Mobile zinc increases rapidly in the retina after optic nerve injury and regulates ganglion cell survival and optic nerve regeneration

Abstract

Retinal ganglion cells (RGCs), the projection neurons of the eye, cannot regenerate their axons once the optic nerve has been injured and soon begin to die. Whereas RGC death and regenerative failure are widely viewed as being cell-autonomous or influenced by various types of glia, we report here that the dysregulation of mobile zinc (Zn²⁺) in retinal interneurons is a primary factor. Within an hour after the optic nerve is injured, Zn²⁺ increases several-fold in retinal amacrine cell processes and continues to rise over the first day, then transfers slowly to RGCs via vesicular release. Zn²⁺ accumulation in amacrine cell processes involves the Zn²⁺ transporter protein ZnT-3, and deletion of slc30a3, the gene encoding ZnT-3, promotes RGC survival and axon regeneration. Intravitreal injection of Zn²⁺ chelators enables many RGCs to survive for months after nerve injury and regenerate axons, and enhances the prosurvival and regenerative effects of deleting the gene for phosphatase and tensin homolog (pten). Importantly, the therapeutic window for Zn²⁺ chelation extends for several days after nerve injury. These results show that retinal Zn²⁺ dysregulation is a major factor limiting the survival and regenerative capacity of injured RGCs, and point to Zn²⁺ chelation as a strategy to promote long-term RGC protection and enhance axon regeneration.

Keywords: amacrine cell; cell death; chelation; exocytosis; neuroprotection.

Fig. 1.

Optic nerve injury leads to rapid elevation of Zn²⁺ in the retina. (A) Low-magnification images of mouse retinas stained by Zn²⁺-selenite AMG from an untreated control mouse (Ctrl, PBS-injected) and 1 d post-NC (pNC). Toluidine blue-stained section at right shows retinal layers. Areas between black arrows show the IPL. (Scale bar, 50 µm.) This panel shows composites of multiple images taken at the same exposure and magnification spliced together. (B and C) Zn²⁺ accumulation visualized in retinal cross-sections by AMG (B) or the fluorescent Zn²⁺ sensor ZP-1 (C). (Scale bars, 25 µm.) Boxed areas: Cellular staining in the GCL. (Scale bar, 10 µm.) Signals are eliminated by the Zn²⁺ chelators TPEN and ZX1. (D–F) Quantitation of AMG staining in the IPL (D, n = 12, 9, 14, 9, 6, 6 retinas per group), positively stained cells in the GCL (E, n = 6, 6, 8, 9, 6, 6 retinas per group), and ZP-1 signal in the IPL (F, n = 7, 7, 5, 6, 7, 5, 6, 6 retinas per group). One-way ANOVA with Bonferroni post hoc tests. *P < 0.05, **P < 0.01, ***P < 0.001 compared with uncrushed controls; ^†P < 0.05, ^†††P < 0.001 compared with 6h pNC. INL, inner nuclear layer; ONL, outer nuclear layer; OPL, outer plexiform layer; RPE, retinal pigment epithelium; T, TPEN; Z, ZX1.

Fig. S1.

Rapid accumulation of Zn²⁺ in the IPL and delayed appearance in GCL somata. (A and B) Quantification of the AMG signal in the IPL using different developing times for AMG staining. Signals in the control retina and in the retina ipsilateral to NC both increased with longer developing periods (A) but the ratio between two groups remained nearly constant (B). n = 4 in each group. (C) Quantitation of the AMG signal in the IPL after NC (n = 12, 9, 8, 14, 7, 9, 6) and (D) quantitation of AMG⁺ cells in the GCL after NC (n = 6, 6, 6, 8, 6, 9, 6; one-way ANOVA with Bonferroni post hoc tests). *P < 0.05, **P < 0.01, ***P < 0.001 compared with uncrushed controls. (E and F) Quantitation of the AMG signal in the IPL (E) and of positively stained cells in the GCL (F) at the indicated time points after NC with 100 μM TPEN or ZX1. E, n = 6–14; F, n = 6–9. Two-way ANOVA with Bonferroni’s post hoc test, *P < 0.05, ***P < 0.001 compared with uncrushed controls; ^††P < 0.01, ^†††P < 0.001, decrease compared with NC-alone group at the same time-point. Values in A, C, and E are normalized to normal control retinas. All bars show mean ± SEM.

