Fig 1.
The CD34+ fraction of HMEC-1 cultures contains the VEGF-induced angiogenic sprouting activity.
(A, left panel) Representative images of spheroids that were generated from HMEC-1 were either unsorted or FACS-sorted, based on CD34 cell surface expression (CD34- and CD34+ populations). Spheroids were embedded in collagen gel supplemented with (+) or without (-) VEGF. Scale bar = 200 μm. (A, right panel) Flow cytometry dot plot demonstrating gating for HMEC-1 sorting based on CD34 expression. (B) The number of sprouts per spheroid and the mean sprout length were quantified using Image J. Error bars represent standard deviation. *Significantly different from unstimulated control (VEGF-) with P < 0.05.
Fig 2.
Silencing of CD34 expression in HMEC-1.
(A) Representative histograms showing the intensity of anti-CD34 (left panel) or anti-podocalyxin (right panel) labeling of parental HMEC-1 compared to an isotype control (grey peak). (B) Quantitative PCR (qPCR) analysis of CD34 and PODXL mRNA levels and (C) flow cytometric analysis of CD34 protein expression on the membrane of HMEC-1 72 h after transfection with CD34-targeting (siCD34) or non-targeting (NT) (siNT) siRNA. Stimulation with VEGF induced CD34 plasma membrane expression in both siNT-treated and siCD34 treated HMEC-1. However, CD34 expression in siCD34-treated HMEC-1 was less than 30% as compared to expression levels in siNT-treated cells. *Significantly different from non-stimulated cells with P < 0.05.
Fig 3.
Effect of CD34 silencing on sprouting, cell migration and invasion.
(A, left panel) Representative images of spheroids that were generated from HMEC-1 transfected with either a non-targeting siRNA (siNT) or siCD34 and subsequently embedded in collagen gel in the presence or absence of VEGF-A. (A, right panel) Spheroids were analyzed at 24 h after embedding and the number of sprouts per spheroid and average sprout length were quantified using Image J. Results were expressed relative to values of siNT transfected cells without VEGF. (B, left panel) siNT- and siCD34-transfected HMEC-1 were grown until confluent. Scratches were made using a pipette tip and images were taken at 0 and 6 h after scratching. (B, right panel) The percentage of width of the scratch filled with cells was quantified over time. (C, left panel) Representative images of HMEC-1 transfected with siNT or siCD34, located at the lower side of the Boyden filter after invasion through Matrigel visualized by DNA staining (i,iii) or Giemsa (ii,iv) staining at 20 h after seeding. (C, right panel) Cells invading the Matrigel were quantified by counting the number of nuclei per microscopic field. Error bars represent standard deviation, except for Fig 3C (right panel) where they represent 95% confidence interval; *Significantly different from siNT control with P < 0.05.
Fig 4.
Endothelial tip cell filopodia in developing retinal vessels express CD34 but not podocalyxin.
Immunofluorescence staining of retinal whole-mounts of wild type (Wt) mice at P5. Both CD34 and podocalyxin (PODXL) are expressed on developing vasculature. CD34 but not podocalyxin is expressed on endothelial tip cell filopodia. Isolectin B4 staining (green) was used to visualize tip cell filopodia. Scale bar = 10 μm.
Fig 5.
CD34 deletion reduces the number of endothelial tip cell filopodia.
(A) Whole-mount isolectin B4 staining showing the angiogenic front of wild type (Wt) (i,iii) and Cd34-/-(ii,iv) mouse retinas at P5. Scale bar (i,ii) = 20 μm. Tip cell filopodia were marked (green dots) and normalized to vessel length (red lines, see insets iii, iv). (A, right panel) Quantification of filopodia per 100 μm vessel length. Error bars represent standard errors of the mean. * Significantly different from wild type with P < 0.05.
Fig 6.
Retinal vessels develop normally in Cd34-/- mice under normoxic conditions.
(A) Isolectin B4 staining (green) of retinal whole-mounts of wild type (Wt) and Cd34-/- mice under normoxic conditions at P25. The plexuses of retinal vessels in both the inner and outer retina are similar in Wt and Cd34-/- mice at P25. Scale bar = 50 μm. (B) Immunofluorescence staining of podocalyxin (PODXL, red) and isolectin B4 (green) of whole-mounts of retinas harvested from wild type (Wt) and Cd34-/- mice at P5. The arrows in the merge (yellow) indicate tip cell filopodia that stain with isolectin B4 but do not co-express podocalyxin. Podocalyxin is not expressed on endothelial tip cell filopodia in either wild type (Wt) or Cd34-/- mice. Scale bars = 100 μm (sprout front) or 25 μm (tip cell).
Fig 7.
Loss of CD34 limits formation of pathological neovascularization in oxygen-induced retinopathy (OIR).
(A) Representative retinal images showing total retinal area and regions of vaso-obliteration (VO) in OIR model for wild type (Wt) and Cd34-/- mice at P12 and P17 and neovascularization (NV) at P17. Scale bar = 50 μm (B) Ratios of VO to total retinal area indicated no significant difference in VO between wild type and Cd34-/- mice. (C) At P17, neovascularization was significantly decreased in Cd34-/- mice as compared to wild type mice (*P < 0.01). Error bars represent standard deviation. * Significantly different from wild type with P < 0.05. (D) Isolectin B4 labeling of retinas harvested at P25 after OIR revealed that epi-retinal tufts in wild type mice aggregated as large continuous areas of neovascularization whereas in Cd34-/- mice, the considerably smaller tufts did not aggregate. Scale bar = 50 μm.