研究背景

卵巢上皮癌（Epithelial ovarian cancer，EOC）是死亡率最高的女性生殖系统恶性肿瘤，其中，高级别浆液性癌（High-grade serous ovarian cancer，HGSOC）约占卵巢上皮癌的60-70%，是最常见的病理组织学类型^[1-4]。临床卵巢癌的患者在诊断时约有70%的已经进入临床晚期(III和IV期)^{[5, 6]}；约70%的晚期患者在经过初次手术和化疗后疾病在3年内复发^[7]；随着复发次数的增加，所有患者都会产生化疗耐药性，导致疾病不断进展，晚期卵巢癌的5年生存率仅为29%。

有研究指出西方国家的死亡率正在小幅度下降^{[8, 9]}，但大多数学者均认为在过去20年中卵巢癌的生存率并没有显著增加^{[3, 10, 11]}。并且，这种降低并非由于出现更有效的治疗措施，而与生活方式改变相关，包括避孕药的使用和绝经后激素替代疗法的减少等^[9]。肿瘤细胞对于化疗药物产生耐药性是导致卵巢癌患者生存期短、预后差的主要原因。探索和发现卵巢高级别浆液性癌的耐药机制，寻找有效的治疗药物对于改善疾病预后具有重要意义。

肿瘤干细胞（Cancer stem cells，CSCs）是一组与正常干细胞相似的具有干细胞特性的癌细胞亚群，存在于大多数组织学类型的原发恶性肿瘤中，具备自我更新、分化、以及增殖的潜能，并且进一步参与所有肿瘤的进展和肿瘤细胞的恶性行为，包括侵袭、转移、复发、耐药等^[12]。CSCs被证实存在于包括上皮性卵巢癌在内的各种实体肿瘤中，如脑、肝、肺、结肠、乳腺、卵巢、胰腺、前列腺、黑色素瘤、头颈部和膀胱等^[13-15]。从理论上讲，CSCs多数处于细胞周期的静止期，因此对于化疗药物存在先天的抵抗能力，可以在肿瘤组织明显缩小的情况下存活，即使是单个CSC也能够重建整个肿瘤。因此，CSCs理论符合卵巢癌本身的生物学特征，并可解释HGSOC患者对化疗耐药的原因^[16-20]。

目前已证实，CSCs在包括脑、头颈部、肺、乳腺、肝、胃、胰腺、卵巢、淋巴瘤、结直肠、前列腺和宫颈癌等多种肿瘤类型的治疗中对化疗显现出抵抗力^[21]。CSCs的耐药机制尚不完全清楚，目前研究结果包括：（1）CSCs的DNA损伤修复相关通路可高表达，以减少细胞死亡^[16]；（2）过表达BCL2抗凋亡蛋白家族可保护CSCs免于凋亡，并增强增殖能力^{[22, 23]}；（3）CSCs在细胞膜高表达药物外排泵，如ABC转运蛋白ABCB1和ABCG2、MDR蛋白，促进化疗药物排出，使得细胞内药物浓度无法达到作用阈值^[24]；（4）肿瘤微环境可帮助CSCs对化疗产生耐药，如成纤维细胞分泌的肿瘤相关活性物质、CSCs周围维持低水平的活性氧状态都有助于CSCs在化疗过程中存活^[24-26]。有关针对卵巢癌CSCs的相关研究还鲜有报道，探讨卵巢癌CSCs的耐药机制以及干预措施是改善HGSOC患者不良预后的方向之一。

本课题组前期已从HGSOC患者的SP细胞（肿瘤干细胞）中通过微量蛋白组学技术筛选到特征性跨膜蛋白30A（Transmembrane protein 30A，缩写为TMEM30A或CDC50A）。体内和体外实验发现，该分子阳性的原代细胞的增殖、分化、转移、以及耐药等能力显著增强，证实了CDC50A是HGSOC肿瘤干细胞的全新表面标志物，并且与HGSOC患者的预后（PFS和OS）相关。进一步研究发现，高表达和沉默CDC50A可显著影响卵巢癌细胞系的增殖和耐药表型，提示它不仅仅是一个表面标记蛋白，更是一个功能性蛋白，或可调节卵巢癌CSCs的耐药功能。进一步开展的高通量转录组测序，试图对其调控具体机制进行深入探讨。

