Figure 1.
Differential but partially overlapping roles of HIF-1α and HIF-2α in retinal vascular development.
A to L. Confocal images of flat-mounted retinas stained with IB4-Alexa 594. A to C and G to I are representative images from areas midway between the optic nerve head and periphery; D to F and J to L each displays one of the 4 pedals of flat-mounted retinas. Images were tiled from multiple panels due to the size of retinas. Genotypes at the Hif-α and Rosa26 loci are indicated on top and to the left of the images, respectively. For example, panels G and J are Hif-1αf/f/Rosa26CreERT2. All mice were treated with tamoxifen by daily oral gavage at P1 through P3 (40 mg/kg in corn oil). At P8, retinas were dissected and stained with IB4-Alexa 594. Note that vascular development was reduced in H and K relative to G and J, respective, and further reduced in I and L. Hyaloid vessels were removed during dissection. Scale bars: A and G, 100 µm; D and J, 500 µm. M. Quantification of vascular branches. LoxP modifications at Hif-1α and Hif-2α loci are indicated below the bar graph, and presence or absence of Rosa26CreERT2 (R26CreERT2) is indicated by solid or open bars, respectively. Quantification was carried out with the assistance of NIH ImageJ. To average out local variations, vascular branches were counted in 3 rectangular areas (0.2×0.5 mm2) areas midway between the optic nerve head and the periphery, with an example shown in D. Where a single pedal was not cut wide enough to encompass all three rectangular areas, the third area was counted in an adjacent pedal. Average values from three areas were used as one data point for statistical analysis. n = 6. * p<0.05; ** p<0.01; *** p<0.001.
Figure 2.
Apparently normal development of the primary layer of retinal vasculature in Hif-1αf/f/Tie2Cre and Hif-2αf/f/Tie2Cre mice.
A and B. Confocal images of flat-mounted P7 retinas from tdTomato reporter mice carrying Tie2Cre transgene. B is expanded from the red rectangle in A. Clearly visible vascular patterns indicate that Tie2Cre is fully functional in retinal blood vessels. C to F. P8 retinas stained with IB4-Alexa 594. Areas in red rectangles are shown at higher magnifications below each source image. Images are representative of 4 to 6 mice in each group. Scale bars are 400 µm in A, 125 µm in B, and 500 µm in C–F.
Figure 3.
Reduced retinal astrocyte development in Hif-2αf/f/Rosa26CreERT2 mice.
A to L. Confocal images showing retinal astrocytic networks of P8 neonatal mice. All mice were treated with tamoxifen by daily oral gavage at P1 through P3. At P8, retinas were dissected, fixed, and stained with rabbit anti-GFAP followed by goat anti-rabbit IgG-Alexa 488. Boxed areas in A–C and G–I are shown at higher magnifications below each source image. Scale bars are 500 µm in A and G and 100 µm in D and J. M. Percentage of retinal area occupied by GFAP+ cells. Quantification was carried out midway between the optic nerve and periphery (white boxes). Equivalent areas in three different pedals were counted for each retina, and the average value was used as one data point. Hif-1a and Hif-2a alleles are indicated below each bar; presence of R26CreERT2 is indicated by solid bars. n = 6; * p<0.05; *** p<0.001.
Figure 4.
Dramatically reduced retinal vascular development in Hif-2αf/f/GFAPCre mice.
All retinas were dissected at P8, fixed and flat-mounted, and stained with IB4-Alexa 594. Development of the primary retinal vascular beds was analyzed by confocal microscopy. A and B. Hif-1αf/f and Hif-1αf/f/GFAPCre mice. C to F. GFAPCre (C), Hif-2αf/f (E), and Hif-2αf/f/GFAPCre (D and F) mice. Areas in red rectangles were expanded and shown below each main image. Out of 22 Hif-2αf/f/GFAPCre mice analyzed, 2 had remnant amounts of blood vessels, as shown in D, whereas the rest were completely devoid of any vascular development (F). Hyaloid vessels were removed during dissection. Scale bar represents 500 µm, and all main images are in the same magnification scale.
Figure 5.
Precocious but tapered development of the astrocytic network in Hif-2αf/f/GFAPCre mice.
