Histone Deacetylase-Mediated Müller Glia Reprogramming through Her4.1-Lin28a Axis Is Essential for Retina Regeneration in Zebrafish

Summary

Histone deacetylases (Hdacs) play significant roles in cellular homeostasis and tissue differentiation. Hdacs are well characterized in various systems for their physiological and epigenetic relevance. However, their significance during retina regeneration remains unclear. Here we show that inhibition of Hdac1 causes a decline in regenerative ability, and injury-dependent regulation of hdacs is essential for regulating regeneration-associated genes like ascl1a, lin28a, and repressors like her4.1 at the injury site. We show selective seclusion of Hdac1 from the proliferating Müller glia-derived progenitor cells (MGPCs) and its upregulation in the neighboring cells. Hdacs negatively regulate her4.1, which also represses lin28a and essential cytokines to control MGPCs proliferation. Interestingly, Hdacs' inhibition reversibly blocks regeneration through the repression of critical cytokines and other regeneration-specific genes, which is also revealed by whole-retina RNA sequence analysis. Our study shows mechanistic understanding of the Hdac pathway during zebrafish retina regeneration.

Graphical Abstract

Highlights

• •Hdac1, along with other Hdacs, is a key regulator of retina regeneration in zebrafish
• •Hdacs regulate MGPCs' formation through Her4.1/Lin28a/let-7 miRNA axis
• •Hdacs' inhibition reversibly blocks MGPCs' proliferation and retina regeneration
• •Hdacs/Her4.1 interplay regulates essential cytokines during retina regeneration

Molecular Mechanism of Gene Regulation; Molecular Neuroscience; Transcriptomics; Model Organism

Introduction

Unlike mammals, vertebrates of piscine and amphibian groups possess very remarkable regenerative potential in almost all tissue types and organs and thus have been studied extensively for a greater understanding of the molecular mechanisms underlying tissue regeneration (Ail and Perron, 2017, Gemberling et al., 2013, Goldman, 2014, Kyritsis et al., 2012, Mokalled et al., 2016, Poss et al., 2002, Rabinowitz et al., 2017, Singh et al., 2012, Wan and Goldman, 2016). Zebrafish, a piscine member, is an excellent model to study the molecular mechanisms of regeneration in complex tissues like brain and retina. Retina regeneration in zebrafish depends mostly on the de-differentiation of Müller glia (MG), which then reprogram themselves to produce progenitor cells with stem cell-like properties (MG-derived progenitor cells [MGPCs]). Later, these MGPCs differentiate into various neurons of the retina and the MG itself with the help of different contributing factors (Fausett and Goldman, 2006, Ramachandran et al., 2010b). These include growth factors (Gramage et al., 2015, Russell, 2003, Wan et al., 2012, Zhao et al., 2014), cytokines (Wan et al., 2014, Zhao et al., 2014), gene transcription factors (Nelson et al., 2012, Ramachandran et al., 2010a, Ramachandran et al., 2012b, Thummel et al., 2010, Wan et al., 2014), epigenome modifiers (Powell et al., 2012, Powell et al., 2013), cell-cycle regulators (Luo et al., 2012, Ramachandran et al., 2011, Ramachandran et al., 2012b), and differentiation factors (Munderloh et al., 2009), which get dynamically regulated with precise orchestration among themselves, leading to the restoration of normal vision (Sherpa et al., 2008). Conversely, mammalian retina often undergoes reactive gliosis upon injury, which eventually leads to scar formation (Bringmann et al., 2009). Although good knowledge of the molecular regulatory networks that form the foundation of regeneration cascade exist in zebrafish, several important gene regulatory networks remain enigmatic. One such unknown territory is the involvement of epigenome modifiers like histone deacetylases (Hdacs). Unraveling crucial gene regulatory network underlying retina regeneration, with temporal and spatial precision, would be possible only through an understanding of the epigenome modifiers such as Hdacs and the subsequent genes involved. Finally, we anticipate that the cellular and molecular regulatory mechanisms that bring about regeneration in the zebrafish could pave the way for curing mammalian retinal damage.

Hdacs usually function as catalysts for the deacetylation of acetyl-L-lysine side chains of histone proteins by forming transcriptional co-repressor complexes, which typically enable the alteration of chromatin structure and repress gene transcription. There are 4 major classes of Hdacs with diverse functions including the modifications of non-histone targets in normal cell biology (Gregoretti et al., 2004). Furthermore, HDAC inhibitors are of immense interest to the research and medical communities, which focus on tackling their anti-cancer properties to develop them as potential tools for cancer treatment (Lombardi et al., 2011, Witt et al., 2009). Hdac1 knockout is embryonically lethal in mice and is known to cause cell proliferation defects through the upregulation of cyclin-dependent kinase inhibitors like p21 and p27 (Lagger et al., 2002, Montgomery et al., 2007, Witt et al., 2009). The decline in Hdac1 is also associated with a decrease in overall Hdac activity exhibited by other Hdacs, which is usually not complemented by overexpression of Hdac2 or Hdac3 (Lagger et al., 2002, Montgomery et al., 2007, Witt et al., 2009). These attributes along with its near-ubiquitous expression make Hdac1 unique, which demands further exploration in regeneration biology of the retina. Hdac1 being part of multi-protein nuclear complexes that are important for causing transcriptional repression and epigenetic landscaping to inhibit the expression of neuronal-specific genes in rest of the tissues (Huang et al., 1999) made us explore the global gene regulations that occur in MGPCs mediated through Hdacs with a particular focus on Hdac1. Furthermore, MG reprogramming leading to induction of MGPCs, with little neuronal characteristics, during retina regeneration is associated with essential expression of pluripotency-inducing factors (Gorsuch et al., 2017, Ramachandran et al., 2010a, Reyes-Aguirre and Lamas, 2016), which could also be targets of epigenome modifiers like Hdacs.

