第一部分  BCL-2抑制剂在儿童急性B淋巴细胞白血病中耐药机制的研究

背景

急性淋巴细胞白血病（acute lymphoblastic leukemia， ALL）是一种早期分化受阻、能够侵犯骨髓、血液和髓外部位的淋巴样细胞恶性增殖性肿瘤，在儿童和成人中均可发病，但是主要发生在儿童，发病高峰在1到4岁，约占该年龄段所有肿瘤的80%。尽管在过去的30年里，诊疗技术的不断提升使当代儿童ALL的5年总生存率普遍可达90%以上，但复发/难治性ALL患儿的预后仍然很差。BCL-2抑制剂是一种新型的抗肿瘤药物，主要通过直接靶向促进细胞存活的BCL-2蛋白来杀伤肿瘤细胞，其具备的高效性、特异性和作为口服药物的良好的生物利用度使之在多种血液系统恶性肿瘤中得到了广泛应用。目前，Venetoclax已被美国食品和药物管理局(Food and Drug Administration， FDA)批准用于治疗慢性淋巴细胞白血病(chronic lymphocytic leukemia， CLL)和急性髓系白血病(acute myeloid leukemia， AML)。在ALL中，Venetoclax的抗肿瘤活性也在部分亚型中得到证实。然而，随着BCL-2抑制剂的广泛应用，获得性耐药也随之出现，因此，寻找新的治疗靶点来克服耐药性是提升临床治疗效果和改善患者预后的重要策略，而确定药物的耐药机制以及探索新的治疗靶点则在很大程度上依赖于肿瘤耐药细胞系模型的构建。因此，本研究旨在利用急性B淋巴细胞白血病（B-cell acute lymphoblastic leukemia， B-ALL）细胞系RS4;11构建对BCL-2抑制剂Navitoclax和Venetoclax耐药的细胞系模型,再结合RNA-sequencing(RNA-seq)技术及Clustered Regularly Interspaced Short Palindromic Repeats-CRISPR associated protein 9（CRISPR/Cas9）技术来研究诱发B-ALL细胞对BCL-2抑制剂耐药的机制。

方法

采用小剂量低浓度递增、间歇诱导的方法构建耐药细胞系，以低浓度10 nmol/L Navitoclax和1 nmol/L Venetoclax作为诱导的起始培养浓度，经过6个月的间歇培养，通过MTT法检测不同药物浓度下RS4;11亲本细胞系和耐药细胞系的活率，从而确认对BCL-2抑制剂耐药的细胞系RS4;11 Navitoclax-Resistant（RS4;11 N-R）细胞系和RS4;11 Venetoclax-Resistant（RS4;11 V-R）细胞系的成功构建。之后采用流式细胞术检测Navitoclax和Venetoclax对RS4;11亲本细胞系和耐药细胞系细胞凋亡的影响，采用EDU检测Navitoclax和Venetoclax对RS4;11亲本细胞系和耐药细胞系细胞周期的影响。为验证耐药细胞系的耐药机制是否与MCL-1的过表达有关，采用MTT检测亲本细胞系和耐药细胞系对MCL-1抑制剂S63845的敏感性，通过细胞凋亡和细胞周期实验验证MTT检测的结果。采用western blot实验检测BCL-2家族抗凋亡蛋白的表达。对RS4;11耐药细胞系和亲本细胞系进行RNA-seq，通过对测序结果进行分析，得到显著表达的差异基因，随后进行Gene Ontology(GO)和Kyoto Encyclopedia of Genes and Genomes(KEGG）富集分析,初步探索对BCL-2抑制剂可能的耐药机制。通过RT-qPCR和Western blot对显著表达的差异基因从mRNA表达水平和蛋白表达水平进行验证。利用CRISPR/Cas9技术敲除显著表达的差异基因，随后利用western blot实验检测目的基因是否成功敲除，利用MTT实验检测基因敲除对耐药细胞系耐药性的影响。

