The human ESCC cell lines (TE1, KYSE150, and KYSE170) and the human normal esophageal epithelial cell line (HEEC) were obtained from the China Center for Type Culture Collection (CCTCC, Wuhan, China). The ESCC cell lines were maintained in RPMI 1640 medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; Invitrogen), while HEEC cells were cultured according to the manufacturer’s instructions. These cells were detected and identified as free mycoplasma and bacteria infection during the past three months. The aforementioned cells were cultured in a humidified incubator at 37°C under 5% CO₂ atmosphere. For thapsigargin (Tg) (#586005, MedChemExpress, Monmouth Junction, NJ, USA) or GSK2606414 (HY-18072, MedChemExpress) treatment, cells at 70% confluence were incubated with the specified concentrations of Tg or GSK2606414 for indicated durations, using DMSO (0.1%) as vehicle control.

Total RNA was extracted using TRIzol reagent (Solarbio, Beijing, China), and cDNA was synthesized with the MightyScript First Strand cDNA Synthesis Master Mix (Sangon Biotech, Shanghai, China), following the manufacturer’s protocol. Real-time PCR analyses were conducted with primer sets as described in Supplementary Table S1 , employing SYBR Green real-time fluorescent quantitative PCR premix (Sangon Biotech) on the StepOne plus Real-Time PCR System (Applied Biosystems, Waltham, Massachusetts, USA). Relative gene expression levels were analyzed using the 2^–ΔΔCT method, with GAPDH serving as an internal reference for normalizing gene expression.

TE1 cells, treated with Tg or DMSO, were subjected to RNA-Sequencing analysis using the Illumina platform at Sangon Biotech. Differentially expressed genes (DEGs) were identified using DEseq R, applying the following criteria: |log2 (fold change)| > 1.

Lipofectamine 2000 (Invitrogen) was used to perform the transient transfection. The siRNAs for PERK, IRE1α, ATF6, CREDL2, APMAP and siControl were purchased from GenePharma (Shanghai, China). Following transfection with 50 nM siRNA, cells were analyzed at 48 h post-transfection. For Tg treatment, transfected cells were re-plated and allowed to attach before being treated with either 100 nM Tg (+) or DMSO (−) for 12 h prior to analysis. The sequences of these siRNAs were listed in Supplementary Table S2 . The pcDNA3.1-CRELD2 vector was acquired from Sangon Biotech. The human full-length cDNA fragments of ATF4 and APMAP were amplified and subsequently inserted into pcDNA3.1 vector (Invitrogen) respectively. The primers used were described in Supplementary Table S3 . The transfection efficiency was assessed by qRT-PCR.

The CRELD2 promoter was cloned into the pGL3-Basic vector (Promega, Madison, WI, USA). KYSE150 cells were plated in 6-well plates and the pGL3-CRELD2 promoter luciferase plasmid was co-transfected with pcDNA3.1-ATF4, pcDNA3.1 empty plasmid and SV40 Renilla luciferase plasmid. After 48 h, luciferase activity was measured sequentially in each well using the Dual-Luciferase Reporter Assay System (Promega) and normalized to Renilla luciferase activity (control). The primers for various CRELD2 promoter fragments were listed in Supplementary Table S4 .

For co-immunoprecipitation (CO-IP), the treated cells were lysed using NP-40 lysis buffer (Solarbio) supplemented with phenylmethanesulfonyl fluoride (PMSF) (Solarbio). Cellular proteins were clarified by centrifugation at 12,000 rpm. Protein-protein interactions were cross-linked by incubating the lysate supernatants with the indicated antibodies for 3 h at 4°C with constant rotation. Normal rabbit IgG (Abbkine, Wuhan, China) without antigenicity was used as a negative control. The complexes were then incubated with precleared protein A/G agarose beads (MedChemExpress) overnight at 4°C under gentle rotation. The immunoprecipitation complexes were boiled with sample loading buffer after three washes with NP-40 buffer. Proteins were separated by SDS-PAGE, transferred to polyvinylidene fluoride membranes (Millipore, Sigma, Burlington, MA, USA), and blocked with 5% skim milk in Tris-buffered saline (TBS). Immunodetection was performed with the specific primary antibodies for overnight incubation at 4 °C, followed by incubation with HRP-conjugated secondary antibodies for 1 h at room temperature. The protein bands were visualized via enhanced chemiluminescence (Zenbio).

