Wnt Signaling Promotes Müller Cell Proliferation and Survival after Injury

Abstract

Purpose.: Müller glia respond to retinal injury by a reactive gliosis, but only rarely do mammalian glial cells re-enter the cell cycle and generate new neurons. In the nonmammalian retina, however, Müller glia act as stem/progenitor cells. Here, we tested the function of Wnt signaling in the postinjury retina, focusing on its ability to influence mammalian Müller cell dedifferentiation, proliferation, and neurogenesis.

Methods.: A 532 nm frequency doubled neodymium-doped yttrium aluminum garnet (Nd:YAG) laser was used to create light burns on the retina of Axin2^LacZ/+ Wnt reporter mice. At various time points after injury, retinas were analyzed for evidence of Wnt signaling as well as glial cell response, proliferation, and apoptosis. Laser injuries also were created in Axin2^LacZ/LacZ mice, and the effect of potentiated Wnt signaling on retinal repair was assessed.

Results.: A subpopulation of mammalian Müller cells are Wnt responsive and, when Wnt signaling is increased, these cells showed enhanced proliferation in response to injury. In an environment of heightened Wnt signaling, caused by the loss of the Wnt negative regulator Axin2, Müller cells proliferated after injury and adopted the expression patterns of retinal progenitor cells (RPCs). The Wnt-responsive Müller cells also exhibited long-term survival and, in some cases, expressed the rod photoreceptor marker, rhodopsin.

Conclusions.: The Wnt pathway is activated by retinal injury, and prolonging the endogenous Wnt signal causes a subset of Müller cells to proliferate and dedifferentiate into RPCs. These data raised the possibility that transient amplification of Wnt signaling after retinal damage may unlock the latent regenerative capacity long speculated to reside in mammalian neural tissues.

Introduction

The retina is a neural circuit that converts light into electrical and chemical signals, and informs the brain about the visual world. Loss of retinal cells through injury or disease results in visual impairment and blindness, conditions that affect over 50 million people in the United States. ¹ One strategy for restoring sight is based on the regenerative capabilities of nonmammals who respond to retinal injury in a unique way: in fish ² and amphibians, ^3,4 and to a lesser extent in chick hatchlings, ^5,6 Müller glia re-enter the cell cycle, dedifferentiate into retinal progenitor cells (RPCs), and generate new neurons and glia that reform functional connections and restore visual function (reviewed by Karl and Reh ⁷ ). None of these steps occurs spontaneously in the mammalian retina. Instead, mammalian Müller cells show minimal proliferation after injury and do not dedifferentiate. ⁸ Instead, they increase production of vimentin, glutamine synthetase (GS), and glial fibrillary acidic protein (GFAP), and result in a reactive gliosis. ^9,10  

Another major difference between the mammalian retina and its regenerating nonmammalian counterpart is cell survival: in fish, approximately 20% of cells that proliferate after injury survive for more than 10 days, ^11,12 but this percentage is reduced in chick hatchlings, ^5,13 and is nonexistent in mammals. ⁸ Nonetheless, some mammalian Müller cells maintain expression profiles of RPCs. ^14,15 This finding has led to the hypothesis that a subset of Müller cells act as stem/progenitor cells of mammalian retina and that with a sufficient stimulus, these cells could give rise to new neurons to restore vision (reviewed previously ^15–17 ). 

A key regulator of nonmammalian RPC proliferation is the Wnt signaling pathway. ¹⁸ In lower vertebrates, Wnt pathway stimulation results in the expansion of an RPC population in the ciliary marginal zone (CMZ). ^19–23 Conversely, Wnt pathway inhibition depletes CMZ progenitors. ^21,24,25  

A variety of Wnt reporter mice have been used to identify sites of Wnt signaling in the mammalian retina, with conflicting results. For example, some well-characterized Wnt reporter strains (e.g., TOPgal, BATgal) showed no evidence of X-gal staining in the postnatal retina, while others, including TCF/Lef^{LacZ 26} and Lrp5^{LacZ 27} exhibit robust, and sometimes overlapping patterns of reporter activity. Wnt and non-Wnt (Norrin) ligands also were expressed in the mammalian retina, ^28,29 which raises the obvious question of whether Wnt signaling is affecting primarily the neurons, Müller glia, or vasculature in the adult mammalian retina. 

