Blood vessel formation requires the integrated regulation of endothelial cell proliferation and branching morphogenesis, but how this coordinated regulation is achieved is not well understood. Flt-1 (vascular endothelial growth factor [VEGF] receptor 1) is a high affinity VEGF-A receptor whose loss leads to vessel overgrowth and dysmorphogenesis. We examined the ability of Flt-1 isoform transgenes to rescue the vascular development of embryonic stem cell–derived flt-1−/− mutant vessels. Endothelial proliferation was equivalently rescued by both soluble (sFlt-1) and membrane-tethered (mFlt-1) isoforms, but only sFlt-1 rescued vessel branching. Flk-1 Tyr-1173 phosphorylation was increased in flt-1−/− mutant vessels and partially rescued by the Flt-1 isoform transgenes. sFlt-1–rescued vessels exhibited more heterogeneous levels of pFlk than did mFlt-1–rescued vessels, and reporter gene expression from the flt-1 locus was also heterogeneous in developing vessels. Our data support a model whereby sFlt-1 protein is more efficient than mFlt-1 at amplifying initial expression differences, and these amplified differences set up local discontinuities in VEGF-A ligand availability that are important for proper vessel branching.
Pathways that shape the first blood vessels to form during development are used subsequently for physiological and pathological angiogenesis. The VEGF-A signaling pathway affects blood vessel formation in complex ways. Our work focuses on how the flt-1 (VEGFR-1) receptor impacts vascular development. We showed that flt-1 negatively regulates endothelial proliferation, while it positively modulates sprouting morphogenesis. To further define the mechanism(s) of flt-1 action, ES cell-derived flt-1 mutant vessels were genetically rescued with transgenes that expressed individual flt-1 isoforms in developing vessels. Flt-1 mutant clones with a randomly integrated membrane-tethered PECAM promoter-driven mflt-1 transgene showed partial rescue of endothelial proliferation and branching morphogenesis. Similar clones expressing a soluble form of flt-1 (sflt-1) rescued proliferation to similar levels, but they were significantly more efficient in the rescue of branching morphogenesis. We then generated flt-1 mutant rescue clones in which the transgene was targeted to the ROSA locus, to overcome transgene insertion site expression differences. Comparison of these clones showed that both transgenes rescued proliferation to similar levels, while sflt-1 showed increased rescue of vessel branching. Finally, a form of mflt-1 lacking the cytoplasmic signaling domain showed similar rescue profiles to the mflt transgene in both assays. These data support a model in which flt-1 acts as a ligand sink in development, and indicate that the spatial presentation of VEGF-A to developing is modulated by interactions with the soluble form of the receptor.
Macrophages are phagocytic cells that help to mediate the inflammatory response, and they also associate with tissue remodelling during mammalian development. Macrophages first develop in the yolk sac at day 10 of mouse embryogenesis, then from the aorta, gonads and mesonephros (AGM) region a day later. Both of these early haematopoietic centres contribute progenitor cells to the fetal liver; however, yolk sac-derived cells are not capable of definitive haematopoiesis. This difference raises the possibility that yolk sac macrophages have unique molecular requirements. Colony-stimulating factor 1 (csf1), a cytokine required for the differentiation, proliferation and survival of most macrophages, signals through a high-affinity receptor, csf1r, that is expressed on macrophages and their precursors. Csf1-null mice (csf1op/csf1op) are viable but have multiple abnormalities that are thought to be direct or indirect effects of reduced macrophage numbers, since mutant mice have only 5–30% of normal macrophage numbers as adults (Wiktor-Jedrzejczak et al, 1990). Most aspects of the phenotype can be rescued by injection of exogenous csf1 or by inserting a CSF1 transgene (Cecchini et al, 1994; Ryan et al, 2001). Although the postnatal effects of the csf1op mutation are well-understood, the effects on early embryonic development are less clear. Female reproductive defects hinder analysis of prenatal development in the absence of csf1, since mutant embryos generally develop in heterozygous females that provide maternal csf1 (Pollard et al, 1991). However, analysis of the csf1op mutation in late gestation and early postnatal mouse development from csf1-null mothers confirmed that some macrophages develop in the absence of csf1(Roth et al, 1998). The importance of csf1 signalling at the first site of embryonic macrophage development, the yolk sac, has not been investigated. To study the effects of loss of csf1 on embryonic macrophages without the complicating effects of maternally-derived csf1, embryonic stem (ES) cell lines were isolated from blastocysts of csf1op/+ heterozygous mice intercrosses. One wild-type line, ES 4·5, one heterozygous line (csf1op/+), ES 3·9, and two homozygous (csf1op/csf1op) lines, ES 3·4–5 and 3·6, were generated. Differentiation of mouse ES cells recapitulates aspects of early yolk sac development (Bautch, 2001). During ES cell differentiation, a primitive vasculature and several haematopoietic lineages form. Specifically, primitive (nucleated) erythrocytes and embryonic macrophages differentiate from blood islands under non-supplemented ES cell culture conditions. ES cell-derived embryonic macrophages appear identical to yolk sac-derived embryonic macrophages (Inamdar et al, 1997). The ES cell lines were differentiated to day 10, which is the peak of macrophage production. Wild-type, heterozygous and homozygous mutant ES cells cultures were fixed and stained with antibodies to Mac-1 or F4/80 to visualize macrophages and platelet/endothelial cell adhesion molecule 1 (pecam1) to visualize the vasculature, or