For analysis of plasma membrane protein expression, plasma membrane was isolated using the MinuteTM Plasma Membrane Protein Isolation and Cell Fractionation Kit (SM-005 Invent Biotechnologies, Plymouth, MN, USA) following the manufacturer’s protocol. Protein expression levels were then determined by Western blot analysis.

The relative protein expression levels were quantified by densitometric analysis of Western blot bands using Image J software and normalized to β-actin or ATP1A1. Triplicate measurements were obtained for every tested parameter.

The β-actin (380624), ATF4 (381426), ATF6 (162665), APMAP (860886), FN1 (R381177), N-cadherin (380671), cyclin D1 (CCND1) (380999), Smad2 (R25742), Smad3 (R25743), p-Smad2 (310079), p-Smad3 (R22919), STAT3 (R22785), p-STAT3 (381552), AKT (R23412), p-AKT (310021), p-TAK1 (252059), TAK1 (R382046), p-NFκB p65 (310013), ATP1A1(R380790) antibodies, and the secondary antibodies (511103 and 511203) were obtained from Zenbio. Antibodies against APMAP (25953-1-AP, Proteintech, Wuhan, China), ZEB2 (67514-1-Ig, Proteintech, Wuhan, China), XBP1s (647501, Biolegend, San Diego, CA, USA), TGFBR1 (ab235578, Abcam, Cambridge, UK), CRELD2 (sc-365168, Santa Cruz Biotechnology, Dallas, Texas, USA), TAB2 (A9867, ABclonal Technology, Wuhan, China), TAB1 (RT1603, HuaBio, Hangzhou, China) were also respectively used. Rabbit IgG (KTD105-CN) antibody was obtained from Abbkine. The β-catenin (PK02151) antibody and the specific secondary antibody used to avoid the influence of heavy/light chain (M21008F) were obtained from Abmart (Shanghai, China). All the antibodies were diluted according to the manufacturer’s instructions.

Cell proliferation ability was assessed using MTS and colony formation assays. For the MTS assay, the 1 × 10³ cells were inoculated in 96-well plates. The Cell Titer 96^® Aqueous One Solution Cell Proliferation Assay Kit (20 µL) (Promega) was added to the culture medium at 0, 24, 48, 72, and 96 h and incubated for 2 h. The absorbance of each well was detected at 490 nm. For the colony formation assay, the 5 × 10³ treated cells were seeded in triplicate in 6-well plates and routinely cultured for approximately 7 days. The colonies were then fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. Colonies composed of 50 cells or more were manually counted.

Cell migration ability was measured by wound healing assay. The treated cells were inoculated into 6-well plates. A scratch wound was created with a sterile 200 µL pipette tip when the cells had grown to 80% confluence, followed by overnight starvation in serum-free medium. The cells were monitored by capturing images at the same fields for 0 and 24 h using an inverted microscope. The percentage of the real-time scratched area was quantified using Image J software.

Invasion assays were conducted using transwell plates (8 µm; Corning Costar, Corning, NY, USA). Briefly, the 1 × 10⁵ transfected cells in 200 µL of RPMI 1640 medium were seeded into the upper compartments of transwells precoated with Matrigel (BD Biosciences, San Jose, CA), while 600 µL of medium supplemented with 10% FBS was added to the lower compartments of the chambers. After 24 h of culture, the cells in the upper chamber were removed, and those on the lower surface of the chamber were fixed with 4% formaldehyde and stained with 0.1% crystal violet. The cells that traversed the membrane were counted under light microscopy at a minimum of 5 predetermined positions per membrane.

The treated KYSE150 cells were seeded on the coverslips (NEST, Wuxi, China) at the appropriate density. The cells were fixed in 4% paraformaldehyde for 20 min and permeabilized with 0.1% Triton X-100 for 15 min. After washing with PBS, the cells were blocked with 2% bovine serum albumin, and then incubated with the rabbit anti-human polyclonal antibody APMAP (Cusabio, Wuhan, China, dilution at 1:100) at 4°C for 24 h. Finally, the cells were incubated with a fluorescent secondary antibody for 1 h at room temperature. Nuclei were stained with DAPI. The protein labeling was visualized under a confocal microscope.

Cell cycle distribution was analyzed by propidium iodide (PI; Multi Sciences, Hangzhou, China) mono-staining. Post-transfection cells were collected and resuspended to a concentration of 1×10⁶ cells/mL. After being washed twice with ice-cold PBS, cell suspensions were incubated with 500 μL of PI working solution for 30 minutes at 4°C in the dark. Cell cycle profiles were subsequently acquired using a FC-500 type flow cytometer (Beckman, Pasadena, USA).