To address this question we used a strain of mice in which LacZ is inserted into the Axin2 gene. ³⁰ Axin2^LacZ/LacZ progenitor cells have a significantly increased sensitivity to a Wnt stimulus and an extended duration of Wnt signaling, ³¹ which provides a genetic model to study how potentiated Wnt signaling affects retinal homeostasis and repair. The transient nature of the amplified Wnt response in Axin2^LacZ/LacZ mice also is important because it avoids the complication of other strategies that constitutively activate the Wnt pathway, or disrupt other signaling pathways in addition to Wnt. 

Materials and Methods

Animals

All experimental procedures followed an approved protocol by the Stanford Committee on Animal Research and were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Axin2^LacZ/LacZ mice were obtained from Walter Birchmeier, PhD. Axin2^lacZ/LacZ and Axin2^LacZ/+ mice were maintained in 129 and C57/B6 background. Albino Axin2^lacZ/LacZ and Axin2^LacZ/+ mice were obtained by backcrossing to CD1 mice. 

Retinal Laser Injury Model

The PASCAL laser system (Topcon, Tokyo, Japan) provided 532 nm optical radiation from a diode-pumped, continuous wave, frequency-doubled neodymium-doped yttrium aluminum garnet (Nd:YAG) laser coupled into a multimode step index optical fiber. Variation of the beam intensity within the top-hat area did not exceed 7%. 

Mice (age 2–3 months) were anesthetized using ketamine hydrochloride (8 mg/kg, intraperitoneal [IP]) and xylazine (1.6 mg/kg, IP) 15 minutes before the procedure. Pupillary dilation was achieved by 1 drop of 1% tropicamide and 2.5% phenylephrine hydrochloride. Topical tetracaine 0.5% was instilled in the eye before treatment. A cover slip was used to focus the laser-aiming beam on the retina. 

Laser burns were titrated as described previously, which exhibited blanching of the fundus pigmentation without frank whitening. ³² Two laser burns were applied to each eye: one superior and one inferior to the optic nerve separated from the nerve by at least 2 lesion diameters. To examine gene expression by quantitative RT-PCR, 12 burns were created per eye. Based on the titrations, the laser setting used to achieve a “light” burn was 400 μm aerial diameter, 120 mW power, and 25 ms pulse duration. 

Optical Coherence Tomography (OCT)

OCT was performed with the Heidelberg Spectralis (Heidelberg Engineering, Heidelberg, Germany). A Volk digital high magnification lens (Volk Digital High Mag Lens; Volk Optical, Mentor, OH) was adhered to the OCT to focus the image on the mouse posterior pole. At predetermined times after laser injury, OCT images were acquired of the laser lesions and measured using an image processing program (ImageJ; National Institutes of Health, Bethesda, MD). Lesion measurements were corrected using the aerial to fundus magnification ×0.39. ³³ Two independent, blinded observers measured images and an average of the two measurements was taken. 

Bromodeoxyuridine (BrdU) Labeling

For cell proliferation analysis, BrdU labeling reagent (Invitrogen, Carlsbad, CA) was injected intraperitoneally according to the manufacturer's instructions. BrdU was labeled for 24 hours before harvesting the eyes. For the BrdU pulse tracing experiment, BrdU was given to mice at postinjury day 2 and labeled for 24 hours. The eyes were harvested on postinjury day 21. 

Retinal Explant Cultures

Retinal explant culture was prepared as described previously. ³⁴ In control cases, 0.1% PBS was added to the media; in other cases, either Wnt3a protein (200 ng/ml; R&D Systems, Minneapolis, MN) or R-spondin (200 ng/ml; R&D Systems) was added to the media. Retinal explants were cultured for 48 hours. The explants were fixed with 4% paraformaldehyde (PFA), embedded in OCT, and cut into 10 μm thick sections for analyses. 

Sample Preparation, Histology, X-Gal, and TUNEL Staining

Eyes were enucleated and fixed in 4% PFA. Samples were embedded in OCT and cut into 10 μm thick radial sections. For the PAS staining, sections were treated with 0.5% periodic acid for 5 minutes, then incubated into Schiff's Reagent (Sigma-Aldrich, St. Louis, MO) for 30 minutes, counterstained with hematoxylin for 3 minutes and processed to mount. X-gal staining was performed as described previously. ³⁵ TUNEL staining was done by using an in situ cell death detection kit (Roche, Indianapolia, IN) following instructions of the manufacturer. 

Immunohistochemistry

Cryosections were permeabilized with 0.2% Triton-100, blocked with 5% goat serum, and incubated with primary antibodies and secondary antibodies at room temperature for 1 hour. For information about antibodies see Supplementary Table S1 (see Supplementary Material and Supplementary Table S1). 