Here we unravel the unique roles played by Hdacs in regulating the expression of various regeneration-associated essential genes like mycb, ascl1a, lin28a, insm1a, and her4.1 during MG reprogramming and induction of MGPCs. We also show the global changes in gene expression with compromised Hdac function during regeneration, through a whole retina RNA sequencing (RNA-seq) analysis. Furthermore, we unravel a few novel gene regulatory networks that connect Hdacs with Delta-Notch signaling, lin28a, and a few essential cytokines. We also show the importance of Hdacs in regulating her4.1 to restrict lin28a expression, which is a necessity for the translation of various genes like ascl1a, and mycb through the suppression of let-7 microRNA. Our study, therefore, unravels the hitherto unknown facts about Hdacs to bring about the induction of MGPCs for successful retina regeneration.

Results

Hdacs Are Necessary for MGPCs' Induction after Retinal Injury

There are different methods to study regeneration, of which needle poke is quicker, more natural, and less traumatic to the fishes. In this method, we injure the retina of fish using a 30G needle after administering deep anesthesia and harvest the eye at various time points. Analysis of hdacs gene expression following retinal injury revealed that hdac3, hdac5, and hdac6 get rapidly induced and hdac1, hdac4, hdac7, and hdac9 get downregulated (Figure 1A). The mRNA in situ hybridization (ISH) analysis revealed a panretinal expression of hdac1 in the uninjured retina, which declines drastically after an injury. Although the mRNA of hdac1 reappeared near the injury spot at 4 days post-injury (dpi), interestingly, it showed exclusion from the majority of bromodeoxyuridine⁺ (BrdU⁺) cells and co-localized with only a fraction of progenitors (Figures 1B, S1A, and S1B). We saw a similar pattern of expression for hdac3, hdac4, hdac5, hdac6, and hdac9 (Figures S1C–S1L). The cell count data of hdac1-expressing cells showed an increased co-localization with proliferating cell nuclear antigen (PCNA) compared with the ones having just BrdU (Figure S1B). PCNA, a molecule with a longer half-life and an ability to stay detectable even after cell-cycle exit (Bologna-Molina et al., 2013, Kimmel and Meyer, 2010, Mandyam et al., 2007), could efficiently be used as an indicator of post-proliferative status as well. From our results, the increased co-localization of hdac1 with PCNA compared with BrdU pulse-labeled cells suggests that hdac1 gets upregulated in post-proliferative MGPCs. Interestingly, immunofluorescence showed a drastic decline in Hdac1 protein expression in MGPCs at 4 dpi (Figure 1C), but we found a higher percentage co-labeling with BrdU, at 2 and 6 dpi (Figure S2A). However, no significant temporal variation was seen in Hdac1 protein levels at different time points post injury in the whole retinal extract (Figure S2B).

Figure 1.

Hdacs' Inhibition Suppresses the Formation of MGPCs

(A) qPCR assay shows injury-dependent regulation of hdacs. *p < 0.04, n.s, not significant.

(B) ISH microscopy of hdac1 expression in uninjured state and at various time points after retinal injury.

(C) Immunofluorescence (IF) microscopy shows that Hdac1 expression is mostly absent from BrdU⁺ cells at the site of injury, in the retina at 4 dpi. Arrowheads indicate co-labeled Hdac1⁺ and BrdU⁺ cells, arrows mark Hdac1⁻ but BrdU⁺ cells.

(D) qPCR assay of hdac1 from GFP-positive and GFP-negative sorted cells from 1016tuba1a: GFP transgenic retina at 4 dpi. ^∗p<0.002.

(E and F) MO-based hdac1 knockdown using two different MOs separately decreases BrdU⁺ cells, relative to the control MO (E), which accounts for 80% reduction in cell number (F) in 2 dpi retina; ANOVA test, *p < 0.001. MOs were delivered at the time of injury, electroporated immediately after injury, and harvested at 2 dpi.

(G) hdac1 inhibition using antisense MO (0.5 mM) causes upregulation of acetylated histone 4 compared with control (0.5 mM), revealed by IF microscopy in 4 dpi retina.

N = 6 biological replicates in all experiments unless specified. Scale bars, 10 μm in (B, C, E, and G). Error bars are SD. ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. The asterisk marks the injury site in (B, C, E, and G). See also Figures S1 and S2.

We further conducted experiments using transgenic fish that marks proliferating MGPCs in the retina. For this, a 1016tuba1a: GFP transgenic zebrafish, which is well characterized to mark proliferating MG-derived progenitors after an injury (Fausett and Goldman, 2006), was used. Furthermore, the 1016tuba1a: Cre-ERT2/β-actin-LCLG double transgenic fish was used in previous studies to identify the lineage of the MG-derived progenitors in regenerating zebrafish retina (Ramachandran et al., 2010b, Ramachandran et al., 2012a). Using this 1016tuba1a: GFP transgenic zebrafish, we sorted the GFP-positive and GFP-negative cell populations from the injured retina to explore the levels of hdac1 mRNA. Interestingly, we found significant downregulation of hdac1 mRNA in the GFP-positive population, compared with the rest of the cells (Figure 1D), suggesting that GFP-positive MGPCs need to have a declined hdac1 expression for a successful regenerative response of the retina. These observations indicate that Hdac1-mediated gene regulation is prevalent both in MGs that are yet to dedifferentiate and enter the cell cycle and in post-proliferative MGPCs.

To investigate the role of Hdac1 during regeneration, hdac1-targeting morpholino (MO), which blocked its protein expression, was used (Figures S2C and S2D). The hdac1-targeting MOs were delivered separately and electroporated at the time of injury and harvested at 2 dpi, which resulted in 80% decline in the number of proliferating cells (Figures 1E, 1F, S2E, and S2F). Transfection of hdac1 and gfp reporter mRNA along with the respective MO that does not bind onto the delivered mRNA could rescue the reduction in cell proliferation in the retina at 2 dpi (Figures S3A and S3B). We also found an enhanced proliferation with gfp-hdac1 fusion mRNA transfection in these rescue experiments (Figures S3A and S3B).

