Abstract Chemotherapy remains a pivotal strategy in the treatment of pancreatic ductal adenocarcinoma (PDAC). Nonetheless, the emergence of drug resistance has limited the clinical efficacy of chemotherapeutic agents, especially gemcitabine. Here, we identified tetraspanin-15 (TSPAN15), a member of the tetraspanin family, that is frequently overexpressed in human PDAC and is correlated with tumor progression and poor prognosis. Elevated levels of TSPAN15 are involved in mediating gemcitabine resistance of in cancer cells, primarily by inhibiting ferroptosis. Knocking down TSPAN15 increases the sensitivity of PDAC cells to gemcitabine in vitro and in vivo by increasing the susceptibility of cancer cells to ferroptosis. Mechanistically, TSPAN15 directly interacts with integrin-β1 (ITGB1) and maintains its stability by inhibiting ITGB1 ubiquitination. This interaction activates the downstream p-FAK/p-AKT/p-mTOR axis and promotes the expression of glutathione peroxidase 4 (GPX4), a central negative regulator of ferroptosis, ultimately attenuating gemcitabine-induced ferroptosis in PDAC cells. Venetoclax, a newly identified targeted inhibitor of TSPAN15, exhibits synergistic efficacy when combined with gemcitabine for treating PDAC both in vitro and in vivo. This study reveals, for the first time, a major clinically relevant chemoresistance mechanism in PDAC mediated by TSPAN15 in sustaining ITGB1/p-FAK/p-AKT/p-mTOR-GPX4 signaling and tuning ferroptosis, revealing its potential as a viable therapeutic target for chemosensitization. Keywords: Pancreatic ductal adenocarcinoma, Tetraspanin-15, Gemcitabine, Chemoresistance, Ferroptosis Graphical abstract [35]Image 1 [36]Open in a new tab Highlights * • TSPAN15 drives chemoresistance by suppressing gemcitabine-induced ferroptosis in pancreatic ductal adenocarcinoma cells. * •TSPAN15 stabilizes ITGB1 through ubiquitination inhibition, forming a functional partnership critical for chemoresistance. * •TSPAN15-ITGB1 axis activates the p-FAK/p-AKT/p-mTOR signaling, upregulating GPX4 to enhance cellular antioxidant defense. * •Venetoclax targeting TSPAN15 synergizes with gemcitabine, demonstrating potent anti-tumor efficacy in PDAC models. 1. Introduction Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal solid tumor, with a dismal 5-year survival rate of 8–13 %[[37]1]. Owing to the absence of early symptoms, most patients with PDAC are diagnosed in the advanced unresectable stage[[38]2]. Therefore, gemcitabine-based chemotherapeutic regimens remain the cornerstone modality for PDAC treatment [[39]3]. However, widespread gemcitabine resistance is a constant challenge since it occurs in most treated patients and results in unsatisfactory therapeutic effects and poor prognoses. Several studies have investigated the underlying mechanisms of gemcitabine resistance in PDAC; however, the intrinsic process remains elusive[[40]4,[41]5]. Thus, an in-depth investigation of the mechanism of gemcitabine resistance and development of novel treatment strategies are of utmost importance for improving the prognosis of PDAC patients. The mechanisms of cell death are diverse, including apoptosis, necrosis, pyroptosis and autophagy, all of which are crucial for fundamental physiological processes of development and tissue homeostasis[[42]6,[43]7]. Effectively inducing cell death is a critical approach for overcoming the resistance of numerous chemotherapeutics and targeted preparations and enhancing drug effects[[44]8]. Ferroptosis is a novel form of nonapoptotic cell death characterized by iron-dependent peroxidation of the lipid membrane induced by the accumulation of significant amounts of reactive oxygen species (ROS)[[45]9]. Ferroptosis resistance in cancer cells is also favored by the increased antioxidant capacity of these cells[[46]10,[47]11]. Accumulated evidence indicates that ferroptosis plays a significant role in the fate of cancer cells and the response to various cancer treatments, such as chemotherapy, radiotherapy, and immunotherapy[[48][12], [49][13], [50][14]]. Tumor cells with drug resistance and a high metastatic tendency show increased susceptibility to ferroptosis, indicating that targeting the negative regulators of ferroptosis may further render chemoresistant cancer cells susceptible to ferroptotic cell death[[51]15]. Glutathione peroxidase 4 (GPX4), a central regulator of ferroptosis, can convert lipid hydroperoxides into lipid alcohols via reduced glutathione, thereby reducing lipid peroxidation and preventing ferroptosis[[52]16]. The upregulation or activation of the intracellular GPX4 level can induce cell resistance to ferroptosis, suppress the therapeutic effects of drugs, and ultimately result in tumor resistance to chemotherapeutics[[53]17]. Therefore, inactivation of GPX4 via genetic or pharmacological approaches may be a promising strategy to induce ferroptosis in cancer cells and overcome chemoresistance in various cancer cells[[54]18]. Tetraspanins are conserved small proteins characterized by four typical hydrophobic transmembrane domains (TM1-TM4), two extracellular loops (ECL1 and ECL2), and one intracellular loop[[55]19]. Generally, tetraspanins interact with other neighboring membrane proteins to form tetraspanin-enriched microdomains (TEMs). TEMs are considered specific functional structures that promote protein‒protein interactions, including those involving membrane receptors, adhesion molecules, and signal transduction molecules, and are involved in diverse biological processes[[56]20,[57]21]. Tetraspanin-15 (TSPAN15), a member of the tetraspanin family, was originally identified as a direct interaction partner of A disintegrin and metalloprotease 10 (ADAM10) which promotes the cleavage of N-cadherin in cell lines and animal models [[58]22]. In recent years, increasing evidence has suggested that TSPAN15 plays an important role in regulating cancer cell metastasis, enhancing cancer stem cell-like properties and inducing chemoresistance in esophageal squamous cell carcinoma and intrahepatic cholangiocarcinoma[[59]23,[60]24]. High TSPAN15 expression is reportedly associated with poor prognosis in several malignancies[[61]25,[62]26]. However, there are very few reports regarding whether and how TSPAN15 regulates malignant characteristics, particularly the chemoresistance of pancreatic cancer. In the present study, we revealed that TSPAN15 was the most significantly upregulated protein in PDAC tissues and that high expression of TSPAN15 in PDAC tissues was associated with tumor progression and poor patient prognosis. Next, we found that TSPAN15 was involved in regulating the sensitive of cancer cells to gemcitabine in a ferroptosis-dependent manner. Mechanistically, TSPAN15 could increase the expression of GPX4, a central negative regulator of ferroptosis, by directly sustaining integrin-β1 (ITGB1) stability and activating the p-FAK/p-AKT/p-mTOR signaling axis, ultimately promoting resistance to gemcitabine-induced ferroptosis in cancer cells. Targeted inhibition of TSPAN15 represents a viable approach for overcoming chemotherapy resistance in PDAC via the induction of ferroptosis. 2. Results 2.1. TSPAN15 is highly expressed in PDAC and is significantly correlated with tumor progression and poor prognosis To identify dysregulated proteins in human PDAC, we previously performed TMT-based quantitative proteomics analysis in five human PDAC tumor samples and paired adjacent normal tissues[[63]27]. Using a cutoff of log[2] (T/N fold change) ≥ 1 or ≤ −1 with p < 0.05, we identified 49 differentially expressed proteins, of which 34 were upregulated and 15 were downregulated. Notably, TSPAN15 was the most significantly upregulated protein in PDAC tissues among all identified upregulated proteins ([64]Fig. 1 A). Furthermore, we verified the mRNA level of TSPAN15 in PDAC via the Cancer Genome Atlas (TCGA) RNA sequencing dataset (GEPIA; [65]http://gepia. cancer-pku.cn/) and different GEO datasets. The results indicated that the level of TSPAN15 was significantly greater in PDAC tissues than in normal tissues and that high levels of TSPAN15 were significantly correlated with tumor progression and poor overall survival (p = 0.03) ([66]Fig. 1B–D, [67]Fig. S1 A-C). Fig. 1. [68]Fig. 1 [69]Open in a new tab TSPAN15 is highly expressed in PDAC tissues and is positively correlated with disease progression. (A) Five human PDAC tissue samples and paired adjacent normal pancreatic tissues were analyzed via TMT quantitative proteomics technology. Volcano diagram representing the differentially expressed proteins in human PDAC tissues vs. adjacent normal tissues, in which data were sorted according to a log[2] (T/N fold change) ≥ 1 or ≤ 1 and p < 0.05. TSPAN15 was the most significantly upregulated protein in PDAC tissues. (B) The mRNA level of TSPAN15 in pancreatic cancer tissues (T) vs. normal pancreatic tissues (N) was confirmed by analyzing the GEPIA database ([70]http://gepia.cancer-pku.cn/). (C) Kaplan–Meier overall survival curves for patients with low (blue) or high (red) expression of TSPAN15 mRNA in the PDAC via the GEPIA database. (D) TSPAN15 mRNA expression levels were significantly associated with the tumor stage of patients with PDAC