RT-PCR and Quantitative RT-PCR

Total RNA was extracted using TRIzol (Invitrogen). cDNA was synthesized by using SuperScript III First-Strand Synthesis Kit (Invitrogen) according to the instructions of the manufacturer. cDNA was used as a template in each PCR reaction. RT-PCR and quantitative PCR reactions were performed as described previously. ³¹ For detailed information about primer sequences see Supplementary Table S2 (see Supplementary Material and Supplementary Table S2). 

Statistical Analysis

Results were presented as the mean ± SD. Student's t-test was used to quantify differences described in this article. P ≤ 0.05 was considered to be significant. 

Results

Wnt Ligands Can Activate Beta Catenin–Dependent Wnt Signaling in the Retina

Multiple Wnt ligands are expressed in the adult retina. ²⁶ Using RT-PCR, we confirmed these previous observations and also showed that non-Wnt ligands, including Norrin, are expressed in the adult retina (reported by Ye et al., ³⁶ Fig. 1A). Because Norrin can activate beta catenin–dependent Wnt signaling in the retina through the Frizzled/Lrp receptor complex, ^28,29 this left open the possibility that X-gal staining in the Axin2^LacZ/+ retina (Fig. 1B) could be attributed to the activity of a non-Wnt ligand. To clarify the contribution of Wnt ligands to retinal homeostasis, we turned to an explant system. 

Figure 1. 

 

Wnt and non-Wnt ligands activate Wnt signaling in the retina. (A) RT-PCR analyses for Norrin and Wnt ligands expressed in the adult retina. (B) X-gal staining of retinal explants from Axin2^LacZ/+ mice treated with PBS and stained for X-gal (n = 3). (C) Retinal explants from Axin2^LacZ/+ mice treated with Wnt3a and stained for X-gal (n = 3). (D) Retinal explants from Axin2^LacZ/+ mice treated with PBS, then co-immunostained for beta-galactosidase (red) and Sox9 (green); arrow indicates double-positive cells. (E) Retinal explants from Axin2^LacZ/+ mice treated with Wnt3a, then co-immunostained for beta galactosidase (red) and Sox9 (green); arrows indicate double-positive cells. (F) Frizzled5 (green) and GS (red) are co-localized in a subset of cells in the INL; arrows indicate double-positive cells. (G) Retinal explants from Axin2^LacZ/+ mice treated with PBS, then stained for X-gal (n = 3). (H) Retinal explants from Axin2^LacZ/+ treated with R-spondin and stained with X-gal (n = 3). IPL, inner plexiform layer. Scale bars: 50 μm (B–E, G, H), 25 μm (F).

First, we showed that purified Wnt3a protein resulted in an increase in X-gal staining in the inner nuclear layer (INL) and ganglion cell layer (GCL) of retinal explants harvested from Axin2^LacZ/+ mice (Figs. 1B, 1C; n = 3). To identify the Wnt-responsive cells, we used double immunostaining and showed that the X-gal^+ve cells co-express the Müller cell marker, Sox9 ³⁷ (Figs. 1D, 1E). We confirmed that Müller cells were Wnt responsive by showing co-expression of the Müller marker GS and the Wnt receptor Frizzled 5 (Fig. 1F; also reported by Liu and Nathans ³⁸ ). 

We also used a biochemical approach to demonstrate that endogenous Wnt ligands activate Wnt signaling in the retina. The secreted protein R-spondin can activate beta catenin-dependent Wnt signaling, but only in the presence of Wnt ligands. ^39,40 We demonstrated that endogenous Wnt ligands were present and functioned in the retina by treating Axin2^LacZ/+ retinal explants with R-spondin alone, and then assaying for beta galactosidase activity using X-gal staining. Compared to PBS-treated Axin2^LacZ/+ explants (Fig. 1G, n = 3), R-spondin–treated retinal explants showed considerably more beta galactosidase activity in the GCL, INL, and the outer nuclear layer (ONL; Fig. 1H, n = 3). Collectively, these data demonstrated that endogenous Wnt ligands are responsible for at least some of the beta catenin–dependent Wnt signaling observed in the retina, and that the Wnt responsive cell population is composed at least partially of Müller glia. 