We then decided to explore the specific effect of hdac1 knockdown on cell proliferation in the early and late proliferative phases of regeneration. For this, we performed late hdac1 knockdown experiments in which the retina is allowed to regenerate until 2 or 4 dpi and then electroporated with hdac1 MO and harvested at 4 or 6 dpi, respectively (Figures S3D–S3I). Although we delivered the MO during injury, it remains dormant in the vitreous until electroporated to get entry into the retina enabling hdac1 knockdown. In both these sets, we found a decline in the MGPCs marked by PCNA compared with the control MO group, as observed in the 0–2 dpi hdac1-knockdown groups. MO-mediated hdac1 repression also caused an upregulation of acetylated histone 4 (Ac.H4) in the retina, which demonstrated the efficacy of Hdac inhibition (Figure 1G). These findings suggest the existence of hdac1-mediated gene regulation in normal reprogramming and perpetuation of retina regeneration.

Furthermore, we explored the effect of the global decline in Hdac activity through well-characterized pharmacological inhibitors of Hdacs, valproic acid (VPA) and trichostatin A (TSA) separately (Bolden et al., 2006, Xu et al., 2007). We found that proliferation decreased up to 80% in VPA or TSA, and we reasoned that Hdacs could be associated with MGPCs induction in the retina, at 4 dpi (Figures 2A and 2B). The reduction in the number of MGPCs seen with VPA or TSA treatment was not because of apoptosis (Figures S4A–S4C) or DMSO in the drug as the solvent (Figures S4D and S4E). Furthermore, we observed increased levels of Ac.H4 by VPA and TSA treatment in the post-injured retina (Figures 2C, 2D, and S4F). These results suggest the existence of Hdac-mediated gene regulatory pathway, which is necessary for the induction of retinal MGPCs during regeneration.

Figure 2.

Hdacs and Her4.1 Mutually Inhibit Each Other during MGPCs-Mediated Retina Regeneration

(A and B) Immunofluorescence (IF) microscopy on retinal sections shows a reduction in the number of proliferating MGPCs following Hdacs' inhibition by VPA and TSA at 4 dpi (A), which is quantified in (B); ANOVA test, *p < 0.0001 in (B).

(C and D) Western blot analysis revealed upregulated acetylated histone 4 with Hdacs inhibition at 2 dpi retina (C) and IF microscopy at 4 dpi (D), compared with uninjured and DMSO-treated control retina.

(E) RT-PCR (upper) and qPCR (lower) of her4.1 mRNA in DAPT-treated retina, compared with uninjured (UC) and 2 dpi retina. *p < 0.001.

(F and G) RT-PCR (upper) and qPCR (lower) (F) and ISH and IF microscopies in 4 dpi retina (G) show that Hdac blocker VPA/TSA increases her4.1 induction compared with respective controls. Arrowheads indicate co-labeled her4.1⁺ and BrdU⁺ cells, arrows mark her4.1⁺ but BrdU⁻ cells in (G). *p < 0.0001 in (F).

(H) Single-cell-stage embryos were injected with her4.1:luciferase vector along with Renilla luciferase mRNA for normalization, which were then treated with TSA for 24 hr before lysing for quantification of her4.1 promoter activity by dual luciferase assay. *p < 0.003.

(I) Fluorescence in situ hybridization and IF microscopy show co-localization of her4.1 and hdac1 with BrdU⁺ MGPCs in 4 dpi retina. Arrowheads indicate her4.1⁺ and hdac1⁻ cells, arrows mark her4.1⁻ but hdac1⁺ cells.

(J) qPCR of hdac1, hdac4, and hdac9 mRNAs with blockade of Notch signaling by DAPT treatment, in 2 dpi retina; *p < 0.002.

(K) Schematic experimental timelines for her4.1 MO electroporation in the retina.

(L) qPCR of hdac1 in the her4.1 knockdown retina as per the experimental timeline in (K); *p < 0.03.

N = 4 biological replicates in all experiments. Scale bars, 10 μm in (A, D, and G) and 20 μm in (I). The asterisk marks the injury site in (A, D, G, and I). ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. See also Figures S2, S4, and S6.

We also performed a whole-retina RNAseq analysis, to explore the changes in global gene expression that occur with the compromised activity of Hdacs. For this, we used total retina isolated from zebrafish of, uninjured (UC), 12 hours post injury (hpi), 4 dpi, and VPA-treated 4 dpi experimental groups. The VPA-mediated Hdacs' inhibition revealed several transcription factors that are up- or downregulated when compared with 12 hpi and 4 dpi (Table S2; Figures S4G and S4H) (GEO accession: GSE98094), including several regeneration-associated genes, namely, ascl1a, mycb, lin28a, foxn4, and insm1a (Goldman, 2014, Kaur et al., 2018, Ramachandran et al., 2010a, Ramachandran et al., 2011, Ramachandran et al., 2012b). We further evaluated the molecular mechanisms of regulation of a few of these genes with regard to normal and deregulated activity of Hdacs.

Hdacs-Her4.1 Interplay Governs MGPCs Proliferation

Next, we explored the possible reasons for the MGPCs induction to be diminished upon Hdacs' suppression. One such candidate gene was a Delta-Notch effector gene her4.1. We chose her4.1 mainly because of its negative regulation on retina regeneration. Forced expression of notch intracellular domain (nicd), which upregulates her4.1, is known to repress MGPCs induction completely, whereas pharmacological inhibition of γ-secretase through the drug DAPT (N-[N-(3,5-difluorophenylacetyl)-L-alanyl]-S-phenylglycine-t-butyl ester), which decreases her4.1 levels, caused an enhancement of the zone of proliferation in injured retina (Conner et al., 2014, Wan et al., 2012) (Figures 2E and S4I).

Interestingly, with Hdacs' inhibition, we found a dose-dependent increase in her4.1 expression in RT-PCR, qPCR, and mRNA ISH in both VPA- and TSA-treated post-injured retina (Figures 2F and 2G). The blockade of Hdacs reflects a decline in BrdU pulse-labeled MGPCs with a concomitant increase in her4.1 levels in the retina (Figure 2G). Furthermore, the injection of her4.1:gfp-luciferase reporter in zebrafish embryos exposed to TSA confirmed the her4.1 gene upregulation by the inhibition of Hdacs (Figure 2H). Moreover, the VPA/TSA-treatment-associated increase in her4.1 expression could be caused by the upregulation of delta and notch genes (Figures S5A and S5B), as discussed earlier (Figures 2F–2H).