according to the TCGA database. (E) Western blot analysis was performed to detect the expression level of TSPAN15 in 6 pairs of PDAC tissues and adjacent tissues (left panel); the protein band intensity was measured with ImageJ software and normalized to that of β-actin (right panel). (F) Immunohistochemical (IHC) staining of TSPAN15 protein expression in human PDAC tissues (tumor) and adjacent normal tissues (normal) (scale bars: 100 μm). (G) Statistical data for the immunohistochemical score of the TSPAN15 protein in PDAC tissues (T, n = 88) and adjacent normal tissues (N, n = 82). (H, I) TSPAN15 protein expression levels were significantly associated with distant tumor metastasis (H) and AJCC stages (I). (J, K) Kaplan‒Meier analyses of overall survival curves and disease-free survival curves. Patients with high TSPAN15 expression (n = 43) had significantly lower overall survival rates and disease-free survival rates than patients with low TSPAN15 expression (n = 45). All error bars represent the means ± SDs; ns: not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001. To further investigate the expression of TSPAN15 at the protein level in PDAC tissues, we first compared the level of TSPAN15 in 6 pairs of PDAC tissues and adjacent normal tissues via Western blot analysis. The results indicated that TSPAN15 expression was higher in PDAC than in paired normal tissues ([71]Fig. 1 E). Moreover, a tissue microarray of primary PDAC and matched nontumor tissues were studied via immunohistochemical (IHC) staining. The results also confirmed that the expression of TSPAN15 was significantly elevated in tumor tissues compared with adjacent normal tissues ([72]Fig. 1F and G). The TSPAN15 level in tumor tissues dramatically increased with tumor progression and was significantly correlated with distant metastasis ([73]Fig. 1H and I). Furthermore, higher expression levels of TSPAN15 were significantly associated with poor overall survival (p = 0.036) and disease-free survival (p = 0.014) in PDAC patients ([74]Fig. 1J and K, [75]Supplementary Table 2). 2.2. TSPAN15 promotes pancreatic cancer cell sphere formation and confers gemcitabine resistance in vitro and in vivo Given the unique characteristic that PDAC tissue contains 90 % nonmalignant stroma, we analyzed 24 published single-cell transcriptomic datasets of primary PDAC tissues to identify the cell types expressing TSPAN15[[76]28]. The results indicated that TSPAN15 was mainly expressed by malignant cancer cells (F[77]ig. S2 A, B). Therefore, we selected PDAC cell lines for subsequent functional experiments both in vitro and in vivo. To assess the role of high-levels TSPAN15 in cancer cell malignant progression, we analyzed the expression levels of TSPAN15 in different PDAC cell lines and found that the TSPAN15 expression level was greater in PANC-1 cell lines than in MIAPaCa-2 cell lines ([78]Fig. 2 A). We subsequently used MIAPaCa-2 cells to stably overexpress TSPAN15 and PANC-1 cells to knockdown TSPAN15 and investigate the role of TSPAN15 in PDAC malignant progression ([79]Fig. S2 C). The results indicated that neither the knockdown nor the overexpression of TSPAN15 significantly affected the proliferation or colony formation of tumor cells in vitro ([80]Fig. S2 D-G). However, we found that the knockdown of TSPAN15 significantly inhibited the sphere formation efficiency, as reflected by the size and number of spheres formed by PANC-1 cells ([81]Fig. 2 B). In contrast, TSPAN15 overexpression promoted the sphere formation efficiency of MIAPaCa-2 cells ([82]Fig. 2 C). Since chemoresistance is a key feature of cancer stemness, we were curious whether TSPAN15 was involved in the modulating chemoresistance of PDAC cells. Our data revealed that the drug sensitivity of the PANC-1 cells to gemcitabine was significantly greater in TSPAN15-knockdown cells than in PANC-1-shNC cells ([83]Fig. 2D, F). Additionally, PANC-1-shTSPAN15 cells treated with gemcitabine exhibited a significantly increased rate of cell death (Annexin-Ⅴ^-, 7-AAD^+), although the number of apoptotic cells (Annexin-Ⅴ^+, 7-AAD^+/Annexin-Ⅴ^+, 7-AAD^-) remained unaltered ([84]Fig. 2 H), suggesting that the survival mechanism mediated by TSPAN15 may not be dependent on the apoptotic pathway in PDAC cells. Conversely, when TSPAN15 was overexpressed in MIAPaCa-2 cells, the opposite effects were observed ([85]Fig. 2E, G, I). Furthermore, we also evaluated TSPAN15 expression levels in different PANC-1 cell lines and showed that TSPAN15 expression was higher in gemcitabine-resistant PANC-1 cell lines than in wild-type PANC-1 cell lines ([86]Fig. 2 J). Following continuous treatment of TSPAN15-knockdown PANC-1 cells with a low dose of gemcitabine for 3 weeks, the expression of TSPAN15 was restored ([87]Fig. 2 K). Fig. 2. [88]Fig. 2 [89]Open in a new tab TSPAN15 promotes pancreatic cancer cell sphere formation and gemcitabine resistance in vitro. (A) The expression levels of TSPAN15 in different PDAC cell lines were assessed by Western blot analysis (top). The relative expression of TSPAN15 was quantified by normalization to that of β-actin (bottom). (B, C) Representative tumor-sphere images (up) and quantification (down) of PANC-1-shNC/shTSPAN15 cells and MIAPaCa-2-vector/TSPAN15 cells cultured with sphere-forming medium (scale bars, 50 μm). (D, E) PANC-1 cells with stable knockdown of TSPAN15 (D) and MIAPaCa-2 cells with stable overexpression of TSPAN15 (E) were treated with gemcitabine at different concentrations for 48 h, and relative cell viability was then measured via the CCK-8 assay. (F, G) Representative images and quantification of the clonogenic assay for PANC-1-shNC/shTSPAN15 and MIAPaCa-2-vector/TSPAN15 cells treated with different concentrations of gemcitabine. (H, I) The indicated cells were treated with 5 μM (PANC-1) or 2.5 μM (MIAPaCa-2) gemcitabine for 48 h and then labeled with annexin-V-APC and 7-AAD, followed by flow cytometry to detect percentages of Annexin-Ⅴ-APC^+ or 7-AAD^+ cells among PANC-1-shNC/shTSPAN15 or MIAPaCa-2-vector/TSPAN15 cells treated with DMSO or gemcitabine for 48 h. Left, representative results of annexin-V-APC/7-AAD staining. Right, quantitative analysis. (J) Western blot analysis showing the expression level of TSPAN15 in PANC-1 cells and gemcitabine-resistant PANC-1 cells (PANC-1-Gem) (continuously treated with 200 nM gemcitabine for 12 weeks) (left). The relative expression of TSPAN15 was quantified via normalization to that of β-actin (right). (K) Western blot analysis showing the expression of TSPAN15 in PANC-1-shNC, PANC-1-shTSPAN15 and PANC-1-shTSPAN15 cells after treatment with 200 nM gemcitabine for 3 weeks (left). The relative expression of TSPAN15 was quantified by normalization to that of β-actin (right). All error bars represent the means ± SDs; ns: not significant; ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001. To further demonstrate the role of TSPAN15 in gemcitabine resistance, we collected drug screening data from the Genomics Drug Sensitivity in Cancer (GDSC) data portal and found that the TSPAN15 high-expression cohort of PDAC cell lines presented notable resistance to multiple chemotherapy agents, including gemcitabine, cisplatin, paclitaxel and erlotinib ([90]Fig. S2 H). Analyses of the TCGA database also revealed that the mRNA levels of TSPAN15 in PDAC tissues were positively correlated with several cell stemness-related markers (CD44, NANOG, LGR5, and PROM1) ([91]Fig. S2 I). These results provide strong evidence that TSPAN15 has a trivial effect on the growth and proliferation of PDAC cells but it can significantly promote tumor cell stemness and resistance to gemcitabine in vitro. To validate these findings in vivo, we established a xenograft PDAC mouse model by injecting PANC-1-shNC or PANC-1-shTSPAN15 cells (1 × 10^6) into the right flank regions of nude mice. We found that the knockdown of TSPAN15 could somewhat inhibited the growth of PANC-1 cells growth in terms of xenograft size and weight, but the inhibition was not significant. However, when we reduced the number of injected cells by 100-fold (1 × 10^4), the incidence and weight of xenografts in the PANC-1-shTSPAN15 group significantly decreased ([92]Fig. 3A–C), which is consistent with our in vitro results. In addition, IHC analysis of tumor xenograft samples also revealed that the expression of Ki67 and a marker of stemness CD44 were significantly inhibited by TSPAN15 downregulation ([93]Fig. 3 D). As anticipated, TSPAN15 knockdown also markedly increased the sensitivity of PANC-1 cells to gemcitabine treatment and decreased the shTSPAN15 tumor volume and weight in vivo ([94]Fig. 3E–H). Together, our results suggest a novel role for TSPAN15 in maintaining PDAC stemness and contributing to drug resistance in PDAC cells. Fig. 3. [95]Fig. 3 [96]Open in a new tab Targeting TSPAN15 inhibited tumorigenicity and increased the sensitivity of PANC-1 cells to gemcitabine in vivo. (A) A total of 1 × 10^6 PANC-1-shNC/shTSPAN15 cells or dilutions of 100-fold (1 × 10^4) were subcutaneously inoculated into nude mice to determine their tumorigenicity. (B) The stem cell frequency was estimated via an online tool available at [97]http://bioinf.wehi.edu.au/software/elda. (C) Statistical analysis of xenograft tumor weights in nude mice at the study endpoint after transplantation of the indicated number of PANC-1-shNC/shTSPAN15 cells. (D) Representative images of immunohistochemical staining of serial xenograft tumor samples stained for Ki67 and CD44 are shown (scale bars, 50 μm). (E) Schematic representation of the drug administration schedule for the xenograft tumor model. (F–H) Representative images (F) and statistical analysis of xenograft tumor volumes (H) and weights (G) in nude mice after implantation of 1 × 10^6 PANC-1-shNC/shTSPAN15 cells, followed by intraperitoneal injection of DMSO or gemcitabine (10 mg/kg) every 3 days (n = 6 mice per group). All error bars represent the means ± SDs; ns: not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001. 