HB-EGF is necessary and sufficient for Müller glia dedifferentiation and retina regeneration (original) (raw)

Dev Cell. Author manuscript; available in PMC 2013 Feb 14.

Published in final edited form as:

PMCID: PMC3285435

NIHMSID: NIHMS342390

Molecular and Behavioral Neuroscience Institute and Department of Biological Chemistry, University of Michigan, Ann Arbor, MI 48109, USA

Corresponding author: Dan Goldman, ude.hcimu@namoruen, ph734-936-2057, fax734-936-26990

Summary

Müller glia (MG) dedifferentiation into a cycling population of multipotent progenitors is crucial to zebrafish retina regeneration. The mechanisms underlying MG dedifferentiation are unknown. Here we report that heparin-binding epidermal-like growth factor (HB-EGF) is rapidly induced in MG residing at the injury site and that proHB-EGF ectodomain shedding is necessary for retina regeneration. Remarkably, HB-EGF stimulates the formation of multipotent MG-derived progenitors in the uninjured retina. We show that HB-EGF mediates its effects via an EGFR/MAPK signal transduction cascade that regulates the expression of regeneration-associated genes, like ascl1a and pax6b. We also uncover an HB-EGF/Ascl1a/Notch/hb-egfa signaling loop that helps define the zone of injury-responsive MG. Finally, we show that HB-EGF acts upstream of the Wnt/β-catenin signaling cascade that controls progenitor proliferation. These data provide a link between extracellular signaling and regeneration-associated gene expression in the injured retina and suggest strategies for stimulating retina regeneration in mammals.

Introduction

Adult neural stem cells hold great promise for restoring nerve function following injury or disease. Like astrocytes, these cells are derived from radial glia. However, unlike astrocytes, neural stem cells are confined to specific niches. Activation of endogenous stem cells and dedifferentiation of astrocytes into neural stem cells may provide a means for stimulating nervous system self-repair. However, mechanisms underlying stem cell activation and astrocyte dedifferentiation are poorly understood. The zebrafish retina is an important model for revealing these mechanisms since the retina is able to mount a robust regenerative response following injury and this response is dependent on the dedifferentiation of Müller glia (MG), a radial glia-like cell that normally does not generate neurons during development, but can regenerate neurons after retinal injury (Bernardos et al., 2007; Fausett and Goldman, 2006; Fimbel et al., 2007; Ramachandran, 2010a; Thummel et al., 2008).

Although MG populate the mammalian retina and share many characteristics with those found in zebrafish, they only exhibit a limited regenerative potential under special conditions (Karl et al., 2008; Ooto et al., 2004; Osakada et al., 2007; Takeda et al., 2008; Wan et al., 2008). In general, MG of the mammalian retina exhibit a reactive gliotic response to injury that often causes more harm than good (Bringmann et al., 2009). If one could stimulate mammalian MG to respond to retinal injury like their zebrafish counterparts, they could potentially be used to regenerate lost cells in people suffering from blinding diseases or injuries. Therefore it is of great importance to understand the mechanisms by which zebrafish MG dedifferentiate into multipotent retinal progenitors.

Although a number of genes have been identified that are necessary for MG dedifferentiation (Fausett and Goldman, 2006; Fausett et al., 2008; Kassen et al., 2007; Qin et al., 2009; Ramachandran et al., 2010a; Raymond et al., 2006; Thummel et al., 2010), mechanisms underlying their activation have remained elusive. Interestingly, CNTF can stimulate MG proliferation in the uninjured retina, but seems to be neuroprotective in the damaged retina where it inhibits MG proliferation (Kassen et al., 2009). It is not known if CNTF stimulates the expression of dedifferentiation-associated genes in the uninjured retina.

We recently reported that Ascl1a is necessary for MG dedifferentiation and retina regeneration by activating a Lin-28/let-7 signaling pathway (Fausett et al., 2008; Ramachandran et al., 2010a). Interestingly, an Ascl1a/Delta/Notch molecular circuitry maintains retinal progenitors during development of the mammalian retina (Nelson et al., 2009) and Notch signaling components are re-activated during zebrafish retina regeneration (Raymond et al., 2006; Yurco and Cameron, 2007). However, it is not clear if Ascl1a mediates Notch signaling component gene induction in the injured retina. In addition, the consequences of Notch signaling in the injured zebrafish retina remain unknown.

We hypothesize that MG monitor retinal health and when damage occurs, MG secrete factors to stimulate their dedifferentiation by activating ascl1a, notch and other signaling cascades that mediate retina regeneration. Here we report that MG-derived HB-EGF stimulates MG dedifferentiation via an epidermal growth factor receptor (EGFR)/ mitogen-activated protein kinase (MAPK) signaling pathway that impinges on regeneration-associated genes like ascl1a, c-mycb and pax6b, and activates Notch and Wnt/β-catenin signaling pathways. In addition, we uncovered an HB-EGF/Ascl1a/Notch/hb-egfa regulatory feedback loop that helps define the zone of dedifferentiated MG. Importantly; we found that proHB-EGF ectodomain shedding was necessary and sufficient to stimulate MG dedifferentiation into a proliferating population of multipotent progenitors in the injured and uninjured retina. These results indicate that HB-EGF directs MG dedifferentiation following retinal injury and suggest that MG themselves influence their regenerative capacity.

Results

hb-egfa is rapidly induced in the injured retina and necessary for MG dedifferentiation

We hypothesize that MG respond to retinal injury by releasing factors that stimulate their dedifferentiation and initiate retina regeneration. To begin identifying these factors we screened genes encoding epidermal growth factor receptor (EGFR) ligands β_-cellulin_, egf, tgfa and hb-egf for injury-dependent gene induction. Of these, only hb-egfa was highly induced in the injured retina (hb-egfb was undetectable) and this induction could be detected as early as 1 hr post retinal injury (hpi) (Figure 1A). This represents the earliest gene induction in the injured retina recorded to date. In situ hybridization assays combined with BrdU immunofluorescence showed that injury-induced hb-egfa expression was confined to BrdU+ MG-derived progenitors at the injury site (Figure 1B). Morpholino-modified antisense oligonucleotide (MO) knockdown of HB-EGFa showed that its expression was necessary for the generation of proliferating MG-derived progenitors (Figure 1C and 1D).

