整合基因组学和机器学习分析确定SPOP为与鼻咽癌肿瘤免疫特征相关的候选生物标志物

文献信息

中文摘要

背景： 鼻咽癌（NPC）中免疫治疗的疗效受到免疫逃逸异质性机制的限制。识别能够预测该表型的稳健分子生物标志物对于患者分层至关重要。本研究旨在从预定义的甲基化相关基因集中识别候选基因，并采用整合计算和实验方法评估其与NPC肿瘤免疫特征的关联。
方法： 我们对NPC转录组数据集与精选的甲基化相关基因集进行了综合分析。将差异表达基因与预定义的甲基化相关基因集取交集，随后进行基于机器学习的候选基因优先级排序。通过机器学习流程对候选基因进行优先级排序，优先考虑同时具有诊断能力和与肿瘤免疫微环境生物学相关性的基因。在保留的四个基因中，SPOP被优先选为下游验证的主要候选基因，并进一步评估其诊断性能、免疫相关性和体外功能作用。
结果： 机器学习驱动的分析将SPOP、ZNF667、TADA2B和ABLIM1优先列为具有诊断和预后相关性的候选基因，其中SPOP被选中进行后续功能验证。SPOP与估计的CD8⁺ T细胞浸润呈正相关，并与免疫相关特征的改变有关。实验验证表明，SPOP敲低抑制NPC细胞增殖并增加细胞凋亡。在肿瘤细胞-T细胞共培养中，SPOP敲低与更强的肿瘤细胞杀伤以及更高的IFN-γ和IL-2水平相关。分子分析进一步显示，SPOP调节伴随SOCS3表达和STAT3磷酸化的变化，而CHX追踪实验提示SOCS3蛋白稳定性发生改变。
结论： 这项整合研究确定SPOP为与NPC细胞表型和肿瘤-免疫相互作用改变相关的候选调节因子。SPOP调节伴随SOCS3丰度和STAT3磷酸化的变化，而SPOP敲低改变了T细胞共培养系统中的肿瘤细胞存活和细胞因子相关读数。需要进一步的生化、挽救和体内研究。

英文摘要

BACKGROUND: The efficacy of immunotherapy in nasopharyngeal carcinoma (NPC) is limited by heterogeneous mechanisms of immune evasion. Identifying robust molecular biomarkers that predict this phenotype is crucial for patient stratification. This study aimed to identify candidate genes derived from a predefined methylation-related gene set and to evaluate their association with tumor-immune features in NPC using an integrative computational and experimental approach.
METHODS: We performed a comprehensive analysis of NPC transcriptomic datasets combined with a curated methylation-related gene set. Differentially expressed genes were intersected with the predefined methylation-related gene set, followed by machine-learning-based candidate prioritization. Candidate genes were prioritized through a machine learning pipeline, prioritizing genes with both diagnostic power and biological relevance to the tumor immune microenvironment. Among the four retained genes, SPOP was prioritized as the lead candidate for downstream validation and was further evaluated for its diagnostic performance, immune relevance, and functional role in vitro.
RESULTS: The machine learning-driven analysis prioritized SPOP, ZNF667, TADA2B, and ABLIM1 as candidate genes with diagnostic and prognostic relevance, with SPOP selected for subsequent functional validation. SPOP was positively correlated with estimated CD8⁺ T-cell infiltration and was associated with altered immune-related features. Experimental validation showed that SPOP knockdown inhibited NPC cell proliferation and increased apoptosis. In tumor cell-T-cell co-culture, SPOP knockdown was associated with greater tumor-cell killing and higher IFN-γ and IL-2 levels. Molecular analyses further showed that SPOP modulation was accompanied by changes in SOCS3 expression and STAT3 phosphorylation, whereas CHX chase assays suggested altered SOCS3 protein stability.
CONCLUSION: This integrative study identifies SPOP as a candidate regulator associated with NPC cell phenotypes and altered tumor-immune interactions. SPOP modulation was accompanied by changes in SOCS3 abundance and STAT3 phosphorylation, while SPOP knockdown altered tumor-cell survival and cytokine-related readouts in the T-cell co-culture system. Further biochemical, rescue, and in vivo studies are required.