本研究在此基础上，基于卵巢癌细胞系，验证了高通量转录组测序的结果，证实CDC50A可影响多条下游耐药通路。通过免疫共沉淀（IP）及质谱技术，寻找CDC50A相互结合的蛋白，选择与测序筛选出的NOTCH以及Hedgehog耐药通路唯一相关的分子Numb进行进一步研究，并通过co-IP及泛素化等试验探究CDC50A是如何与Numb蛋白相互作用的。在CDC50A、Numb过表达细胞系中明确CDC50A-Numb下游的入核转录调控因子为Gli 1，进一步通过定量PCR筛选Gli 1可影响哪些耐药通路，并通过体内及体外实验验证CDC50A-Numb-Gli 1可导致卵巢癌细胞对顺铂以及PARPi耐药，而Gli 1的抑制剂治疗有效。最后利用TCGA数据库数据明确CDC50A高表达是影响患者预后的因素，并在患者中验证CDC50A-Numb-Gli 1是否可导致顺铂及PARPi耐药。

研究方法

分析患者原代细胞RNA-Seq数据，探究CDC50A与耐药相关通路的关系。对患者原代CDC50A+及CDC50A-细胞高通量转录组测序数据（RNA Sequencing,RNA-Seq）重新进行差异基因富集分析。流式细胞术分选浆液性卵巢癌细胞系SKOV3，OVCAR3以及CaOV4中CDC50A+以及CDC50A-的细胞，并利用RT-qPCR技术验证RNA-Seq结果。构建CDC50A过表达质粒，同时包装shCDC50A沉默慢病毒并进行SKOV3+shCDC50A及OVCAR3+shCDC50A稳转株筛选。过表达及沉默效果均经过RT-qPCR和蛋白免疫印迹技术验证。SKOV3及OVCAR3细胞系过表达及沉默CDC50A后经过RT-qPCR再次验证RNA-Seq结果。

筛选膜蛋白CDC50A导致卵巢癌顺铂耐药的下游蛋白并探究其相互作用机制。SKOV3细胞内过表达CDC50A分子，利用IP技术将CDC50A蛋白及其结合的蛋白免疫共沉淀下来并通过质谱技术对蛋白胶条进行检测，筛选出可能与CDC50A相互作用、并且与上一步骤中NOTCH及Hedgehog耐药通路唯一相关的分子Numb。通过IP，co-IP及细胞爬片免疫荧光技术验证CDC50A是否与Numb分子相互结合。构建Numb分子PTB结构域过表达质粒和PPR结构域过表达质粒，co-IP技术明确CDC50A与Numb相结合的结构域组分。SKOV3及OVCAR3细胞过表达及沉默CDC50A后，观察CDC50A对内源性Numb蛋白表达水平的影响，并利用细胞泛素化及IP试验探究CDC50A是否可调节Numb蛋白泛素化降解。CCK8试验在细胞水平明确Numb对耐药表型的影响。鉴于CDC50A并非泛素化相关分子，co-IP及泛素化等试验验证泛素化酶DDB1与CDC50A及Numb分子间的作用关系，并通过CCK8试验明确DDB1分子对耐药表型的影响和CDC50A与DDB1是否有协同耐药作用。

明确CDC50A-Numb的下游入核转录因子，探究其影响哪些耐药分子的表达水平并验证其抑制剂是否可治疗CDC50A高表达所致的顺铂及PARPi耐药。Gli 1是Hedgehog通路中关键转录因子，有研究提示Numb蛋白可影响Gli 1分子的泛素化降解，因而本章内容围绕Gli 1蛋白进行探究。在SKOV3及OVCAR3细胞系中过表达Numb，CDC50A或DDB1，通过蛋白免疫印迹法研究Gli 1表达水平的变化。通过co-IP试验探究Numb与Gli 1蛋白间的相互作用，进而利用细胞泛素化试验研究Numb、CDC50A和DDB1与Gli 1泛素化降解之间的关系。CCK8试验验证Gli 1是否是CDC50A顺铂耐药的下游入核转录因子。RT-qPCR筛选CDC50A和Gli 1相关的耐药分子，并通过蛋白免疫印迹法在CDC50A过表达及沉默细胞系中进行验证。PDX试验探究CDC50A-Numb-Gli 1-BRCA1是否可与卵巢癌奥拉帕利耐药相关，并且通过CCK8试验验证Gli 1抑制剂是否可以治疗CDC50A过表达所致的顺铂及奥拉帕利耐药。