结果：

Navitoclax和Venetoclax的MTT实验结果证明了RS4;11 N-R和RS4;11 V-R细胞系的成功构建。细胞凋亡和细胞周期实验结果表明，亲本细胞系被Navitoclax和Venetoclax显著抑制（p<0.001），而耐药细胞系的凋亡和增殖基本不受药物影响。MCL-1抑制剂S63845的MTT实验结果表明，相比于亲本细胞系，2种耐药细胞系对S63845更加耐药。而对BCL-2家族抗凋亡的BCL-2、BCL-XL和MCL-1蛋白的western blot结果发现，这些蛋白的表达水平在亲本和耐药细胞系中没有显著性差异。RNA-seq结果显示，E1A binding protein p300（p300, EP300）在耐药细胞系中的表达较亲本细胞系升高，而随后进行的western blot和RT-qPCR结果则分别从蛋白水平上和mRNA水平上证明了耐药细胞系中存在EP300的表达上调。通过western blot实验我们证明了CRISPR/Cas9技术成功将2种诱导构建的对BCL-2抑制剂耐药的细胞系（RS4;11 N-R和RS4;11 V-R细胞系）及2种天然对BCL-2抑制剂耐药的细胞系（Reh和Nalm6细胞系）中的EP300基因敲除，并且MTT实验结果显示，EP300基因敲除后，上述4种对BCL-2抑制剂耐药的细胞系对BCL-2抑制剂（Navitoclax和Venetoclax）、BCL-XL抑制剂（A1331852）和MCL-1抑制剂（S63845）的敏感性都提高了，提示引起B-ALL细胞系对BCL-2抑制剂耐药的机制可能与EP300的表达上调有关。

结论

本研究通过构建体外耐药细胞系模型，结合RNA-seq测序技术和CRISPR/Cas9技术确定了EP300的过表达可能是导致B-ALL细胞系对BCL-2抑制剂耐药的机制。

第二部分  基于生物信息学方法预测儿童复发性急性淋巴细胞白血病预后相关的铁死亡基因

背景

急性淋巴细胞白血病(acute lymphoblastic leukemia，ALL)是儿童最常见的恶性血液病，尽管危险度分层诊疗方案的实施和治疗方案的优化使当代儿童ALL的治愈率已超过90%，然而仍有相当多的患儿出现化疗耐药和肿瘤复发。由于化疗药物主要通过诱导细胞凋亡发挥抗白血病作用，因此，白血病细胞抗凋亡活性的增加是最常见的复发机制之一。铁死亡作为一种铁依赖性细胞死亡，由脂质过氧化产物的异常积累引起。由于它不同于其他类型的程序性死亡，例如细胞凋亡、坏死性凋亡和自噬性细胞死亡，因此，诱发肿瘤细胞的铁死亡成为了新的肿瘤治疗策略，特别是对于对传统疗法反应不佳的侵袭性、复发性恶性肿瘤。研究表明，诱导铁死亡可能是治疗复发ALL的有效方法，因此，我们的研究旨在确定儿童复发性ALL预后相关的铁死亡相关基因（ferroptosis-related genes，FRGs）并构建预后风险模型。

方法

我们从Therapeutically Applicable Research to Generate Effective Treatments(TARGET)数据库获得256个儿童ALL患者的临床信息和基因表达数据，并将患儿分为复发组和非复发组。从铁死亡FerrDb数据库中获得268个FRGs，结合复发组和非复发组儿童ALL患者之间的差异表达基因，从而得到差异表达的FRGs（differentially expressed FGRs，DEFRGs）。随后，将具有完整临床信息的191个患者样本随机分为训练组（n=96）和验证组（n=95），对训练组进行DEFRGs的单因素Cox回归分析和Lasso回归分析来确定预后相关的DEFRGs，随后进行多因素Cox回归分析和风险模型的构建，根据构建的风险模型将训练组分为高、低风险两组，并同时在验证组和整个样本组中来验证风险模型的准确性。比较高、低风险组中的差异表达基因并对其进行Gene Ontology(GO)和Kyoto Encyclopedia of Genes and Genomes(KEGG）富集分析，以对高、低风险组之间差异表达基因的功能进行分析。