Astrocyte development was examined by anti-GFAP IF staining. A and B. Anti-GFAP staining at P0. In A, GFAP+ cells are mostly limited to the optic nerve head (lower left). In Hif-2αf/f/GFAPCre mice, GFAP+ structures extended much further. White lines demarcate the borderline between strongly GFAP+ and the rest of retinal areas. Red lines mark the approximate average position of the white line. Boxed areas are shown in higher magnifications in the insets. C and D. Anti-GFAP staining at P3. Upper and lower insets are expanded from corresponding boxes. E to H. Anti-GFAP stained retinas at P8 shown at low (E and F) and high (G and H) magnifications. Scale bars, 500 µm in A to F, 50 µm in G and H. I. Percentage (%) of GFAP+ areas. Quantifications were carried out in white boxes in A, B, E and F, and lower boxes in C and D. Three such areas were quantified per mouse and average values were used as one data point. J. Quantification of GFAP+ wavefronts to the optic nerve, measured as distances between red curves and optic nerves. n = 5. * p<0.05, ** p<0.01, *** p<0.001.
Figure 6.
Reduced number of PDGFRα+ astrocytes in Hif-2αf/f/GFAPCre mice.
A to D. Retinas dissected at P0 and P3 were subject to anti-PDGFRα IF staining. Wavefronts of PDGFRα+ astrocytes are marked by white lines. Red lines represent the approximate average positions of the zigzag white lines. In C, no red line is provided because the front of PDGFRα+ astrocyte network was relatively even. Areas indicated by white boxes were quantified for percentage of PDGFRα+ tissues. E. Distance from optic nerve head to the front of PDGFRα+ areas (white line in C, or red lines in A, B, and D). F. Percentage of PDGFRα+ retinal areas. Quantifications were carried for areas indicated by white boxes, taking the average of three such areas from the same mouse as one data point. Scale bars are 500 µm. n = 4. ** p<0.01.
Figure 7.
Depletion of Pax2+ astrocyte progenitors in Hif-2αf/f/GFAPCre mice.
A to D. Anti-Pax2 IF staining of retinas at P0 (A and B) and P4 (C and D). The abundance of Pax2+ cells was dramatically reduced in Hif-2αf/f/GFAPCre mice at both P0 and P4. Scale bars are 500 µm. E to J. Double IF staining of P0 retinas with rabbit anti-Pax2 and rat anti-GFAP, followed by goat anti-rabbit IgG-Alexa 488 and donkey anti-rat IgG-Cy3. In Hif-2αf/f retinas, Pax2+ (green) are abundantly present in the vicinity of strongly GFAP+ (red) optic nerve head, but mature GFAP+ astrocytes are virtually absent. In Hif-2αf/f/GFAPCre mice, Pax2+ cells less abundantly present off the optic nerve, accompanied by many GFAP+ cells. Scale bars in E to J are 200 µm. K. Percentage (%) of retinal areas occupied by Pax2+ cells. n = 5. * p<0.05, *** p<0.001.
Figure 8.
Proliferation and apoptosis assays.
A to H. Proliferation assays. Mice were injected with BrdU at P3, and retinas were dissected after an hour. BrdU incorporation in astrocyte progenitors was detected by anti-BrdU and anti-Pax2 double IF staining. A to F show representative images from areas indicated in G and H. Both Pax2+ and BrdU+ cells were counted with the assistance of NIH ImageJ program. I. Proliferation index was calculated as % of Pax2+ cells that were also BrdU+. Cre refers to GFAPCre. n = 4. No significant difference was found between Hif-2αf/f and Hif-2αf/f/GFAPCre mice. J to L. Apoptosis assay by anti-active Caspase 3 IF staining. J and K are P3 retinas. No apoptotic cells were detected in either Hif-2αf/f or Hif-2αf/f/GFAPCre mice. L is from mice treated with 75% oxygen for 16 hours between P7 and P8. The same anti-active Caspase 3 staining procedure detected large numbers of apoptotic cells in oxygen-treated mice. Images are representative of data from 4 mice in each group. Scale bars are 50 µm for A to F, 500 µm for G and H, and 50 µm for J to L.
Figure 9.
Model for HIF-2α regulated retinal astrocyte differentiation.
In wild-type mice, physiological level of HIF-2α may partially suppress differentiation of mature astrocytes (green stars) from their progenitors (orange ovals). Thus, while a proportion of astrocyte progenitors differentiate into non-proliferative mature astrocytes, the rest may undergo active proliferation to replenish progenitor populations. HIF-2α deficiency disturbs the balance between progenitor proliferation and astrocyte differentiation. Accelerated astrocyte differentiation may cause rapid loss of astrocyte progenitors due to inadequate time for proliferation, leading to unsustainable astrocyte development.