探讨CDC50A与卵巢癌转移及预后的关系，并在TCGA数据库验证上述耐药机制。基于SKOV3及OVCAR3细胞系，通过RT-qRCR及蛋白免疫印迹法验证本课题组先前的研究结论——CDC50A可影响Bmi-1，β-catenin，APC，E-Cadherin，Vimentin，TGF-β1，NOTCH1以及Oct4的表达水平。进一步通过细胞爬片免疫荧光技术显示过表达CDC50A细胞中转移关键分子E-Cadherin，Vimentin的表达情况。对比TCGA数据库中374例卵巢浆液性肿瘤患者和GDC数据库中180位无肿瘤对照患者的RNA测序数据，探究CDC50A表达水平与卵巢癌的关系。利用374例卵巢癌患者测序数据及临床数据，探究高表达CDC50A是否可影响患者预后。最后利用上述TGCA数据库数据及癌症药物敏感性基因组学(GDSC）验证CDC50A-Gli 1通路可导致卵巢癌患者对顺铂及PARPi耐药。

研究结果

验证RNA-Seq结果，证实CDC50A可影响多条下游耐药通路。重新分析患者原代CDC50A+及CDC50A-细胞RNA-Seq数据，发现CDC50A+细胞内Wnt、NOTCH等肿瘤发生信号通路，肿瘤增殖、转移相关通路以及糖蛋白代谢等通路显著富集增高，这些信号通路均与肿瘤耐药相关。此外，肿瘤干细胞关键通路Hedgehog通路的关键分子在CDC50A+的细胞内也显著富集。流式细胞术分选三株卵巢癌细胞系的CDC50A+细胞比例均较少，且有差异，SKOV3细胞系CDC50A⁺细胞比例最少（平均0.2%），OVCAR3细胞系稍多（平均1.2%），CaOV3细胞系最多（平均7.2%）。RT-qPCR的整体趋势基本符合RNA-seq结果，即在CDC50A⁺的细胞中，Wnt，NOTCH及Hedgehog三条通路被激活且与干细胞相关的标志物、细胞膜药物外排泵及抗凋亡等耐药相关分子过表达。有显著变化的分子在NOTCH通路以及Wnt通路中较多，而在Hedgehog通路中偏少且Gli 1分子总是存在显著差异，提示其可能是CDC50A作用的下游分子之一。成功构建CDC50A过表达质粒和shCDC50A沉默慢病毒后，在过表达及沉默CDC50A的细胞系通过RT-qPCR再次验证了RNA-Seq以及流式验证试验的结果，且Gli 1的变化趋势再次显示出它可作为潜在的下游调节分子。

CDC50A与DDB1协同作用，促进Numb分子泛素化降解，从而降低其蛋白水平，导致卵巢癌顺铂耐药。IP及质谱共发现413个蛋白同时存在于三个重复的实验组，其中仅46个蛋白在实验组特异性与CDC50A结合。GO富集分析发现它们主要集中在蛋白结合，细胞间识别与连接，膜筏，转录，DNA损伤修复，以及离子转运相关通路中，其中蛋白功能分析显示仅Numb分子与上一步骤中NOTCH及Hedgehog通路密切相关。IP，co-IP及细胞爬片免疫荧光均证实CDC50A可与Numb分子相互结合。在验证Numb分子PTB结构域过表达质粒和PPR结构域过表达质粒的过表达效果后，co-IP证实CDC50A与Numb分子PTB结构域相互结合而不与PPR结构域结合。外源性过表达CDC50A后，细胞内源性Numb水平显著下调，而沉默CDC50A后，内源性Numb水平显著上升，进一步泛素化试验明确CDC50A可通过增加Numb泛素化降解，从而降低后者蛋白水平。CCK8试验显示Numb过表达可逆转CDC50A导致的卵巢癌细胞顺铂耐药。co-IP证实泛素化酶DDB1可与CDC50A和Numb分子相互结合，进一步的泛素化试验明确DDB1可增加Numb泛素化降解且CDC50A与DDB1存在协同作用。CCK8试验中，DDB1与卵巢癌细胞顺铂耐药相关并可被Numb过表达逆转耐药表型。此外，DDB1与CDC50A有协同作用，共同促进卵巢癌细胞顺铂耐药。

