Search

Blood vessels supplying tumors are often dysfunctional and generally heterogeneous. The mechanisms underlying this heterogeneity remain poorly understood. Here, using multicolor lineage tracing, in vivo time-lapse imaging and single cell RNA sequencing in a mouse glioma model, we identify tumour-specific blood endothelial cells that originate from cells expressing the receptor for colony stimulating factor 1, Csf1r, a cytokine which controls macrophage biology. These Csf1r lineage endothelial cells (CLECs) form up to 10% of the tumour vasculature and express, besides classical blood endothelial cell markers, a gene signature that is distinct from brain endothelium but shares similarities with lymphatic endothelial cell populations. in silico analysis of pan-cancer single cell RNAseq datasets highlights the presence of a comparable subpopulation in the endothelium of a wide spectrum of human tumours. We show that CLECs actively contribute to sprouting and remodeling of tumour blood vessels and that selective depletion of CLECs reduces vascular branching and tumour growth. Our findings indicate that a non-tumour resident Csf1r-positive population is recruited to tumours, differentiates into blood endothelial cells to contribute to vascularization and, thereby, tumour growth.

SUMMARYBlood vessels supplying tumors are often dysfunctional and generally heterogeneous. The mechanisms underlying this heterogeneity remain poorly understood. Here, using multicolor lineage tracing, in vivo time-lapse imaging and single cell RNA sequencing in a mouse glioma model, we identify tumour-specific blood endothelial cells that originate from cells expressing the receptor for colony stimulating factor 1, Csf1r, a cytokine which controls macrophage biology. These Csf1r lineage endothelial cells (CLECs) form up to 10% of the tumour vasculature and express, besides classical blood endothelial cell markers, a gene signature that is distinct from brain endothelium but shares similarities with lymphatic endothelial cell populations. in silico analysis of pan-cancer single cell RNAseq datasets highlights the presence of a comparable subpopulation in the endothelium of a wide spectrum of human tumours. We show that CLECs actively contribute to sprouting and remodeling of tumour blood vessels and that selective depletion of CLECs reduces vascular branching and tumour growth. Our findings indicate that a non-tumour resident Csf1r-positive population is recruited to tumours, differentiates into blood endothelial cells to contribute to vascularization and, thereby, tumour growth.

We have previously shown that PIXSCAN, our X-ray micro-CT prototype based on the XPAD2 photon counting detector was useful to monitor the development of tumors at low irradiation dose in a murine model of lung cancer (WMIC2008). In order to challenge PIXSCAN in conditions where contrast was less favorable for tumor detection, we have investigated the possibility to monitor brain tumor development in the context of an orthotopic glioblastoma xenograft model. Although endogenous X-ray contrast between pathological and normal tissue was too low to detect millimetric glioblastoma, more than 75% of tumors (n=9) could be successfully revealed by intravenous injection of iodinated contrast agent (200µl IOMERON 350). Patterns and properties of tumor vessels obviously favored focal accumulation of iodine and resulted in ~60HU extra absorption inside the tumor versus healthy brain. This imaging protocol was then used to test whether an anti-angiogenic treatment already used for colorectal and lung cancers (Avastin) would also reduce the tumor burden in glioblastoma. Avastin was delivered to two groups of mice at two different stages of the disease, namely 2 or 4 weeks after tumor implantation: during 3 weeks mice received twice a week intravenous injection of Avastin diluted in PBS (10mg/kg) or the same volume of PBS alone. Tomographic reconstructions of their brain were performed at the end of the treatment protocol in the presence of iodinated contrast agent. In control groups, tumors could be detected in most of the mice (80%, n=5 and 75%, n=4) whereas these percentages significantly decreased to 40% (n=5) and 25% (n= 8) for the groups receiving late and early Avastin treatment, respectively. These results could either support a significant tumor size reduction by Avastin or an Avastin-induced change in vessel permeability preventing iodine accumulation hence imaging sensitivity. To decide between these two alternatives we used additional imaging modalities. As implanted glioblastoma cells had been transfected with a GFP expressing vector, we performed on the same animals optical epifluorescence imaging and in vivo two-photon microscopy in addition to the CT scan. Similar millimetric green tumors were observed in both treated and non treated mice suggesting that Avastin mainly modulates vessels' permeability. The sum of information given by these 3 complementary techniques are instrumental for a better knowledge of glioblastoma physiopathology and their treatment.