Laser Injury Activates Müller Cells, Leading to a Reactive Gliosis

To explore the role of Wnt signaling in retinal repair, we used a laser injury model that creates a well-delineated zone of damage (Fig. 2A). We used a 532 nm frequency doubled Nd:YAG laser to create light burns on the retina. Within 60 minutes of injury, the site of damage could be visualized ophthalmoscopically (Fig. 2B). Spectral domain (SD)-OCT showed hyperreflectivity extending from the RPE to the outer plexiform layer (OPL) throughout the ONL (Fig. 2C, n = 10). After 2 days, SD-OCT continued to show hyperreflectivity that now was restricted to the ONL; the photoreceptor layer showed normal lower reflectivity (Fig. 2D, n = 10). Within 24 hours of the injury, histologic analyses demonstrated a continuous ONL with variable nuclear thickness and architecture, photoreceptor discontinuity, and RPE loss and hypertrophy (Fig. 2E, n = 3). 

Figure 2. 

 

Acute injury induces a reactive gliosis in the mammalian retina. (A) Schematic cross-section through the mouse eye indicating location of the laser injury. (B) Fundus image of laser injuries, superior and inferior to optic nerve; arrow indicates laser injury site. The plane of OCT scanning is indicated by the green line. (C) OCT scan of the superior burn detect laser damage (brackets) 1 hour (n = 10) and (D) 2 days after laser injury (n = 10). (E) PAS staining confirms focal site of damage (n = 3). (F) Apoptotic cells, indicated by TUNEL staining in intact retina (n = 4). (G) Apoptotic cells are localized to the ONL on postinjury day 1 (n = 4). (H) The number of apoptotic cells is reduced by the second day after injury (n = 4). (I) 7 days after injury, only minimal TUNEL staining is detectable (n = 3). (J) Cells in the intact, adult neural retina fail to incorporate BrdU over a 24-hour period (n > 10). (K) Using albino mice, a few BrdU-positive cells are detectable in the RPE of the intact adult eye (n > 10). (L) BrdU incorporation is elevated significantly following laser injury in the neural retina (n > 10), and (M) in the RPE and choroid (n > 10). (N) Ki67 immunostaining reveals the pattern of injury-induced cell proliferation (n = 3). (O) Quantitative RT-PCR analyses show elevated expression of CyclinD1 on postinjury day 1 (n = 3; *P = 0.01, t-test). (P, Q) By day 3, minimal cell proliferation is detected in the neural retina (arrows, n = 3), and (Q) in the RPE and choroid (n = 3). (R) GFAP immunostaining (green) is isolated to the GCL of the intact retina (n = 3). (S) Two days after injury GFAP expression (green) is increased within the injury site (n = 3). (T) Quantitative RT-PCR analysis for GFAP (n = 3; *P = 0.01, t-test) indicates that injury stimulates GFAP expression. (U) GFAP (green) still is expressed strongly in the injury site 7 days after injury (n = 3). ON, optic nerve. Scale bars: 100 μm (C, D, H), 50 μm (E–G, I–U).

The zone of damage caused by the laser was spatially and temporally localized. Using TUNEL staining to identify apoptotic cells, ⁴¹ we showed that after laser injury, cell apoptosis was located predominantly in the ONL (compare control, Fig. 2F to Fig. 2G; n = 4). TUNEL staining was maximal in the ONL between 1 and 2 days after injury (Figs. 2G, 2H; n = 4); by postinjury day 7, cell death in the neural retina was complete (Fig. 2I, n = 3). This temporal pattern of cell death is similar to that described in other retinal injury models. ⁴²  

The laser injury induced localized cell proliferation in the retina. Relative to controls, where cell proliferation is nonexistent in the retina (Figs. 2J, 2K; n > 10) and minimal in the RPE (Fig. 2K, n > 10), BrdU^+ve cells were evident following laser injury: 24 hours after laser injury, a few BrdU^+ve nuclei were observed (not shown), but when BrdU was injected 2 days after laser injury, the number of BrdU^+ve cells was increased dramatically (Figs. 2L, 2M; n > 10). Ki67 immunostaining and quantitative RT-PCR for CyclinD1 confirmed the location and the amplitude, respectively, of this proliferative burst (Figs. 2N, 2O; n = 3). Only a small percentage of the BrdU^+ve cells co-expressed the microglial marker Iba-1 ⁴³ (Supplementary Fig. S1, n = 3 [see Supplementary Material and Supplementary Fig. S1]). The injury-induced re-entry of cells into the cell cycle was short-lived: only a few cells were labeled when BrdU was administered 3 days after laser injury (Figs. 2P, 2Q; n = 3). 