CDC50A通过降低Numb蛋白水平，抑制后者对Gli 1的泛素化降解作用，升高细胞内Gli 1蛋白水平，增加耐药相关分子的转录，导致卵巢癌对顺铂及PARPi耐药。过表达Numb降低细胞内Gli 1蛋白的水平，而过表达CDC50A或DDB1分子显著升高细胞内Gli 1蛋白水平。co-IP试验证实Numb可与Gli 1分子相互结合。泛素化试验证实Numb可促进Gli 1蛋白泛素化降解，而CDC50A和DDB1则可下调Gli 1蛋白泛素化降解。NOTCH通路，ABC药泵以及DNA损伤修复相关通路中的BRCA1，ABCC2，ABCC5，APC，XPC，NOTCH1，NOTCH2，DLL1，以及ABCG2是CDC50A-Numb-Gli 1调节的下游耐药分子。CCK8试验证实CDC50A可导致卵巢癌细胞对顺铂及奥拉帕利耐药，且这种耐药表型可被Gli 1抑制剂缓解。PDX药敏试验证实高表达CDC50A, GLi 1以及BRCA1分子的组织对尼拉帕利耐药，且相比于用药前，用药后的组织中CDC50A, GLi 1及BRCA1显著富集，而Numb水平下调。

CDC50A与卵巢癌转移相关并显著影响患者预后，在TCGA数据库中CDC50A-Gli1耐药通路得到验证。CDC50A细胞可显著升高Bmi-1，β-catenin，APC，Vimentin，TGF-β1，NOTCH1以及Oct4的表达，降低E-Cadherin的表达，促进肿瘤细胞上皮间质转化和转移。肿瘤患者中CDC50A表达水平显著高于健康对照人群（P=0.0045），且CDC50A高表达组的肿瘤患者预后要显著差于低表达组（HR 1.885, 95%CI 1.226-2.899, P=0.00388）。在TCGA数据库中，CDC50A和DDB1与Numb表达水平呈负相关，与Gli 1表达水平正相关，且CDC50A和Gli 1与BRCA1表达水平呈正相关，间接证实了本研究中CDC50A与DDB1协同作用降低Numb水平，进而增加Gli 1及BRCA1蛋白表达的作用通路。IC50预测试验中，CDC50A高表达患者对顺铂及Olaparib治疗不敏感，Gli 1高表达患者对顺铂治疗不敏感且对Olaparib治疗耐药，间接佐证CDC50A可导致卵巢癌顺铂及PARPi耐药，而Gli 1抑制剂治疗可能有效的结论。

研究结论

CDC50A可以上调多条耐药通路的表达，包括NOTCH通路、Wnt通路、Hedgehog通路、细胞膜药物外排泵及抗细胞凋亡通路等，证实CDC50A是一个功能蛋白，可导致卵巢癌细胞耐药。

CDC50A通过与Numb分子PTB结构域相互结合，竞争性抑制Numb与下游分子结合发挥作用。CDC50A与DDB1具有协同作用，可促进Numb分子泛素化降解，降低Numb蛋白水平。此外，CDC50A及DDB1导致的卵巢癌细胞耐药表型可被Numb过表达逆转。因此，Numb是CDC50A耐药的下游分子。

Numb可与Gli 1分子相互结合，促进后者泛素化降解。CDC50A及DDB1通过降低Numb的水平以及竞争性抑制Numb与Gli 1分子结合，降低Gli 1泛素化降解并增加其蛋白水平。CDC50A-Numb-Gli 1通路可以增加耐药相关的DNA损伤修复通路，ABC药泵以及NOTCH通路的转录水平，导致卵巢癌对顺铂及PARPi耐药，而Gli 1抑制剂有治疗作用。

