Fibroblasts are non-haematopoietic structural cells that define the architecture of organs, support the homeostasis of tissue-resident cells and have key roles in fibrosis, cancer, autoimmunity and wound healing¹. Recent studies have described fibroblast heterogeneity within individual tissues¹. However, the field lacks a characterization of fibroblasts at single-cell resolution across tissues in healthy and diseased organs. Here we constructed fibroblast atlases by integrating single-cell transcriptomic data from about 230,000 fibroblasts across 17 tissues, 50 datasets, 11 disease states and 2 species. Mouse fibroblast atlases and a Dpt^IRESCreERT2 knock-in mouse identified two universal fibroblast transcriptional subtypes across tissues. Our analysis suggests that these cells can serve as a reservoir that can yield specialized fibroblasts across a broad range of steady-state tissues and activated fibroblasts in disease. Comparison to an atlas of human fibroblasts from perturbed states showed that fibroblast transcriptional states are conserved between mice and humans, including universal fibroblasts and activated phenotypes associated with pathogenicity in human cancer, fibrosis, arthritis and inflammation. In summary, a cross-species and pan-tissue approach to transcriptomics at single-cell resolution has identified key organizing principles of the fibroblast lineage in health and disease.

成纤维细胞是定义器官结构的非造血结构细胞，支持组织驻留细胞的稳态，并在纤维化、癌症、自身免疫和伤口愈合¹中发挥关键作用。最近的研究已经描述了单个组织¹内的成纤维细胞异质性。然而，该领域缺乏在健康和患病器官组织中以单细胞分辨率对成纤维细胞进行表征。在这里，我们通过整合来自 17 个组织、50 个数据集、11 个疾病状态和 2 个物种的约 230,000 个成纤维细胞的单细胞转录组数据来构建成纤维细胞图谱。小鼠成纤维细胞图谱和Dpt ^IRESCreERT2敲入小鼠在组织中鉴定出两种通用的成纤维细胞转录亚型。我们的分析表明，这些细胞可以作为一个储存库，可以在广泛的稳态组织和疾病中的活化成纤维细胞中产生专门的成纤维细胞。与来自扰动状态的人类成纤维细胞图谱的比较表明，成纤维细胞转录状态在小鼠和人类之间是保守的，包括与人类癌症、纤维化、关节炎和炎症的致病性相关的通用成纤维细胞和活化表型。总之，一种跨物种和泛组织的单细胞分辨率转录组学方法已经确定了健康和疾病中成纤维细胞谱系的关键组织原则。