An external file that holds a picture, illustration, etc. Object name is nihms342390f1.jpg

Injury-dependent hb-egfa induction is necessary for MG dedifferentiation and proliferation

(A) RT-PCR showing the temporal expression pattern of genes encoding EGFR ligands in the uninjured and injured retina. (B) In situ hybridization and BrdU immunofluorescence shows hb-egfa is induced in proliferating progenitors in the injured retina (* marks the injury site). (C) MO-mediated HB-EGFa knockdown reduces the number of proliferating MG-derived progenitors in the injured retina at 4 dpi (those cells that do proliferated do not contain the lissamine-labeled MO, arrowheads). (D) Quantification of (C) at different hb-egfa MO concentrations. *P<0.01. Error bars are standard deviation. Scale bars, 50 µm. Abbreviations: ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer; MO, antisense morpholino-modified oligonucleotide.

HB-EGF stimulates MG dedifferentiation into a multipotent progenitor in the uninjured retina

To investigate if HB-EGF was able to stimulate MG dedifferentiation we took advantage of 1016 tuba1a:gfp transgenic fish that report MG dedifferentiation and proliferation by restricting GFP expression to MG-derived progenitors (Fausett and Goldman, 2006). We tested recombinant soluble human HB-EGF because of its commercial availability and its ability to activate the zebrafish EGFR (Goishi et al., 2003). Human recombinant HB-EGF is encoded by amino acids 1–148 of pro-HB-EGF which is a membrane spanning protein that undergoes ectodomain shedding, releasing a soluble extracellular portion that contains an EGF-like domain that binds the EGFR (Schneider and Wolf, 2009). This domain shares 63% amino acid identity with zebrafish HB-EGF, including 6 highly conserved and critical cysteine residues (Figure S1). Recombinant HB-EGF or PBS + BSA (vehicle) was intravitreally injected into the eye once daily for 2 days, being careful not to damage the retina, which in vehicle-injected retinas would be reflected by increased MG GFP expression and BrdU incorporation. Three days later and 3 hrs prior to sacrifice, fish received an intraperitoneal (IP) injection of BrdU. Retinas were then processed for GFP, BrdU and glutamine synthetase (GS, MG marker) immunofluorescence. Remarkably, HB-EGF induced widespread GFP expression and BrdU incorporation throughout the retina in cells that spanned the inner nuclear layer (INL) and morphologically resembled MG (Figure 2A and 2B). GS immunofluorescence confirmed that these GFP+/BrdU+ cells were indeed derived from MG (Figure 2B). RT-PCR and in situ hybridization assays showed HB-EGF activated a similar set of early response genes as those induced following retinal injury (Figures 2C, 2D and S2A). In contrast, intravitreal injection of EGF had no detectable effect on MG proliferation.

An external file that holds a picture, illustration, etc. Object name is nihms342390f2.jpg

HB-EGF stimulates MG dedifferentiation into multipotent retinal progenitors in the uninjured retina

(A) Diagram of experimental protocol and low magnification photomicrographs showing HB-EGF injection into the eye of 1016 tuba1a:gfp fish stimulates GFP expression and BrdU incorporation throughout the INL of the uninjured retina. Scale bar, 150 µm. (B) High magnification view of (A) showing GFP+/BrdU+ cells express the MG marker glutamine synthetase (GS). Scale bar, 50 µm. (C) RT-PCR and (D) in situ hybridization assays showing HB-EGF stimulates the expression of genes characteristic of dedifferentiated MG. In (D), the PBS + BSA treated retinas did not show an in situ hybridization signal above background for any of the probes tested – data only shown for the ascl1a probe. Scale bars, 50 µm. (E) The matrix metalloprotease inhibitor, GM6001, suppresses injury-dependent proliferation of MG-derived progenitors and this regulation is rescued by soluble HB-EGF. (F) Quantification of the effects observed in (E). (G) BrdU and retinal cell type-specific immunfluorescence shows HB-EGF generated progenitors in the uninjured retina at 14 days post injection are multipotent and generate all major retinal cell types. Scale bar, 50 µm. (H) Quantification of regenerated cell types observed in (G). *P<0.05. Error bars are standard deviation. Abbreviations: as in Figure 1. See also Figures S1 and S2.

The above data suggested that soluble HB-EGF was sufficient to mediate MG dedifferentiation and proliferation. To determine if membrane spanning proHB-EGF also was able to signal MG proliferation we inhibited ectodomain shedding by immersing fish in water containing the pan-metalloproteinase inhibitor GM6001. GM6001 was previously shown to inhibit HB-EGF ectodomain shedding and signaling (Stoll et al., 2010; Xu et al., 2004). We found that GM6001 inhibited injury-induced MG proliferation and this inhibition was rescued with soluble HB-EGF (Figure 2E and 2F). Therefore, proHB-EGF ectodomain shedding appears to be required for its action on MG.

We used a BrdU lineage tracing strategy to test if HB-EGF-induced dedifferentiated MG in the uninjured retina were multipotent. For these experiments HB-EGF was injected daily for 2 days into the eyes vitreous and 3 days later fish received an IP injection of BrdU before being sacrificed 10 days later (Figure 2A). Immunofluorescence was used to identify cells that co-label with anti-BrdU and anti-retinal cell type-specific antibodies. We found that HB-EGF-induced progenitors generated zpr1+ photoreceptors, GS+ MG, HuC/D+ amacrine cells, PKC+ bipolar cells and Zn5+ ganglion cells in the uninjured retina (Figure 2G). Although HB-EGF stimulated the formation of all major retinal cell types, quantification revealed that HB-EGF-induced progenitors preferentially differentiate into photoreceptors (Figure 2H, compare with DMSO treated injured retinas in Figure 6B). Therefore, HB-EGF-induced progenitors exhibit characteristics typical of injury-induced MG-derived retinal stem cells.