Statistical analysis and graphing were conducted using GraphPad Prism 8.0 (La Jolla, CA, USA). The quantitative data, derived from at least three independent biological replicates. Comparisons between two groups were performed using Student’s t-test, while multiple group comparisons were analyzed by one-way ANOVA analysis. Data analysis was presented as mean ± standard deviation (S.D). Statistical significance was defined as *p < 0.05, ** p< 0.01.

To investigate the appropriate conditions for inducing ER stress in ESCC cells in vitro, we selected Tg as an inducer. Initially, TE1 cells were respectively treated with 50 nM, 100 nM, and 200 nM Tg for 12 h. The mRNA and protein expression levels of UPR effectors XBP1, ATF4, and ATF6 were notably increased, particularly at 100 nM ( Figure 1A ), thus 100 nM Tg was determined as the optimal concentration. TE1 cells were further incubated with 100 nM Tg for 12 h and 24 h, and the alterations in the UPR marker expression levels indicated that 12 h was the optimal treatment duration ( Figure 1B ). To ascertain the potential role of ER stress in ESCC cells, we performed RNA-sequencing analysis on TE1 cells treated with 100 nM Tg for 12 h, using DMSO-treated cells as a control (15), and CRELD2 was one of the most significantly differentially expressed genes. Further analysis of the UALCAN database (http://ualcan.path.uab.edu) revealed significant upregulation of CRELD2 in esophageal cancer ( Figure 1C ). The increased mRNA and protein expression levels of CRELD2 were also detected in three ESCC cell lines compared with normal esophageal epithelial cells (HEEC) ( Figures 1D, E ). Moreover, the upregulation of CRELD2 at mRNA and protein expression levels in response to ER stress was confirmed in both KYSE150 and TE1 cells, correlating with the expression of XBP1, ATF4, and ATF6 ( Figures 1F, G ). Given the biphasic nature of ER stress responses in tumor cells, we performed extended Tg treatment time-course (0-96 h) experiments in KYSE150 and TE1 cells. The expression level of CRELD2 peaked at 12 hours and subsequently declined during the late ER stress phase (48-96 hours) ( Supplementary Figure S1A ). This phase-specific expression pattern suggests that CRELD2 may act as an early ER stress-inducible gene, primarily responsible for regulating early-phase ER stress response in ESCC cells.

To recognize which branch of the UPR contributes to CRELD2 induction, as depicted in Figures 2A, B , three siRNAs (targeting PERK, IRE1α, and ATF6) were employed to individually silence the three ER transmembrane sensors in KYSE150 and TE1 cells. The depletion of PERK not only downregulated its canonical target ATF4 but also concomitantly impaired Tg-induced CRELD2 upregulation at both mRNA and protein expression levels ( Figures 2A–D ). Consistent with PERK pathway involvement, GSK2606414 (a PERK kinase inhibitor)-mediated blockade significantly decreased both mRNA and protein expression levels of ATF4, with a corresponding reduction in CRELD2 expression levels in KYSE150 and TE1 cells ( Figures 2E, F ). In contrast, the knockdown of IRE1α and ATF6 disrupted the induction of XBP1 and HSPA5 but not CRELD2 ( Figures 2A, B ). ATF4, a canonical target of PERK, plays a critical role as a transcription factor in ER stress. To explore the role of ATF4 in regulating transcription and expression of CRELD2, we initially overexpressed ATF4 in KYSE150 and TE1 cells, which was sufficient to increase the mRNA and protein expression levels of CRELD2 ( Figures 2G, H ). A putative ATF4 binding site in CRELD2 promoter region (-1149 ~ -1135 bp) was further predicted using JASPAR database ( Figure 2I ), and luciferase assays showed that ATF4-mediated activation of the CRELD2 promoter reporter construct was significantly dependent on this region ( Figure 2J ). These findings demonstrate that PERK/ATF4 pathway promotes transcription and expression of CRELD2 in response to ER stress in ESCC cells.