2.3. TSPAN15 is involved in regulating ferroptosis in PDAC cells To elucidate the mechanism by which TSPAN15 maintains stemness and drug resistance in PDAC cells, we performed RNA sequencing in PANC-1-shNC and PANC-1-shTSPAN15 cells after treatment with gemcitabine for 48 h. A total of 2238 transcripts were significantly differentially expressed (2-fold change, p < 0.05) in PANC-1-shTSPAN15 cells, including 1260 upregulated genes and 978 downregulated genes ([98]Fig. 4 A). We selected all the DEGs for KEGG pathway enrichment analysis and found that several pathways were enriched, among which the ferroptosis, peroxisome, and oxidative phosphorylation pathways attracted our attention ([99]Fig. 4 B). Fig. 4. [100]Fig. 4 [101]Open in a new tab TSPAN15 regulates ferroptosis in PDAC cells. (A) Heatmap representing the genes differentially expressed between PANC-1-shNC and PANC-1-shTSPAN15 cells following treatment with gemcitabine (Gem, 2.5 μM) for 48 h, as determined by RNA-sequencing analysis, in which data were sorted as log[2] (fold change) ≥ 1 or ≤ 1 and p < 0.05. (B) KEGG pathway enrichment analysis of DEGs in PANC-1-shNC-Gem/shTSPAN15-Gem cells based on RNA-sequencing data. (C) Western blot analysis revealed the expression level of GPX4 in PANC-1-shNC/shTSPAN15 and MIAPaCa-2-vector/TSPAN15 cells (left). The relative expression of GPX4 was quantified via normalization to that of β-actin (right). (D, E) The CCK-8 assay was used to assess the relative viability of PANC-1-shNC/shTSPAN15 (D) and MIAPaCa-2-vector/TSPAN15 (E) cells after treatment with 5 μM erastin alone or in combination with the 10 μM ferroptosis inhibitor liproxstatin-1 (Lip) for 48 h. (F, G) Representative images (left) and quantification (right) of the results of the clonogenic assay of PANC-1-shNC/shTSPAN15 (F) and MIAPaCa-2-vector/TSPAN15 (G) cells treated with 5 μM erastin alone or in combination with the 10 μM ferroptosis inhibitor liproxstatin-1 (Lip) for 7 days. (H, I) Representative FACS images of the lipid ROS levels in PANC-1-shNC/shTSPAN15 (H) and MIAPaCa-2-vector/TSPAN15 (I) cells treated with 2.5 μM erastin alone or in combination with 10 μM ferroptosis inhibitor liproxstatin-1 (Lip) for 12 h. (J, K) Representative transmission electron microscopy images of PANC-1-shNC/shTSPAN15 (J) and MIAPaCa-2-vector/TSPAN15 (K) cells treated with erastin (10 μM) or additional liproxstatin-1 (10 μM) for 12 h. All error bars represent the means ± SDs; ns: not significant; ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Ferroptosis is a novel form of cell death that results from the production of iron-dependent reactive oxygen species (ROS) from excessive lipid peroxidation. Previous studies have demonstrated that GPX4 is a central regulator of ferroptosis, as it converts lipid hydroperoxides into lipid alcohols via reduced glutathione, thereby reducing lipid peroxidation and preventing ferroptosis. Our results revealed that knocking down TSPAN15 in PANC-1 cells significantly suppressed the expression of GPX4, and increasing the expression of TSPAN15 in MIAPaCa-2 cells increased GPX4 expression ([102]Fig. 4 C), suggesting that TSPAN15 may participate in regulating ferroptosis in PDAC cells. To verify this hypothesis, we treated PANC-1/MIAPaCa-2 cells with the ferroptosis inducer erastin (5 μM) either alone or in combination with the ferroptosis inhibitor liproxstatin-1 (10 μM). The results suggested that, compared with the PANC-1-shNC group, erastin treatment significantly decreased the cells viability and clone formation in the PANC-1-shTSPAN15 group. Moreover, the ferroptosis inhibitor liproxstatin-1 dramatically rescued this phenomenon ([103]Fig. 4D, F). Additionally, the level of ROS in PANC-1/MIAPaCa-2 cells was detected by flow cytometry, which revealed that the baseline levels of ROS in PANC-1-shTSPAN15 cells were slightly greater than those in PANC-1-shNC cells ([104]Fig. 4 H, [105]Fig. S3A). However, after erastin treatment, the ROS levels in PANC-1-shTSPAN15 cells increased more prominently than those in PANC-1-shNC cells, while concurrent administration of liproxstatin-1 largely rescued the aforementioned effect ([106]Fig. 4 H). Meanwhile, morphological changes in mitochondria, including a smaller volume, thicker membrane and fewer mitochondrial cristae, were also observed in PANC-1-shTSPAN15 cells, and these phenomena were reinforced following erastin treatment ([107]Fig. 4 J). These results were reversed in MIAPaCa-2 cells after was overexpressed TSPAN15 ([108]Fig. 4E, G, I, K, [109]Fig. S3B). Taken together, these experimental results confirmed that TSPAN15 is involved in the regulation of ferroptosis in PDAC cells. 2.4. TSPAN15 regulates the sensitivity of PDAC cells to gemcitabine by modulating ferroptosis To further verify that TSPAN15 regulates the sensitivity of PDAC cells to gemcitabine by modulating ferroptosis, we added various cell death inhibitors, including ferrostatin-1 (Fer1) and liproxstatin-1 (Lip), two specific ferroptosis inhibitors, necrostatin-1 (Nec1), a necrosis inhibitor, bafilomycin A1(Baf-A1), an autophagy inhibitor, and Z-VAD-fmk (Z-VAD), an apoptosis inhibitor, to cancer cells in an attempt to rescue the potential cell death induced by gemcitabine. As shown in [110]Fig. 5A, necrostatin-1 and Baf-A1 did not alleviate the cell death induced by gemcitabine in PANC-1 cells after the knockdown of TSPAN15. Z-VAD moderately inhibited gemcitabine-induced death in PANC-1-shTSPAN15 cells. However, ferrostatin-1 and liproxstatin-1 significantly attenuated gemcitabine-induced cell death in PANC-1-shTSPAN15 cells. In addition, the ferroptosis inhibitor liproxstatin-1 significantly restored the colony formation ability of PANC-1-shTSPAN15 cells treated with gemcitabine ([111]Fig. 5 B). The rate of cell death (Annexin-Ⅴ^-, 7-AAD^+) and the level of ROS accumulation induced by gemcitabine in PANC-1-shTSPAN15 cells also decreased following treatment with liproxstatin-1 ([112]Fig. 5D, F; [113]Fig. S4A). Furthermore, TEM images of the mitochondrial structure revealed that liproxstatin-1 partially reversed the morphological changes in the mitochondria, including a smaller volume, thicker membrane and fewer mitochondrial cristae in PANC-1-shTSPAN15 cells after treatment with gemcitabine ([114]Fig. 5 H). Since the level of lipid peroxidation reflects the degree of ferroptosis, we then investigated the level of lipid peroxidation in PDAC cell lines. The results indicated that the baseline levels of MDA and the GSSG/GSH ratio were not dramatically different between PANC-1-shNC and PANC-1-shTSPAN15 cells, whereas the MDA levels and the relative ratio of GSSG/GSH were significantly increased in PANC-1-shTSPAN15 cells following treatment with gemcitabine ([115]Fig. S4D–G). In addition, we also performed the above experiments in MIAPaCa-2 cells overexpressing TSPAN15, and as expected, increasing the expression of TSPAN15 in MIAPaCa-2 cells led to the opposite results ([116]Fig. 5C, E, G, I, [117]Fig. S4B–C). Taken together, these results strengthened the effect of TSPAN15 on gemcitabine resistance through regulating ferroptosis in PDAC cells. Fig. 5. [118]Fig. 5 [119]Open in a new tab TSPAN15 regulates the sensitivity of PDAC cells to gemcitabine by modulating ferroptosis. (A) The relative viability of PANC-1-shTSPAN15 cells was detected via a CCK-8 assay following treatment with gemcitabine (Gem, 5 μM) in the presence or absence of various agents: ferrostatin-1 (Fer1, 10 μM), necostatin-1 (Nec1, 1 mM), liproxstatin-1 (Lip, 10 μM), bafilomycin A1(Baf-A1, 10 μM) or z-VAD-FMK (z-VAD, 10 μM) for 48 h. (B, C) Representative images (left) and quantification (right) of the results of the clonogenic assay of PANC-1-shNC/shTSPAN15 (B) and MIAPaCa-2-vector/TSPAN15 (C) cells treated with 0.5 μM gemcitabine alone or in combination with 10 μM ferroptosis inhibitor liproxstatin-1 (Lip) for 7 days. (D, E) The indicated cells were treated with 5 μM gemcitabine in the presence or absence of liproxstatin-1 (Lip, 10 μM) for 48 h and then collected for labeling with annexin-V-APC and 7-AAD, followed by flow cytometry to detect the percentages of Annexin-Ⅴ-APC^+, 7-AAD^+ PANC-1-shNC/shTSPAN15 or MIAPaCa-2-vector/TSPAN15 cells. Left, representative results of annexin-V-APC/7-AAD staining. Right, quantitative analysis. (F, G) Representative FACS images of the lipid ROS levels in PANC-1-shNC/shTSPAN15 (F) and MIAPaCa-2-vector/TSPAN15 (G) cells treated with 5 μM gemcitabine alone or in combination with 10 μM ferroptosis inhibitor liproxstatin-1 (Lip) for 12 h. (H, I) Representative transmission electron microscopy images of PANC-1-shNC/shTSPAN15 (H) and MIAPaCa-2-vector/TSPAN15 (I) cells treated with gemcitabine (5 μM) or additional liproxstatin-1 (10 μM) for 12 h (scale bars, 500 nm). All error bars represent the means ± SDs; ns: not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. 