An external file that holds a picture, illustration, etc. Object name is nihms342390f6.jpg

Notch signaling regulates the differentiation of MG-derived progenitors

(A, C) Diagram of the experimental protocol and representative photomicrographs showing BrdU and retinal cell type-specific immunofluorescence. (B, D) Quantification of double-labeled cell types. *P<0.01. Error bars are standard deviation. Scale bar, 50 µm. Abbreviations: as in Figure 1.

HB-EGF signals MG dedifferentiation via an EGFR/MAPK/Ascl1a signaling cascade

HB-EGF is a ligand for the EGFR, which often links extracellular signals with changes in gene expression via a MAPK signaling pathway (Lemmon and Schlessinger, 2010). Therefore we investigated if EGFR activation and MAPK signaling were required for MG dedifferentiation in the injured retina. For this analysis we exposed injured retinas to the EGFR inhibitor PD153035 (Fry et al., 1994; Fry et al., 1997), the MAPK (ERK1 and 2) inhibitors PD98059 (Dudley et al., 1995) or SL327 (Atkins et al., 1998) and DMSO in separate experiments. Fish also received a pulse of BrdU 3 hrs prior to sacrifice at 4 days post retinal injury (dpi). BrdU immunofluorescence showed that EGFR inhibition reduced the number of BrdU+ progenitors by ~75% while MAPK inhibition reduced progenitors by ~65–75% (Figure 3A and 3B). These inhibitors also blocked injury-dependent induction of regeneration-associated genes like ascl1a, lin28, pax6b, deltaA, notch1 and her4 (Figure 3C and S3). These data suggest that an EGFR/MAPK signaling pathway mediates MG dedifferentiation in the injured retina.

An external file that holds a picture, illustration, etc. Object name is nihms342390f3.jpg

HB-EGF and retinal injury stimulate MG dedifferentiation via an EGFR/MAPK/Ascl1a signaling cascade

(A) EGFR inhibition (PD153035) or MAPK inhibition (PD98059, SL327) reduces the number of proliferating progenitors generated following retinal injury. (B) Quantification of (A). *P<0.01. (C) RT-PCR shows that EGFR or MAPK inhibition blocks induction of genes associated with MG dedifferentiation. (D) EGFR or MAPK inhibition reduces the number of proliferating progenitors generated following HB-EGF treatment of the uninjured retina. (E) HB-EGFa knockdown suppresses induction of regeneration associated genes in the injured retina. [hb-egf MO] is in mM. C-MO is 0.5 mM. (F) Ascl1a knockdown has little effect on hb-egfa induction in the injured retina (MOs are 0.25 mM). (G) Real-time PCR quantification of (E and F) normalized to uninjured value. The hb-egf MO is at 0.5 mM and ascl1a MOs are at 0.25 mM. *P<0.01. (H) ascl1a and hb-egfa in situ hybridization assays along with BrdU immunofluorescence show co-localization in the injured retina. (I) MO-mediated knockdown of Ascl1a in the presence of HB-EGF suppresses HB-EGF-dependent generation of BrdU+ progenitors. BrdU+ progenitors that do form in the Ascl1a knockdown retina generally lack lissamine labeled MO (arrowheads). Control MO does not block the generation of BrdU+ progenitors (arrows). (J) Quantification of BrdU+ cells in (I). *P<0.01. Error bars are standard deviation. All scale bars, 50 µm. Abbreviations: as in Figure 1. See also Figure S3.

To determine if the EGFR/MAPK signaling pathway also mediates HB-EGF’s effect on the uninjured retina we delivered HB-EGF into the vitreous of an uninjured retina and immersed fish in water +/− the EGFR inhibitor PD153035 or the MAPK inhibitors PD98059 or SL327; fish then received an IP injection of BrdU 3 hrs prior to sacrifice 4 days later. EGFR or MAPK inhibition suppressed progenitor formation (BrdU+ cells) in HB-EGF treated retinas (Figure 3D). Together these data suggested that HB-EGF/EGFR/MAPK signaling is a very early event during retina regeneration that controls MG dedifferentiation, at least in part, by regulating expression of regeneration-associated genes.

We next investigated if HB-EGF might signal in a paracrine and/or autocrine manner by examining the cells expressing EGFR. We found that most dedifferentiated and proliferating MG express egfr mRNA (Figure S2B). Since HB-EGF is expressed in essentially all dedifferentiated and proliferating MG (Figure 1B), these data suggest HB-EGF may signal in both a paracrine and autocrine manner.

Injury-dependent induction of ascl1a, pax6b and c-mycb are characteristic of dedifferentiated MG (Fausett et al., 2008; Ramachandran et al., 2010a). Previous studies showed Ascl1a controls c-mycb expression, and that Ascl1a and Pax6b act in parallel but independent pathways (Ramachandran et al., 2011). Here we report that knockdown of HB-EGFa blocks injury-dependent ascl1a and pax6b gene induction (Figure 3E, 3G and S3), while Ascl1a knockdown had little effect on injury-dependent induction of hb-egfa (Figure 3F and 3G). Therefore, HB-EGFa acts upstream of Ascl1a and Pax6b. Consistent with this regulatory hierarchy, hb-egfa and ascl1a are co-expressed in dedifferentiated MG-derived progenitors (Figure 3H) and MO-mediated Ascl1a knockdown suppressed BrdU+ progenitor formation in response to HB-EGF stimulation (Figure 3I and 3J). These results indicate that injury-dependent induction of the HB-EGF/EGFR/MAPK signaling cascade impinges on ascl1a and pax6b gene expression to control MG dedifferentiation and retina regeneration.