reacted with benzidine to visualize haemoglobinized erythrocytes (Fig 1). The number of Mac-1 positive cells was dramatically decreased in the homozygous mutant lines compared to wild-type or heterozygous lines (compare Fig 1C to panels A and B). Cultures stained with a second macrophage-specific antibody, F4/80, showed a similar reduction of F4/80 positive cells in the mutant ES lines (Fig 1D and E). The quantitative reduction in macrophages was estimated by fluorescence-activated cell sorting analysis of day 10 cultures stained with Mac-1. On average, 3% of the total cell counts from wild-type cultures and 5% of heterozygous (csf1op/+) cultures were Mac-1 positive. Both csf1op/csf1op mutant ES cell lines had significantly fewer Mac-1 positive cells than heterozygous or wild-type cultures, with each line containing <0·5% Mac-1 positive cells. These results establish that embryonic macrophages require csf1 for expansion and survival from their earliest development in the yolk sac. This finding is consistent with the expression pattern of the only known csf1receptor (csf1r), which is expressed on early embryonic macrophages (unpublished observations). csf1op/csf1op ES cell cultures have few macrophages. ES cells were differentiated in vitro to day 10 under non-supplemented culture conditions as described (Bautch, 2001). Paraformaldehyde-fixed cultures were reacted with antibodies as described previously (Inamdar et al, 1997; Bautch, 2001). Panels A–C, double label confocal images with Mac-1 (red) and pecam (green); panels D-E, double label epifluorescence images with F4/80 (red) and pecam (green); panels F and G were reacted with benzidine to reveal primitive erythrocytes (blue). A, D, F Wild-type cultures; B, csf1op/+ culture; C, E, G csf1op/csf1op cultures. Note the Mac-1+ cells (arrows A–C), and the F4/80+ cells (arrows D) that are rare or absent in the csf1op/csf1op mutant background. It was found that macrophage development from the AGM region of day 11·5 csf1op/csf1op mutant embryos was compromised (Minehata et al, 2002), so it is likely that both yolk sac-derived and AGM-derived macrophages require csf1 for their development. However, it was also found that precursors of other haematopoietic cells and endothelial differentiation were affected by csf1. To determine whether loss of csf1 led to reduction of multiple haematopoietic lineages, the ES cultures were reacted with benzidine to visualize primitive erythrocytes. Benzidine staining of homozygous mutant cultures was similar to that of wild-type cultures, indicating that primitive erythrocytes develop in the absence of csf1 (Fig 1F and G). Close visual inspection also revealed no significant difference in the vasculature among the different genotypes (Fig 1A–E). This apparent discrepancy in results could reflect intrinsic differences between yolk sac- and AGM-derived cells, or it could reflect different culture and assay conditions. However, our results correlate with the restriction of csf1r expression to individual cells with macrophage-like morphology in the ES cell cultures. To determine when embryonic macrophage development was compromised by the lack of csf1 in ES cell cultures, RNA expression of macrophage markers was analysed using reverse transcription polymerase chain reaction (Fig 2). This semi-quantitative assay showed that the early marker myeloperoxidase and the later marker integrin alpha M (also known as CD11b) both had weak or absent amplification signals in mutant cultures. In contrast, the expression of integrin beta 2 (also known as CD18), the beta subunit of Mac-1 that is also associated with other integrin ligands, was not down-regulated in the csf1op/csf1opmutant ES cell cultures. These results suggest that embryonic macrophage development is affected at relatively early stages of differentiation and maturation by the lack of csf1. SFFV proviral integration 1 (Sfpi1, also known as PU.1) is a transcription factor required for the maturation of several haematopoietic lineages including macrophages, and targeted mutation of SFPI1 also aborts embryonic macrophage development at a similar developmental stage (Olson et al, 1995). This is expected given that sfpi1 is thought to regulate CSF1R expression. However, unlike the csf1op/csf1op ES cell cultures that have rare Mac-1 positive cells (Fig 2C), differentiated SFPI1−/− ES cells completely lack Mac-1- and F4/80-expressing macrophages [(Olson et al, 1995) and unpublished observations]. This result suggests that csf1 mediates only a subset of the effects of sfpi1 in the embryonic macrophage lineage. Reverse transcription-polymerase chain reaction (RT-PCR) analysis of day 10 ES cultures and controls. Total RNA was isolated from each line and RT-PCR was performed as described (Olson et al, 1995) for 35 cycles, and the products were separated on an 8% acrylamide gel and visualized using ethidium bromide. The genes analysed are on the left, and the size of the PCR products is on the right. Lane 1, no DNA; Lane 2, RAW 264·7 cells; Lane 3, Py-4-1 cells; Lane 4, Py-YSA cells; Lane 5, day 10 line 3.6 ES cells (csf1op/csf1op); Lane 6, day 10 line 3·4–5 ES cells (csf1op/csf1op); Lane 7, day 10 line RI (+/+) ES cells. MPO, myeloperoxidase; ITGAM, integrin alpha M; ITGB2, integrin beta 2; ACTB, beta-actin. Others have shown that csf1 is required for adult and late-gestational embryonic macrophage development. The reduction of early macrophage markers in mutant ES cell cultures indicates that the earliest stages of embryonic macrophage development also require csf1; therefore, we conclude that csf1 is critical for macrophage development and survival throughout life. Furthermore, with the increasing interest in in vitro stem cell systems, a better understanding of the requirements for macrophage development from ES cells can provide a basis for development of haematopoietic therapeutics. This work was supported by NIH grant HL43174 to V.L.B.
