Transcriptome mapping of bone fracture identifies Notch signaling as an important regulator of skeletal repair

We present a transcriptomic analysis that provides a better understanding of regulatory mechanisms within the healthy and injured periosteum leading to novel translational approaches for bone healing. The focus of this work is on the early regulatory control of bone healing by completing a transcriptomic analysis of forming periosteal callus cells on day 3 post fracture. Based on our previous work indicating that induced Notch1 signaling in osteoprogenitors leads to better healing, we contrasted samples in which Notch 1 intracellular domain (NICD1) is overexpressed by periosteal stem/progenitor cells with control unperturbed periosteum. We determined molecular mechanisms and changes in skeletal stem/progenitor cells (SSPC) and other cell populations within callus including hematopoietic lineages. Notch ligands were differentially expressed in endothelial and mesenchymal populations, with Dll4 restricted to endothelial cells while Jag1 expressed by various mesenchymal populations. When targeting deletion of Dll4 in SSPCs using a-smooth muscle actin (aSMACreER) in mesenchymal cells there was no phenotype, while deletion in EC using Cdh5CreER exhibited negative effects on the early fracture healing. Translation of these observations into clinically relevant model of bone healing revealed the positive effects of combination of Notch ligands delivery with currently used osteogenic inducer BMP2. Overall design: Single cell RNA sequencing (sc RNA-seq) was performed from intact and fractured periosteal cells (3 days after the fracture) in aSMACre+/NICD1 and Cre-/NICD1 mice. To induce NICD1 overexpression Cre- and Cre+ mice were injected with tamoxifen (75 mg/kg) on a day of fracture and 2 dpf (or if intact, 3 and 1 day before sacrificing animals). Three animals of each group were sacrificed and periosteum (intact) or periosteal callus (fractured) were pooled and digested as previously described (13, 42) using Collagenase P (Roche, IN, USA), hyaluronidase (Sigma Aldrich, St Louis, MO, USA) digestion for 1 h on 37°C with shaking. Live, mesenchymal (CD45-) and hematopoietic (CD45+) cells were sorted and single cell RNA-seq performed (10x Genomics). Periosteal cells were digested, stained for CD45-eFluor 450, Ter119-APC and propidium iodide to exclude dead cells. CD45-Ter119- and CD45+Ter119+ were sorted (BD FACSSymphonyTM S6 Cell Sorter) and twelve thousand live cells were loaded into 10X Chromium X controller with ~10,000 cells cell barcoded for RNA-seq using a Chromium Single Cell 3' Reagent Kits v3.1. Libraries were sequenced on a NovaSeq 6000 S4 flow cell lane (Illumina) with an average sequencing depth of 100,000 reads per cell for CD45- and 50,000 reads per cell for CD45+. For FASTQ generation and alignments, Illumina basecall files were converted to FASTQs using bcl2fastq v2.20.0.422 (Illumina) and FASTQ files associated with the gene expression libraries aligned to the mm10 genome using the version 3.1.0 Cell Ranger count pipeline (10x Genomics).