Fig. S2.

Changes of retinal Zn²⁺ after NC or intraocular injection of exogenous Zn²⁺. (A–C) Image (A) and quantification (B and C) of Zn²⁺ levels in the experimental retina ipsilateral to NC (left) and in the contralateral side (Ctrl, right). A small but significant increase in Zn²⁺ levels was detected in the IPL contralateral to the NC was observed 1 d pNC (A and B), but was no longer evident at 3 d in either the IPL (A and B) or GCL (A and C) (n = 5 retinas in each group; one-way ANOVA with Bonferroni post hoc tests). **P < 0.01, ***P < 0.001 compared with normal retinas; unpaired t test, ^†P < 0.05, ^††P < 0.01, ^†††P < 0.001, significance of decrease compared with experimental retinas ipsilateral to NC (Scale bar, 25 μm). (D–F) Image (D) and quantification (E and F) of AMG signal at 1 d and 3 d after intraocular injection of ZnCl₂ (100 μM, 1 mM). No change was observed in either the IPL (1 d, D and E) or GCL (3 d, D and F) compared with normal retina injected with saline (n = 8 retinas in each group). (Scale bar, 25 μm.) (G and H) Images (G) and quantitation (H) of flat-mounted retinas stained with a βIII-tubulin antibody to visualize surviving RGCs 2 wk after intraocular injection of ZnCl₂ (1 mM). ns, not significant; (n = 6 normal retinas, n = 8 retinas injected with ZnCl₂₎. (Scale bar, 50 μm.) Values in B and E are normalized to normal control retina. All graphs show mean ± SEM).

Fig. 2.

Zn²⁺ accumulation in the IPL and subsequent transfer to cells of the GCL: role of ZnT-3. (A) ZnT-3 immunostaining in retinas of slc30a3^+/+ and slc30a3^−/− littermates. (Scale bar, 25 μm.) (B) Quantitation of ZnT-3 immunostaining in the IPL before and after NC in wild-type mice with and without TPEN treatment (normalized to normal control; n = 10, 8, 8, 6). One-way ANOVA, **P < 0.01, ***P < 0.001 respectively compared with uncrushed controls; ^††P < 0.01, ^†††P < 0.001 compared with 1d pNC. (C) Quantitation of ZnT-3 expression in IPL of slc30a3^−/− retinas and slc30a3^+/+ littermates (normalized; n = 10, 8; 8, 6 retinas per group). Unpaired t test, ***P < 0.001 compared with slc30a3^+/+ littermate controls. (D–F) Images (D) and quantitation of AMG staining in IPL (E, n = 6 retinas per group) and GCL (F, cells per 14-μm section; n = 6 retinas per group) of slc30a3^−/− and slc30a3^+/+ littermates. Unpaired t test. **P < 0.01, ***P < 0.001 compared with slc30a3^+/+ littermate controls. (Scale bar, 25 μm.) (G–I) TeNT blocks vesicular release of Zn²⁺, causing continued Zn²⁺ build-up in the IPL and diminished accumulation in cells of the GCL. (G) Images show AMG staining 3 d after intraocular injection of TeNT (20 nM). Note reduced number of Zn²⁺-positive cells in the GCL. Deletion of slc30a3, the gene encoding ZnT-3, eliminates Zn²⁺ accumulation in IPL. (Scale bar, 50 μm.) (H and I) Quantitation of Zn²⁺-positive cells in the GCL (H, cells per 14-μm section) and intensity in the IPL (I, normalized). n = 6, 7 in H and n = 12, 4, 6, 7, 6, 6 in I. Unpaired t test, ***P < 0.001, comparison between indicated groups. All data represent mean ± SEM.

Fig. S3.