We continued our characterization of the injury response in control mice. As expected, Müller glial cells were activated by acute injury. ^2,12,44 Relative to their levels in the intact retina, expression of the Müller cell markers GFAP, GS, and vimentin were increased significantly after damage (Figs. 2S–U; n = 3 and Supplementary Fig. S1, n = 3 [see Supplementary Material and Supplementary Fig. S1]). We used GFAP (Fig. 2U, n = 3), as well GS and vimentin (Supplementary Fig. S1, n = 3 [see Supplementary Material and Supplementary Fig. S1), immunostaining to confirm that the injury resulted in a reactive gliosis. This is in agreement with previous reports on the mammalian response to other types of retinal injuries (reviewed by Jadhav et al. ⁴⁵ ). 

Endogenous Wnt Signaling Is Upregulated by Retinal Injury

We characterized how Wnt signaling was affected by damage to the retina. Relative to signaling in the intact retina, the Wnt pathway was upregulated significantly by laser injury (Fig. 3A, n = 3). We mapped the spatial and temporal distribution of the Wnt response using Axin2^LacZ/+ mice and found X-gal staining increased, specifically at the injury site (Figs. 3B, 3C; n = 10). We compared Wnt signaling in control Axin2^LacZ/+ mice with Axin2^LacZ/LacZ mice. Deletion of both copies of the negative Wnt regulator Axin2 leads to a prolonged, amplified Wnt response in these mice. ³¹ Using quantitative RT-PCR for exon1 of the Axin2 gene, we demonstrated increased expression levels of the Wnt target gene in Axin2^LacZ/LacZ mice (Fig. 3D, n = 3). X-gal staining demonstrated that the primary site of enhanced Wnt signaling was the injured ONL (Figs. 3E, 3F; n = 10). 

Figure 3. 

 

A neuroprotective function for Wnt signaling in the retina. (A) Quantitative RT-PCR analyses demonstrate significantly higher Axin2 exon1 expression in injured retinas versus intact retinas (n = 3, *P < 0.05). (B) Low and (C) high magnification images of the injured Axin2^LacZ/+ retina stained for X-gal (n = 10). (D) Axin2 exon1 expression is elevated significantly in the injured Axin2^LacZ/LacZ retina compared to the injured Axin2^LacZ/+ retina (n = 3, *P < 0.01). (E) Low and (F) high magnification images of the injured Axin2^LacZ/LacZ retina stained for X-gal (n = 10). (G) Quantification of the sizes of the injuries using pixel counts from the OCT images, made immediately after injury. The injury size did not vary between Axin2^LacZ/+ and Axin2^LacZ/LacZ mice (n = 10). (H) TUNEL staining demarcates dying cells on postinjury day 1 in Axin2^LacZ/+ (n = 4) and (I) Axin2^LacZ/LacZ mice (n = 6). (J) TUNEL staining on postinjury day 2 in the Axin2^LacZ/+ retina (n = 4) and (K) in the Axin2^LacZ/LacZ retina (n = 3). (L) TUNEL staining on postinjury day 4 in the Axin2 ^LacZ/+ retina (n = 4) and (M) in the Axin2^LacZ/LacZ retina (n = 4). (N) TUNEL staining on postinjury day 7 in the Axin2^LacZ/+ retina (n = 4) and (O) in the Axin2^LacZ/LacZ retina (n = 4). (P) Quantification of the pixel counts of TUNEL staining between injured Axin2^LacZ/+ and Axin2^LacZ/LacZ retina (n > 3, *P > 0.05). (Q) GFAP expression on postinjury day 2 in the Axin2^LacZ/+ retina (n = 3) and (R) in the Axin2^LacZ/LacZ retina (n = 3). (S) Nestin expression on postinjury day 2 in the Axin2^LacZ/+ retina (n = 3), and (T) in the Axin2^LacZ/LacZ retina (n = 3). Scale bars: 50 μm (B–M).

Elevated Wnt signaling in the injured retina correlated with decreased apoptotic death of retinal neurons. The laser injuries were of the same initial size (Fig. 3G, n = 10), but beginning on postinjury day 1 and continuing throughout the injury response, TUNEL staining was reduced in the retinas of Axin2^LacZ/LacZ mice (Figs. 3H–O; n > 3). Quantification of the TUNEL staining in Axin2^LacZ/LacZ and Axin2^LacZ/+ retinas confirmed the reduction in cell death in mice with elevated Wnt signaling (Fig. 3P, n > 3). In addition, Axin2^LacZ/LacZ mice showed higher expression levels of GFAP and Nestin on postinjury day 2 (Figs. 3Q–T, n = 3). 

