Erythro-myeloid progenitors can differentiate from endothelial cells and modulate embryonic vascular remodeling

Erythro-myeloid progenitors (EMPs) were recently described to arise from the yolk sac endothelium, just prior to vascular remodeling, and are the source of adult/post-natal tissue resident macrophages. Questions remain, however, concerning whether EMPs differentiate directly from the endothelium or merely pass through. We provide the first evidence in vivo that EMPs can emerge directly from endothelial cells (ECs) and demonstrate a role for these cells in vascular development. We find that EMPs express most EC markers but late EMPs and EMP-derived cells do not take up acetylated low-density lipoprotein (AcLDL), as ECs do. When the endothelium is labelled with AcLDL before EMPs differentiate, EMPs and EMP-derived cells arise that are AcLDL+. If AcLDL is injected after the onset of EMP differentiation, however, the majority of EMP-derived cells are not double labelled. We find that cell division precedes entry of EMPs into circulation, and that blood flow facilitates the transition of EMPs from the endothelium into circulation in a nitric oxide-dependent manner. In gain-of-function studies, we inject the CSF1-Fc ligand in embryos and found that this increases the number of CSF1R+ cells, which localize to the venous plexus and significantly disrupt venous remodeling. This is the first study to definitively establish that EMPs arise from the endothelium in vivo and show a role for early myeloid cells in vascular development.


SUPPLEMENTAL MATERIAL
Video S1 -Time lapse of Csf1r-derived cell division and entry into embryonic circulation.
Csf1r MeriCreMer Rosa26 tdT embryos were pulsed with tamoxifen at E7.75, harvested at E8.5 and injected with Alexafluor488 conjugated AcLDL to label endothelial cells (green). Embryos were cultured overnight in the presence of 10 µM 4-hydroxytamoxifen and time-lapsed the following morning on an environment-controlled microscope stage. Confocal laser scanning microscope was used to take an image every 6 minutes for a total time (t) of 10 hours. In the middle of the frame, a tdT + cell (white arrow) appears with an endothelial phenotype in the vessel wall, then rounds and buds up from the endothelium, and begins to divide (t=10.1 hr). In the top right of the frame, a tdT + cell divides into two cells (yellow arrowheads), and one of the daughter cell enters into circulation. Scale bar: 50µm.
Video S2 -Csf1r-derived cell circulate but also patrol in the vessel wall.
Csf1r MeriCreMer Rosa26 tdT embryos were pulsed with tamoxifen at E7.75, harvested at E9.0 and injected with Alexafluor488 conjugated AcLDL to label endothelial cells (green). Confocal laser scanning microscope was used to take an image every 6 minutes. An EMP (red) intravasates through the vessel wall (green) and enters circulation (white arrow). The cell is observed moving within the vessels at speeds much lower than surrounding circulating cells that can be detected as red streaks. Scale bar: 25µm.
Video S3 -Csf1r-derived cell patrolling in the vessel wall.
Csf1r MeriCreMer Rosa26 tdT embryos were pulsed with tamoxifen at E7.75, harvested at E9.0 and injected with Alexafluor488 conjugated AcLDL to label endothelial cells (green). Confocal laser scanning microscope was used to take an image every 6 minutes. An EMP (red) is observed moving within the vessels. Cells in circulation appear as streaks in single frames due to their high velocity. In the last frames, the EMP disappears, possibly through circulation or by extravasating out of the plane of focus. Scale bar: 25µm. Figure S1 -Yolk sac-derived EMPs contribute to neonatal tissue resident macrophages and vascularizing retinas, where they home to regions of angiogenic sprouts, remodeling and anastomosis. Csf1r MeriCreMer Rosa26 tdT embryos were pulse with tamoxifen at E7.75, such that all yolk sac derived EMPs are labeled with tdTomato throughout fetal and neonatal development. Organs were harvested from 6-day-old pups (P6). Yolk sac-derived tdT + cells (red) were present in all organs harvested including the spleen, lungs, liver, intestine, and retinas (A to M). Developing P6 retinas were counterstained with IsolectinB4 to visualize blood vessels (J-M, green). tdT + cells were widely distributed throughout the retina (D-E), and associated with the radial periphery, a site of continuous angiogenesis (J, white arrow); anastomosing loops (K, arrows) and angiogenic sprouts (L). In the more mature regions of the network, tdT + cells often Supplemental Figure 3 -Representative scatter plots of flow cytometry from E8.5 isolated yolk sacs. Yolk sacs were isolated from embryos (in this case, 10 somites), treated with Collagenase (0.5mg/mL) and DNAseI (100µg/mL) for 35 min, mechanically dissociated, blocked and stained with annotated antibodies. Either 7AAD PerCP (shown here) or EF780 viability dye was used to exclude cell debris and non-viable cells. SSC-A vs FSC-A scatterplot is representative of all viable, single cell population derived from E8.5 yolk sacs. All experiments, controls and compensation were performed using either FACS Canto II or FACS AriaIII analyzer. All flow cytometry analyses of yolk sacs were performed in a similar manner.
Supplemental Figure 5 -Two phenotypes are present at 10 somites depending on location within the yolk sac. Csf1r iCre Rosa26 tdT embryos were harvested at E8.5 (10 somites). Yolk sacs were immunostained for PECAM-1. tdT + cells located in the blood islands were for the most part rounded in morphology, whereas tdT + cells that developed in the vascular plexus distal to the blood islands had elongated endothelial morphology and co-localized with the vessel wall. Scale bar: 100 µm. In early vascular developmental, a spatial separation of arterial and venous blood vessels is present within the yolk sac. At E9.5, the arterial plexus has large branched vessels that are visible in brightfield images whereas in the venous regions, vessels are smaller (A). The difference can be more easily identified after immunostaining for an endothelial cell marker, in this case PECAM-1 (B). The large arterial vessels are separated by regions of honeycomb-shaped capillaries. The venous vasculature has smaller vessels where the majority of vessels align parallel to one another. The border between the regions is not exact based on these criteria; however, analysis was limited to regions well within either the arterial or venous plexus. Brightfield image taken from 1 . PECAM-1 image previously published in 2 .

Supplemental Figure 8 -Morphological measurements of arterial and venous plexus after CSF1-Fc injection.
Embryos were injected with PBS (control) or CSF1-Fc at E8.5 and cultured for 24 hours. Yolk sacs were then immunostained for CD31 and imaged. The density of branchpoints, average segment length, mean vessel thickness and the density of avascular holes was quantified using Biologic Image Analysis software. Arterial and venous vascular networks were identified based on location within the yolk sac. No differences in these metrics were observed between control and CSF1-Fc injected, or between the arterial and venous plexus.