CDC50A高表达促进卵巢癌转移，并影响患者的生存。TCGA数据库浆液性卵巢癌患者的数据证实了CDC50A-Numb-Gli 1通路可导致顺铂及PARPi耐药的理论。

Background

Epithelial ovarian cancer (EOC) is the most lethal malignant tumor of the female reproductive system with the highest mortality, and the most common pathology is high-grade serous ovarian cancer (HGSOC), accounting for approximately 60-70% of ovarian cancer ^[1-4]. About 70% of patients with clinical ovarian cancer have entered the advanced stage (III and IV) at the time of diagnosis ^{[5, 6]}; about 70% of patients with advanced disease relapse within three years after primary surgery and chemotherapy ^[7]; with recurrences, all patients develop resistance to chemotherapy, leading to progression and a five-year survival rate of 29% for advanced ovarian cancer.

Although western countries reported a slight decline in mortality, most scholars agree that its survival rate has not increased significantly in the past 20 years ^{[3, 8-11]}. Moreover, lifestyle changes, such as the contraceptives use and postmenopausal hormone replacement therapy after doctor assessment and supervision, instead of EOC treatment improvement, may lead to mortality decline ^[9]. Given that chemotherapeutic resistance is the main reason for poor prognosis, it is crucial to explore and discover the resistance mechanism and find effective drugs.

Cancer stem cells (CSCs), a fraction of tumor cells with stem cell characteristics, exists in most types of cancers and have self-renewal, proliferation and differentiation potential, involved in malignant behavior, such as the invasion, metastasis, recurrence and drug resistance ^[12]. CSCs have been found in various solid tumors including EOC, e.g.brain, liver, lung, colon, breast, ovary, pancreas, prostate, melanoma, head and neck, and bladder ^[13-15]. Theoretically, CSCs own innate resistance to chemotherapy because they are in stationary phase of cell cycle, and can survive in a significant reduction of tumor tissues after chemotherapy ^[17-19]. And even a single CSC is able to rebuild the entire tumor on account of proliferation and differentiation ability ^{[16, 20]}. Therefore, CSCs accord with the biological features of EOC and would explain why the drug-resistant mechanism.

The chemotherapy-resistance of CSCs has been confirmed in the brain, head and neck, lung, breast, liver, stomach, pancreas, ovary, lymphoma, colorectal, prostate, and cervical cancer ^[21]. The mechanism remains unclear and current research suggested that: (1) DNA damage repair related pathways can be highly expressed in CSCs to reduce cell death ^[16]; (2) Overexpression of anti-apoptotic protein family-like BCL2 can protect CSCs from apoptosis and enhance proliferation ability ^{[22, 23]}; (3) highly express drug efflux pumps, such as ABC transporter and multidrug resistance (MDR) proteins, promote drugs excretion to limit the intracellular drug concentration to decrease effect ^[24]; (4) Tumor microenvironment (tumor-related active substances secreted by fibroblasts and low level of reactive oxygen) contribute to CSCs survival during chemotherapy ^[24-26]. Additionally, no there is no effective treatment for EOC-CSCs currently. As a result, exploring the drug resistance mechanism of EOC-CSCs and looking for its therapeutic agents are the directions to improve the poor prognosis of HGSOC patients.

Previously, our research group has found out Transmembrane protein 30A (TMEM30A or CDC50A) by screening SP cells (tumor stem cells) membrane proteins of HGSOC patients using microproteomics technology. We then identified CDC50A as a novel surface marker of HGSOC-CSCs when Vivo and in vitro experiments verified that CDC50A+ primary cells had significantly enhanced proliferation, differentiation, metastasis, and drug resistance characteristics, leading to poor prognosis (PFS and OS). Further studies revealed that CDC50A overexpression or silence significantly affected proliferation and drug resistance phenotype of EOC cell lines, suggesting that CDC50A is not only a surface marker protein but also a functional protein, which may be a critical molecular to regulate drug resistance. Despite high-throughput transcriptional sequencing (RNA sequcing, RNA-Seq), the regulatory mechanisms remain unclear.

