expression similarity from those with weak connections (Fig. 5D). WGCNA performed using transcriptomic data across 5 stages. (A) Sample tree and corresponding stages. (B) Hierarchical clustering dendrogram of genes. (C) Module-trait relationships. Each row represents module eigengenes obtained from WGCNA, with each column corresponding to a stage. (D) Heatmap illustrating topological overlap matrix (TOM) among all modules. Red and deep areas represent high overlap, lighter colors represent decreasing overlap. Functional Enrichment of Overlapping Genes Between Green Module and DEGs. The green module was highly associated (r = 0.91) with pre-calcification (EM) stage (Fig. 5C, 6A), underscoring its central role in this early phase. Within green module, we identified 1,078 genes with negative values of both Gene Significance (GS) and Module Membership (MM; Supplementary Table S2). Notably, genes residing in upper-right quadrant were considered as key genes, including ATP1A1, ATP2A3, CALB1, BPI fold containing family B member 3 (BPIFB3), TS variant 5 (ETV5), and Osteopontin (SPP1), and their GS and MM values approached −1 (Fig. 6B). Apart from negative genes in green module, there were over 2000 genes with positive values. This is why the green module generally shows a positive correlation with EM. An overlap between the 1,078 negative module genes with the 706 down-regulated DEGs from EM vs RG2 comparison yielded 486 consensus genes (Fig. 6C). KEGG pathway analysis revealed that these shared genes were significantly enriched in critical biological processes (p < 0.05), including amino acid and sphingolipid metabolism, endoplasmic reticulum protein processing, neuroactive ligand-receptor interaction, cytokine-cytokine receptor interaction, and the calcium signaling pathway (Fig. 6D). An interaction network between genes and KEGG/GO terms further elucidated a complex regulatory landscape, illustrating multiple genes involved in ion transport (Fig. 6E). Genes and functional analysis associated with green module. (A) Heatmap of Eigengene adjacency for green module. (B) Scatter plot of gene significance and module membership for green module. (C) Overlap between negatively correlated genes in WGCNA green module and down-regulated DEGs in EM vs RG2 comparison. (D) KEGG analysis of overlapped genes (p < 0.05). (E) Interaction network between overlapped genes and functional enrichment analysis. Significant Go terms and KEGG pathways are presented (p < 0.05). Protein-protein interaction (PPI) network was constructed using the 486 consensus genes. Top 10 hub genes were identified and heat shock 70 kDa protein 5 (HSPA5) showed the highest degree of connectivity (Supplementary Table S3), indicating its central position with the most extensive interactions, and it may be the most critical regulator in network. HSPA5 is postulated to modulate the endoplasmic reticulum stress response, potentially through its functional associations with BCL2 (apoptosis regulator) and ATP2A3. Additionally, the capability of calcium binding of CALB1 may be influenced by ATP2A3 and brain derived neurotrophic factor (BDNF, Fig. 7). PPI network illustrates the top 10 nodes ranked by Degree method along with the first-stage nodes and the shortest path using overlapped genes. Genes colored from red to orange represent hub genes, red indicates highest node score. The rest genes colored from dark to light indicate genes associated with hub genes, they were continuously mapped using log2|fold change|. Functional Profiling of Genes in Salmon Module. The correlation coefficient between the salmon module and RG3 was 0.68 (Fig. 5C, Fig. 8A). This module contained 443 genes with positive GS and MM values (Supplementary Table S2), including Fibronectin 1 (FN1), collagen family members COL4A1 and COL4A2, integrin family members ITGA1, ITGA9 and ITGB5, fibroblast growth factor 10 (FGF10) and vascular endothelial growth factor C (VEGFC), receptor tyrosine kinases FGFR1 and VEGFRKDRL (Fig. 8B). And most of them were also in Supplementary Table S1. The 443 genes were significantly enriched in metabolism, endoplasmic reticulum protein processing, ECM receptor interactions, muscle cytoskeleton, and focal adhesions (p < 0.05, Fig. 8C). Genes and functional analysis associated with salmon module. (A) Heatmap of Eigengene adjacency for salmon module. (B) Scatter plot of gene significance and module membership for salmon module. (C) KEGG analysis of filtered genes in salmon module (p < 0.05). The key genes identified above played distinct roles at different stages and can serve as candidate genes for eggshell quality traits, providing molecular markers for selection in layer chicken breeding and identification of targets for nutritional regulation. Comparison of calcium concentration in uterine fluid across different stages As mineralization proceeded, calcium concentration in uterine fluid gradually increased across stages (Fig. 9). No significant difference was observed between the EM and INI groups (p = 0.12). The RG1 group exhibited significantly higher calcium concentration than EM group (p = 0.014), while the RG2 and RG3 groups showed extremely significantly higher calcium concentrations than the EM group (p < 0.01). The variation of calcium in uterine fluid was consistent with calcium absorption in uterus, based on expression of CALB1 and vitamin D receptor (VDR). Variation of calcium concentration in uterine fluid across five stages. Ns represents no significance (p > 0.05). *P ≤ 0.05, **P ≤ 0.01. Discussion As shown in Fig. 10, eggshell formation was a continuous process of input and output, which is constantly self-regulated. Calcium ions and carbon dioxide underwent transport and reaction after being absorbed by uterine cells, then they were transported into the extracellular uterine fluid. The efficiency of calcium absorption depends on the interaction of CALB1 (Kamata et al., 2020), VDR and 1,25(OH)₂D₃ (Miller, 2017). Meanwhile, the primary binding agents for the inorganic components of the eggshell are organic substances. During rapid mineralization phase, large amounts of eggshell matrix proteins were synthesized within uterine cells and were transported into uterine fluid to participate in the formation of calcified layer. The endoplasmic reticulum, this bustling factory, not only needs to promptly process misfolded proteins to maintain normal protein synthesis and secretion functions, but also to regulate calcium ion channels to maintain intracellular calcium homeostasis during periods of massive calcium influx and efflux. Therefore, endoplasmic reticulum stress, protein glycosylation, calcium transport, and homeostasis are key factors influencing eggshell formation. Abnormalities in these