ZnT-3 expression and cellular colocalization. (A) Quantification of ZnT-3 intensity in the IPL of experimental retinas ipsilateral to the injured optic nerve (left) compared with the contralateral side (right). No increase was detected in the contralateral side at either 1 or 3 d pNC (n = 5 in each group; one-way ANOVA with Bonferroni post hoc tests.) ***P < 0.001 compared with normal eyes; unpaired t test, ^††P < 0.01, ^†††P < 0.001, decrease compared with experimental eyes. (B) Quantitation of ZnT-3 intensity in the IPL 1 d after intraocular injection of ZnCl₂ (100 μM, 1 mM). No change was seen compared with the normal retina (injected with 0.9% saline). n = 8 in each group. (C) Images of retinal cross-sections immunostained for ZnT-3 and βIII-tubulin (TUJ-1 antibody) to visualize ZnT-3 expression in RGCs 3 d post-NC; (Lower) enlargements of the framed areas. [Scale bars, 25 μm (lower magnification), 10 μm (higher magnification).] (D) Quantitation of ZnT-3 intensity in GCL 3 d post NC (n = 5 in each group; one-way ANOVA with Bonferroni post hoc tests). **P < 0.01 compared with normal retinas. (E–G) ZnT-3 colocalizes with amacrine cell markers in the IPL 1 d after NC. (E) Confocal images of retinal cross-sections show overlap of ZnT-3 with the amacrine cell markers GAD65/67 (E), but not the bipolar cell marker PKCα (F). (G) Quantitative analysis (Pearson’s r value) shows a significantly higher colocalization of ZnT-3 with GABAergic synapses (VGAT, GAD65/67) than either glutamatergic synapses (VGLUT1, PKCα) or the Müller cell marker CRALBP (as shown in Fig. 3; n = 10 retinas per group; one-way ANOVA, Bonferroni’s post hoc test), *P < 0.05, ***P < 0.001 compared with VGAT; ^†††P < 0.001 compared with GAD65/67. [Scale bars, 25 μm (A, lower magnification), 5 μm (A, higher magnification), 50 μm (B, lower magnification), 5 μm (B, higher magnification).] Values in A, B, and D are normalized to normal control retina. All bars show mean ± SEM.

Fig. 3.

ZnT-3 is localized in amacrine cell processes. (A–C) Confocal images through retinal cross-sections show a strong overlap of ZnT-3 with the amacrine cell marker VGAT (A), but much less overlap with the bipolar cell marker VGLUT1 (B) or the Müller cell marker CRALBP (C). [Scale bars, 25 μm (lower-magnification), 5 μm (higher-magnification).] (D) Colocalization frequency (Mander’s value, tM) of ZnT-3 with cell type-specific markers (see also Fig. S3 E–G). n = 10 retinas per group, one-way ANOVA with Bonferroni post hoc tests. **P < 0.01, ***P < 0.001 compare with VGAT; ^†††P < 0.001 compare with GAD65/67. All bars show mean ± SEM.

Fig. S4.

Cellular localization of Zn²⁺ elevation. (A–F) Early localization in amacrine cell terminals detected using the fluorescent Zn²⁺-specific sensor ZP-1 (green), cell-type specific antibodies, and AAV6 to selectively express mCherry (red) in amacrine cells. One day pNC, mCherry encoded by AAV6 partially overlaps with ZP-1 (A) and with the amacrine cell markers VGAT (B) and GAD65/67 (C) (green), but not with βIII-tubulin, an RGC marker detected with the TUJ-1 antibody (D) or the bipolar cell markers VGLUT1 or PKCα (E and F, respectively). (G and H) At 3 d, mCherry encoded by AAV2 colocalizes with ZP-1 (G) and βIII-tubulin, a marker for RGCs (H). (Scale bars in A, B, G, and H: Upper rows, 25 μm; Lower rows, 5 μm; in C, D, E, and F: Upper rows, 50 μm; Lower rows, 5 μm.)

Fig. S5.