Potentiated Wnt Signaling Results in Müller Cell Proliferation and Generate Retinal Progenitor-Like Cell at the Site of Injury

Laser injury induced cell proliferation in the neural retina. We analyzed serial sections through injured retinas, and stained each section for BrdU. Three separate injuries were evaluated on postinjury day 2. In Axin2^LacZ/+ mice, we identified, on average, 30 BrdU^+ve cells/injury. In Axin2^lacZ/LacZ mice, we found, on average, 58 BrdU^+ve cells/injury (data not shown). 

In the retinas of Axin2^LacZ/LacZ mice, Müller cells proliferated more after injury. Compared to controls (Figs. 4A, 4C; n = 3), numerous BrdU^+ve/Sox9^+ve cells were evident in the injured ONL of Axin2^LacZ/LacZ mice (Figs. 4B, 4D; n = 3). Some of these BrdU^+ve co-expressed Pax6 (Figs. 4F, 4H; n = 3), indicating that they were RPCs. This was in stark contrast to controls, where no BrdU^+ve /Pax6^+ve cells were detectable (Figs. 4E, 4G; n = 3). The RPC marker Nestin also was increased in Axin2^LacZ/LacZ mice as compared to controls (Figs. 4I, 4J; n = 3) and some of these Nestin^+ve cells also were BrdU^+ve (Fig. 4J, arrows; n = 3). Thus, in response to a heightened endogenous Wnt signal, Müller cells re-enter the cell cycle after injury and adopt the expression patterns of RPCs. 

Figure 4. 

 

Injury-induced Wnt signaling stimulates retinal progenitor cell proliferation. (A, C) In the Axin2^LacZ/+ retina (n = 3), BrdU^+ve (red) cells do not colocalize with Sox9 (green) immunostaining (n = 3). (B, D) In the injured Axin2^LacZ/LacZ retina some BrdU^+ve (red) cells coexpress Sox9 (green; arrows indicate yellow, double-positive cells, n = 3). (E, G) In the Axin2^LacZ/+ retina BrdU^+ve (red) cells do not colocalize with Pax6 (green, n = 3). (F, H) In the injured Axin2^LacZ/LacZ retina some BrdU^+ve (red) cells coexpress Pax6 (green; arrows indicate yellow cells, n = 3). (I) In the injured Axin2^LacZ/+ retina, BrdU^+ve (red) cells do not colocalize with Nestin (green) immunostaining (n = 3). (J) In the injured Axin2^LacZ/LacZ retina some BrdU^+ve cells co-express Nestin (arrows, n = 3). Scale bars: 50 μm (A, B, E, F, I, J), 25 μm (C, D, G, H).

Amplified Wnt Signaling Leads to Long-Term Cell Survival after Injury

In the mammalian retina, cells that proliferate in response to injury do not survive past one or two mitotic cycles. ⁸ We evaluated cell survival in control and Axin2^LacZ/LacZ mice by first creating a laser injury and then delivering BrdU on postinjury day 1. Mice were evaluated for the presence of BrdU^+ve cells in the retina 48 hours and 21 days after injury. 

BrdU^+ve cells were found in the RPE and choroid, INL, and ONL on postinjury day 2 (Figs. 5A, 5B; n = 10 for Axin2^LacZ/+ ; n = 14 for Axin2^LacZ/LacZ ). After 21 days, BrdU^+ve cells were not detectable in the control retina (Fig. 5C, n = 7), but in the Axin2^LacZ/LacZ mice, BrdU^+ve cells still were present in the INL, ONL, and RPE (Fig. 5D, n = 7). Some of these BrdU^+ve surviving cells expressed Sox9 (Figs. 5E, 5F; n = 3). Thus, in an environment of amplified Wnt signaling, more cells proliferated in response to injury, survived for longer periods of time, and eventually expressed markers of a Müller cell fate. 

Figure 5. 

 

Amplified Wnt signaling leads to long-term cell survival in the mammalian retina. Following laser injury, mice received an injection of BrdU; 24 hours later, BrdU^+ve (red) cells are detected in the RPE, INL, and ONL of (A) Axin2^LacZ/+ mice (n = 10) and (B) Axin2^LacZ/LacZ mice (n = 14). (C) On postinjury day 21, BrdU^+ve (red) cells are detected in the RPE, but not in the neural retinas of Axin2^LacZ/+ mice (n = 7). Box indicates the injury site. (D) BrdU^+ve (red) cells are detected in the RPE and in the INL of Axin2^LacZ/LacZ retina even after 21 days (n = 7). Box indicates the injury site. (E) No BrdU^+ve (red) cells co-express Sox9 in Axin2^LacZ/+ retina after 21 days (n = 3). (F) Some BrdU^+ve (red) cells are Sox9^+ve in Axin2^LacZ/LacZ retina after 21 days (n = 3). (G, I) BrdU^+ve (red) cells do not colocalize with Rhodopsin in Axin2^LacZ/+ retina after 7 days; boxed area is shown in higher magnification in ([I], n = 7). (H, J, J') In Axin2^LacZ/LacZ mice some BrdU^+ve cells co-express Rhodopsin on postinjury day 7 (n = 4). Box indicates the injury site. (K) PAS staining on postinjury day 21 in the Axin2^LacZ/+ retina (n = 6). (L) In a subset of Axin2^LacZ/LacZ mice PAS staining indicates that the cellular integrity is re-established on postinjury day 21 (n = 4). Scale bars: 50 μm (A–D, G, H, K, L), 25 μm (E, F, I–J').