We further explored if the increased her4.1 levels seen in TSA-treated retina influenced the hdac1 mRNA levels in the post-injured retina. We found an anticipated decline in hdac1 mRNA levels revealed by mRNA ISH in TSA-treated 4 dpi retina (Figure S5C). These results suggest the existence of Her4.1-mediated repression of hdac1 expression. In another scenario, the Hdac1 protein levels also declined in the uninjured retina with VPA treatment in a dose-dependent manner (Figure S5D).

Further analysis by double mRNA ISH of hdac1 and her4.1 showed their co-expression in a limited number of cells at 4 dpi (Figure 2I). We also found several hdac1-positive cells that are her4.1 negative, and vice versa. Furthermore, the DAPT-mediated inhibition of Delta-Notch signaling, which brings down her4.1 levels (Conner et al., 2014, Wan et al., 2012) (Figures 2E and S4I), caused an enhancement of hdac1, hdac4, and hdac9 expression at 2 dpi (Figure 2J). It is interesting to note that hdac1, hdac4, and hdac9 showed a similar temporal expression pattern post-retinal injury (Figure 1A). This observation supports the possibility that these three genes are regulated similarly. Furthermore, to confirm the influence of Her4.1 on hdac1 expression, we employed MO-based gene knockdown approach. We analyzed the efficacy of her4.1-targeting MO in zebrafish embryos by co-injection with a construct harboring MO-binding sequence appended to GFP along with her4.1 or control MO (Figure S2D). Similarly, we saw upregulation of hdac1 expression with the her4.1 knockdown from 0 to 4 dpi (Figures 2K and 2L). Notably, the hdac1 expression from its earlier panretinal nature, immediately after the injury, stayed restricted to the neighboring cells of MGPCs. This type of hdac1 expression could be a post-proliferative phenomenon seen from its increased co-labeling with PCNA (Figure S1B). Her4.1-mediated gene repression may be a potential cause for this. This speculation is also because of the increased her4.1 expression in neighboring cells of actively proliferating MGPCs in the injured retina at 4 dpi (Figures S5E and S5F). At later stages of regeneration, this type of Her4.1 expression probably necessitates the stringent regulation of Hdac1 expression in the neighboring cells of actively proliferating MGPCs. It is mainly evident from the decline in the hdac1 levels in a her4.1-repressed scenario from 3 to 5 dpi (Figures 2K and 2L). It is also important to note that if her4.1 is inhibited immediately after the retinal injury, we found an increase in hdac1 expression (Figure 2L). These findings suggest that Her4.1 has a role in hdac1 repression at early stages that is just opposite at later stages of regeneration. These results presumably indicate a mutual but differential regulation between Her4.1 and Hdac1 during various phases of regeneration.

lin28a Expression Mediated through Hdacs Is Necessary during Retina Regeneration

Furthermore, we probed for the expression pattern of regeneration-associated genes like mycb, ascl1a, and insm1a in VPA/TSA-treated retina. Interestingly, we found an increase in the expression of genes mycb and ascl1a, and a sharp decline in insm1a levels, in the VPA-treated injured retina (Figures 3A–3D) and luciferase assay done in zebrafish embryos (Figure 3E). The decrease in insm1a, a transcriptional repressor (Ramachandran et al., 2012b), also may contribute to enhanced mycb and ascl1a levels. The ascl1a and mycb were also regulated similarly with the hdac1 knockdown in the retina (Figures S6A–S6C). Surprisingly, the increased mRNA levels of ascl1a and mycb seen in Hdac-inhibited retina did not show a corresponding increase in their protein levels, which instead decreased (Figures 3F–3H). We speculated that lack of Lin28a-mediated suppression of let-7 microRNA could be the possible reason for reduced protein translation of Ascl1a and Myc from their mRNAs, as described earlier (Ramachandran et al., 2010a). We found a significant reduction in lin28a expression in Hdacs-inhibited retina after injury (Figures 3I, 3J, S6A, S6D, and S6E). Furthermore, the cell proliferation results in rescue experiments (Figure S3A and S3B) may also be because of changes in lin28a mRNA levels (Figure S3C). The reduction in lin28a levels caused an anticipated increase in the let-7 microRNA in TSA/VPA-treated retina (Figure 3K). The abundant let-7 microRNA could block the translation of regeneration-associated mRNA such as ascl1a and mycb, which justified the reduced protein levels of Ascl1a and Mycb (Figures 3F–3H).

Figure 3.

Hdacs' Repression Causes Upregulation of ascl1a and mycb, and Repression of insm1a and lin28a

(A and B) RT-PCR (upper) and qPCR (lower) show upregulation of mycb (A) and ascl1a (B) with VPA-mediated Hdacs' repression, relative to the control in 2 dpi retina; *p < 0.0001.

(C) ISH microscopy reveals that ascl1a and mycb mRNAs show an expanded zone of expression at 4 dpi, with VPA.

(D) RT-PCR (upper) and qPCR (lower) show downregulation of insm1a in Hdacs' blockers TSA (left) and VPA (right) by qPCR, relative to the control in 2 dpi retina. *p < 0.003.

(E) Hdacs' repression using TSA inhibits luciferase activity in 24 hours post fertilization embryos injected with insm1a:gfp-luciferase vector; *p < 0.0002.

(F and G) Western blot analysis reveals a decline in Ascl1a and Myc protein levels in 2 dpi retina treated with VPA (F) and TSA (G), compared with control. GS, glutamine synthetase; it is the loading control.

(H) Densitometry analysis of western blots from (F), done in triplicates. *p < 0.003, n.s, not significant.

(I and J) RT-PCR (upper) and qPCR (lower) show a decline in lin28a expression at 2 dpi retina (I), confirmed by ISH microscopy (J), with VPA-mediated Hdacs' repression, compared with the control in 4 dpi retina; ANOVA test, *p < 0.0002 in (I). N = 3 biological replicates.