The AMG signal in the optic nerve (ON) increases slightly after NC and is unaffected by intraocular injections of chelators. (A) Longitudinal sections through the optic nerve after NC with or without intraocular injections of the Zn²⁺ chelator TPEN. Asterisks indicate injury site. (Scale bar, 200 μm.) (B) Quantitation of the AMG signal proximal to and 2-mm distal to the crush site. (n = 5, 5, 4, 4, 4). One-way ANOVA with Bonferroni’s post hoc test, **P < 0.01, ***P < 0.001 compared with uncrushed controls; ns, not significant. (C) AMG staining in longitudinal sections through the optic nerves (ON) of slc30a3^−/− and littermate controls (slc30a3^+/+) before and 1 d after NC. Asterisks denote the injury site. (Scale bar, 200 μm.) Note that A and C represent composites of multiple images taken at same exposure and spliced together. (D) Quantitation of AMG signals in the optic nerve proximal to (Prox.) and 2-mm distal to (Dist.) the crush site before and 1 d after NC. (n = 5, 6; 6, 6; 6, 6). Upaired t test, ^††P < 0.01, ^†††P < 0.001, compared with slc30a3^+/+ littermates. Values in B and D are normalized to normal controls. All bars show mean ± SEM.

Fig. 4.

Elimination of the vesicular Zn²⁺ transporter ZnT-3 promotes RGC survival and axon regeneration. (A and B) RGC survival. (A) Retinal whole-mounts immunostained for βIII-tubulin to visualize RGCs in retinas of normal control mice or in mice 2 wk pNC, with or without deletion of slc30a3, the gene for ZnT-3. No general abnormalities were observed in the retinas of mice lacking slc30a3 (with or without NC). (Scale bar, 50 μm.) (B) Effect of slc30a3 deletion on RGC survival: quantitation. n = 8, 10; unpaired t test, ***P < 0.001 compared with slc30a3^+/+ controls. (C and D) Quantitation (C) and images (D, GAP-43 immunostaining) showing effects of slc30a3 deletion on optic nerve regeneration 2 wk pNC. n = 8, 10; unpaired t test, ***P < 0.001 compared with slc30a3^+/+ controls. (Scale bar, 200 μm.) Asterisks denote injury site. D shows composites of multiple images taken at the same exposure and magnification spliced together.

Fig. 5.

Chelating Zn²⁺ enhances RGC survival. (A–F) Retinal whole-mounts immunostained for βIII-tubulin to visualize RGCs in normal control mice (A) or 2 wk pNC with treatments as indicated. (Scale bar, 50 μm.) (G) Quantitative results. [Chel, chelators; Pretreat, additional injection given 1 d before NC; n = 11, 6, 6, 6 (TPEN); n = 11, 7, 8, 6 (ZX1)]. One-way ANOVA, Bonferroni post hoc tests. ***P < 0.001 compared with NC alone; ^††P < 0.01, ^†††P < 0.001 for decrease compared with chelator-alone group. (H) Combined effects of TPEN treatment and slc30a3 deletion. n = 8, 10, 10; one-way ANOVA with Bonferroni post hoc tests, ***P < 0.001 compared with slc30a3^+/+ controls. (I and J) Chelator effects on cleaved caspase-3 and the antiapoptotic protein Bcl-xL 5 d after NC (images are in Fig. S6 I and J). Values in J are normalized to uncrushed controls. n = 6 per group, one-way ANOVA with Bonferroni post hoc tests. **P < 0.01, ***P < 0.001 compared with uncrushed controls; ^††P < 0.01, ^†††P < 0.001 compared with NC-alone group. All bars show mean ± SEM.

Fig. S6.