2.5. TSPAN15 promotes resistance to gemcitabine-induced ferroptosis in PDAC cells by activating the p-FAK/p-AKT/p-mTOR-GPX4 axis The above data provide solid evidence that TSPAN15 regulates the sensitivity of PDAC cells to gemcitabine by modulating ferroptosis. However, the potential mechanism by which TSPAN15 regulates this process remains obscure. We therefore performed RNA sequencing in PANC-1-shNC and PANC-1-shTSPAN15 cells (without Gem treatment), and a total of 582 differentially expressed transcripts were identified and subjected to gene set enrichment analysis ([120]Fig. S5A). The PI3K-AKT signaling, ferroptosis and peroxisome pathways were significantly enriched ([121]Fig. 6 A, [122]Fig. S5B), and these pathways were also enriched in the genes differentially expressed between PANC-1-shNC and PANC-1-shTSPAN15 cells after treatment with gemcitabine ([123]Fig. 4 B), which suggests that the PI3K-AKT signaling pathway may be involved in TSPAN15-regulated gemcitabine-induced ferroptosis in PDAC cells. To verify this result, the phosphorylation levels of key proteins of the PI3K-AKT pathway were analyzed via Western blotting. The phosphorylation of FAK, AKT and mTOR significantly decreased after TSPAN15 was knocked down in PANC-1 cells, whereas the total protein levels of these proteins remained unchanged ([124]Fig. 6 B). In contrast, when TSPAN15 was overexpressed in MIAPaCa-2 cells, the phosphorylation levels of FAK, AKT and mTOR were significantly increased ([125]Fig. S5C). Moreover, knockdown of TSPAN15 significantly inhibited the expression of GPX4 and the stemness markers CD44 and NANOG in PANC-1 cells, whereas overexpression of TSPAN15 in MIAPaCa-2 cells resulted in the opposite effect ([126]Fig. 6 B, [127]Fig. S5C). These results collectively demonstrated that TSPAN15 regulates gemcitabine-induced ferroptosis in PDAC cells by modulating the p-FAK/p-AKT/p-mTOR-GPX4 signaling axis. Fig. 6. [128]Fig. 6 [129]Open in a new tab TSPAN15 directly sustains the protein stability of ITGB1 and activates the p-FAK/p-AKT/p-mTOR-GPX4 signaling axis to promote gemcitabine-induced ferroptosis resistance in pancreatic cancer cells. (A) Gene set enrichment analysis (GSEA) demonstrating that PI3K-AKT pathway related genes were significantly enriched following TSPAN15 knockdown in PANC-1 cells (p = 0.029). (B) The protein levels of ITGB1, p-FAK, p-AKT, p-mTOR, GPX4, NANOG and CD44 in PANC-1-shNC/shTSPAN15 cells were assessed by Western blotting (left). The relative expression of proteins was quantified via normalization to that of β-actin (right). (C) The relative mRNA expression levels of ITGB1 in PANC-1-shNC and PANC-1-shTSPAN15 cells were detected via RT-PCR. (D) The physical interaction of endogenous TSPAN15 with ITGB1 was detected by co-IP with anti-TSPAN15 and anti-ITGB1 antibodies in PANC-1 cells. (E) The physical interaction of exogenous TSPAN15 with ITGB1 was monitored by co-IP with anti-Ha antibody or anti-Flag antibody in HEK293T cells transfected with TSPAN15-Ha- and/or ITGB1-Flag-expressing plasmids. (F) Immunofluorescence assay showing the colocalization (yellow) of TSPAN15 (green) and ITGB1 (red) in PANC-1-shNC and PANC-1-shTSPAN15 cells (scale bar: 10 μm). (G) The turnover of ITGB1 proteins in PANC-1-shNC/shTSPAN15 cells was measured by CHX (150 μg/ml) treatment and then detected by Western blot analysis. (H) Representative Western blot showing TSPAN15 and ITGB1 expression levels in PANC-1-shNC/shTSPAN15 cells treated with MG132 (20 μM) for 6 h (top). The relative expression of proteins was quantified by normalization to that of β-actin (bottom). (I) Co-IP with an anti-ITGB1 antibody and representative a Western blot for ubiquitin and ITGB1 in PANC-1-shNC/shTSPAN15 cells. (J) The protein levels of p-FAK, p-AKT, p-mTOR, GPX4, NANOG and CD44 in MIAPaCa-2-vector and MIAPaCa-2-TSPAN15 cells treated with or without 100 ng/ml ITGB1 neutralizing antibody OS2966 for 24 h were assessed by Western blotting (left). The relative expression of proteins was quantified by normalization to that of β-actin (right). All error bars represent the means ± SDs; ns: not significant; ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001. 2.6. TSPAN15 maintains the stability of ITGB1 by inhibiting its ubiquitination and degradation, thereby activating the p-FAK/p-AKT/p-mTOR-GPX4 axis Integrin subunit beta 1 (ITGB1) is a member of the integrin (ITG) family that is closely related to the PI3K/AKT signaling pathway and has been widely studied in tumors, including hepatocellular carcinoma and cervical cancer. Using the STRING database, we found that TSPAN15 may interact with ITGB1 ([130]Fig. S5D). Furthermore, Western blotting confirmed that the knockdown of TSPAN15 in PANC-1 cells significantly decreased the protein level of ITGB1, whereas the overexpression of TSPAN15 in MIAPaCa-2 cells significantly increased the level of ITGB1 ([131]Fig. 6 B, [132]Fig. S5C). However, RT‒qPCR analysis revealed that the level of TSPAN15 did not affect the ITGB1 mRNA level in PDAC cells ([133]Fig. 6 C). This finding led us to hypothesize that TSPAN15 may interact with ITGB1 to maintain its stability within PDAC cells. To verify this hypothesis, we performed coimmunoprecipitation analyses and revealed physical interactions between endogenous TSPAN15 and ITGB1 in PDAC cells and between exogenously introduced Ha-tagged TSPAN15 and Flag-tagged ITGB1 in HEK293T cells ([134]Fig. 6D and E). Moreover, the immunofluorescence data further demonstrated that TSPAN15 and ITGB1 were significantly colocalized especially on the membrane, in PANC-1 and MIAPaCa-2 cells. However, after the knockdown of TSPAN15 in PANC-1 cells, the protein expression level of ITGB1 protein and its colocalization with TSPAN15 on the membrane were significantly decreased ([135]Fig. 6 F, [136]Fig. S5E). Next, we examined whether the introduction of TSPAN15 could increase the stability of the ITGB1 protein. Cycloheximide (CHX), a protein synthesis inhibitor, was used to determine the effect of TSPAN15 on ITGB1 stability. Our results revealed that TSPAN15 depletion decreased ITGB1 expression with shortened half-lives ([137]Fig. 6 G). In contrast, TSPAN15 overexpression stabilized ITGB1 expression by prolonging the half-lives of both proteins ([138]Fig. S5F). To further evaluate the relationship between ITGB1 protein stability mediated by TSPAN15 and the proteasome system, we utilized MG132, a 26S proteasome inhibitor. Notably, treatment with MG132 resulted in the accumulation of the ITGB1 protein, as shown in PANC-1-shNC/PANC-1-shTSPAN15 cells. Additionally, ITGB1 protein levels induced by proteasome inhibition could not be further decreased by TSPAN15 knockdown ([139]Fig. 6 H, [140]Fig. S5G). We subsequently investigated whether TSPAN15 maintains ITGB1 stability by regulating its ubiquitination. Compared with PANC-1-shNC, the inhibition of TSPAN15 led to an increase in ITGB1 ubiquitination and a decrease in ITGB1 protein levels ([141]Fig. 6 I). Conversely, overexpression of TSPAN15 in MIAPaCa-2 cells resulted in reduced ubiquitination of ITGB1 compared with that in MIAPaCa-2-vector cells ([142]Fig. S5H). These results indicated that TSPAN15 interacts with ITGB1, leading to the deubiquitination of ITGB1 and thereby maintaining its stability. To further demonstrate that TSPAN15 regulates p-FAK/p-AKT/p-mTOR-GPX4 signaling axis by stabilizing ITGB1 and thereby regulating ferroptosis in PDAC cells, MIAPaCa-2-TSPAN15 cells were treated with the ITGB1 neutralizing antibody OS2966. We found that OS2966 reversed the increased phosphorylation of FAK, AKT and mTOR caused by TSPAN15 overexpression in MIAPaCa-2 cells. Similarly, OS2966 was also effective at inhibiting the high expression of GPX4, CD44 and NANOG in MIAPaCa-2-TSPAN15 cells ([143]Fig. 6 J). In addition, OS2966 treatment significantly reversed the resistance to ferroptosis induced by gemcitabine or erastin and promoted ROS accumulation in MIAPaCa-2-TSPAN15 cells ([144]Fig. S6). Taken together, these data demonstrated that TSPAN15 promotes resistance to gemcitabine-induced ferroptosis in pancreatic cancer cells by directly sustaining the protein stability of ITGB1 and activating the p-FAK/p-AKT/p-mTOR-GPX4 signaling axis. 2.7. Venetoclax is a novel TSPAN15 inhibitor that exhibits synergistic efficacy with gemcitabine to induce ferroptosis in pancreatic cancer cells Next, we used molecular docking to further explore the direct binding of TSPAN15 and ITGB1, and the results revealed that the binding sites of TSPAN15 included mainly the Q211, D170, L177, Y200, S172, and P174 amino acid residues and that the binding sites of the ITGB1 protein included the R676, K619, R562, N564, and A589 amino acid residues via salt bridges, hydrogen bonds, and hydrophobic interactions ([145]Fig. 7 A). Next, we