(K) qPCR analysis of the let-7a microRNA shows an upregulation with VPA/TSA treatment in 4 dpi retina. *p < 0.001.

(L) TSA-treated 3 dpa fin shows a reduction in blastema compared with DMSO control.

(M) qPCR analysis of the VPA-treated fin blastema shows an increase in her4.1 and decrease in lin28a mRNA levels at 6 dpa compared with control. *p < 0.002.

Scale bars, 10 μm in (C and J) and 500 μm in (L). The asterisk marks the injury site in (C and J). ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. See also Figures S3–S6.

Interestingly, the regenerating caudal fin blastema also showed a decrease in cell mass in response to TSA-mediated Hdac inhibition (Figure 3L). Furthermore, we observed increased levels of Ac.H4 by VPA/TSA treatment in 6 days post amputation (dpa) fin blastema (Figure S6F) as found in the post-injured retina (Figures 2C, 2D, and S4F). Interestingly, Hdac inhibition-mediated decline in fin blastema also showed increased her4.1 and decreased lin28a levels (Figure 3M), similar to that seen in the retina (Figures 2F and 3I). These results suggest the existence of a possibly conserved Hdac-dependent gene regulatory pathway during the regeneration of various tissues.

Inhibition of Hdacs, the facilitators of transcriptional repression, caused a decline in the expression of genes like lin28a, which seemed a puzzle. However, our previous results suggested the existence of a potential transcriptional repressor such as Her4.1, which in TSA-treated retina stays upregulated (Figures 2F and 2G). This enhancement of her4.1 could mediate the decline in the expression of lin28a.

Her4.1 Directly Represses the Expression of lin28a in Injured Retina

Since we found an upregulation of her4.1 and downregulation of lin28a in Hdacs-inhibited retina, we explored whether Her4.1 had any direct effect on the expression levels of lin28a. We first examined the impact of reduced her4.1, caused by DAPT treatment, on lin28a promoter activity in zebrafish embryos, injected with lin28a:gfp-luciferase construct. The luciferase assay showed an increased promoter activity in a dose-dependent manner (Figure 4A), which supported this view. To ascertain further the involvement of Her4.1 in lin28a expression, we examined its promoter for potential Her/Hes binding sites (Kageyama et al., 2007) and found two putative binding sites (Figure 4B). At first, we validated if the her4.1:gfp-luciferase construct showed upregulation of promoter activity with nicd mRNA co-injected in zebrafish embryos. We found an expected upregulation of her4.1 promoter activity in a dose-dependent manner (Figure 4C). We then co-injected zebrafish embryos with lin28a:gfp-luciferase reporter bearing either wild-type or mutated Her/Hes binding sites on lin28a promoter separately, along with increasing concentrations of nicd mRNA. We found that NICD overexpression caused a dose-dependent downregulation of lin28a promoter activity in the wild-type construct, which was absent in the mutant. These results showed that Her/Hes binding onto lin28a promoter negatively contributes to its functionality (Figure 4D). Fluorescence ISH analysis revealed that many her4.1-expressing cells showed weak lin28a levels, and vice versa (Figure 4E). This observation also suggested the existence of a Her4.1-mediated lin28a repression in retinal MGPCs. Furthermore, the knockdown of her4.1 during the active proliferative phase of regeneration also confirmed these results. We found that blockade of her4.1 from 3 to 5dpi caused an increase in lin28a levels in the retina (Figures 4F and 4G). Furthermore, the late knockdown of her4.1 also caused a significant increase in the BrdU⁺ MGPCs in the retina (Figures 4H–4J), probably because of increased lin28a expression as demonstrated before (Elsaeidi et al., 2018, Ramachandran et al., 2010a). These results supported the earlier conjecture that Hdacs upregulated the expression of lin28a through downregulation of a repressor like Her4.1.

Figure 4.

Her4.1 Restricts the Zone of MGPCs by Suppressing lin28a Expression

(A) Notch inhibition by DAPT stimulates luciferase activity in 24 hours post fertilization (hpf) embryos injected with lin28a:gfp-luciferase vector. *p < 0.0001.

(B) Diagram of lin28a promoter with putative Her/Hes binding sites.

(C) NICD overexpression stimulates luciferase activity in 24 hpf embryos injected with her4.1:gfp-luciferase vector compared with control. *p < 0.0002.

(D) NICD regulates lin28a promoter-driven reporter activity in zebrafish embryos, co-injected with lin28a:gfp-luciferase construct and nicd mRNA, through the Her/Hes binding sites on the lin28a promoter, relative to 0 ng nicd mRNA; *p < 0.005; n.s, not significant.

(E) Fluorescence in situ hybridization microscopy shows mutual exclusion of her4.1 and lin28a, and IF marks BrdU⁺ MGPCs in 4 dpi retina. Arrowheads indicate co-labeled her4.1⁺ and lin28a⁺ cells, arrows indicate her4.1⁺ but lin28a⁻ cells.

(F) Schematic of experimental timelines for her4.1 MO electroporation.

(G) qPCR of lin28a in her4.1 knockdown retina as per experimental timeline in (F). ^∗p<0.001.

(H–J) Experimental timelines for her4.1 MO electroporation (H); immunofluorescence microscopy shows increased number of BrdU⁺ cells in retinal sections electroporated with lissamine-tagged her4.1 MO as per H (I); quantification of BrdU⁺ cells seen with late knockdown of her4.1 in (I) (J); *p < 0.0002 in (J).

N = 3 biological replicates in all experiments. Scale bars, 10 μm in (E and I). The asterisk marks the injury site in (E and I). ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. See also Figures S2–S6.