To investigate the potential effects of CRELD2 on the malignant biological behavior of ESCC cells in vitro, the pcDNA3.1-CRELD2 plasmid and si-CRELD2 were respectively employed to upregulate or downregulate the expression of CRELD2 in KYSE150 and TE1 cells ( Figure 3A ). Overexpression of CRELD2 markedly increased viability and colony-forming ability of KYSE150 and TE1 cells detected by MTS and colony formation assays ( Figures 3B, C ; Supplementary Figures S2A, B ). Moreover, overexpression of CRELD2 promoted KYSE150 and TE1 cells migration and invasion, as evidenced by wound healing and Transwell assays ( Figures 3D, E ; Supplementary Figures S2C, D ). Conversely, knockdown of CRELD2 produced the opposite effects ( Figures 3B–E ; Supplementary Figures S2A–D ). Flow cytometric cell cycle analysis revealed increased S-phase entry in CRELD2-overexpressing cells, whereas knockdown of CRELD2 caused G1 phase accumulation ( Figure 3F ; Supplementary Figure S2E ). Collectively, our findings demonstrate that CRELD2 may drive malignant phenotypes in ESCC cells in vitro.

Given the evident oncogenic role of CRELD2, we further performed functional assays to determine whether CRELD2 mediates ER stress-regulated ESCC progression. Knockdown of CRELD2 significantly attenuated the Tg-induced proliferation of KYSE150 and TE1 cells ( Figures 4A, B ; Supplementary Figures S3A, B ). Likewise, the enhanced migration and invasion abilities of KYSE150 and TE1 cells resulting from Tg treatment were also diminished by the knockdown of CRELD2 ( Figures 4C, D ; Supplementary Figures S3C, D ). These findings demonstrate that CRELD2 is critical for ER stress-driven malignant progression in ESCC.

We subsequently explored the underlying mechanism of CRELD2 in ESCC progression. Current studies support that CRELD2 is not only an endoplasmic reticulum-resident protein but also a secreted protein (10), and the subcellular localization of CRELD2 suggests a potential role for CRELD2 in trafficking proteins from the ER (16). The CRELD2 binding chaperones were identified via LC-MS/MS assay in chondrocytes and osteoblasts (16), and among the publicly reported potential binding proteins, adipocyte plasma membrane-associated protein (APMAP) caught our attention due to its localization and evident function. The physical interaction of CRELD2 with APMAP was verified by co-immunoprecipitation (CO-IP) assay in KYSE150 and TE1 cells, and the binding effect was further enhanced in Tg-treated cells ( Figure 5A ). Further analysis of the UALCAN database revealed significant upregulation of APMAP in esophageal cancer ( Figure 5B ), and the increased mRNA and protein expression levels of APMAP were also detected in three ESCC cell lines compared with HEEC cells ( Figures 5C, D ).

We further investigated the functional interplay between CRELD2 and APMAP. As shown in Figure 5E , neither overexpression nor knockdown of CRELD2 significantly altered mRNA or protein expression levels of APMAP in KYSE150 and TE1 cells. Moreover, the mRNA expression level of APMAP was not affected by Tg treatment ( Figure 5F ). Immunofluorescence was then employed to investigate whether CRELD2 could influence the location of APMAP, and as shown in Figure 5G , APMAP was located mainly on the cell membrane, and the fluorescence of APMAP on the membrane was significantly increased in Tg-treated cells, whereas this effect was partially blocked by the knockdown of CRELD2. Western blot analysis of membrane fractions revealed significantly increased APMAP localization at the plasma membrane following Tg treatment, which was partially reversed in CRELD2-deficient cells ( Figure 5H ). These data suggest that CRELD2 may act as a molecular chaperone for APMAP and promote APMAP localization to the cell membrane under ER stress condition.

To further investigate the biological function of APMAP in ESCC cells, si-APMAP was used to knock down the expression of APMAP in KYSE150 and TE1 cells and the knockdown efficiency was verified using qRT-PCR method ( Figure 6A ). The knockdown of APMAP impaired the proliferation, migration, and invasion abilities of KYSE150 and TE1 cells, as detected by MTS, colony formation, wound healing, and transwell assays ( Figures 6B–E ). In summary, APMAP may function as a cancerous gene in ESCC.

Since the ER stress-inducible CRELD2 can increase the membrane localization of APMAP, and both genes exhibit significant oncogenic effects, we further explored the potential mechanisms of APMAP in ESCC. Given the critical role of epithelial-mesenchymal transition (EMT) in cell migration and invasion (17), we systematically evaluated the impact of APMAP on key EMT and proliferation markers, including FN1, N-cadherin, ZEB2, VIMENTIN, SNAIL2, ZEB1, TWIST1, CDH1, SNAIL1, CCND1, CCNE1, and KI-67 ( Figure 7A ). Knockdown of APMAP in KYSE150 and TE1 cells consistently downregulated expression of FN1, N-cadherin, ZEB2, and CCND1 at both mRNA ( Figure 7A ) and protein levels ( Figure 7B ). These results suggest that APMAP may promote migration and invasion of ESCC by regulating EMT and enhances proliferation through regulating CCND1. Furthermore, overexpression or knockdown of CRELD2 in KYSE150 and TE1 cells recapitulated corresponding expression changes in APMAP-regulated genes ( Supplementary Figures S4A, B ). Given the established role of CRELD2 in promoting APMAP membrane localization, our findings collectively suggest that APMAP may act as the downstream executor of CRELD2-mediated signaling, transducing upstream stress responses into the regulation of EMT and proliferation.