employed TargetMol to conduct a virtual screening of 2808 compounds based on the predicted interaction sites of TSPAN15 and ITGB1 to identify potential inhibitors of TSPAN15 ([146]Supplementary Table 3). The top 10 compounds based on docking scores were selected for in vitro cellular experiments to validate their activity ([147]Fig. 7 B). We found that among the 10 compounds tested, only venetoclax could significantly increase the sensitivity of PANC-1 cells to gemcitabine, whereas its monotherapy did not effectively alter the survival and colony formation of PDAC cells ([148]Fig. 7C–E). In addition, the drug ZIP synergy scores were calculated via the online SynergyFinder software. As shown in [149]Fig. 7 F, the white rectangle indicates the region of the maximum synergistic area, and the results showed that combined treatment with gemcitabine and venetoclax had highly synergistic effects in inhibiting tumor proliferation (ZIP synergy scores >10). Moreover, venetoclax inhibited the phosphorylation of FAK, AKT, and mTOR and the expression of ITGB1 and GPX4 in PANC-1 cells in a concentration gradient-dependent manner ([150]Fig. 7 H). Venetoclax treatment also dramatically reversed the increased phosphorylation of FAK, AKT, and mTOR as well as the increased levels of ITGB1 and GPX4 caused by TSPAN15 overexpression in MIAPaCa-2 cells ([151]Fig. 7 G). All of these results were in accordance with those obtained by directly knocking down TSPAN15 or treating PDAC cells with ITGB1 neutralizing antibodies. In addition, venetoclax treatment also increased ITGB1 ubiquitination and decreased ITGB1 protein levels, which was similar to the results of direct knockdown of TSPAN15 in PANC-1 cells ([152]Fig. 7 I). To further evaluate venetoclax-targeted interactions, we performed a cellular thermal shift assay (CETSA). Drug‒protein interactions were examined in the native cellular environment based on the ligand-induced changes in protein thermal stability via a CETSA. Venetoclax was able to bind and stabilize the TSPAN15 protein in PANC-1 cells ([153]Fig. 7 J), confirming the direct interaction between venetoclax and the TSPAN15 protein. Fig. 7. [154]Fig. 7 [155]Open in a new tab Venetoclax is a TSPAN15 inhibitor that promotes gemcitabine-induced ferroptosis by inhibiting the TSPAN15-dependent p-FAK/p-AKT/p-mTOR-GPX4 signaling axis in pancreatic cancer cells. (A) Predicted binding complex models of TSPAN15 and ITGB1 via the AlphaFold protein docking tool ([156]https://alphafold.ebi.ac.uk/). (B) Relative cell viability of PANC-1 cells was detected by CCK-8 assay following treatment with gemcitabine (Gem, 5 μM) alone or in combination with the top 10 compounds for 48 h (BC-48: demecarium bromide, AG1341: nelfinavir, RDX: tenapanor, TD: revefenacin, Vene: venetoclax, Dabig: dabigatran etexilate mesylate, Dacla: daclatasvir, Pante: pantethine, Saqui: saquinavir). (C) Venetoclax binds to the active pocket of TSPAN15 through the formation of three hydrogen bonds with residues C177, K201 and Y200. (D) The relative viability of PANC-1 and MIAPaCa-2 cells treated with 5 μM gemcitabine or 10 μM venetoclax alone or a combination of both for 48 h was detected via a CCK-8 assay. (E) Representative images (left) and quantification (right) of the results of the clonogenic assay of PANC-1 and MIAPaCa-2 cells treated with 5 μM gemcitabine or 10 μM venetoclax alone or a combination of both for 7 days. (F) Heatmaps depicting the combination drug responses of gemcitabine and venetoclax in PANC-1 cells. Relative cells viability was determined via a CCK-8 assay after treatment with gemcitabine and venetoclax for 48 h, and SynergyFinder ([157]https://synergyfinder.fimm.fi) was subsequently used to estimate drug interactions via ZIP synergy scores. ZIP synergy scores < −10 indicated antagonism, scores from −10 to 10 suggested addition, and scores >10 indicated synergism. (G) Western blot assays were conducted to assess the expression of ITGB1, p-FAK, p-AKT, p-mTOR and GPX4 in MIAPaCa-2-vector cells and MIAPaCa-2-TSPAN15 cells with or without venetoclax treatment (25 μM, 48 h) (left). The relative expression of proteins was quantified by normalization to that of β-actin (right). (H) Western blot assays were conducted to assess the expression of ITGB1, p-FAK, p-AKT, p-mTOR and GPX4 in PANC-1 cells treated with 5 μM or 10 μM venetoclax for 48 h (left). The relative expression of proteins was quantified by normalization to that of β-actin (right). (I) Co-IP with an ITGB1 antibody and a representative Western blot for ubiquitin and ITGB1 in PANC-1 cells after treatment with or without 10 μM venetoclax for 24 h. (J) Cellular thermal shift assays (CETSAs) were performed to confirm the binding specificity of venetoclax with TSPAN15 in the cell lysates of PANC-1 cells. Plots represent the signal intensity of TSPAN15 normalized to the signal intensity at 46 °C. All error bars represent the means ± SDs; ns: not significant; ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001. To assess the clinical relevance of these results, we conducted an animal study in which venetoclax was orally administered in combination with gemcitabine. We found that venetoclax treatment alone had no effect on tumor growth or tumor weight, but it could synergize with gemcitabine treatment and significantly inhibited tumor growth and tumor weight ([158]Fig. 8A–C). Furthermore, immunohistochemical staining revealed that venetoclax along or combined with gemcitabine treatment significantly downregulated the expression of GPX4, ITGB1, CD44 and Ki67 in tumor samples ([159]Fig. 8 D). Both venetoclax and gemcitabine were well tolerated with no significant weight loss in the mice in the monotherapy and combination treatment groups ([160]Fig. 8 E). These results indicate that venetoclax, as a novel inhibitor of TSPAN15, exhibits synergistic efficacy when combined with gemcitabine by inducing ferroptosis in pancreatic cancer cells. Fig. 8. [161]Fig. 8 [162]Open in a new tab Venetoclax treatment enhances the sensitivity of PDAC xenograft tumors to gemcitabine in vivo. (A–C) Representative images (A) and statistical analysis of xenograft tumor volumes (B) and weights (C) in nude mice after implantation of 1 × 10^6 PANC-1 cells, followed by intraperitoneal injection of DMSO or gemcitabine (10 mg/kg), oral administration of venetoclax (100 mg/kg), or a combination of both treatments every 3 days (n = 10 mice per group). (D) Representative images of immunohistochemical staining of serial xenograft tumor samples stained for GPX4, ITGB1, CD44 and Ki67 are shown (scale bars, 50 μm). (E) The body weights of the mice were measured every three days. All error bars represent the means ± SDs; ns: not significant; ∗∗p < 0.01 and ∗∗∗∗p < 0.0001. 3. Discussion As one of the notoriously aggressive cancers with the highest mortality, the diagnosis and treatment strategies for PDAC are far from satisfactory. Although gemcitabine-based chemotherapy has been well recognized as the mainstay therapy for advanced and metastatic pancreatic cancers for decades, the development of chemoresistance severely limits its treatment efficiency[[163]3]. Therefore, further exploration of the possible molecular mechanisms are necessary to identify new targets and biomarkers for individualized treatment and prognosis prediction. In the present study, we identified TSPAN15, which was significantly overexpressed in PDAC tissues, via a proteomics study. TSPAN15 belongs to a family of six tetraspanins (TSPAN5, TSPAN10, TSPAN14, TSPAN15, TSPAN17 and TSPAN33) that have been identified as specific ADAM10-interacting protein partners[[164]29]. The TSPAN15/ADAM10 scissor complex is involved in regulating multiple physiological processes by cleaving a wide variety of membrane-anchored substrates, including Notch receptors, amyloid precursor protein, cadherins and growth factors[[165]30]. Previous studies have reported that TSPAN15 interacts with beta-transducin repeat containing E3 ubiquitin protein ligase (BTRC) to promote esophageal squamous cell carcinoma metastasis by activating NF-κB signaling[[166]24]. Barbaud et al. reported that TSPAN15 regulates ADAM10-dependent N-cadherin cleavage in invasive bladder cancer cells and promotes cancer progression and metastasis[[167]31]. However, the roles and detailed mechanism of TSPAN15 in PDAC progression remain unclear. In this study, we identified TSPAN15, which is significantly overexpressed in PDAC tissues and is correlated with tumor progression and poor prognosis. Although the expression level of TSPAN15 had a negligible effect on tumor cell proliferation, we unexpectedly found that the knockdown of TSPAN15 markedly inhibited the stemness of cancer cells and increased their sensitivity to gemcitabine, whereas the overexpression of TSPAN15 had the opposite effect. In addition, Yoshizumi et al. also reported that TSPAN15 could enhance cancer stem cell-like properties and induce chemoresistance by promoting ADAM10 activation and activating the Notch1 signaling pathway in intrahepatic cholangiocarcinoma[[168]23], which, when combined with our results, suggests that TSPAN15 is a potential target for improving the chemoresistance of cancer cells. The causes of gemcitabine