Hdac Inhibition Reversibly Suppresses MGPCs' Formation and Differentiation

We further investigated whether inhibition of Hdacs had a long-lasting impact on retina regeneration. To address this, we performed two sets of experiments. In the first one, fish were dipped in VPA from the time of injury until 4 dpi, immediately an intraperitoneal injection of BrdU was administered, and the eyes were harvested at 8 dpi (Figure 5A). In the second set, we performed the same experiment, except that the BrdU pulsing was given at 8 dpi and eyes were harvested after 5 hr (Figure 5D). These experiments were intended to demarcate the difference in performance of MGPCs with Hdacs' inhibition and after its withdrawal. In the first set, we found a decline in BrdU⁺ cells at 8 dpi as a result of reduced proliferation at 4 dpi with VPA treatment. Interestingly, we found a significant increase in PCNA-positive cells in 8 dpi as opposed to control (Figures 5B and 5C). In the second set, we found a significant increase in BrdU⁺ cells, which mostly co-labeled with PCNA⁺ cells at 8 dpi (Figures 5E and 5F). These observations support the view that in the absence of Hdacs' inhibition the BrdU⁺ cells seen at 8 dpi are newly formed MGPCs.

Figure 5.

Hdacs' Inhibition Reversibly Suppress MGPCs' Formation and Differentiation

(A–C) An experimental timeline showing VPA treatment from 0 to 4 dpi, BrdU pulse on 4 dpi, and harvest on 8 dpi (A), which shows the re-induction of PCNA⁺ and minimal BrdU⁺ proliferating cells revealed by immunofluorescence (IF) microscopy, compared with control (B), quantified in (C); ANOVA test, *p < 0.01 in (C).

(D–F) An experimental timeline showing VPA treatment from 0 to 4 dpi and a BrdU pulse done on 8 dpi (D) shows co-labeling of BrdU⁺ and PCNA⁺ cells in the re-induced MGPCs in VPA-treated retina and minimal proliferating cells in control, which is revealed by IF microscopy (E), which is quantified in (F); ANOVA test, *p < 0.01 in (F).

(G–J) The temporal evaluation of re-induced MGPCs, which are labeled with BrdU in an experiment timeline (G), which reveals that the regeneration occurs similar to control, revealed by IF microscopy (H), and cell quantification indicated its similarity to control in 9 and 60 dpi (I), with a preferential localization bias toward ONL than INL and GCL in VPA experimental regime compared with control (J); ANOVA test, *p < 0.01, **p < 0.02 in (J).

N = 6 biological replicates in all experiments. Scale bars, 10 μm in (B, E, and H). Error bars are SD. The asterisk marks the injury site in (B, E, and H). ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. See also Figures S1–S4.

We further investigated, whether these newly formed MGPCs restore their functionality at a later stage of regeneration. For this, we labeled the newly formed MGPCs, after VPA withdrawal, with BrdU for 3 days. These BrdU cells were then traced up to 60 dpi and compared with those at 9 dpi in another similar experiment. The 9 dpi data would give us an idea about the approximate number of BrdU-positive cells that appeared in the retina after VPA withdrawal, which we then compared with the 60 dpi data. We found that the MGPCs migrated to various cell layers of the retina even at 60 dpi and showed only a marginal decline in cell number when compared with control, where there was no drug treatment (Figures 5G–5J).

Hdac-Her4.1 Axis Is Necessary for the Essential Cytokine Regulation during MGPCs Formation

Cytokines released at the site of injury trigger the regeneration cascade in the retina (Elsaeidi et al., 2014, Gramage et al., 2015, Wan et al., 2014, Zhao et al., 2014). We reasoned that the reintroduction of MGPCs after TSA/VPA withdrawal might involve the active participation of cytokines, as Hdacs could regulate them in other systems (Larsen et al., 2007, Leoni et al., 2005, Li et al., 2008). To decipher this, we first examined if regeneration-associated cytokines are affected by TSA/VPA treatment. We found that the majority of the cytokines like il6, il11a, il11b, lepa, lepb, and crlf and receptors like il11ra, lepr, and lifra were significantly inhibited in 2 dpi retina treated with TSA/VPA (Figures 6A and 6B), which may also explain the lack of enough MGPCs for regeneration. Moreover, these cytokines, which get upregulated in the injured retina, were downregulated with Hdacs' inhibition, revealed in a whole-retina RNA-seq analysis, at 4 dpi (Figure S6G and Table S1). We then investigated whether these cytokines possibly contribute to the reappearance of MGPCs and normal regeneration after withdrawal of Hdacs' inhibition. To address this, we designed an experimental timeline in which zebrafish were given TSA dip treatment from the time of injury till 4 dpi and shifted to fresh water for 2 more days and the eyes harvested at 6 dpi (Figure 6C). The cytokines showed a drastic increase, more than the normal levels found in 2 or 6 dpi retina (Figures 6D and 6E). We speculated whether the increased her4.1 levels caused by TSA/VPA treatment contributed to the decline in the levels of cytokines, as a potential reason for reduced proliferation after retinal injury. In silico analysis revealed several Hes/Her binding sites (Kageyama et al., 2007) in the regulatory sequences of various cytokines addressed in this study (Figure S6H). We then decided to probe if this Her4.1-mediated repression exists on cytokine gene regulations in injured retina. We performed knockdown of her4.1 at 3 dpi and analyzed the levels of cytokines at 5 dpi (Figure 6F). Interestingly, we found upregulation of almost all cytokines with the her4.1 knockdown in the retina (Figure 6G). We reasoned that the Her4.1-mediated repression of cytokines could get relieved with TSA withdrawal. In a confirmative result, we found that the enhanced levels of her4.1 with TSA/VPA treatment would decline progressively after the drug withdrawal (Figure 6H). This observation could be because of auto–negative feedback regulation of Her4.1 on its transcription (Figure 6I) (Wilson et al., 2016), similar to other Hes/Her proteins' auto-inhibitory regulations (Giudicelli et al., 2007, Hubaud and Pourquie, 2014). These results suggest that new injury is not necessary for the observed cytokine re-induction, as long as Her4.1 expression stays low after drug withdrawal. Furthermore, many regeneration-associated genes like ascl1a, lin28a, mycb, and sox2 reappeared, probably influenced by the cytokines' surge (Figures 6J–6L), causing an overall standard regenerative response.

Figure 6.

