研究背景及研究目的

肿瘤是多种具有异质性的组织、细胞及分子组成的一个复杂生态系统，从而能够在机体微环境强大的选择压力下促进自身的适应和进化。这种瘤内异质性在肿瘤的生长、侵袭和转移中起到了重要的作用，但目前对于瘤内肿瘤细胞亚型的鉴定仍没有很好的方法。因此本研究利用单细胞测序技术分析基因组拷贝数变异情况，旨在深入解析微卫星稳定结直肠癌（microsatellite stable colorectal cancer，MSS CRC）的瘤内异质性。

研究方法

我们首先从基因表达数据库（Gene Expression Omnibus, GEO）中获得一个包含19名韩国MSS CRC患者的26例组织样品的单细胞转录组测序数据集（准入号为GSE132465，简称为SMC数据集），接着利用R平台的Seurat包对其进行数据降维和细胞分群注释，使用CopyKAT包推断肿瘤细胞中不同的体细胞拷贝数变异（somatic copy number alteration，SCNA）模式，使用monocle2、SCENIC、CellChat等算法包或软件探究其各自的分子特征及肿瘤细胞与免疫微环境的相互作用。随后，下载另一个单细胞测序数据集（准入号为GSE144735，由5名来自比利时的MSS CRC患者的15个组织样品组成，简称为KUL3数据集）对我们的发现进行验证。此外，基于单细胞测序中发现的肿瘤细胞染色体SCNA特征，我们结合TCGA数据集中结直肠癌转录组测序数据，利用生存分析鉴定出预后相关基因，并在本中心NCC（National Cancer Center，国家癌症中心）队列收集的24名患者转录组测序数据和GEO的GSE39582数据集中进行了验证。

研究结果

我们发现MSS CRC的肿瘤细胞具有三种主要的SCNA模式（分别命名为C1，C2和C3）。在这三类肿瘤细胞中，C2包含了大部分二倍体细胞且有着相对最低的拷贝数变异负荷，C1以13号和20号染色体明显扩增为主要特点，而C3则呈现与C1相反的特征。在生物学行为方面，C1主要表现为抗原加工提呈、细胞凋亡和铁死亡等相关通路活性增强，C2为能量代谢改变和上皮间质转化活跃，而C3则主要参与肿瘤微环境中异常的血管生成过程。同时这三类肿瘤细胞也呈现出不同的突变特征，如C1中多见TP53突变，而C2则携带更多的KRAS和BRAF突变。此外我们发现上皮特异性基因OLFM4的高表达与MSS CRC患者更好的预后相关，且与更早的病理分期、N分期、M分期相关。

研究结论

我们从拷贝数变异的角度，在单细胞层面解析了MSS CRC的瘤内异质性，发现CRC细胞有三种SCNA模式，且具有各不相同的分子特征和免疫微环境构成。

Background and objective

With functionally heterogeneous cells, tumors comprise a complex ecosystem to promote tumor adaptability and evolution under strong selective pressure from the given microenvironment. Diversifying tumor cells or intratumor heterogeneity is essential for tumor growth, invasion, and immune evasion. However, there is still no reliable method for classifying tumor cell subtypes. We aim to explore intratumor heterogeneity based on copy number variation in microsatellite stable colorectal cancer (MSS CRC) with single-cell sequencing data.

Methods

To characterize the somatic copy number alteration (SCNA) of MSS CRC in a single-cell profile, we analyzed 26 tissue samples from 19 Korean MSS CRC patients (GSE132465, the SMC dataset) and then explored the molecular features and the composition of the tumor immune microenvironment with Seurat, CopyKAT, monocle2, SCENIC, and CellChat R packages. We further verified our findings with 15 tissue samples from 5 Belgian MSS CRC patients (GSE144735, the KUL3 dataset). Besides, we identified a prognosis-related signature in TCGA dataset, and we verified the clinical features of this gene with the NCC (National Cancer Center) cohort and the GSE39582 cohort.

Results

We employed single-cell RNA sequencing data to identify three types of tumor cells in MSS CRC based on their SCNA characteristics. Among these three SCNA patterns, C1 had significant chromosome 13 and chromosome 20 amplification, whereas C3 was the polar opposite of C1. And C2 contained most diploid cells and had a lower SCNA burden. In terms of biological function, C1 was associated with antigen processing and presentation, apoptosis, and ferroptosis while C2 was linked to ATP metabolic process as well as epithelial-mesenchymal transition. And angiogenesis in the tumor microenvironment was related to C3. Furthermore, the three types of tumor cells harbored different mutations: more mutant TP53 in C1, and more mutant KRAS as well as BRAF in C2. Besides, we discovered an epithelial-specific signature OLFM4, the enhanced expression of which was associated with a better prognosis and an earlier stage, N stage as well as M stage in MSS CRC patients.

Conclusions

Based on copy number characteristics, we interpreted intratumor heterogeneity and identified three types of tumor cells with variable SCNA characteristics in MSS CRC. In tumor microenvironment, tumor cells with these three SCNA patterns showed distinct biological functions and molecular features.