processes are associated with a decline in uterine function in laying hens (especially during late laying period), which can lead to thinner, more fragile eggshells and early culling of layers. Schematic diagrams of molecular mechanisms during eggshell calcification. Black arrows indicate ion transportation or promotion of biological processes. Red arrows indicate the growth orientation of crystal. During eggshell formation, changes in the concentration of calcium ions in uterine fluid correlated with the expression patterns of key genes such as CALB1, VDR (calcium binding), TRPA1 (calcium influx), and ATP2B1 (calcium efflux). In other words, the uterus precisely regulated the uptake and utilization of calcium ions by autonomously modulating the expression of these genes, indicating that uterus had varying calcium requirements during different stages of mineralization. This demonstrated that indiscriminate addition of calcium is a common phenomenon in production. Precisely adjusting the levels of calcium and vitamin D according to the physiological stages of eggshell mineralization is a viable measure for improving calcium absorption and utilization, reducing feed waste, and lowering production costs. Numerous genes associated with eggshell matrix proteins have been identified in this study, most of them were significantly upregulated during mineralization. OC-116, directly participated in eggshell mineralization (Marie et al., 2015; Riou et al., 2017; Poyatos Pertiñez et al., 2020). OC-17 is present in high concentrations in uterine fluid at various stages (Gautron et al., 1997). Retinoic acid receptor responder 1, RARRES1 (OCX-32) may participate in the deposition of vertical crystalline layer and cuticle layer (Arias et al., 2001). BPIFB3 (OCX-36) is also an active glycoprotein during the mineralization phase (Moreau et al., 2022). However, OVAL exhibits a completely opposite expression pattern (Fig. 4), and studies have reported that its concentration was the highest in uterine fluid during early stages of mineralization (Gautron et al., 1997; Moreau et al., 2022). Proteins exert a dual effect on calcium carbonate deposition. When protein concentrations are excessively high, they bind to crystal surfaces and block growth sites, thereby inhibiting crystallization (Nys et al., 2004). In other words, when the egg has not yet entered the uterus, highly expressed OVAL inhibits calcium carbonate formation. Once mineralization begins in earnest, its expression is downregulated. This precise, targeted process further demonstrated that uterus is an organ capable of autonomous regulation. Conclusion This study elucidated the regulatory mechanisms governing eggshell deposition in the uterine portion of the hen’s oviduct, and can provide a theoretical foundation for management, nutritional regulation, and prevention of eggshell defects in large-scale layer farming. The key candidate genes identified in this study can be used in subsequent breeding programs to develop laying hen breeds resistant to eggshell thinning and to screen for high and stable egg production in hens, thereby extending the laying period, reducing culling rates, and improving overall economic efficiency in poultry farming. In addition, egg production in laying hens is inherently a process that places a strain on hen’s body. By adhering to principles of animal welfare and protection, implementing phased nutritional supplementation and environmental management, and optimizing cage housing conditions to prevent heat stress, it may be possible to improve uterine health and accomplish the transition from traditional, experience-based farming to precision farming based on physiological principles. Data availability statement Raw data were not deposited in an official repository. The data that support the study findings are available from the authors upon request. Author contribution Xiaofan Yu: Conceived and initially designed the study, organized and coordinated the experimental work, performed the animal euthanasia procedures and uterine tissue collection, conducted the experiments, and prepared the initial manuscript. Xinye Zhang: Collected uterine fluid samples, organized and recorded the experimental data, and prepared the visual presentation of the transcriptomic analysis results. Yalan Zhang: Processed the raw transcriptomic sequencing data, including quality control, differential gene expression analysis, and functional enrichment analysis. Xing Chen: Surveyed and evaluated the experimental site and assisted in determining the breed and number of experimental animals, dissected animals. Mengyuan Wang: Performed weighted gene co-expression network analysis of the transcriptomic data and measured calcium concentrations in the relevant samples. Xiaolong Ma: Generated and optimized the gene–pathway network diagrams using Cytoscape and its related plugins. Yuhui Xie: Repeated the quality control and differential gene screening procedures for the transcriptomic data and assessed the consistency and reliability of the analytical results. Dehe Wang: Reproduced the analyses underlying all result figures and independently checked the consistency and reliability of the reported results. Yuhang Peng: Established standards for animal selection. Prepared the final mechanistic model illustrating the regulatory relationships proposed in this study. Zhonghua Ning: Provided the experimental animals, contributed to the refinement of the overall experimental design, supervised the research, and contributed to the interpretation of the findings. Lujiang Qu: Grouped experimental animals. Contributed to the scientific guidance of the study, critically revised and polished the manuscript, and reviewed and approved the final version. Declaration of competing interest All authors declare no conflicts of interest. Disclosures The authors declare no competing interests. During the preparation of this work authors did not use any AI and AI-assisted technologies. Acknowledgments We are sincerely grateful to all participants involved in this work and all the research staff and students working on the project. This work was supported by National Key Research and Development Program of China (2024YFD2000300) and Beijing Rural Revitalization Agricultural Science and Technology Project (NY2502130125). Footnotes Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.psj.2026.107300. Contributor Information Zhonghua Ning, Email: ningzhh@cau.edu.cn. Lujiang Qu, Email: quluj@163.com. Supplementary materials References Associated Data Supplementary Materials Data Availability Statement Raw data were not deposited in an official repository. The data that support the study findings are available from the authors upon request.