The HB-EGF/EGFR/MAPK/Ascl1a signaling cascade regulates injury-dependent induction of Notch signaling components

Ascl1 impacts Notch signaling during retina development by regulating the expression of genes encoding Notch ligands (Nelson et al., 2009; Nelson and Reh, 2008). Notch signaling components are also induced in the injured adult zebrafish retina (Raymond et al., 2006; Yurco and Cameron, 2007). Therefore we investigated if HB-EGF/MAPK-dependent induction of Ascl1a in the injured retina also regulated the expression of Notch signaling components. We used RT-PCR, Real-time qPCR and in situ hybridization assays to confirm and extend previous studies showing Notch signaling component gene expression in the injured retina (Figures 4A and S4). In general, notch1, deltaA, deltaB, deltaC and deltaD mRNAs were induced at the injury site within 6–24 hpi (Figures 4A, S4A and S4B). In situ hybridization assays combined with BrdU immunofluorescence showed notch1 was predominantly induced in BrdU+ proliferating MG-derived progenitors (Figure S4B, arrowheads), while ligand encoding delta genes were preferentially induced in BrdU− cells immediately neighboring BrdU+ progenitors (Figure S4B, arrows). To determine if the BrdU−/delta+ cells were differentiating, we co-stained retinal sections with anti BrdU and anti-HuC/D antibodies (Figure S4C). This analysis showed the expected HuC/D expression in amacrine and ganglion cells while the BrdU−/delta+ cells were HuC/D− indicating that these cells are not differentiating ganglion or amacrine cells. However we cannot rule out that these delta+ cells are differentiating photoreceptors, MG or other retinal interneurons, or the possibility that they are cycling progenitors where delta expression is restricted from the S phase of the cell cycle.

An external file that holds a picture, illustration, etc. Object name is nihms342390f4.jpg

HB-EGF and Ascl1a regulate Notch signaling component genes in the injured retina

(A) Real-time PCR shows injury-dependent induction of Notch signaling components. (B) RT-PCR shows HB-EGF activates Notch signaling component genes in the uninjured retina. (C) RT-PCR shows MO-mediated HB-EGFa knockdown suppresses injury-dependent induction of Notch signaling component genes. (D) Real-time PCR quantification of (C). *P<0.01. (E) RT-PCR shows Ascl1a knockdown suppresses injury-dependent induction of Notch signaling component genes. (F) Real-time PCR quantification of (E). *P<0.01. (G) Ascl1a knockdown suppresses HB-EGF-dependent induction of notch1 and her4 mRNAs. Error bars are standard deviation. Scale bar, 50 µm. Abbreviations: as in Figure 1. See also Figures S4 and S5.

The her4 gene is a downstream target of Notch; thus her4 promoter activity can serve as a probe for Notch signaling (Takke et al., 1999). We found that her4 gene expression was induced within 6 hpi (Figures 4A and S4A) and used her4.1:ERT2CreERT2;β-actin2:loxP-mCherry-loxP-gfp double transgenic fish (Boniface et al., 2009) to show that this induction was a result of increased her4 promoter activity in proliferating MG-derived progenitors (Figure S4D). These data suggest Notch signaling is activated in MG-derived progenitors.

RT-PCR analysis of uninjured retinas treated with HB-EGF indicated that mRNAs encoding Notch signaling components deltaA, notch1 and her4 are regulated in an HB-EGF-dependent fashion (Figure 4B). Indeed, MO-mediated HB-EGF knockdown in the injured retina prevented the induction of these genes (Figures 4C, 4D and S5A) as did EGFR or MAPK inhibition (Figure 3C). Furthermore, MO-mediated Ascl1a knockdown suppressed both HB-EGF and injury-dependent induction of Notch signaling component genes (Figures 4E, 4F, 4G, S5B and S5C). Therefore, the HB-EGF/EGFR/MAPK/Ascl1a signaling cascade activates Notch signaling component gene expression in the injured retina.

Notch signaling feeds back to inhibit hb-egfa and ascl1a gene expression to limit the zone of proliferating progenitors in the injured retina

To investigate if Notch signaling is necessary for injury-dependent MG dedifferentiation and proliferation we took advantage of the γ-secretase inhibitor DAPT, which prevents Notch cleavage and subsequent signaling by its intracellular domain, NICD (Geling et al., 2002). We inhibited Notch signaling in 1016 tuba1a:gfp transgenic fish that express GFP in proliferating MG-derived progenitors (Fausett and Goldman, 2006). Surprisingly, we found that DAPT treatment expanded the zone of dedifferentiated (GFP+) and proliferating (BrdU+) progenitors in the injured retina (Figure 5A, data shown for 4 dpi, similar results were observed at 2 dpi). Quantification revealed a 3–4-fold increase in GFP+ and BrdU+ cells that was accompanied by an expansion of the area these cells occupied (Figure S6A). RT-PCR (Figure 5B) and in situ hybridization assays (Figure 5C and 5E) showed that DAPT stimulated expression of regeneration-associated genes like ascl1a and hb-egfa and that this expression was restricted to the injury site. Consistent with the idea that HB-EGF mediates the effects of DAPT on MG proliferation we found that MAPK or EGFR inhibition suppressed DAPT-dependent expansion of MG proliferation (Figure 5F). ascl1a:gfp transgenic fish showed that injury and DAPT-dependent induction of ascl1a mRNA was a result of increased ascl1a promoter activity (Figures 5D and S6B) that was confined to MG-derived progenitors (Figure S6B and S6C). Importantly, DAPT did not stimulate MG dedifferentiation or proliferation in uninjured retinas. Collectively our data suggests that an HB-EGF/Ascl1a/Notch/hb-egfa signaling loop modulates the response of MG to injury and helps define the zone of dedifferentiated MG in the injured retina.