In this study, we verified the results of RNA-Seq and confirmed that CDC50A could affect multiple downstream drug resistance pathways based on EOC cell lines. Besides, immunoprecipitation (IP) and mass spectrometry were used to find the CDC50A-binding protein, and Numb, as the only protein closely related to NOTCH and Hedgehog pathway, was selected to further research. Co-IP and ubiquitin experiments were used to explore how CDC50A interacts with Numb protein. Further, in CDC50A and Numb overexpressed cell lines, Gli 1 was identified as the downstream transcription regulator, and further quantitative PCR was adopted to screen which drug resistance pathways Gli 1 could regulate. In vivo and in vitro experiments confirmed that CDC50A-Numb-Gli 1 pathway can lead to cisplatin and PARPi resistance of ovarian cancer cells, while Gli 1 inhibitor is effective in treatment. Finally, sequencing data and clinical information of serous ovarian cancer patients in TCGA database were used to confirm that CDC50A high expression was an adverse factor for prognosis, and to verify whether CDC50A-Numb-Gli 1 could lead to cisplatin and PARPi resistance.

Methods

1. Analyse RNA-Seq data and explore the relationship between CDC50A and drug-resistance pathways. Differential gene enrichment analysis was performed for RNA-Seq data of primary CDC50A+ and CDC50A- cells. CDC50A+ and CDC50A- cells from HGSOC cell lines SKOV3, OVCAR3, and CaOV4 were separated by flow cytometry, and real-time quantitative PCR (RT-qPCR) was adopted to verify the RNA-Seq result. The CDC50A overexpression plasmid was constructed, and shCDC50A lentivirus was packaged and prepared, followed by SKOV3+shCDC50A and OVCAR3+shCDC50A stable strains screening. After examining overexpression and silence efficacy by the RT-qPCR and the Western Blot (WB), RNA-Seq data was re-tested in overexpression and silence ovarian cell lines.

2. Screening downstream proteins of membrane protein CDC50A leading to cisplatin resistance in ovarian cancer and exploring their interaction mechanism. CDC50A was overexpressed in SKOV3, and IP was used to immunoprecipitate CDC50A and its binding protein analyzed by mass spectrometry. Numb, which may interact with CDC50A and is uniquely related to NOTCH and Hedgehog resistance pathways in the previous step, was selected. The binding of CDC50A to Numb was then verified by IP, CO-IP, and Immunocytochemistry (ICC). The PTB domain and PPR domain overexpression plasmids were constructed, and the binding domain of Numb to CDC50A was estimated. The effect of CDC50A on the expression level of endogenous Numb protein was observed by the WB in CDC50A overexpressed and silenced SKOV3 and OVCAR3, and whether CDC50A could regulate Numb protein ubiquitin degradation was investigated by cell ubiquitin and IP. CCK8 confirmed the effect of Numb on drug-resistant phenotypes at the cellular level. Since CDC50A is not a ubiquitin-related molecule, co-IP and ubiquitin experiments were conducted to verify the interaction between DDB1 and CDC50A and Numb. The CCK8 test was conducted to determine the influence of DDB1 on cis-platinum resistance and whether CDC50A and DDB1 have synergistic effects.

3. Identify downstream transcriptional factor of CDC50A-Numb, explore its effects on drug-resistant molecules, and verify if its inhibitors can treat cisplatin and PARPi resistance induced by CDC50A high expression. Gli 1 is an essential transcription factor in Hedgehog pathway. Studies have suggested that Numb can affect the ubiquitination and degradation of Gli 1. Therefore, this chapter focuses on Gli 1 protein. Numb, CDC50A or DDB1 were overexpressed in SKOV3, and OVCAR3 and WB showed the changes of Gli 1 expression. The Numb and Gli 1 binding was investigated by co-IP, and the relationship between Numb, CDC50A, and DDB1 and the Gli 1 ubiquitination was investigated by cell ubiquitin and IP. CCK8 verified whether Gli 1 was a downstream molecular for cisplatin resistance induced by CDC50A. CDC50A and Gli 1-related drug resistance molecules were screened by RT-qPCR and confirmed by WB in CDC50A overexpression and silenced cell lines. The PDX trial tested whether CDC50A-Numb-Gli 1-BRCA1 was associated with Niraparib resistance in ovarian cancer. CCK8 trials verified whether Gli 1 inhibitors could treat cisplatin - and Olaparib- resistance ovarian cancer.  