resistance are complex and may be related to various physiological processes. Previous investigations have revealed that numerous mechanisms and pathways can contribute to the development of gemcitabine resistance, including the lack of nucleoside transporters (NTs), the dysfunction of certain nucleoside enzymes, epithelial–mesenchymal transition (EMT), and tumor microenvironment interactions[[169]4,[170]32]. The cutting-edge solution for overcoming gemcitabine chemoresistance is to target primary resistance by enhancing its transport and phosphorylation[[171]33]. In recent years, an increasing number of studies have reported that inducing ferroptosis in cancer cells could increase the efficacy of cancer chemotherapy[[172]13]. Song et al. reported that the ubiquitin protein ligase E3 component n-recognin 5 (UBR5) mediates colorectal cancer chemoresistance by attenuating ferroptosis via Lys 11 ubiquitin-dependent stabilization of Smad3-SLC7A11 signaling. Targeted inhibition of UBR5 in combination with a ferroptosis inducer synergistically sensitizes colorectal cancer to oxaliplatin-induced cell death and controls tumor growth[[173]34]. In hepatocellular carcinoma, Chen et al. verified that the inhibition of phosphoseryl-tRNA kinase (PSTK) could increase HCC cell sensitivity to chemotherapeutic treatment by inducing ferroptosis in cancer cells[[174]35]. We found that TSPAN15 regulated the expression of GPX4 in PDAC cells. As a key regulator of ferroptosis, GPX4 plays a crucial regulatory role in the process of ferroptosis, with its unique function of converting lipid hydroperoxides into nontoxic lipid alcohols, thereby inhibiting lipid peroxidation[[175]36]. Recent studies have demonstrated that targeting GPX4 is a potential strategy to overcome tumor resistance to immunotherapy and chemotherapy in breast and gastric cancer[[176]37,[177]38]. Additionally, in nasopharyngeal carcinoma, blockade of GPX4 could significantly increase the chemosensitivity of tumor cells[[178]39]. Our study demonstrated, for the first time, that TSPAN15 is a key regulator of ferroptosis in PDAC cells by regulating the expression of GPX4 and influencing the accumulation of ROS. Targeting TSPAN15 can significantly inhibit the expression of GPX4 and enhance gemcitabine-induced ferroptosis in pancreatic cancer cells. With respect to the molecular mechanism of TSPAN15, previous studies have focused primarily on TSPAN15 acting as an ADAM10 partner to assist ADAM10 in proteolytic cleavage of the extracellular region of many surface molecules[[179]22]. In our study, we identified a novel role of TSPAN15 in PDAC cells; specifically, TSPAN15 can directly interact with ITGB1 and sustain its stability. This mode of action dramatically extends the functional repertoire of TSPAN15 in mammalian cells. ITGB1 is a family of integrins, and a series of studies have revealed its role in maintaining the stemness of tumor cells and promoting tumor metastasis and chemotherapy/radiation resistance by participating in the transduction of various intracellular signaling pathways[[180]40,[181]41]. A previous study reported that β-1,4-N-acetyl-galactosaminyltransferase 1 (B4GALN) can directly interact with ITGB1 to inhibit its ubiquitin-independent proteasomal degradation, resulting in the activation of FAK/AKT pathways and promoting hepatocellular carcinoma stemness and progression[[182]42]. In head and neck squamous cell carcinoma (HNSCC), Zhao et al. reported that tensin 4 (TNS4) mediated the interaction between ITGB1 and ITGB5 which triggered FAK/Akt signaling pathway activation and promoted tumorigenesis[[183]43]. These findings suggest that stable expression of ITGB1 is critical for mediating the activation of the FAK/Akt signaling pathway and promoting the tumor malignant characteristics of tumors. Our results also verify that TSPAN15 is a pivotal interaction partner of ITGB1 and maintains its stability by mediating the deubiquitination of ITGB1, which further activates the p-FAK/p-AKT/p-mTOR-GPX4 signaling axis and promotes resistance to gemcitabine-induced ferroptosis in pancreatic cancer cells. However, several ubiquitination-associated proteins, such as CUL5, SOCS3, and USP9x, have been identified as regulators of ITGB1 ubiquitination and stability in cancer cells[[184]44,[185]45]. The detailed mechanism by which TSPAN15 regulates the ubiquitination or deubiquitination of ITGB1 by interacting with it remains an open question that warrants further exploration. Moreover, given our results of molecular docking, we discovered multiple potential direct interaction sites between TSPAN15 and ITGB1. Most importantly, we identified an inhibitor, venetoclax, which targeted the interaction sites between TSPAN15 and ITGB1. Venetoclax treatment significantly promoted gemcitabine-induced ferroptosis and sensitized PDAC cells to gemcitabine therapy by inhibiting the p-FAK/p-AKT/p-mTOR-GPX4 signaling axis. Venetoclax, a small molecule inhibitor of Bcl-2, has been approved by the US Food and Drug Administration for the treatment of naive elderly patients with acute myeloid leukemia (AML) who are unfit for conventional intensive chemotherapy. However, few reports on its therapeutic effect on solid tumors exist[[186]46,[187]47]. In this study, we first identified venetoclax as a potential inhibitor of TSPAN15 and blocked the interaction of TSPAN15 with ITGB1, thereby inhibiting the p-FAK/p-AKT/p-mTOR-GPX4 signaling pathway and ultimately increasing the sensitivity of PDAC cells to gemcitabine by inducing ferroptosis. For the first time, our study introduces venetoclax into the field of targeted therapy for solid tumors, which may expand the potential applications of venetoclax in cancer treatment pending future clinical trials. In conclusion, our study identified a novel regulatory role of TSPAN15 as a specific binding partner of ITGB1, sustaining its stability and regulating gemcitabine-induced ferroptosis by activating the p-FAK/p-AKT/p-mTOR-GPX4 signaling axis. Furthermore, targeting TSPAN15 with its potential inhibitor venetoclax enhanced the effectiveness of gemcitabine-based chemotherapy against PDAC. As a novel connector between chemoresistance and ferroptosis in PDAC, TSPAN15 may hold the potential to identify patient populations that may benefit from the pharmacological induction of ferroptosis. TSPAN15-targeted strategies may have broad therapeutic applications for developing effective anticancer treatments. 4. Methods 4.1. Clinical samples Tissue microarrays comprising 88 PDAC tumors and 82 adjacent pancreatic tissues were acquired from Outdo Biotech (HPanA170Su05, Shanghai, China). Six Fresh tissue samples including PDAC and paired adjacent pancreatic tissues were obtained from patients who underwent primary pancreatectomy. This study was approved by the Institutional Review Board of the Ethics Committee of the Zhejiang Provincial People's Hospital (No. QT2024278) and was conducted in accordance with the guidelines of the Declaration of Helsinki. 4.2. Cells and cell culture The human pancreatic cancer cell lines PANC-1, MIAPaCa-2, SW1990, ASPC-1, BXPC-3 and the human embryonic kidney cell line HEK-293T were acquired from the American Type Culture Collection (ATCC, Rockville, MD, USA). Among these, PANC-1, MIAPaCa-2, and SW1990 cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 % fetal bovine serum (BC-SE-FBS01C, Biochannel, Nanjing, China), 100 U/ml penicillin and 100 μg/ml streptomycin at 37 °C with 5 % CO[2]. The ASPC-1, BXPC-3 and HEK-293T cells were cultured in Rosewell Park Memorial Institute (RPMI)-1640 medium, also supplemented with 10 % fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin under the same conditions. These cell lines were authenticated through short tandem repeat analysis, and tested negative for mycoplasma contamination. 4.3. Cell transfection and vector construction To achieve stable overexpression of TSPAN15 in MIAPaCa-2 cells, full-length human TSPAN15 gene was subcloned into lentivirus vector pLent-EF1a–FH–CMV-GFP-Puro (GenePharma, Shanghai, China) to package pLent-TSPAN15-lentiviral particles, and the emptied vector pLent-empty was used as controls. MIAPaCa-2 was transfected with pLent-TSPAN15 or pLent-empty lentiviral particles and selected with puromycin for 2 weeks. The MIAPaCa-2 with stable overexpression of TSPAN15 and emptied vector were named MIAPaCa-2-TSPAN15 and MIAPaCa-2-vector, respectively. To generate TSPAN15 stable knockdown PANC-1 cell lines, we utilized the pLenti-CMV-Puro vector (GenePharma, Shanghai, China) to insert control or TSPAN15-targeting shRNA templates. Lentiviral particles expressing these shRNAs were used to infect PANC-1 cells, and selected with puromycin for 2 weeks. The PANC-1 with stable knockdown of TSPAN15 and control were named PANC-1-shTSPAN15-1, PANC-1-shTSPAN15-2 and PANC-1-shNC. The knockdown and overexpression effects of TSPAN15 in PANC-1 and MIAPaCa-2 cells were validated by Western blot analysis after treated with puromycin for 2 weeks. 