An external file that holds a picture, illustration, etc. Object name is nihms342390f5.jpg

Notch inhibition stimulates MG proliferation and expands the zone of dedifferentiated MG in the injured retina via induction of hb-egfa gene expression

(A) DAPT treatment of 1016 tuba1a:gfp transgenic fish results in an expansion of the zone of dedifferentiated MG (GFP+/BrdU+) in the injured retina. Scale bar, 50 µm. (B) RT-PCR (gel) and Real-time PCR (4 dpi, graph) showing DAPT induces ascl1a, hb-egfa and egfr mRNA expression in the injured retina. **P<0.01 for ascl1a and hb-egfa; *P<0.05 for egfr. (C) In situ hybridization showing DAPT expands ascl1a mRNA expression in the injured retina (* marks the injury site). Scale bar, 50 µm. (D) ascl1a:gfp transgenic fish exhibit injury- and Notch-dependent regulation of transgene expression. Scale bar, 50 µm. (E) DAPT stimulates hb-egfa mRNA expression in proliferating MG-derived progenitors of the injured retina. Scale bar, 50 µm. (F) MAPK inhibition (PD98059 and SL327) and EGFR inhibition (PD153035) suppress DAPT-dependent expansion of MG proliferation in the injured retina. *P<0.01. (G) BrdU immunofluorescence on retinal sections reveals BrdU incorporation in the injured retinas of WT and hsp70:gal4;uas:nicd-myc transgenic fish over-expressing NICD. Scale bar, 50 µm. *P<0.01. (H) In situ hybridization assays for hb-egfa and ascl1a mRNA expression in WT and hsp70:gal4;uas:nicd-myc fish over-expressing NICD (* marks the injury site). Scale bar, 50 µm. (I) RT-PCR shows ascl1a, hb-egfa and pax6b mRNA expression are inhibited by NICD over-expression. (J) NICD-myc overexpression suppresses HB-EGF-induced cell proliferation in the uninjured retina. BrdU+ cells in HB-EGF-treated retinas over-expressing NICD-myc are generally NICD negative (arrows). Scale bar is 150 µm in low magnification images and 50 µm in higher magnification images. Error bars are standard deviation. Abbreviations: as in Figure 1. See also Figure S6.

The observation that Notch signaling inhibits MG dedifferentiation and proliferation in the injured zebrafish retina was surprising in light of reports indicating Notch signaling stimulated proliferation of MG-derived progenitors in birds and mammals (Del Debbio et al., 2010; Ghai et al., 2010; Hayes et al., 2007). Therefore, to further investigate the role of Notch signaling in the injured zebrafish retina we took advantage of hsp70:gal4;uas:nicd-myc double transgenic fish that allow one to conditionally express an NICD-Myc fusion protein via heat shock-dependent Gal4 expression (Scheer et al., 2001). NICD over-expression in MG following retinal injury was confirmed by anti-Myc and anti-GS co-immunfluorescence along with widespread expression of the NICD target gene, her4 (Figure S6D and S6E). In contrast to our results with DAPT, NICD over-expression reduced the number of injury-induced BrdU+ progenitors (Figure 5G), suppressed the expression of ascl1a and hb-egfa (Figure 5H and 5I) and reduced the number of progenitors generated in response to HB-EGF treatment (Figure 5J). Those cells that remained BrdU+ generally lacked NICD (Figures 5J and S6D, arrows).

Surprisingly, at 2 dpi we observed little effect of NICD over-expression on BrdU+ progenitor formation (Figure 5G). This suggested that Notch signaling may inhibit progenitor expansion, but not the initial MG division to generate a progenitor. Indeed, the size of neurogenic clusters was reduced from ~8 cells to ~2 cells in hsp70:gal4;uas:nicd-myc double transgenic fish over-expressing NICD (Figure S6F). Quantification of the number of neurogenic clusters comprised of 3 or less cells, identified 6% in injured retinas from wild type fish that increased to 70% in injured retinas from fish over-expressing NICD. In addition, NICD overexpression from 0–2 dpi had little effect on the initiation of progenitor proliferation, while NICD overexpression from 3–4 dpi dramatically suppressed progenitor proliferation (Figure S6G). This NICD-dependent reduction in neurogenic cluster size was reminiscent of the effects of Pax6a or Pax6b knockdown (Thummel et al., 2010). Therefore, we investigated if Notch signaling regulated Pax6 expression. Interestingly, we found that NICD over-expression inhibited pax6b expression (Figure 5I), which may contribute to the reduced proliferation observed in injured NICD-treated retinas. Consistent with this idea, Pax6a or Pax6b knockdown dramatically suppressed DAPT-dependent expansion of MG proliferation and reduced neurogenic cluster size from ~8 cells/cluster in control fish to ~2 cells/cluster following Pax6 knockdown (Figure S6H).

Notch signaling regulates differentiation of MG-derived progenitors

To determine if Notch signaling also affected progenitor migration and differentiation we treated fish with DMSO or DAPT from 0–4 dpi. At 4 dpi fish received an IP injection of BrdU and 10 days later, fish were sacrificed and immunofluorescence was used to identify the cell types that were regenerated from the BrdU+ progenitors. Interestingly, DAPT treatment stimulated an increase in the number of BrdU+ MG (GS+) in the INL with a concomitant decreased in BrdU+ photoreceptors (Zpr1+), amacrine/ganglion cells (HuC/D+) and bipolar cells (PKC+) (Figure 6A and 6B). DAPT treatment from 0–2 or 5–7 dpi had little effect on progenitor differentiation suggesting a fairly narrow window of time when Notch influences progenitor differentiation.

We used a similar lineage tracing strategy to further characterize the effects of Notch signaling on progenitor differentiation in hsp70:gal4;uas:nicd-myc fish where NICD was over-expressed from 0–4 dpi. In contrast to DAPT treatment, NICD over-expression resulted in an increase in BrdU+ photoreceptors (Zpr1+) at the expense of MG, bipolar and ganglion cells (Figure 6C and 6D). Interestingly, NICD overexpression had a similar effect when overexpressed 0–2 dpi and 5–7 dpi, suggesting it can drive progenitor differentiation at different times following retinal injury (Figure S6I and S6J). These differentiated cells appeared to emanate from a small population of progenitors that remained in the NICD overexpressing fish at 4 dpi (Figure S6K). Together these data suggest that Notch signaling not only regulates MG dedifferentiation and proliferation, but also regulates progenitor differentiation.