On postinjury day 7 we found a few BrdU^+ve cells in the retinas of Axin2^LacZ/LacZ mice that costained for the rod photoreceptor marker, rhodopsin. ^46,47 In Axin2^LacZ/+ control mice we failed to detect any rhodopsin ^+ve/BrdU^+ve cells in the injury site; rather, rhodopsin expression was limited to the outer segments of the photoreceptors (Figs. 5G, 5I; n = 7). The BrdU^+ve cells that coexpressed rhodopsin were located primarily in the ONL (Figs. 5H, 5J, 5J'; n = 4). By postinjury day 21, 40% (4/10) of the injuries generated in Axin2^LacZ/LacZ mice showed evidence of a restoration of cellular integrity in the retina; in Axin2^LacZ/+ mice this response was never observed (0/6; Figs. 5K, 5L). 

Discussion

The mammalian retina appears to harbor a latent regenerative potential that many believe can be exploited in a therapeutic strategy to restore vision (reviewed previously ^7,44,48 ). To date, most strategies have focused on transplantation approaches to replace missing or damaged cells of the retina, but even the most successful of these show only 0.1% of the transplanted cells actually migrate into the retina. ⁴⁹  

Rather than transplantation, an alternative approach uses the potential of developmental signals to promote tissue regeneration (reviewed by Harada et al. ⁵⁰ ). Wnt proteins are powerful and ubiquitous stem cell self-renewing factors, and the strength of such a strategy rests in the central role that Wnts has in retinal development ^21,25,51 and in neural injury repair. ⁵²  

Wnt Signaling Is Part of the Body's Natural Response to Injury

Beta catenin–dependent Wnt signaling is activated by injury. Regardless of the type of tissue involved (i.e., cornea, ⁵³ skin, ⁵⁴ muscle, ⁵⁵ lung, ⁵⁶ intestine, ⁵⁷ pancreas, ⁵⁸ bone ³¹ ), or the kind of injury (e.g., chemical or thermal damage; freezing, crushing or cutting; or inhalation), the endogenous Wnt pathway is activated at the site of damage. 

Here, we expanded this to show that laser injury also upregulates endogenous Wnt signaling in the adult retina, specifically at the site of damage (Fig. 3). Similar to chemical and light-induced damage (e.g., as reported by Fischer and Reh, ⁵ and Takeda et al. ⁴⁸ ), this injury model is sufficient to activate Müller cells but has an advantage in that it leaves the majority of the retina intact (Fig. 2). 

Wnt Signals Are Essential for Tissue Repair

In a wide array of injury models, investigators have demonstrated that inhibiting endogenous Wnt signaling curtails tissue repair. ^59–63 This also is true in the retina: using nonmammalian models of retinal injury, investigators have demonstrated that here, too, Wnt signaling is essential for a regenerative response. ⁶⁴ These nonmammalian models provide a roadmap of sorts, whereby the program of regeneration can be outlined and then potentially recapitulated in mammalian tissues. ⁶⁵  

Wnts and Retinal Regeneration

Whether endogenous Wnt signals are required for mammalian retinal healing has been more difficult to ascertain, in part because in mammals the repair process terminates in a reactive gliosis. ^9,10 Nonetheless, investigators have shown that Wnt signals act as a mitogenic stimulus to Müller cells ^66,67 and at least in vitro, a subset of these proliferating cells can differentiate into photoreceptors after a retinoid ⁴⁷ or a hedgehog-conditioned media ⁶⁸ stimulus. Collectively, these data hint at an important role for Wnt signaling in potential regeneration of the mammalian retina. 