结果

我们从256个儿童ALL样本中筛选了43个DEFRGs，并且这些基因在复发组中显著上调。对训练组使用单因素Cox分析筛选了28个与总生存(overall survival, OS)相关的DEFRGs。然后，Lasso回归分析筛选出8个预后相关的基因并用于构建预后风险模型，而进一步的多因素Cox回归分析确定了出5个最佳的预后基因（MT1G、GPT2、TFAP2C、GCLM和HSBP1）。对训练组、验证组和整个样本组中的高、低风险组计算风险评分和进行生存分析，结果显示，在不同分组中高风险评分均与不良预后相关。受试者工作特征(receiver operating characteristic curve，ROC)曲线显示出良好的预测准确性（训练组、验证组和整个样本组的预后模型的AUC值在1年时分别为0.817、0.87和0.918，在3年时分别为0.652、0.770和0.782，在5年时分别为0.743、0.825和0.851）。因此，我们构建的预后风险模型在训练组、验证组和整个样本组中得到了成功验证。对高、低风险组进行的差异分析和富集分析则提示ALL患者的预后与免疫活性相关。对高、低风险组进行的单样本基因富集分析(single-sample gene set enrichment analysis, ssGSEA)的结果则表明，在高风险组中存在免疫细胞浸润和免疫相关通路的激活。

结论

在儿童复发性ALL中构建的FRGs的预后风险模型能够有效地将患者分为高风险和低风险两组并准确预测患者预后，并且高、低风险组预后的差异可能与肿瘤的免疫浸润相关，这对于预测临床ALL患者预后和优化个体化治疗具有重要意义。

Part1  Mechanisms of resistance to BCL-2 inhibitors in childhood B-cell acute lymphoblastic leukemia

Background

Acute lymphoblastic leukemia is a malignant proliferation of lymphoid cells blocked at an early stage of differentiation that can invade bone marrow, blood, and extramedullary sites. It can occur in both children and adults, but mainly occurs in children, with the peak incidence between 1 and 4 years old, accounting for about 80% of all cancers in this age group. Although in the past 30 years, the continuous improvement of diagnosis and treatment technology has made the 5-year overall survival rate of contemporary ALL children generally reach more than 90%, the prognosis of patients with refractory or relapsed ALL is still poor. BCL-2 inhibitor is a novel type of anti-tumor drug, which mainly kills tumor cells by directly targeting the BCL-2 protein that promotes cell survival. Its high efficiency, specificity and good bioavailability as an oral drug make it widely used in various hematological malignancies. Currently, Venetoclax has been approved by Food and Drug Administration (FDA) for the treatment of chronic lymphocytic leukemia (CLL) and acute myeloid leukemia(AML). In ALL, the antitumor activity of Venetoclax has also been validated in some ALL subtypes. However, with the clinical application of BCL-2 inhibitors, acquired drug resistance also appears. Therefore, finding new therapeutic targets to overcome drug resistance is an important strategy to promote clinical treatment effects and improve patient prognosis. Identifying the mechanism of drug resistance and exploring new therapeutic targets largely depend on the construction of tumor drug-resistant cell line models. Therefore, our study aims to use the B-cell acute lymphoblastic leukemia cell line RS4;11 to construct two cell line models that are resistant to BCL-2 inhibitors-Navitoclax and Venetoclax, respectively, and then combine RNA sequencing (RNA-seq) technology and Clustered Regularly Interspaced Short Palindromic Repeats-CRISPR associated protein 9 （CRISPR/Cas9）technology to study the potential mechanisms of drug resistance to BCL-2 inhibitors.