HB-EGF activates a Wnt/β-catenin signaling cascade to control proliferation of MG-derived progenitors

We recently found that activation of a canonical Wnt/β-catenin signaling cascade is necessary for the generation of a proliferating population of retinal progenitors in the injured retina (Ramachandran et al., 2011). Consistent with the idea that HB-EGF mediates the effects of retinal injury on the Wnt signaling pathway, we found that HB-EGF stimulated β-catenin nuclear accumulation in MG derived progenitors (Figure 7A) and gfp mRNA expression in MG of the Wnt signaling reporter fish TOP:dGFP (Figure 7B) (Dorsky et al., 2002). Pharmacological destabilization of β-catenin with pyrvinium (Thorne et al., 2010), dramatically reduced the number of HB-EGF induced BrdU+ progenitors (Figure 7C and 7D), and suppressed HB-EGF-dependent induction of regeneration-associated genes (Figure 7E). In addition, Ascl1a knockdown suppressed HB-EGF-induced nuclear accumulation of β-catenin (Figure 7F). These results establish Ascl1a and Wnt/β-catenin signaling as downstream effectors of HB-EGF. Furthermore the observation that Ascl1a expression is necessary for β-catenin induction and Wnt signaling is necessary for ascl1a induction suggests a positive feedback loop where low levels of Wnt signaling contributes to the injury-dependent induction of regeneration-associated genes and that the resulting Ascl1a expression further stimulates Wnt signaling to levels where β-catenin can be readily detected.

An external file that holds a picture, illustration, etc. Object name is nihms342390f7.jpg

HB-EGF activates a Wnt/β-catenin signaling cascade

(A) β-catenin immunofluorescence shows HB-EGF stimulates β-catenin stabilization in proliferating progenitors of the uninjured retina. Scale bar, 50 µm. (B,C) Pyrvinium (Pyrvin)-mediated inhibition of Wnt/β-catenin signaling suppresses HB-EGF-dependent generation of proliferating progenitors in the uninjured retina. Error bars are standard deviation. *P<0.01. Scale bar, 50 µm. (D) RT-PCR shows pyrvinium suppresses HB-EGF-dependent induction of regeneration associated genes in the uninjured retina. (E) In situ hybridization and BrdU immunofluorescence shows HB-EGF-dependent induction of transgene expression in TOP:dGFP Wnt signaling reporter fish is inhibited by pyrvinium (pyrvin). (F) Ascl1a knockdown suppresses HB-EGF-dependent β-catenin stabilization and BrdU incorporation in the uninjured retina. (G) Summary of signaling cascades and genes regulated by HB-EGF.

Discussion

MG dedifferentiation into a cycling population of multipotent retinal progenitors underlies successful retina regeneration in zebrafish. Mechanisms that convey cellular damage to MG and stimulate their dedifferentiation have remained elusive. Here we report that HB-EGF, an EGFR ligand, is necessary and sufficient for inducing MG dedifferentiation into a cycling population of retinal progenitors. HB-EGF appears to stimulate MG dedifferentiation via a MAPK signaling pathway that regulates expression of regeneration-associated genes like ascl1a, d_eltaA_, pax6b and c-mycb. Although HB-EGF can stimulate ascl1a expression via a MAPK signaling pathway, we also found that Ascl1a feeds back and inhibits hb-egf expression via a Notch signaling pathway. Inhibition of this Notch signaling pathway in the injured retina expands the zone of MG-derived progenitors, presumably due to increased HB-EGF that stimulates ascl1a expression. In addition to regulating MG dedifferentiation and proliferation, we also found that Notch signaling contributes to progenitor differentiation, stimulating photoreceptor formation at the expense of glia. HB-EGF also activated the Wnt/β-catenin signaling cascade which we recently showed was necessary for MG proliferation (Ramachandran et al., 2011). Finally, we showed that HB-EGF is sufficient to induce MG dedifferentiation and the generation of multipotent retinal progenitors in the uninjured retina. A summary of these findings is presented in Figure 7G.

HB-EGF is an injury-induced EGFR ligand that stimulates the formation of multipotent retinal progenitors

HB-EGF is a member of a family of ligands that activate the EGFR and include EGF, TGFα, β-cellulin, epiregulin and amphiregulin (Harris et al., 2003). HB-EGF is well suited for defining the zone of dedifferentiated MG at the injury site. It is first synthesized as a transmembrane protein (pro-HB-EGF) and can be cleaved by metalloproteinases (MMPs) within a juxtamembrane domain on the cell surface to release soluble HB-EGF (Higashiyama et al., 1991; Higashiyama et al., 1992; Higashiyama and Nanba, 2005). Although pro-HB-EGF can signal to receptors on its own and neighboring cells in a non-diffusible manner, ectodomain shedding allows soluble HB-EGF to activate the EGFR at a distance. In addition, HB-EGF has a high affinity for heparin and heparin sulfate which can restrict its diffusion resulting in high local ligand concentrations. Using an inhibitor of MMPs we found that pro-HB-EGF ectodomain shedding is required for MG dedifferentiation.

We used soluble recombinant human HB-EGF to investigate its effects on MG dedifferentiation and proliferation in the uninjured retina. Remarkably, HB-EGF stimulated MG dedifferentiation as indicated by the induction of genes like, ascl1a, deltaA, pax6b and c-mycb. Although Ascl1a mediates deltaA and c-mycb expression, the mechanisms underlying injury-dependent induction of ascl1a and pax6b have remained elusive. Our data suggests an HB-EGF/EGFR/MAPK pathway regulates these genes and provides a mechanism linking injury-dependent extracellular cues to the activation of genes known to be necessary for retina regeneration.

