Single‐cell transcriptomics reveals pathogenic dysregulation of previously unrecognised chondral stem/progenitor cells in children with microtia

To the Editor: Microtia, a common craniofacial birth defect worldwide, results from auricular cartilage dysplasia and is accompanied with several unfortunate physical and psychological consequences for children.1 However, existing treatments have adverse outcomes and its pathogenesis is poorly understood.2 Because characterisation of cell type composition and function in auricular cartilage can increase our understanding of microtia,3, 4 we conducted the first single-cell transcriptomic survey in these tissues. We collected auricular cartilage of three children with microtia and six normal controls (NC; two children, four adults), then analysed the transcriptomes of 47 214 total cells after stringent quality control (Figure 1A and Figure S1A–F; Table S1). We detected seven cell types through unsupervised clustering analysis combined with previous reports of cell type-specific marker genes (Figure 1B and Figure S1G). The four cartilage-related cell types included quiescent chondral stem/progenitor cells (CSPCs; expressing EGR15 HES1,6 COL2A1 and CYTL1), which could directly differentiate into chondrocytes (Chonds; COL2A1 and CYTL1); quiescent stromal stem/progenitor cells (SSPCs; EGR1, HES1, COL1A1 and LUM) and stromal cells (SCs; COL1A1 and LUM), whereas other cell types included immune cells, perivascular cells and endothelial cells (Figure 1C,D and Figures S1H,J and S2; Table S2). Gene ontology (GO) terms enriched in the above cell types also showed their specific functions (Figure 1D). Pseudotime trajectory analysis revealed two distinct developmental lineages: CSPCs differentiation into Chonds (chondral lineage), and SSPCs differentiation into SCs (stromal lineage) (Figure 1E). Moreover, immunofluorescence showed that SSPCs were specifically found in the transitional layer and inner perichondrium, whereas CSPCs were specifically located in the chondrium and transitional layer, which differed from published descriptions of ‘cartilage stem cells’ (located in the outer perichondrium; similar to mesenchymal stem cells [MSCs] that specifically co-express CD44 and CD90 and also exhibit the capacity for differentiation into chondrocytes, osteocytes and adipocytes in vitro)7 (Figure 1F and Figure S1I). Subclustering analysis showed four subtypes in the chondral lineage, with CSPCs progressing to early chondrocytes (ECs), to intermediate chondrocytes (ICs) and finally to late chondrocytes (LCs) (Figure 2A,B and Figure S3A,B; Table S3). Stemness markers, chondrocyte differentiation and extracellular matrix (ECM)-related genes were highly expressed in CSPCs, but were gradually downregulated from ECs to LCs, whereas cartilage degeneration-related genes were progressively upregulated from CSPCs to LCs (Figure 2C and Figure S3C). GO analysis supported the above findings (Figures 1D and 2D). Three subtypes were identified in the stromal lineage (Figure 2E and Figure S3D,E; Table S3), with SSPCs differentiating into SC1 or SC2 in parallel (Figure 2F). GO analysis suggested that SC1 could participate in communication with other cell types outside the cartilage, whereas SC2 likely functioned in cartilage formation (Figure 2G,H). Single-cell regulatory network inference and clustering8 identified four major regulons, that is, transcription factors (TFs) and their targets, included SOX8 regulon for CSPCs, NFATC2 regulon for ECs, SOX9 regulon for ICs and LCs and FOXP1 regulon for the stromal lineage, which enriched for cartilage development-related GO terms (Figure 2I–K and Figure S4; Table S4). Focusing on the chondral lineage, we found a higher proportion of CSPCs and ECs, but fewer ICs and LCs, in children compared with adults, among NC group (Figure 3A). Further, approximately half of the differentially expressed genes (DEGs) were shared among at least two subtypes, whereas relatively few DEGs were specific to a single subtype (Figure 3B and Figure S5A,B; Table S5). Upregulated and downregulated DEGs enriched for ageing and ECM formation-related terms, respectively, in most or all subtypes of adults (Figure 3C and Figure S5C,D). Additionally, we found that DEGs in each subtype were significantly enriched in the Aging Atlas9 and GenAge10 databases (Figure 3D), in which the TF, CEBPB, was upregulated in all subtypes in the chondral lineage of adults (Figure 3E). CEBPD, the interaction partner of CEBPB, was also upregulated in each subtype (Figures 3E and Figure S5E). They exhibited the highest co-expression/regulatory activity (Figure S5F), and their regulatory activities were upregulated in all subtypes of adults (Figure 3F). Further gene regulatory network (GRN) analysis confirmed that upregulated DEGs targeted by CEBPB and CEBPD were significantly enriched in ageing-related terms (Figure 3G and Figure S5G,H). When exploring the role of the chondral lineage in microtia, a lower proportion of CSPCs was found in microtia compared with NC children (Figure 4A). Microtia-associated DEGs (MADEGs) were obtained after excluding the influence of gender factor, and CSPCs had the highest percentage of differentially expressed marker genes among all subtypes (Figure S6A–C; Table S6). In particular, CSPCs showed more significant enrichment than other subtypes for ‘ECM structural constituents’, ‘cartilage development’, ‘cellular respiration’ among downregulated MADEGs, and ‘response to oxidative stress’, ‘apoptotic signaling pathway’ among upregulated MADEGs (Figure 4B,C). Time course analysis identified four gene groups potentially influencing the chondral lineage trajectory of microtia (Figure 4D and Figure S6D,E; Table S7). The TF group, predominantly comprised TFs, was responsible for transcriptional regulation (Figure 4D). In this group, genes showed the highest transcription levels in CSPCs, and differences in gene expression were largest in CSPCs differentiating into ECs between microtia and NC children (Figure 4D). To determine which TFs could contribute essential pathogenic functions to microtia development, we used the PageRank algorithm to assess this potential for TFs in GRNs constructed with downregulated or upregulated MADEGs of CSPCs in microtia, respectively (Figure 4E,H). In downregulated GRN, the CSPC-specific SOX8 TF (Figure 2J) had the highest PageRank score and greatest decrease in expression (Figure 4E,F and Figure S7), targeting MADEGs involved in ECM organisation and cartilage development. In upregulated GRN, quiescent stem cell marker EGR1 showed the higher differential expression and PageRank score simultaneously, primarily targeting MADEGs related to the response to oxidative stress (Figure 4H,I and Figure S8). Immunohistochemistry verified that SOX8 decreased, while EGR1 increased, in protein expression in CSPCs of microtia (Figure 4G,J). Here, we systematically annotate cartilage-related cell subtypes in human auricular cartilage by single-cell RNA sequencing. This study also represents the first description of CSPCs in the chondrium, potentially enabling directed chondrogenic differentiation, and shows that dysregulation of cartilage development by SOX8, and dysregulation of the response to oxidative stress by EGR1, in CSPCs may be causative mechanisms driving microtia. While future work will explore these possibilities, these findings build a theoretical foundation for advanced exploration of the microtia pathogenesis and offer new avenues for potential treatments. This project was funded by National Natural Science Foundation of China (81771014, 81800920, 31871325, 32170667) and Natural Science Foundation Project of Shanghai Science and Technology Innovation Action Plan (20ZR1409900). Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.