Methods

The drug-resistant cell line was established by intermittent induction with small doses and low concentrations. For Navitoclax, the initial culture concentration of induction was 10 nmol/L and for Venetoclax, 1 nmol/L. After 6 months of intermittent cultivation, MTT assay was used to detecte the viability of RS4;11 parental cell line and drug-resistant cell lines with varying drug doses to confirme the successful establishment of BCL-2 inhibitor-resistant cell lines, RS4;11 Navitoclax-Resistant（RS4;11 N-R）cell line and RS4;11 Venetoclax-Resistant cell line（RS4;11 V-R）. Then the effect of Navitoclax and Venetoclax on the distribution of cell apoptosis among RS4;11 parental cell line and drug-resistant cell lines was detected by flow cytometry and the effect of Navitoclax and Venetoclax on the cell cycle of RS4;11 parental cell line and drug-resistant cell lines was detected by EDU. In order to verify whether the mechanism of drug-resistant cell line is related to the overexpression of MCL-1, MTT was used to detect the sensitivity of the parental cell line and drug-resistant cell lines to the MCL-1 inhibitor S63845, and cell apoptosis and cell cycle experiments were performed to verify the results of the MTT assay. The expression of anti-apoptotic proteins of BCL-2 family was detected by western blot experiment. RNA-seq technology was performed on RS4;11 drug-resistant cell lines and parental cell line, and the significantly differential expressed genes were obtained by analyzing the sequencing results, followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG） enrichment analysis to initially explore the possible mechanism of BCL-2 drug resistance. Significantly differential expressed genes were verified from mRNA and protein expression level by RT-qPCR and Western blot. CRISPR/Cas9 technology was used to knock out significantly differential expressed genes, and western blot was used to detect whether the target gene was successfully knocked out, and MTT assay was used to detect the impact of gene knockout on drug-resistant cell lines.

Results

The results of MTT experiments of Navitoclax and Venetoclax proved the successful construction of RS4;11N-R and RS4;11V-R cell lines. The results of cell apoptosis and cell cycle experiments showed that the parental cell line was significantly inhibited by Navitoclax and Venetoclax (p<0.001), while the apoptosis and proliferation of drug-resistant cell lines were basically not affected by the drugs. The results of MTT assay of MCL-1 inhibitor S63845 showed that the two drug-resistant cell lines were more resistant to S63845 than the parental cell line. Besides, the western blot detecting of the BCL-2 family anti-apoptosis proteins, BCL-2, BCL-XL and MCL-1, found that there was no significant difference in the expression levels of these proteins between the parental and drug-resistant cell lines. The result of RNA-seq showed that the expression of E1Abindingproteinp300 (p300, EP300) in the drug-resistant cell lines were higher than that in the parental cell line, and the results of subsequent western blot and RT-qPCR proved that the upregulation of EP300 expression in drug-resistant cell lines from protein level and mRNA level, respectively. Western blot experiments proved that the CRISPR/Cas9 technology successfully knocked out the EP300 gene in two cell lines, RS4;11 N-R and RS4;11 V-R, which were successfully induced and constructed to be resistant to BCL-2 inhibitors and two cell lines, Reh and Nalm6, which were naturally resistant to BCL-2 inhibitors. In addition, the results of MTT experiment showed that the above four cell lines resistant to BCL-2 inhibitors became more sensitive to BCL-2 inhibitors (Navitoclax, Venetoclax), BCL-XL inhibitors (A1331852) and MCL-1 inhibitors (S63845) after EP300 gene was knocked out, suggesting that the mechanism of drug resistance of B-ALL cell lines to BCL-2 inhibitors may be related to the up-regulation of EP300 expression.

Conclusions

By establishing drug-resistant cell lines model in vitro, combining with RNA sequencing technology and CRISPR/Cas9 technology, it was determined that the overexpression of EP300 is the drug resistance mechanism leading to the B-ALL cell lines that were resistant to BCL-2 inhibitors.