4. To explore the relationship between CDC50A and metastasis and prognosis of ovarian cancer and verify the drug resistance mechanism in TCGA database. Based on SKOV3 and OVCAR3, RT-qPCR and WB were used to demonstrate our previous research conclusion that CDC50A can affect BMI-1, β-catenin, APC, E-Cadherin, Vimentin, TGF-β1, NOTCH1, and Oct4 expression levels. Furthermore, the expression level of E-cadherin and Vimentin, essential molecules for tumor metastasis, in overexpressed CDC50A cells was checked by ICC. The RNA-Seq data of 374 patients pathologically diagnosed as serous ovarian tumors in the TCGA database and 180 health populations in the GDC database were compared to explore the relationship between CDC50A expression level and ovarian cancer and whether increased CDC50A expression could affect the patients’ prognosis. Finally, the above TGCA database data and cancer drug sensitivity genomics (GDSC) were used to verify if CDC50A-Gli 1 pathway can lead to cisplatin and PARPi resistance in ovarian cancer patients.     

Results

Verify RNA-Seq results and confirm that CDC50A could affect multiple drug resistance pathways. Re-analysis of primary CDC50A+ and CDC50A-cell RNA-Seq data showed tumor signaling pathways such as Wnt and NOTCH pathway, tumor proliferation and metastasis-related pathways, and glycoprotein metabolism pathways in CDC50A+ cells were significantly enriched. In addition, key molecules of Hedgehog pathway, a critical pathway in tumor stem cells, were also significantly enriched in CDC50A+ cells. The proportion of CDC50A+ cells in the three ovarian cancer cell lines was small and varied by flow cytometry: SKOV3 cell line was the least (average 0.2%), the OVCAR3 cell line was slightly more (average 1.2%), and the CaOV3 cell line was the most (average 7.2%). RT-qPCR showed a consistent trend with RNA-Seq data that Wnt, NOTCH, and Hedgehog pathways were activated, stem cell-related markers, drug resistance, and anti-apoptotic molecules were overexpressed in CDC50A+ cells. More molecules had significant changes in the NOTCH and Wnt pathway than the Hedgehog pathway, and Gli 1 (a key protein of the Hedgehog pathway) always had significant differences, suggesting that it may be one of the downstream molecules of CDC50A. The RT-qPCR of overexpressing and silencing CDC50A cell lines was performed after the CDC50A overexpressed plasmid and shCDC50A silenced lentivirus were successfully constructed. The results were basically in line with RNA-Seq and flow cytometry verification results that Gli 1 showed its role as a potential downstream regulatory molecule of CDC50A again.

CDC50A synergies with DDB1 to promote the ubiquitin degradation of Numb, thereby reducing its protein level and leading to drug resistance of ovarian cancer cells. IP and mass spectrometry identified a total of 413 proteins in three repeated experimental groups, of which only 46 proteins were explicitly bound to CDC50A (not shown in the control group). GO enrichment analysis showed mainly protein binding, intercellular recognition and connection, membrane raft, transcription, DNA damage repair, and ion transport-related pathways. Numb was only closely related to NOTCH and Hedgehog pathways in the protein function analysis. IP, CO-IP, and ICC confirmed that CDC50A could bind to Numb, and co-IP confirmed that CDC50A bound to PTB domain of Numb instead of PPR domain. The Numb level was significantly down-regulated after exogenous overexpression of CDC50A, while its level was significantly increased after silencing of CDC50A. The ubiquitin trial confirmed that CDC50A could increase Numb ubiquitin degradation, reducing Numb protein level. CCK8 showed that Numb overexpression could R CDC50A-induced cisplatin resistance in ovarian cancer cells. Co-IP and ubiquitin assay revealed that DDB1 could bind with CDC50A and Numb, and CDC50A and DDB1 had a synergistic effect of increasing Numb ubiquitin degradation. Congruously, the CCK8 test also showed that DDB1 and CDC50A had a synergistic effect of jointly promoting cisplatin resistance of ovarian cancer cells.  