4.4. Cell viability and colony formation assay For the cell viability assay, 8000 cells/well were seeded into 96-well plate for 24 h and then treated with the indicated chemicals for 48 h. Cell viability was measured using the CCK-8 assay with the Cell Counting Kit-8 (MA0218, Meilunbio, Dalian, China), following the manufacturer's instructions. For the colony formation assay, cells were seeded at a density of 1000 cells/well in 12-well plates for approximately 10 days. The indicated chemicals were added at a specific concentration on the third day. Colonies were counted after fixing the cells with methanol followed by staining with 0.5 % crystal violet. All experiments were conducted in triplicate and the results are presented as the mean ± SD. 4.5. Quantitative real-time PCR (qRT-PCR) Total RNA was extracted using TRIzol reagent (Invitrogen,15596026), and cDNA was amplified with a HiFiScript cDNA Synthesis Kit (CW2569 M, Vazyme Biotech Company, China). Real-time PCR was performed in an ABI Quant Studio 5 device (Applied Biosystems, Foster City, CA, USA) with SYBR Green Real-Time PCR Master Mix (A46109, Takara, Dalian, China). The relative mRNA expression levels were determined by the cycle threshold (Ct) normalized against β-actin levels using the 2^−ΔΔCt formula. The primer sequences are provided in [188]Supplementary Table S1. 4.6. Tumor sphere formation assay 200 single cells per well were resuspended in serum-free DMEM/F12 (D0697, Merck, Germany) containing 1 × B27 supplement (12587010, Gibco, USA), 20 ng/mL epidermal growth factor, and 10 ng/ml basic fibroblast growth factor (13256-029, Gibco, USA), and subsequently seeded into 96-well ultralow attachment dishes (3747, Corning, NY, USA). Every three days, the number of spheres was counted and sphere images were collected under the microscope. 4.7. Flow cytometry Cells in logarithmic growth were cultured in 6-well plate (2 × 10^5 cells/well) overnight. The indicated chemicals at a specific concentration were added, and the cells were treated for 24 h. Then, cells were collected and the cell suspension was adjusted to 1 × 10^6 cell/ml. Cells were stained with Annexin-V-APC/7-AAD (AT105, Liankebio, Hangzhou, China) as the manufacturer's instructions. Stained samples were analyzed by flow cytometry (Agilent NovoCyte, Santa Clara, CA, USA). 4.8. Reactive oxygen species (ROS) assay 2 × 10^5 cells/well cultured in 6-well plate were treated with the indicated chemicals at a specific concentration for 24 h. Intracellular reactive oxygen species (ROS) levels were measured by monitoring the oxidation of cell-permeable 20, 70-dichlorofluorescein diacetate (DCFH-DA, Beyotime, Shanghai, China), using flow cytometry at excitation and emission wavelengths of 488 nm and 525 nm by flow cytometry. For ROS fluorescence, cell cultures were treated with 10 μmol/ml DCFDA (ab113851, Abcam, Wuhan, China) for 20 min to detect ROS through confocal fluorescence microscopy. 4.9. Transmission electron microscopy (TEM) imaging TEM was conducted for the direct observation of cell morphology during ferroptosis. After treated with indicated chemicals at a specific concentration, cells were digested with 0.05 % trypsin, collected in a 1.5 ml microcentrifuge tube and immediately centrifuged to remove the trypsin. Then, human serum was added and mixed before resuspension. The cell mass was fixed with 2.5 % glutaraldehyde diluted in phosphate buffer, stored overnight at 4 °C, and stained with 1 % OsO4 for 2 h. The samples were dehydrated using increasing concentrations of alcohol (30 %, 50 %, 70 %, 90 %, 95 %, and 100 %). After embedding in epoxy resin, the samples were sectioned into 60–90 nm slices and stained with lead citrate. Images were examined using a transmission electron microscope (JEM-1400FLASH, JEOL). 4.10. Immunofluorescence staining Cells growing on coverslips in 6-well plates were fixed with 4 % (w/v) paraformaldehyde for 30 min, and permeabilized with 0.2 % Triton X-100 for 10 min. The cells were blocked with 2 % (w/v) bull serum albumin (BSA) for 1 h, and incubated overnight at 4 °C with the anti-TSPAN15 (Orb1755, Biorbyt, Cambridgeshire, UK) and anti-ITGB1 (16396-1-AP, Proteintech, Wuhan, China) primary antibody. After washing with PBS three times, the cells were incubated with the corresponding Alexa Fluor 488 (A-11008, Invitrogen, Waltham, USA) or 594 (A-11005, Invitrogen, Waltham, USA) conjugated secondary antibodies at room temperature for 1 h. Finally, DAPI (4′,6-diamidino-2phenylindole) was utilized to stain the nuclei (100 ng/ml, 10 min in room temperature). The samples were then scanned using a confocal microscope (Zeiss 710 confocal microscope). 4.11. Immunohistochemistry (IHC) The xenograft tumor tissues and human PDAC tissues were preserved in 4 % PFA, followed by paraffin embedding. The paraffin-embedded tissues were cut into 5 μm thick slices. The paraffin-embedded tissue sections were deparaffinized in a series of gradient ethanol baths, rehydrated, and treated with methanol containing 0.3 % hydrogen peroxide for 10 min to block endogenous peroxidase at room temperature. Subsequently, the tissue slides were heated for 30 min in a pH 6.0 antigen retrieval solution to induce antigen retrieval and then incubated overnight with indicated primary antibody at 4 °C overnight. Tissue staining was performed using a Prolink-2 Plus HRP rabbit polymer detection kit (Golden Bridge, Bothell, WA, USA) according to the manufacturer's instructions. Images were captured using Aperio ScanScope CS software (Aperio Technologies, Vista, CA, USA). The results were evaluated based on the intensity and extent of staining by two independent observers who were blinded to the results of the other makers and clinical outcomes. Briefly, the positive staining areas were scored as follows: the staining percentage of positively stained areas over total tissue area were defined as 0 = ≤ 5 %; 1 = 5 %–25 %; 2 = 26 %–50 %; 3 = 51–75 %; and 4 = ≥ 75 %. The intensity was graded as follows: 0, negative; 1 +, weak (yellow); 2 +, moderate (light brown); and 3 +, strong (dark brown). A staining index was used in which a score of 0–6 was considered low expression, and ≥6 was considered high expression. The representative images of the scoring standards for TSPAN15 immunohistochemical staining area and staining intensity are presented in [189]Fig. S7. 4.12. Western blot analysis Cells in logarithmic growth were treated with the indicated chemicals at a specific concentration. Total protein extraction and quantification were performed using RIPA buffer (P0037, Beyotime Biotechnology, China), which contained a protease inhibitor cocktail (S8830, Roche, USA), and a BCA Protein Assay Kit (ZJ102, Epizyme, China). Subsequently, 50 μg of total proteins were separated on 10 % SDS-PAGE gels and transferred to polyvinylidene fluoride (PVDF) membranes (IPVH00010, Millipore). The membranes were blocked with PBS containing 5 % non-fat dry milk for 1 h and then incubated overnight with the primary antibodies at 4 °C. All specific antibodies were listed in [190]Supplementary Table S1. On the next day, the membranes were incubated for 2 h with HRP-conjugated secondary antibodies at room temperature. Protein signals were visualized by using Super ECL Plus (AIWB-006, Affinibody, Wuhan, China) under a chemiluminescent and fluorescent imaging system (Tanon, Shanghai, China). 4.13. Co-immunoprecipitation Cells with 80 % density in 10 mm cell culture dishes were collected and lysed in 1 ml of Cell Lysis Buffer (P0030F, Beyotime Biotchnology, China) containing protease inhibitor cocktail (S8830, Roche, USA) for sonicating on ice, and the supernatant was collected and centrifuged at 10,000 g for 10 min at 4 °C. One-tenth volume of lysate was taken out as input, and the left part was mixed with 50 μl of Protein A/G magnetic beads ([191]B23202, Selleck) and gently rotated at 4 °C for 1 h to reduce nonspecific binding. Subsequently, the supernatant was incubated with indicated primary antibody overnight. The beads were washed three times with IP lysis buffer and boiled with 2 × loading buffer for 10 min. Beads were removed by centrifugation, and the precipitated samples were subjected to immunoblot analysis. Ubiquitination detecting, cells were treated with 10 μM MG132 for 6 h, lysed in NP40 buffer, and boiled at 95 °C for 5 min. Cell lysed for immunoprecipitation assays with 1 μg of anti-ITGB1 antibody followed by Western blot analysis to assess using ubiquitination. 4.14. Cellular thermal shift assay (CETSA) CETSA assesses the binding efficiency of venetoclax to the TSPAN15 protein in tumor cells. PANC-1 cells were treated with either venetoclax or DMSO (as the control group) and subsequently lysed using RIPA buffer resulting in the formation of six distinct groups. The protein extraction solution was heated at temperatures of 36 °C, 48.1 °C, 50.2 °C, 52.3 °C, 54.3 °C, and 56.2 °C for 20 min, respectively. Following this, repeated freezing and thawing cycles were conducted to extract the protein. Western blot analysis was then performed to evaluate the expression of TSPAN15 protein and the internal reference protein at each heating temperature. The heat melt curves for both the venetoclax treatment group and the control group were obtained by analyzing the changes in TSPAN15 expression in " venetoclax treated group/control group" against temperature. 