Part2  Prediction of ferroptosis genes related to the prognosis of relapsed pediatric acute lymphoblastic leukemia based on bioinformatics methods

Background

Acute lymphoblastic leukemia (ALL) is the most common hematological malignancy in children. Although the cure rate of contemporary children with ALL has exceeded 90% due to the implementation of risk stratified diagnosis and the optimization of treatment schemes, there are still a considerable number of children with chemotherapy resistance and tumor recurrence. Since chemotherapy drugs exert anti-leukemic effects mainly by inducing apoptosis, the increase of anti-apoptotic activity of leukemia cells is one of the most common relapse mechanisms. Ferroptosis is an iron-dependent cell death caused by abnormal accumulation of lipid peroxidation products. Because it is different from other types of programmed death, such as apoptosis, necroptosis, and autophagic cell death, inducing ferroptosis in tumor cells has become a new tumor treatment strategy, especially for aggressive, recurrent malignancies that do not respond well to conventional therapies. Studies have shown that induction of ferroptosis may be an effective method for the treatment of relapsed ALL, therefore, our study aimed to identify the prognostic ferroptosis-related genes (FRGs) in children with relapsed ALL and construct a risk assessment model.

Method

We obtained gene expression data of 256 children with ALL from the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) database, and the children were divided into relapse group and non-relapse group. 268 FRGs were obtained from FerrDb database, and the differentially expressed FRGs(DEFRGs)  were obtained by combining the differentially expressed genes (DEGs) between the relapse group and the non-relapse group of children with ALL patients. The 191 patient samples with complete clinical information were randomly divided into a training cohort (n=96) and a testing cohort (n=95). Univariate Cox regression analysis and Lasso regression analysis of DEFRGs were performed on the training cohort to determine prognostic-related DEFRGs, followed by multivariate Cox regression analysis and risk model construction. According to the constructed risk model, the training cohort was divided into high-risk and low-risk groups. and verify the accuracy of the risk model in the testing cohort and the whole sample cohort. DEGs in high and low risk groups were compared and Gene Ontology(GO) and Kyoto Encyclopedia of Genes and Genomes(KEGG） enrichment analysis were performed on them to analyze the function of differential genes between high and low risk groups.

Result

We screened 43 DEFRGs from 256 pediatric ALL samples, and these genes were significantly upregulated in the relapsed group. 28 DEFRGs associated with overall survival (OS) were screened using univariate Cox analysis on the training cohort. Then, Lasso regression analysis screened 8 prognostic-related genes for constructing the prognostic risk model, and further multivariate Cox regression analysis identified 5 optimal prognostic genes (MT1G, GPT2, TFAP2C, GCLM, and HSBP1). Risk scores and survival analysis were calculated for the high and low risk groups in the training cohort, testing cohort, and the whole sample cohort, and the results showed that high risk scores were associated with poor prognosis in each group. The receiver operating characteristic curve (ROC) showed good predictive accuracy (the AUC values of the prognostic model in the training cohort, the testing cohort and the whole sample cohort were 0.817, 0.870 and 0.918 at 1 year, 0.652, 0.770 and 0.782 at 3 years, 0.743, 0.825, and 0.851 at 5 years, respectively,). Therefore, the prognostic risk model that we constructed was successfully validated in training cohort, testing cohort, and whole sample cohort. Difference analysis and enrichment analysis between high-risk and low-risk groups suggested that the prognosis of ALL patients was related to immune activity. Single-sample gene enrichment analysis (ssGSEA) between high-risk and low-risk groups showed that there were immune cell infiltration and activation of immune-related pathways in the high-risk group.

Conclusion

The prognostic risk model of FRGs constructed in pediatric relapsed ALL can effectively divide patients into high-risk and low-risk groups and predict patient prognosis. Besides, the difference in prognosis between high-risk and low-risk groups may be related to tumor immune infiltration, which has great significance for predicting the prognosis of clinical ALL patients and the optimization treatment of individualized