CDC50A leads to resistance to cisplatin and PARPi by decreasing Numb protein level, inhibiting the ubiquitin degradation of Gli 1, increasing intracellular Gli 1 protein level, and enhancing the transcription of drug-resistant molecules. Numb overexpression decreased intracellular Gli 1 protein level, while CDC50A or DDB1 overexpression significantly increased intracellular Gli 1 protein level. The combination of Numb and Gli 1 was identified by co-IP, and Numb can promote the ubiquitination of Gli 1, while CDC50A and DDB1 can down-regulate the ubiquitination of Gli 1. BRCA1, ABCC2, ABCC5, APC, XPC, NOTCH1, NOTCH2, DLL1, and ABCG2 of NOTCH pathway, ABC drug pump, and DNA damage repair related pathway are downstream drug-resistance related molecules of CDC50A-Numb-GLi 1. CCK8 confirmed that Gli 1 inhibitors could alleviate CDC50A induced cisplatin- and Olaparib-resistance in ovarian cancer cells. PDX drug sensitivity test confirmed that tissues with high expression of CDC50A, GLi 1, and BRCA1 were resistant to Niraparib, and CDC50A, GLi 1, and BRCA1 were significantly enriched in patients after treatment compared with those before treatment, while Numb levels were down-regulated. In vivo experiments have shown that tumor tissues overexpressing CDC50A are resistant to cisplatin and Olaparib, while Gli 1 inhibitors can significantly increase the killing effect.

CDC50A overexpression is associated with ovarian cancer metastasis and poor prognosis, and CDC50a-Gli 1 resistance pathway was verified in TCGA database. CDC50A significantly increased the expression of BMI-1, β-catenin, APC, Vimentin, TGF-β1, NOTCH1 and Oct4, and decreased the expression of E-cadherin, promoting epithelial mesenchymal transformation (EMT) and metastasis. The expression level of CDC50A in tumor patients was significantly higher than that in healthy controls (P=0.0045), and the prognosis of patients with high CDC50A expression was significantly worse than that of patients with low CDC50A expression (HR 1.885, 95%CI 1.226-2.899, P=0.00388). TCGA database analyses demonstrated that CDC50A and DDB1 were negatively correlated with Numb expression level and positively correlated with Gli 1 expression level, and CDC50A and Gli 1 were positively correlated with BRCA1 expression level, which indirectly confirmed the synergistic effect of CDC50A and DDB1 to reduce Numb level and increase Gli 1 and BRCA1 protein expression. In the IC50 predictive test, patients with high expression of CDC50A were insensitive to cisplatin and Olaparib, and patients with high expression of Gli 1 were insensitive to cisplatin and resistant to Olaparib, indirectly supporting the conclusion that CDC50A can lead to cisplatin and PARPi resistance in ovarian cancer, while Gli 1 inhibitor treatment may be effective.  

Conclusion

1. CDC50A up-regulates the expression of multiple drug-resistant pathways, including NOTCH, Wnt, Hedgehog, and anti-apoptotic pathways, confirming that CDC50A is a functional protein that can lead to drug resistance in ovarian cancer cells.

2. CDC50A binds to the PTB domain of Numb, which will competitively inhibit the binding of Numb to downstream molecules. Apart from this functional inhibition, CDC50A cooperates with DDB1 to promote the ubiquitin degradation of Numb and reduce its level. In addition, the drug-resistant phenotype induced by CDC50A and DDB1 can be Rd by Numb overexpression. Therefore, Numb is a downstream molecule of CDC50A, contributing to drug resistance.

3. Numb can bind with Gli 1 to promote its ubiquitin degradation. CDC50A and DDB1 can reduce the ubiquitin degradation of Gli 1 and increase its protein level by decreasing Numb level and competitively inhibiting the binding of Numb to Gli 1. CDC50A-Numb-Gli 1 pathway increases the transcription levels of DNA damage repair pathway, ABC drug pump, and NOTCH pathway, leading to cisplatin and PARPi resistance in ovarian cancer. Gli 1 inhibitors are effective against the drug resistance induced by CDC50A overexpression in ovarian cancer.

4. High expression of CDC50A promotes ovarian cancer metastasis and affects patient survival. Data from the TCGA database in serous ovarian cancer patients confirmed the theory that the CDC50A-MNUMB-Gli 1 pathway leads to cisplatin and PARPi resistance.