Hdac-Influenced MGPCs' Formation Is Mediated through Her4.1 Cytokine Axis

(A and B) RT-PCR (A) and qPCR (B) assay revealed injury-induced expression of cytokines in uninjured, 2 dpi, and TSA-treated 2 dpi retina; *p < 0.01 in (B), N = 6 biological replicates.

(C–E) Enhanced expression of cytokines in TSA withdrawal experiments as per the experimental timeline (C), shown by RT-PCR (D) and qPCR (E).

(F) Experimental timelines for late knockdown of her4.1, from 3–5 dpi.

(G) qPCR of various cytokines with 2 concentrations of her4.1 MO electroporated as per experimental timeline in (F). ^∗p<0.001.

(H) Downregulation of her4.1 in TSA withdrawal experiments as per experimental timeline in (C); *p < 0.007.

(I) qPCR of her4.1 with two concentrations of her4.1 MO electroporated as per experimental timeline in (F); *p < 0.004.

(J–L) RT-PCR (J) and qPCR (K and L) were used to assay re-induction of injury-induced genes like mycb, ascl1a, lin28a, and sox2 in uninjured, 2 dpi, and post-TSA treatment, which show significant upregulation of lin28a (K) and moderate increase in the level of other genes (L); ANOVA test, *p < 0.001 in (K).

N = 6 biological replicates in all. Error bars are SD. See also Figures S4–S6.

Hdacs Are Necessary for Normal Distribution of MGPCs Across Retina Layers

As we found the necessity of Hdacs in MGPCs formation, a similar mechanism presumably underlies progenitor differentiation and their migration. We then explored if Hdacs are necessary for the movement of MGPCs across various retinal layers. In an experiment, we labeled almost all MGPCs with BrdU from the time of injury to 5 dpi (Figure 7A). The fish were then exposed to VPA from 7 to 14 dpi, with an aim to block Hdacs during the cell migration/differentiation phase of retina regeneration. Although we found no apparent reduction in the total number of cells at 14 dpi, there was a preferential localization of the migrated cells to the outer nuclear layer, with a marginal decrease in the inner nuclear layer and a significant decline in the ganglion cell layer (Figures 7B and 7C). These findings suggest that Hdacs-driven gene regulation persists even at later phases of retina regeneration, which could influence the migration pattern of MGPCs to various retinal layers.

Figure 7.

Hdacs' Inhibition Causes Selective Bias in MGPCs' Migration toward the Outer Nuclear Layer

(A–C) Determination of the fate of MGPCs in Hdac-inhibited condition as shown in experimental timeline (A), shows a significant increase of cell localization in ONL and a decline in INL and GCL, revealed by immunofluorescence (IF) microscopy (B) and quantified in (C), in the 14 dpi retina. *p < 0.0004 in (C), N = 6 biological replicates.

(D) Signaling pathways and genes that are regulated by Hdacs at different times and stages of retina regeneration.

Scale bar, 10 μm in (B). Error bars are SD. The asterisk marks the injury site in (B). ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. See also Figures S4 and S6.

Discussion

We explored the significance of HDACs, with a distinct focus on Hdac1, during retina regeneration and fin blastema formation in this study. Although the involvement of Hdac1 in progenitor differentiation is known during embryonic retinal development (Yamaguchi et al., 2005), proper cell division, and maintenance of pluripotency of embryonic stem (ES) cells (Jamaladdin et al., 2014), its mechanistic involvement during retina regeneration remained underexplored. Following injury, we found a drastic decline in the transcript levels of hdac1 throughout the retina, which was restored back near MGPCs only at around 4 dpi. On the other hand, the protein level of Hdac1 remained unchanged in MGPCs and their neighboring cells until 4 dpi, a time when it stayed low in expression in MGPCs. Interestingly, at a later time point, 6 dpi, Hdac1 protein again become ubiquitous in all retinal cells. These seemingly contradictory observations regarding transcript and protein levels of Hdac1, found in our study, might hold the key to successful retina regeneration. The presence of Hdac1 protein in all retinal cells except MGPCs suggests the existence of Hdac1-mediated repression of regeneration-associated genes in these cells. Conversely, depletion of Hdac1 from MGPCs during proliferative phase ensures the reversal of this blockade of transcription as a necessity for successful proliferation. The reappearance of Hdac1 and the subsequent inhibition of transcription of regeneration-associated genes in post-proliferated MGPCs suggest its role not only in potentiating MGPCs for successful cell multiplication brought about by its depletion but also in controlling the extent of proliferation at the vicinity of injury. In an analogous situation, HDAC1 in ES cells was reported to selectively target the promoters of pluripotency-inducing factors like Oct4, Sox2, and Nanog as a requirement for controlling their expression (Kidder and Palmer, 2012). Also, depletion of HDAC1 has been shown to cause enrichment of ES cell-specific gene expression. Similar necessary role of HDAC1 is also required for proper differentiation regime of ES cells (Dovey et al., 2010). Given the similarity between ES cells and MGPCs regarding gene expression, we speculate that the relocation of hdac1 from its panretinal appearance to the neighboring cells of MGPCs probably would act similarly. Further studies are necessary to prove the guiding factors that regulate hdac1 expression near MGPCs, and to find the collaborating factors that guide Hdac1 or other Hdacs to specific gene targets to regulate their expression.

Our study also supports another recently discovered phenomenon in mice retina, where the researchers over-expressed Ascl1 along with Hdac inhibition for promoting the regenerative capacity of mouse MG (Jorstad et al., 2017). It is interesting to note that in our zebrafish study too we find a global decline in hdac1, hdac4, hdac7, and hdac9 mRNA levels soon after injury. Furthermore, the mRNA levels are restored back to normalcy, but actively proliferating MGPCs lacked Hdac1 protein, which indicates that its absence is a favorable criterion for normal cell proliferation. However, the panretinal decline in Hdac activity was not a facilitator of MGPCs proliferation in zebrafish, probably because of upregulated Her4.1 and downregulated lin28a levels.

