Single cell transcriptomics reveals marked heterogeneity for intrinsic molecular subtype and cellular function in breast cancer models and clinical tissue. Single cell transcriptomics reveals marked heterogeneity for intrinsic molecular subtype and cellular function in breast cancer models and clinical tissue

Breast cancers exhibit substantial clinical heterogeneity, consisting of a number of intrinsic subtypes that predict prognosis and influence clinical treatment. To explore epithelial cellular heterogeneity within breast cancers we used single cell transcriptomics to examine thousands of cells from an Estrogen Receptor (ER) positive breast cancer cell-line, a treatment-na•ve PDX model of ER+ breast cancer and an untreated primary ER+ tumour. We show that intrinsic subtype, established from bulk tumours, can be ascribed to individual cells and is heterogenous across cells within tumours and cell-lines. We observed significant heterogeneity in the expression and transcriptional activity of ER, an important driver and therapeutic target, which correlated with expression of two known transcriptional co-regulatory factors of ER, Forkhead Box A1 (FOXA1) and Progesterone Receptor (PR). Intriguingly, we show that ER can differentially induce or suppress gene expression in distinct cellular subpopulations. Unsupervised clustering identified two distinct sub-populations of cells, and their putative transcriptional drivers, that possess unique signatures of transcriptional plasticity and de novo resistance to endocrine therapies. Finally, we extended our study to a clinical ER+ breast cancer and highlight the complex epithelial heterogeneity present in primary tumours, recapitulating many of the features observed in the models. Overall, these results suggest a high degree of epithelial heterogeneity within breast cancer that is present in primary tumours, as well as models of disease, including cell lines commonly considered to be homogenous. We show that single-cell transcriptomics can be used to identify sub-populations of cells with phenotypes of potential clinical relevance, and their putative transcriptional drivers. This approach may lead to new insights into de novo drug resistance.

乳腺癌具有显著的临床异质性，包含多种可预测预后并影响临床治疗的固有亚型。为探究乳腺癌内的上皮细胞异质性，本研究采用单细胞转录组学技术，对来自雌激素受体（Estrogen Receptor, ER）阳性乳腺癌细胞系、未接受治疗的ER阳性乳腺癌患者来源异种移植模型（Patient-Derived Xenograft, PDX）以及未接受治疗的原发性ER阳性肿瘤的数千个细胞进行了检测。本研究证实，从实体瘤批量样本中确定的固有亚型可归因于单个细胞，且在肿瘤及细胞系内的不同细胞间存在异质性。我们观察到，作为重要致癌驱动因子与治疗靶点的ER，其表达与转录活性存在显著异质性，且该异质性与ER的两种已知转录共调控因子——叉头框蛋白A1（Forkhead Box A1, FOXA1）以及孕激素受体（Progesterone Receptor, PR）的表达相关。有趣的是，本研究证实ER可在不同的细胞亚群中差异性诱导或抑制基因表达。无监督聚类分析鉴定出两种不同的细胞亚群及其推定的转录调控驱动因子，这两种亚群具有独特的转录可塑性特征以及对内分泌治疗的原发性耐药特征。最后，本研究将分析拓展至临床ER阳性乳腺癌样本，阐明了原发性肿瘤中存在的复杂上皮异质性，并重现了在模型中观察到的诸多特征。总体而言，本研究结果表明，乳腺癌内存在高度的上皮异质性，这种异质性不仅存在于原发性肿瘤中，也存在于各类疾病模型中，包括通常被认为具有均一性的细胞系。本研究证实，单细胞转录组学技术可用于鉴定具有潜在临床相关表型的细胞亚群及其推定的转录调控驱动因子。该研究方法可为原发性耐药机制的研究提供新的见解。