BrdU-based lineage tracing studies showed that HB-EGF-treated MG entered the cell cycle and differentiated into all major retinal cell types. Interestingly, EGF treatment of the mouse retina did not stimulate MG proliferation unless retinas were also damaged (Karl et al., 2008). This differential response in uninjured and injured mouse retina may be due to injury dependent activation of the EGFR (Close et al., 2006). However, even in the damaged mouse retina where EGFR was induced, regeneration was restricted to a small number of amacrine cells (Karl et al., 2008). These distinctly different outcomes of EGFR activation in zebrafish and mice suggests EGFR activation in mice does not completely activate a program of multipotency that allows dedifferentiated and proliferating MG to regenerate all retinal cell types. Indeed, EGF stimulation of the injured mouse retina did not induce Ascl1 gene expression (Karl et al., 2008), which is an essential regeneration-associated gene in zebrafish (Fausett et al., 2008; Ramachandran et al., 2010a).

We found that an EGFR/MAPK signal transduction cascade is essential for zebrafish retina regeneration and mediates the effects of HB-EGF on the uninjured retina. MAPK signaling was previously reported to be required for proliferation of MG-derived progenitors in the postnatal chick retina (Fischer et al., 2009a; Fischer et al., 2009b) and likely mediates the effects of EGF on MG proliferation in the injured mouse retina (Karl et al., 2008). However, the mechanism by which MAPK signaling stimulates MG dedifferentiation and proliferation remains unknown in birds and mammals. Our data suggests that the HB-EGF/EGFR/MAPK signaling pathway regulates genes, like ascl1a and pax6b that are necessary for MG dedifferentiation and progenitor proliferation, along with signaling pathways that impinge on progenitor proliferation (like Notch and Wnt/β-catenin). It is likely that these and other downstream targets of the HB-EGF/EGFR/MAPK signaling cascade will determine the balance between MG dedifferentiation, proliferation and multipotency.

Our data raise the interesting possibility that MG may be a source of secreted factors that stimulate their dedifferentiation in response to injury. It is well known that MG processes make contact with cells throughout the retina and therefore are well positioned to sense retinal damage. In addition, MG participate in phagocytosis of dying retinal photoreceptors (Bailey et al., 2010). Thus, injury-induced perturbations in MG and their interactions with other retinal cell types (alterations in signaling by diffusible factors or cell contact, increased phagocytosis, etc.) may stimulate release of basally produced HB-EGF from MG at the injury site further stimulating MG dedifferentiation and hb-egf gene expression.

Interestingly, HB-EGF is expressed by MG in proliferative vitreoretinopathy which is thought to represent a poorly regulated retinal wound healing process where MG hyperproliferate and cause glial scarring (Hollborn et al., 2005). Therefore, simply expressing HB-EGF may not be sufficient to stimulate retina regeneration in mammals and may even be detrimental if it is uncontrolled. Perhaps other factors are needed to collaborate with HB-EGF to help elicit MG dedifferentiation and multipotency in mammals. Candidates include FGF2 and insulin that in combination stimulate MG proliferation in the postnatal chick retina (Fischer et al., 2002) and CNTF that stimulates MG proliferation in zebrafish (Kassen et al., 2009). Although these factors can stimulate MG proliferation it is not known if they are necessary for MG dedifferentiation and regeneration. Furthermore, it is possible that epigenetic constraints in mammals may prevent these factors from awakening gene expression programs that direct cellular dedifferentiation and multipotency. Further study on the collaboration of HB-EGF with other factors involved in retina regeneration and the role epigenetics plays during retina regeneration are important areas of future investigation.

The role of Notch signaling in retina regeneration

One of the most intriguing consequences of Notch inhibition in the injured zebrafish retina is the expanded zone of dedifferentiated and proliferating progenitors. Notch inhibition did not stimulate MG dedifferentiation and proliferation in the uninjured retina, suggesting that Notch was acting on an injury-induced signaling cascade. Indeed, we identified an HB-EGF/MAPK/Ascl1a/Notch/hb-egfa signaling loop where Notch inhibition stimulates HB-EGF production. Excess HB-EGF will presumably diffuse across a larger area of retina surrounding the injury site and thereby increase the zone of dedifferentiated MG. The HB-EGF/Ascl1a/Notch/hb-egfa feedback loop appears to help maintain a stable level of HB-EGF so that the number of MG recruited to repair the retina is appropriate for the severity of the injury.

It is interesting that DAPT treatment of injured retinas resulted in an increased number of progenitors whose fate was biased towards MG, while NICD over-expression largely inhibited progenitor formation and of those progenitors that did form, NICD biased their differentiation towards photoreceptors. These results are different from the effects of Notch inhibition and activation reported in the developing mouse retina where Notch activity stimulates retinal stem cells to precociously differentiate into MG and Notch inhibition led to an enhancement of photoreceptor production (Jadhav et al., 2006; Scheer et al., 2001; Yaron et al., 2006). These results suggest that in the adult injured retina, Notch signaling affects both MG dedifferentiation and progenitor differentiation and may be more permissive than instructive in specifying cell fate.

The above results are in direct contrast to those reported in the postnatal chick retina and adult mouse retina where Notch signaling stimulated proliferation of MG-derived progenitors (Das et al., 2006; Del Debbio et al., 2010; Ghai et al., 2010; Hayes et al., 2007) and, at least in the postnatal chick retina, inhibited progenitor differentiation (Hayes et al., 2007). However, similar to the injured zebrafish retina, Notch signaling inhibits neural stem cell proliferation in the zebrafish and mouse brain (Chapouton et al., 2010; Imayoshi and Kageyama, 2011; Imayoshi et al., 2010). The reason why Notch signaling has different consequences on proliferation of MG-derived progenitors in zebrafish compared to birds and mice is not clear, nor is it clear if this difference contributes to their different regenerative potentials.