Zn²⁺ chelation or blocking vesicular release of Zn²⁺ promotes RGC survival after optic nerve injury. (A and B) Dose–response relationship between RGC survival (2 wk after NC) and concentration of TPEN (A) or ZX1 (B). Chelators were administered twice, once right after NC (D0) and again 4 d later (D4). Both TPEN and ZX1 show maximal effects at 100 μM. A, n = 11, 5, 6, 5; B, n = 11, 6, 7, 8. One-way ANOVA with Bonferroni’s post hoc test, **P < 0.01, ***P < 0.001 compared with PBS treated group; ^†P < 0.05, compared with TPEN treatment at 20 μM. (C) RGC survival at two distances (1 mm, 2 mm) from the optic disk with PBS vs. TPEN treatment (100 μM injected on days 0 and 4; n = 11, 6; unpaired t test, ***P < 0.001 compared with PBS group. (D) Pure 129S mice show higher RGC survival (∼26.1%) than C57 (16.3%) at 2 wk pNC. n = 4 per group. ***P < 0.001, compared with normal 129S retinas. (E) Representative regions of flat-mounted retinas immunostained for βIII-tubulin to evaluate RGC survival 2 wk after NC with an additional injection of TPEN or ZX1 1 d before NC (pre). (Scale bar, 50 μm.) (F) Two weeks after NC, image from flat-mounted retina showing effect of a third TPEN injection at 7 d (D7) after NC. (Scale bar, 50 μm.) (G and H) Quantitation and image (flat-mounted retina) showing RGC survival 2 wk after NC with intraocular TeNT (20 nM) (n = 8, unpaired t test, ***P < 0.001 compared with PBS group). (Scale bar, 50 μm.) (I and J) Images (retinal cross-sections) showing effects of Zn²⁺ chelators on levels of cleaved caspase-3 and the antiapoptotic protein Bcl-xL 5 d after NC. RGCs are identified by immunostaining for βIII-tubulin. Quantitation is shown in Fig. 5 I and J. (Scale bars, 50 µm.) (K) Effects of combinatorial treatments on RGC survival 2 wk pNC (Zym/CPT, Zymosan plus CPT-cAMP; deln, deletion; one-way ANOVA with Bonferroni’s post hoc test; ns, not significant). Bars show mean ± SEM. (L and M) Long-term effects of ZnT-3 deletion on RGC survival. (L) Portions of flat-mounted retinas stained with an antibody to βIII-tubulin to visualize surviving RGCs 12 wk after NC with or without slc30a3 gene deletion. (Scale bar, 50 μm.) (M) Quantitation of RGC survival at 2 wk and 12 wk with or without slc30a3 gene deletion [n = 8 (slc30a3^+/+, 2 wk), n = 12 (slc30a3^−/−, 2 wk), n = 6 (slc30a3^+/+, 12 wk), n = 8 (slc30a3^−/−, 12 wk); two-way ANOVA with Bonferroni post hoc tests] ***P < 0.001, compared with slc30a3^+/+ littermate controls; ^††P < 0.01, ^†††P < 0.001, significance of decreases compared with 2 wk after NC within the same strain. All data points represent mean ± SEM.

Fig. S7.

Chelation of Zn²⁺ enhances expression of growth-related genes after NC. (A) Gene names and primer sequences for quantitative reverse-transcription PCR. (B) Quantitative RT-PCR for genes encoding proteins associated with axon regeneration (Fn14, ATF3, KLF6, Sprr1a, GAP43) and cell death (DLK, CHOP, Hrk, Puma, Bim). Numbers represent fold-increase relative to intact normal controls 4 d after NC (two retinas per sample, sample number = four per group). *P < 0.05, **P < 0.01, ***P < 0.001 compared with uncrushed control group; ^†P < 0.05, ^††P < 0.01, compared with NC-alone with vehicle control treatment. All bars show mean ± SEM.

Fig. 6.

Zn²⁺ chelation leads to long-term RGC survival and stabilizes the effects of pten deletion. (A) Portions of flat-mounted retinas immunostained for βIII-tubulin to visualize surviving RGCs 12 wk after NC with and without TPEN treatment or pten deletion. (Scale bar, 50 μm.) (B) Quantitation of long-term RGC survival. n = 11, 6, 6, 10 (2 wk); n = 9, 6, 6, 6 (12 wk), two-way ANOVA with Bonferroni post hoc tests; ***P < 0.001 compared with the same treatment at 2 wk after NC; ns, not significant. All data points represent mean ± SEM.

Fig. 7.