We found that the global decline in expression of other hdac genes, namely, hdac4, hdac7, and hdac9, in post-injured retina are restored back to normalcy at later stages of regeneration. This phenomenon may potentially act as a prelude to unblock the transcriptional repression of various pro-proliferative genes like pluripotency-inducing factors to bring about successful regeneration. Since we found a good correlation between the effects of hdac1 knockdown and pan-Hdacs inhibition using TSA/VPA in regenerating retina, we speculate that Hdac1 might be capable of subduing the activity of other Hdacs, similar to that reported in other systems (Lagger et al., 2002, Montgomery et al., 2007, Witt et al., 2009). Increase in the level of acetylated histone4 upon knocking down hdac1 separately or inhibiting all Hdacs through TSA also could justify the overall increase in gene expression of regeneration-associated genes like ascl1a, myca, mycb, and her4.1. We further show that the abundance of her4.1 could account for reduced MGPCs, as is evident from its knockdown in the proliferative phase of regeneration, causing a significant increase in the number of MGPCs.

It is believed that Delta-Notch signaling enables the retina to limit the zone of proliferation near the injury site. Now, considering that hdac1 expression is predominant in the post-proliferative phase of MGPCs, it is logical to find upregulated expression of hdacs upon her4.1 blockade. In other words, decline in her4.1 because of DAPT treatment, causing an increased cell number, could bring about more cells that exit cell cycle, and hence would accompany an increased hdacs level.

It is interesting to note that, upon knocking down hdac1 separately or inhibiting all Hdacs in general, the number of MGPCs decline drastically with a deleterious effect on retina regeneration. This scenario might be possible through a super-abundance of a repressor like her4.1, which we found had a substantial inhibitory effect on lin28a, which is a vital pluripotency-inducing factor. According to a previous study (Ramachandran et al., 2010a), a decrease in Lin28a would ensure the presence of high levels of let-7 microRNA, which could prevent the translation of several regeneration-associated genes such as ascl1a, mycb, and zic2b and components of Sonic hedgehog signaling (Kaur et al., 2018, Ramachandran et al., 2010a). In support of this speculation, this study found that the panretinal Hdacs inhibition brought about an enhancement of let-7 microRNA. Therefore, downregulation of lin28a and upregulation of let-7 microRNA in the Hdacs-inhibited retina could repress the translation of ascl1a and mycb, despite their increased levels of mRNA, resulting in lack of adequate MGPCs formation. This part of the study also provides insights into the significance of depletion of Hdac1 in MGPCs, which in turn would regulate Her4.1 to induce Lin28a, which would be crucial for the successful proliferation of MGPCs.

Cytokine surge near the injury site often facilitates retinal regeneration (Goldman, 2014, Wan and Goldman, 2016). The involvement of cytokines is also prominent in other tissue regeneration (Bouzaffour et al., 2009, Gonzalez-Rosa et al., 2017, Li et al., 2014), which makes the involvement of Hdacs-cytokines link crucial in retina-regeneration-specific gene expression paradigm. In general, the typical Hdacs expression necessitates the injury-induced cytokines as a prelude to successful regeneration, and their inhibition downregulates different regeneration-associated cytokines, thus culminating in reduced MGPCs proliferation. Withdrawal of the block on Hdacs function showed that the suspension of MGPCs proliferation by Hdacs inhibition was reversible, resulting in resumption of regeneration through a reappearance of cytokine surge and pluripotency factors like lin28a, which possibly is brought about by the disappearance of the abundant repressors like her4.1. This regulatory network could lead to the re-induction of MGPCs, which are BrdU-traceable even after 60 days.

The complex mechanism of temporal regulation of retina regeneration through Hdacs is presented in a model comprising the findings from this study (Figure 7D). In differentiated MG, the presence of Hdacs ensures blockade of transcription of pro-proliferative genes like ascl1a and mycb. The reported retinal abundance of let-7 microRNA, which could block the translation of ascl1a and mycb also support this hypothesis (Ramachandran et al., 2010a). Upon injury, through different cellular signaling pathways, pro-proliferative genes like ascl1a are highly induced, which upregulates lin28a, which helps in degradation of let-7 microRNA. Therefore, in de-differentiating MGs, even though Hdacs might be present, gene regulatory networks responsible for differentiation are already afoot ensuring the successful formation of MGPCs. At this stage, Delta/Notch-signaling-mediated induction of her4.1 in the neighboring cells of MGPCs could be responsible for inhibiting lin28a transcription, thus controlling the extension of de-differentiation and proliferation. In this study, the her4.1 knockdown in the proliferative phase of regeneration causes an increase in the MGPCs. In proliferating MGPCs, depletion of Her4.1 is also necessary for building up adequate levels of lin28a and cytokines. Furthermore, to control the extent of regeneration, Her4.1, which is a downstream target of Notch signaling, self-downregulates its expression (Wilson et al., 2016), thus allowing reappearance of Hdacs, which may bring about differentiation of newly generated cells. Finally, our study unravels essential mechanisms by which Hdacs orchestrate these effects through mutual signaling pathways yielded through a tight regulation of Her4.1-Lin28a axis bringing changes in the expression of several regeneration-associated genes and cytokines. This study places Hdacs among critical regulators required for tissue regeneration, providing valuable insights for understanding signaling mechanisms that might help in understanding MG reprogramming in the injured mammalian retina, and also regarding damaged human retinae toward successful repair.

Limitation of the Study

Absence of a well-characterized Her4.1 antibody for use in zebrafish limited a few experimental explorations in this study.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: September 28, 2018

Supplemental Information

Table S2. The List of Transcription Factors Regulated in Normal and VPA-Treated Retina, Related to Figures 1–3, 6, S4, and S6

The table shows a list of transcription factors, transcription co-factors, and chromatin remodeling factors along with their ENSEMBL IDs, obtained from RNA-seq data that are regulated during various time points, and Hdac inhibition with VPA (valproic acid), post-retinal injury.

Table S3. List of Primers Used in the Study, Related to Figures 1–4, 6, S3, S5, and S6

All the primers used in this study are listed along with their names indicating the genes in first column and the 5′ to 3′ nucleotide sequence, in the next column.