Taken together our data suggests that Notch signaling impacts many stages of retina regeneration, playing both early (MG dedifferentiation and progenitor cell proliferation) and late (cell differentiation) roles. Thus it is important to finely tune Notch signaling so each of these roles can run optimally during retinal repair. We suspect that the HB-EGF/Ascl1a/Notch/hb-egfa signaling loop is one mechanism to keep Notch signaling levels in tune with regenerative demands.

In summary we have identified an HB-EGF/EGFR/MAPK signaling cascade that impinges on genes and signal transduction cascades involved in all aspects of retina regeneration. This cascade connects retinal damage with MG dedifferentiation. The injury-dependent localization of hb-egfa gene expression to MG at the injury site suggests MG may direct their own dedifferentiation and proliferation. Importantly, we show HB-EGF is necessary and sufficient for stimulating MG dedifferentiation into a population of cycling multipotent progenitors that can regenerate all retinal cell types.

Experimental Procedures

RNA isolation and PCR

All primers used in this study are listed in Supplemental Experimental Procedures. Total RNA was isolated using Trizol (Invitrogen). cDNA synthesis and PCR reactions were as previously described (Fausett et al., 2008; Ramachandran et al., 2010a). Real-time PCR reactions were carried out in triplicate with ABsolute SYBR Green Fluorescein Master Mix (Thermo Scientific) on an iCycler real-time PCR detection system (BioRad).

Retinal lesion and morpholino (MO)-mediated gene knockdowns

Retinas were injured and electroporated with MOs as previously described (Fausett et al., 2008; Ramachandran et al., 2010a). Briefly, fish were anesthetized and the right retina was poked 4 times, once in each quadrant, using a 30g needle inserted through the sclera to the length of the bevel (~5mm). Lissamine-MOs (Gene Tools, LLC) were delivered at the time of injury using the same needle to poke the retina. Approximately 0.5 µl of MO was injected into the eye and uptake by cells was facilitated by electroporation (Fausett et al., 2008; Ramachandran et al., 2010a).

EGFR, metalloproteinase, MAPK and Notch Inhibitors

The γ-secretase inhibitor, DAPT (Calbiochem) was used to inhibit the Notch signaling pathway. Inhibitor stocks (10 mM) were prepared in DMSO. DAPT was used at a final concentration of 40 µM (Chapouton et al., 2010). MAPK inhibitors PD98059 and SL327 (Sigma) were used at 20 µM, and EGFR inhibitors PD153035 (Calbiochem) and PD158780 (Sigma) were used at 10 µM. Metalloproteinase inhibitor, GM6001, was used at 10 µM. Fish were immersed in fish water containing the inhibitor. Control fish were treated with DMSO (1:200).

Growth factor injections, BrdU incorporation and lineage tracing

Fish were anesthetized and the left eye (control) was injected with 2 µl of vehicle (PBS plus 0.1% BSA) and the right eye was injected with 200 ng of human recombinant HB-EGF (R&D Systems). Solutions were delivered into the eye’s vitreous through the cornea, being careful not to injure the retina. For BrdU incorporation, a solution of 20 mM BrdU was injected (20 µl) intraperitoneally into the anesthetized fish.

Heat shock

The day before retinal injury, Wt and hsp70:gal4 /;uas:nicd-myc fish were immersed in a water bath at 37.5 °C for 2 hr and then returned to their tank at 28 °C. Over the next 2–4 days this heat shock was repeated 3× each day at equally spaced intervals except the duration was reduced to 1 hr at 37.5 °C. Wt fish were heat shocked as controls.

Tissue preparation, immunohistochemistry and in situ hybridization

Adult fish were overdosed with tricaine and the eyes were dissected; the lens was removed and the eye cup was fixed in 4% paraformaldehyde. Fixed samples were prepared for immunofluorescence as previously described (Fausett and Goldman, 2006; Ramachandran et al., 2010a; Ramachandran, 2010). The antibodies used in this study were diluted in PBS plus 0.1% Triton X-100 as previously described (Fausett and Goldman, 2006; Ramachandran et al., 2010a; Ramachandran, 2010). For BrdU staining, sections were treated with 2N HCl at 37 °C for 20 min, rinsed with 0.1M sodium borate solution (pH 8.5) for 10 min, and then processed using standard immunohistochemical procedures.

In situ hybridization was performed as described previously (Barthel and Raymond, 2000). Digoxigenin-labeled RNA probes were prepared using the DIG RNA labeling kit (Roche Diagnostics). Primers for amplifying her4, notch and hb-egfa are listed in Table S1.

Cell quantification and statistical analysis

Cell counts were determined by counting BrdU+ or GFP+ cells in retinal sections visualized using fluorescence microscopy. All sections of the retina were examined and at least 3 individuals were used. The area occupied by GFP+ cells was measured with ImageJ software. Statistical comparisons were conducted using a two-tailed unpaired Student’s _t_-test. Significance was established at p<0.05. Error bars are standard deviation (SD).

Supplementary Material

01

Acknowledgements

This research was supported by NEI grant RO1 EY 018132 from the NIH and grant G2010013 from the American Heath Assistance Foundation-National Glaucoma Research. We thank Bruce Riley (Texas A & M University) for providing hsp70:gal4;uas:nicd-myc transgenic fish; Wenbiao Chen (Vanderbilt University) for providing her4.1: ERT2CreERT2 transgenic fish; Richard Dorsky (University of Utah) for providing TOP:dGFP fish; David Hyde (University of Notre Dame) for providing pax6a and pax6b targeting MOs; David Turner (University of Michigan) for providing anti-myc antibody; James Elder and Stefan Stoll (University of Michigan) for providing HB-EGF and GM6001; Bruce Appel (Vanderbilt University) for providing pBSK-deltaA and deltaD; Julian Lewis (University College London) for providing pBSK-deltaB, deltaC; and We Ge (Chinese University of Hong Kong) for providing zebrafish egfr cDNA. We thank members of the Goldman lab for comments and advice on these studies and Randall Karr for fish care.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References