Zn²⁺ chelation promotes optic nerve regeneration. (A and B) Longitudinal sections through the mouse optic nerve immunostained for GAP-43 2 wk pNC (A) and quantification of results (B) (asterisk: injury site) (Scale bar, 200 μm.) (Chel, chelators; Pretreat, additional injection given one day before NC in B). n = 10, 8, 7, 6 (TPEN) and 10, 7, 6, 6 (ZX1). One-way ANOVA, Bonferroni post hoc tests. **P < 0.01, ***P < 0.001 compared with NC alone; ^†P < 0.05, decrease compared with chelator-alone group; ^#P < 0.05, ^##P < 0.01, increases compared with chelator-alone group. (C and D) Image (C) and quantitation (D, n = 6 per group) showing axon regeneration in mice with intraocular TPEN combined with pten deletion in RGCs. One-way ANOVA with Bonferroni post hoc tests. (Scale bar, 200 µm.) *P < 0.05, **P < 0.01; ns, not significant. (E) Quantitation showing effects of slc30a3 deletion, with and without TPEN treatment, on optic nerve regeneration 2 wk pNC. n = 8, 10, 18; one-way ANOVA with Bonferroni post hoc tests, **P < 0.01, ***P < 0.001 compared with slc30a3^+/+ controls. (F–H) Image (F) and quantitation (G) showing regeneration induced by combining TPEN and pten deletion 12 wk pNC. G, (n = 8, 5, 5, 6). One-way ANOVA with Bonferroni post hoc tests. *P < 0.05, **P < 0.01. F and H, CTB-labeled axons in the optic nerve (F, single 14-µm section) and optic chiasm (H, stack of three 14-µm sections). (Scale bars, 200 μm in main images, 50 µm in enlarged areas.) All bars show mean ± SEM. A, C, F, and H represent composites of multiple images taken at same exposure conditions spliced together.

Fig. S8.

Zn²⁺ chelation promotes axon regeneration: dose–response effects, specificity, and lack of synergy with Zymosan treatment. (A) Longitudinal sections through the optic nerve 2 wk after NC and treatment with chelators with or without equimolar Zn²⁺. Asterisks denote injury site. (Scale bar, 200 μm.) (B and C) Quantitation of axon growth at the indicated distances beyond the crush site 2 wk after NC: dose–response effects of TPEN and ZX1 and elimination of effects by presaturating chelators with equimolar Zn²⁺. B, n = 10, 5, 8, 5, 7, 6; C, n = 10, 6, 7, 8, 6, 5. One-way ANOVA with Bonferroni’s post hoc test, *P < 0.05, **P < 0.01, ***P < 0.001 compared with NC-alone group; ^†P < 0.05, ^††P < 0.01, compared with chelator-alone treatment. All bars show mean ± SEM. (D) Longitudinal sections through the mouse optic nerve two weeks after TPEN treatment combined with Zymosan plus CPT-cAMP (Upper) and effect of pten gene deletion alone (Lower) compared with TPEN treatment combined with pten gene deletion (Fig. 7C). Asterisks denote injury site. (Scale bar, 200 μm.) Complete dataset is shown in Fig. 7D. Note that A and D represent composites of multiple images taken at same exposure and spliced together.

Fig. S9.

Correlation between GAP43 and CTB labeling of regenerating axons. (A and B) Images of doubly labeled (GAP43, green; CTB, red) longitudinal sections through the optic nerve 2 wk after NC with or without the Zn²⁺ chelator TPEN (100 μM, injected at days 0 and 4 pNC). Asterisks denote injury site. (Scale bar, 200 μm.) This panel represents a composite of multiple images taken at same exposure and spliced together. (B) Enlarged images of framed areas in A and merged image of GAP43 and CTB labeled axons. (Scale bar, 50 μm.) (C and D) Quantification of doubly labeled axons in bar graph form (C) and of individual labels represented in a dot-plot (D). Note high correlation between GAP43 and CTB labeling. n = 8, 9; **P < 0.01, ***P < 0.001, unpaired t test. All data points show mean ± SEM.

Fig. 8.

Therapeutic window for Zn²⁺ chelation. (A and B) Surviving RGCs following the same treatments described above. (A) Single treatments, n = 11, 6, 20, 6, 21; (B) multiple treatments, n = 15, 19, 8 retinas per group. (C and D) Number of regenerating axons 0.5 mm from the injury site 2 wk after TPEN treatment (100 μM) at the indicated time points (days, d) after NC. (C) Single injections, n = 10, 8, 10, 8, 11 cases per group. (D) Multiple injections, n = 14, 10, 8 cases per group. Addl, additional TPEN injection. One-way ANOVA with Bonferroni post hoc tests. *P < 0.05, **P < 0.01, ***P < 0.001 compared with PBS-treated controls; unpaired t test, ^†P < 0.05, ^†††P < 0.001 compared with single TPEN treatment; unpaired t test, ^##P < 0.01 compared with TPEN treatments on days 0 and 4. All bars represent mean ± SEM.