4.15. MDA and GSH/GSSG assay GSH/GSSG assay: PANC-1 cells were seeded into 6-well plates at a density of 1 × 10^6 cells per well for 24 h prior to the experiment. Following treatment with specified chemicals at a predetermined concentration, the cells were collected, washed with PBS, and subjected to freeze-thaw breakage using liquid nitrogen and a 37 °C water bath. The total intracellular glutathione (GSH) and oxidized GSH levels were determined using a GSH assay kit (S0053, Beyotime, Shanghai, China) in accordance with the manufacturer's instructions. MDA Assay: PANC-1 and MIAPaCa-2 cells were seeded into 6-well plates at a density of 1 × 10^6 cells per well. Following treatment with specified chemicals at a predetermined concentrations, samples and standards were prepared, and the optical density (OD) was measured at 532 nm. The malondialdehyde (MDA) assay was conducted using a lipid peroxidation MDA assay kit (S0131, Beyotime, Shanghai, China) according to the manufacturer's instructions. MDA content was quantified, and the ratio of MDA levels to protein concentration was subsequently calculated. 4.16. In-vivo experiment and xenograft tumor models All animal experiments were conducted according to the ARRIVE guidelines and approved by the Animal Ethics Committee of Zhejiang Provincial People's Hospital (No. 20241220438528). 6-week-old male BALB/c-Nu mice were purchased from Vital River Laboratory Animal Tec Co. Ltd (Beijing, China) and maintained in an SPF environment. To detect the effect of TSPAN15 on the tumourigenicity of PDAC cells, A total of 1 × 10^6 or 1 × 10^4 PANC-1-shNC cells and PANC-1-shTSPAN15 cells were suspended in 100 μl of PBS separately and subcutaneously implanted into right armpit of nude mice (n = 4, each group). Forty days after the injection, the tumor incidence of each group was observed and the stem cell frequency was estimated using an online tool at [192]http://bioinf.wehi.edu.au/software/elda. All mice were euthanized and the xenografts were excised by dissection and weighed. To investigate the effect of TSPAN15 knockdown on the sensitivity of tumor cells to gemcitabine, PANC-1-shNC cells or PANC-1-shTSPAN15 cells (1 × 10^6 per mouse) were injected into the right axillae of nude mice (n = 12 per group), When the volume of the xenograft reached 100 mm^3, mice were randomly allocated into four groups: 1) PANC-1-shNC + DMSO (n = 6); 2) PANC-1-shTSPAN15 + DMSO (n = 6); 3) PANC-1-shNC + gemcitabine (n = 6); 4) PANC-1-shTSPAN15 + gemcitabine (n = 6). The mice in these groups were treated with 10 mg/kg gemcitabine or equal amounts of DMSO via intraperitoneal injection every three day for 20 days. The weights of the mice and the volumes of the xenograft tumors were calculated based on formula 0.5 × length × width^2 every 3 days following the injection. The tumors were collected and fixed in 4 % paraformaldehyde. For gemcitabine + venetoclax combination therapy, PANC-1 cells (1 × 10^6 per mouse) were injected into the right axillae of nude mice. When the volume of xenograft reached 100 mm^3, mice were randomly allocated into four groups (n = 10 per group): 1) PANC-1 + DMSO; 2) PANC-1 + gemcitabine; 3) PANC-1 + venetoclax; 4) PANC-1 + gemcitabine + venetoclax. Mice were treated with gemcitabine (10 mg/kg intraperitoneally, once every three day) with or without venetoclax (50 mg/kg orally, once every day) for 20 days. The weights of the mice and volumes of the xenograft tumors were calculated based on formula 0.5 × length × width^2 every 3 days following the injection. The tumors were collected and fixed with 4 % paraformaldehyde. 4.17. RNA sequencing PANC-1-shNC cells and PANC-1-shTSPAN15 cells were treated with or without 2.5 μM gemcitabine for 48 h and total RNA was extracted using TRIzol reagent (Invitrogen). Each sample was prepared in triplicate and RNA sequencing and analyses were performed, and sequenced on Illumina Novaseq 6000 sequencing instrument (BGI, China). The clean reads were aligned to the reference genome using HISAT and to reference gene sequences using Bowtie2. Gene expression data were assessed using RSEM software. Differentially expressed genes between PANC-1-shNC cells and PANC-1-shTSPAN15 cells, as well as between PANC-1-shNC cells and PANC-1-shTSPAN15 cells treated with gemcitabine were identified based on the criteria of a log[2] (T/N fold change) ≥ 1 or ≤ −1 and p < 0.05. 4.18. High-throughput virtual screen for potential TSPAN15 inhibitors For the docking study, the structures of TSPAN15 and ITGB1 were downloaded from the Protein Databank ([193]http://www.rcsb.org), and amino acid sequence were derived from UniProt. The protein structure of TSPAN15 was optimized to achieve a reasonable side chain configuration. The compounds were sourced from Approved Drug Library (TargetMol, L1000) and virtual screening workflow was employed utilizing Glide SP (standard precision mode) followed by MMGBSA (Molecular Mechanics Generalized Born Surface Area) for screening. The protein structure was treated as rigid, while small molecules were treated as flexible, with all other parameters set to their default values. The docking results underwent energy optimization, retaining the top 50 % based on docking scoring for subsequent screening. Poses with Glide score of less than −5 kcal/mol and an MMGBSA ΔG less than −40 kcal/mol were selected. Structure-based clustering was performed using MOE (Version 2022.02), where molecular fingerprints were calculated using BIT-MACCS (MACCS Structural Keys (Bit packed)) model, and structural similarity was assessed using Tanimoto SS (Tanimoto Superset/Subset). By visualizing the binding conformations of the molecules with TSPAN15, the molecules that interacted with the protein binding interface were identified. 4.19. Bioinformatics analysis The 24 single cell RNA-seq data of PDAC tissues were obtained from published data[[194]28]. The Seurat (v2.3.0) R toolkit was used to filter the preliminary data with mitochondrial genes expressing >200 genes/cell, each gene expressed in at least 3 cells and >10 % expressed in cells as the screening criteria. Further, the data was standardized using the NormalizeData function (Scale.factor = 10000), and the ScaleData function was used to normalize the cells. The data was then dimensionally reduced through principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE), followed by clustering of the cell populations. Marker genes for each cell population were identified using the FindMarkers function, and all populations were annotated according to the previously reported marker genes for various cell types. The annotation of genes for each cell type is as follows: Ductal cell 2: KRT19, KRT7, TSPAN8, SLPI; Ductal cell 1: AMBP, CFTR, MMP7; Acinar: PRSS1, CTRB1, CTRB2, REG1B; Endocrine cell: CHGB, CHGA, INS, IAPP; Fibroblast: LUM, DCN, ACTA2, PDGFRB, COL1A1; Endothelial cell: CDH5, PLVAP, VWF, CLDN5; Macrophage: AIF1, CD64, CD14, CD68; T cell: CD3D, CD3E, CD4, CD8; B cell: MS4A1, CD79A, CD79B, CD52. 4.20. Statistical analysis Statistical analysis was performed using GraphPad Prism 9 (GraphPad, LaJolla, USA). All measurement data were presented as the mean ± standard deviation (mean ± SD). Differences between two groups were determined by performing an unpaired two tailed Student's t-test or by conducting one-way or two-way ANOVA for multiple comparisons when comparing three or more groups. Chi-square test was used to compare qualitative data. The correlation between two groups was examined using Pearson's correlation. Survival analyses were performed using the Kaplan–Meier method and compared using the log-rank test. P < 0.05 was considered to be statistically significant (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001). Detailed protein lists, specific antibodies and other chemicals and reagents were listed in [195]Supplementary Table S1. [196]Supplementary Table S2 all datasets generated were available from the corresponding author upon request. CRediT authorship contribution statement Peng Nan: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Xiao Wang: Software, Project administration, Methodology, Formal analysis, Data curation. Anan Li: Validation, Software. Yumei Ge: Supervision, Project administration, Conceptualization. Zongting Gu: Visualization, Methodology, Formal analysis, Data curation. Yingying Wang: Writing – original draft, Visualization, Supervision, Data curation. Ran Tao: Writing – review & editing, Validation, Supervision, Project administration, Funding acquisition, Conceptualization. Ethics approval This study was approved by the Institutional Review Board of the Ethics Committee of the Zhejiang Provincial People's Hospital (No. QT2024278) and performed in accordance with the guidelines of the Declaration of Helsinki. Funding This work was supported by grants from the Zhejiang Provincial Major Science and Technology Program for Medical Science of China (WKJ-ZJ-1901 to R Tao), National Natural Science Foundation of China (82403401 to P Nan), Zhejiang Provincial Natural Science Foundation of China under grant (ZCLQ24H1602 to P Nan) and the Zhejiang Provincial Health Science and Technology Project of China (2024KY763 to P Nan). Declaration of competing interest The authors of this manuscript have no conflict of interest. Acknowledgments