Here, we extended these in vitro studies using an in vivo injury model and Axin2^LacZ/LacZ mice to show that elevated Wnt signaling leads to more robust proliferation and cell survival. In this genetic model, where Wnt signaling is amplified through the removal of the negative regulator Axin2, ³¹ we showed that programmed cell death is reduced (Fig. 3), and Müller cells proliferate more and express RPC markers (Fig. 4), similar to what has been observed in a number of nonmammalian models. 

The total number of BrdU cells resulting from a laser retinal injury is, however, very small. Within a single injury created in Axin2^LacZ/+ mice, we found an average of 30 BrdU^+ve cells compared to an average of 58 BrdU^+ve cells in injuries created in Axin2^LacZ/LacZ mice. This constitutes a very small cell population for analysis. Nonetheless, within this small population we found evidence in Axin2^LacZ/LacZ mice of proliferating cells that unequivocally were double-labeled with BrdU and rhodopsin. This was, however, a rare occurrence and, thus, requires further verification using other types of models in which extensive proliferation is observed after injury. 

We also found evidence, in Axin2^LacZ/LacZ mice, of proliferating Pax6^+ve cells that appeared to migrate (e.g., see previously reported results ^69–71 ) from the INL into the injury site (Fig. 4). Other investigators have identified Pax6^+ve stem cells in the INL of animals with regenerating retinas, ⁷² but, to our knowledge, this is the first demonstration of injury-activated BrdU ^+ve/Sox9^+ve RPCs that persisted for at least 3 weeks in the mammalian retina. 

Is Wnt Insufficient for Mammalian Retinal Regeneration?

Amphibians and fish retinas grow throughout life, ⁷³ and the same signals that mediate retinal growth are re-used to stimulate regeneration after injury. ⁶⁴ Other animals (e.g., chicks) stop growing as adults, and only exhibit retinal regenerative abilities when young. ⁷⁴ Mammals appear to have an even lower regenerative potential, which correlates with a very restricted pattern of Wnt signaling in the retina (Fig. 1). 

We speculated that the tight regulation on mammalian Wnt signaling has evolved as a defensive mechanism against oncogenic transformation. Multiple lines of evidence demonstrate that unrestrained Wnt signaling activity causes hyper-proliferation and cancer (reviewed previously ^{75, 76} ). Consequently, there is likely to be strong evolutionary pressure to sequester Wnt signaling in mammalian tissues. As support for this theory, three of the four classes of Wnt signaling inhibitors, including Dkk, Frizzled-related protein and Cerberus, have no homologues in Drosophila, which suggests that evolutionary pressures acts to generate different Wnt inhibitors to control Wnt activity during tissue development and tissue repair. ⁷⁷  

Recently, a direct molecular link has been established between Wnt signaling and the Hippo pathway, which has emerged as a key regulator of tissue growth and patterning in development and disease (reviewed by Bao et al. ⁷⁸ ). In at least one tissue the Hippo pathway restricts beta catenin-dependent Wnt signaling via an indirect interaction between the transcriptional regulator TAZ and the Wnt target, Disheveled. ⁷⁹ It remains to be seen whether Hippo, TAZ, and other components of this kinase cascade modulate cell cycle exit and terminal differentiation in other mammalian tissues, such as the neural retina. As a conceptual framework, however, it is tempting to speculate how an equilibrium between a positive growth stimulus (Wnt) and an inhibitory one (Hippo) could prevent tissue overgrowth in a static situation, such as the intact retina, yet also be activated in postmitotic cells that have lost cell adhesion due to injury. 

A local increase in Wnt signaling at the site of damage may be sufficient to induce proliferation to repair some tissues, but in the mammalian retina a transient Wnt signal clearly is insufficient. Our data from analyses of the injured Axin2^LacZ/LacZ retina raise the intriguing possibility that prolonging the endogenous Wnt signal induced after acute injury may be sufficient to allow Müller cells to re-enter the cell cycle and dedifferentiate. This strategy may unlock the latent regenerative capacity long speculated to reside in mammalian neural tissues. 

Supplementary Materials

pdf S1 

Acknowledgments

A number of students contributed to early analyses of the Axin2^LacZ/LacZ retina, including Eleni Papadoyannis, Jemianne Bautista, and Elena Maden; we thank them all for their contributions. 

Footnotes

 Supported by California Institute for Regenerative Medicine TR1-01249. DJH is a CIRM Bridge Scholar (TB1-01190).

Footnotes

 Disclosure: B. Liu, None; D.J. Hunter, None; S. Rooker, None; A. Chan, None; Y.M. Paulus, None; P. Leucht, None; Y. Nusse, None; H. Nomoto, None; J.A. Helms, None