To further investigate the signaling pathways underlying APMAP-mediated EMT and proliferation, we analyzed the effect of APMAP knockdown on key proteins in EMT- and proliferation-related pathways, including the TGF-β/SMAD, NF-κB, IL6/STAT3, Wnt/β-catenin, and PI3K/AKT signaling pathways ( Figure 7C ). Markedly decreased phosphorylation of SMAD2/3 and NF-κB p65 was observed upon APMAP knockdown in both KYSE150 and TE1 cell, whereas no significant reduction in STAT3 or β-catenin phosphorylation was detected, and AKT phosphorylation remained unaffected ( Figure 7C ). These results suggest that APMAP may promote carcinogenesis primarily through activation of TGF-β/SMAD and NF-κB pathway.

We further investigated the activation of p-SMAD2/3 and the regulatory role of APMAP under ER stress. As shown in Figure 7D , upregulation of p-SMAD2/3 was observed in Tg-treated KYSE150 and TE1 cells, which was partially inhibited by APMAP knockdown, indicating that the TGF-β/SMAD signaling pathway was activated under ER stress and that APMAP was involved in this effect. Given that the binding of CRELD2 to APMAP promotes APMAP localization on the cell membrane under ER stress, we speculated whether APMAP might be involved in regulating the activation or release of TGF-β receptors (TGFBRs) which are also localized on the cell membrane. It has been reported that endogenous TAK1 stably coimmunoprecipitates with TGFBR1 under TGF-β1 unstimulated condition, and TGF-β1 stimulation triggers the dissociation of TAK1 from TGFBR1 and induces phosphorylation and activation of TGFBR1 (18). We speculated that whether APMAP could competitively bind with TAK1 under ER stress and subsequently lead to the dissociation of TAK1 from TGFBR1 to further activate the TGF-β/SMAD signaling pathway. As provided in Figure 7E , endogenous TAK1 was coimmunoprecipitated with TGFBR1 both in KYSE150 and TE1 cells, while Tg treatment attenuated this binding effect. Conversely, the association of APMAP with TAK1 was significantly increased in Tg-treated ESCC cells. An APMAP overexpression plasmid was further constructed ( Supplementary Figures S5A, B ), and the overexpression of APMAP significantly reduced the interaction of TAK1 with TGFBR1 ( Figure 7F ), indicating that APMAP may mediate TGF-β/SMAD activation by binding to TAK1 in competition with TGFBR1 under ER stress condition.

We then investigated the regulatory role of APMAP in the activation of the NF-κB pathway. Phosphorylation-activated TAK1 can transduce signals to several downstream signaling cascades, including the MKK3/6-p38 MAPK cascade, MAPK kinase (MKK) 4/7-JNK cascade, and NF-κB inducing kinase-IκB kinase cascade (19). We speculated that in addition to binding to TAK1, APMAP may promote its phosphorylation and activation, thereby activating the NF-κB pathway. As shown in Figure 7G , knockdown of APMAP attenuated TAK1 phosphorylation, while overexpression of APMAP promoted the phosphorylation of TAK1, suggesting the regulatory role of APMAP in TAK1 phosphorylation and activation. However, APMAP does not possess kinase activity; we therefore conjectured that APMAP may indirectly regulate the phosphorylation of TAK1. It has been demonstrated that TAK1 can be activated by its interaction with TAK1-binding protein 1 (TAB1) and TAK1-binding protein 2 (TAB2). TAK1 forms complexes with TAB1 and TAB2, which contribute to the autophosphorylation of TAK1 (20). The physical interaction of APMAP with TAB1 and TAB2 was verified in KYSE150 and TE1 cells, and this binding effect was enhanced by Tg treatment ( Figure 7H ). These data suggest that APMAP may regulate activation of NF-κB pathway through promoting the phosphorylation of TAK1 by contributing to the formation of ternal complexes of TAK1, TAB1, and TAB2 under ER stress condition.