Müller glia provide essential tensile strength to the developing retina

Abstract

To investigate the cellular basis of tissue integrity in a vertebrate central nervous system (CNS) tissue, we eliminated Müller glial cells (MG) from the zebrafish retina. For well over a century, glial cells have been ascribed a mechanical role in the support of neural tissues, yet this idea has not been specifically tested in vivo. We report here that retinas devoid of MG rip apart, a defect known as retinoschisis. Using atomic force microscopy, we show that retinas without MG have decreased resistance to tensile stress and are softer than controls. Laser ablation of MG processes showed that these cells are under tension in the tissue. Thus, we propose that MG act like springs that hold the neural retina together, finally confirming an active mechanical role of glial cells in the CNS.

Figure 1.

Notch signaling is active in the nascent and mature MG in the developing retina. (A and A’) MG span the entirety of the three neural layers from the inner limiting membrane (ILM) to the outer nuclear layer (OLM). (B) Using Cralbp as an early marker for MG specification in cryosections, we see expression on the apical side of the retina at 48 hpf. (C and D) This expression increases in the INL by 60 hpf (C), and was strongly expressed at 72 hpf in mature MG (D). (E) In whole-mount samples at 35 hpf, many cells within the nascent retina have Notch activity (green), especially on the apical side of the retina (arrowheads). (F) Notch activity at 55 hpf appears in radial cells on the apical side of the retina (arrowheads). (G) Notch activity is restricted to MG at 72 hpf. (H–J) The Notch-positive cells in the mature retina are MG, as shown by the overlap in the TP1:Venus and GFAP:dTomato transgenic constructs (arrows). (K) The transplantation strategy to generate clones of TP1:Venus (green) embryos into the ath5:gapRFP host (magenta). (L–P) Panels from a time-lapse movie of a clone in a living zebrafish retina (n = 6). (L) Notch is active in multiple cells in the clone at 38 hpf. Notch activity is progressively restricted until a single cell maintains Notch on the apical side of the retina at 42 hpf (M). Notch activity increases in this cell at 48 hpf (N), and at 55 hpf (O) it migrates basally and will position itself within the INL and begin to mature by 62 hpf (P). Figures always show the apical side of the retina at the top and basal at the bottom. OPL, outer plexiform layer; IPL, inner plexiform layer. Bars, 10 µm.

Figure 2.

Blocking Notch activity during retinogenesis results in a loss of MG. (A and B) Treatment with DAPT at 45 hpf results in the loss of Notch activity (green) in whole-mount retinas at 72 hpf. (C and D) In cryosections, MG specifically express glutamine synthetase (GS), but after DAPT treatment the retina is completely lacking in GS expression. (E and F) DAPT treatment of double transgenic fish, ath5:gapRFP;GFAP:GFP, results in a complete loss of GFP-expressing MG in the retina of the transgenic line GFAP:GFP. (G) Treatment with DAPT at 50 hpf is too late to completely block MG specification in the central retina (arrow). G′ shows the boxed region. (H and I) There are reduced GFAP:GFP-positive cells in the central retina after Su(H) morpholino injection or treatment with Compound E, another γ-secretase inhibitor, at 96 hpf, similar to DAPT treatments. (J and K) Retinal neurons are specified and patterned properly in the SoFa fish. (L and M) There is no increase in cell death (TUNEL+ cells) in the absence of MG (mean + SEM, n = 10 retinas). Bars: (A–D) 25 µm; (E–G) 20 µm; (G’) 10 µm; (H and I) 30 µm; (J and K) 20 µm; (L) 50 µm.

Figure 3.

Retinas lacking MG show intraretinal tears or “retinoschisis” in the GCL. Confocal images of the living fish retina in vivo. The retina is double transgenic for both vsx1:GFP, labeling most bipolar cells (green), and ath5:gapRFP, strongly labeling retinal ganglion cells in the GCL (red). (A–D) At 72 hpf (A and B) and 96 hpf (C and D), DAPT-treated retinas lacking MG have a specific ripping in the GCL (dashed line) along the basal surface of the retina (white solid line). (E) The Su(H) morphant retina does not have MG, as shown by the loss of GFAP:GFP-positive cells. (F) We did not observe any ripping within the retina of control fish. However, ripping in the GCL occurs in similar proportions in Su(H) morphant and DAPT-treated retinas (n = 50 retinas for each condition). (G and H) Electron microscopy shows local ripping within the GCL along the basal-most portion of the retina (arrowheads). Bars, 20 µm.

Figure 4.

MG act as springs in the retina to hold the neural retina together. (A) To test the biophysical properties of retinas lacking MG, we used AFM to pull on the tissue. (B) Tissue deformation was not significantly different for measurements at 33 hpf or 48 hpf (mean ± SEM [error bars], n = 10 retinas). (C) At 72 hpf, retinas deformed significantly more when no MG were present in the tissue than controls (mean ± SEM [error bars], n = 10 retinas). (D) To test if MG are under strain within the retinal tissue, we applied a laser ablation technique to specifically target the apical or basal processes of GFAP:GFP MG in vivo. Ablating the apical process resulted in its retraction of the process and displacement of the cell body basally to the IPL. Ablation of the basal process resulted in its rapid retraction and in cell displacement apically in the INL. The merge is the overlay between each time point. Red lines delineate the regions of ablation. (E and F) Confocal images showing the expanded GCL labeled in the ath5:gapRFP;vsx1:GFP transgenic fish after DAPT treatment. (G) In the absence of MG, we noted a significant increase specifically in the thickness of the GCL (n = 20 retinas, mean + SEM [error bars], P < 0.01). (H) Stiffness measurements at the time of treatment (∼45 hpf), 72 hpf, and 96 hpf of control and DAPT-treated retinae (n = 20 retinas, mean + SEM [error bars], P < 0.01). IPL, inner plexiform layer; ONL, outer nuclear layer. Bars, 15 µm.

Figure 5.

Our model: Retinal progenitors behave as springs in the nascent tissue as they are radial in nature and support the tissue. As they are depleted in the developing retina, the spring function is passed on to MG in the mature retina. However, the retina lacking MG does not have these “springs.” When under mechanical stress, retinae lacking MG will deform and the tissue will rip, releasing the tension.