Retina regeneration: lessons from vertebrates

Abstract

Unlike mammals, vertebrates such as fishes and frogs exhibit remarkable tissue regeneration including the central nervous system. Retina being part of the central nervous system has attracted the interest of several research groups to explore its regenerative ability in different vertebrate models including mice. Fishes and frogs completely restore the size, shape and tissue structure of an injured retina. Several studies have unraveled molecular mechanisms underlying retina regeneration. In teleosts, soon after injury, the Müller glial cells of the retina reprogram to form a proliferating population of Müller glia-derived progenitor cells capable of differentiating into various neural cell types and Müller glia. In amphibians, the transdifferentiation of retinal pigment epithelium and differentiation of ciliary marginal zone cells contribute to retina regeneration. In chicks and mice, supplementation with external growth factors or genetic modifications cause a partial regenerative response in the damaged retina. The initiation of retina regeneration is achieved through sequential orchestration of gene expression through controlled modulations in the genetic and epigenetic landscape of the progenitor cells. Several developmental biology pathways are turned on during the Müller glia reprogramming, retinal pigment epithelium transdifferentiation and ciliary marginal zone differentiation. Further, several tumorigenic pathways and gene expression events also contribute to the complete regeneration cascade of events. In this review, we address the various retinal injury paradigms and subsequent gene expression events governed in different vertebrate species. Further, we compared how vertebrates such as teleost fishes and amphibians can achieve excellent regenerative responses in the retina compared with their mammalian counterparts.

INTRODUCTION

The regenerative capacity varies remarkably among animals. Mammals often regenerate a limited set of tissues and organs such as the skin, liver, epithelia of kidney, gut and lungs [1–5] and often fail to regenerate complex organs such as the brain, retina or heart. In contrast, the non-mammalian vertebrates such as teleost fishes and amphibians exhibit a robust regenerative capacity even in complex organs after remarkable tissue damage [6, 7]. The teleost model, zebrafish, is extensively used to study organ regeneration, especially the retina [8, 9]. The retina is the neurosensory part of the eye, which originates as an outgrowth of the diencephalon and is thus a part of the central nervous system (CNS). CNS has the least regenerative potential in mammals and hence their retina cannot proceed for successful regeneration. Major debilitating eye diseases like age-related macular degeneration, glaucoma and diabetic retinopathy cause visual impairment due to the loss of retinal neurons. Retina regeneration studies include fish, newts, frogs, embryonic chicks and rodents. It is intriguing to explore the potential of the retina to regenerate after an injury in humans, which could be helped by lessons from cold-blooded vertebrates such as fishes and frogs.

Soon after a retinal injury the Müller glia of the retina reprogram giving rise to a proliferating population of retinal progenitors (Müller glia-derived progenitor cells, MGPCs) that eventually replace the damaged retina with functional retinal neurons and Müller glia. Here, various retinal injury paradigms and molecular mechanisms leading to regeneration in lower vertebrates are compared to chicks and mammals. The cascade of events during retina regeneration is also explained in comparison to higher vertebrates while drawing parallels to developmental biology and cancer. The regenerative capability of primitive vertebrates is significant and complete, probably because of their ability to prevent infection in the aquatic environment and resort to slow regeneration compared to faster wound healing. The terrestrial animals rather indulge in a faster wound healing soon after injury, unlike their aquatic counterparts. The reasons behind this contrasting approach hold the key to robust regenerative capability in non-mammalian vertebrates. Based on the available literature, here, we try to comprehend and contrast the mammalian and non-mammalian approaches to retina regeneration and highlight the benefits imparted during mammalian retina regeneration while adopting gene expression paradigms from non-mammalian vertebrates.

INJURY PARADIGMS

Retinal damage happens by various mechanisms that target either the whole retina or specific retinal cell layers. The different methods of inducing retinal injury include mechanical, chemical, light-induced or genetic ablation in transgenic animals (Figs 1 and 2; Table 1).

Figure 1.

A diagrammatic representation of retinal architecture. The innermost layer facing the vitreous is the ganglion cell layer (GCL), comprising cell bodies of ganglion cells whose axonal extensions give rise to the optic nerve. The GCL is connected to the inner nuclear (INL) via the inner plexiform layer (IPL), where cytoplasmic extensions of ganglion cells connect with bipolar cells via amacrine cells as interneurons (amacrine cells shuffle between INL and GCL). The INL has three neuronal cell types (amacrine cells, bipolar cells and horizontal cells) and one non-neuronal cell, Müller glia. The INL is connected to the outer nuclear layer (ONL) via another synaptic layer, the outer plexiform layer (OPL). The cytoplasmic extensions of bipolar cells are connected to photoreceptor cells with horizontal cells acting as interneurons. The rod and cone photoreceptor cells constitute the ONL, the outer fragment of whose is covered with RPE.

Figure 2.

A pictorial representation of different retinal damage methods. (a) The different parts of the eye are shown along with three primary mechanisms inducing damage in various retinal cell types, which lead to restoration in regeneration competent animals like zebrafish or embryonic chicks and amphibians. (b) The three different modes of physical damage involve injury mediated with a 30G needle, removal of a small retinal patch with a micro knife and subretinal injection to induce retinal detachment. (c) Damage to different retinal layers caused by intravitreal injection of varying concentrations of chemical agents. These chemical agents cause cell death by disrupting the ion channels. (d) The damage to photoreceptor cells was induced after keeping the fish in the dark chamber for a long time followed by short-term exposure to intensely bright light. Light exposure causes cellular damage through chemical, mechanical or thermal mechanisms.

Table 1.

The injury paradigms used in different animal models along with the retinal cell types or layers affected

Animal	Injury paradigm	Cells damaged
Fish	Mechanical
	Cryoinjury [206]	All the retinal cells
	Needle poke [26]	All the retinal cells
	Surgical excision [29, 30]	All the retinal cells
	Chemical
	ATP [207]	Photoreceptors, ganglion cells
	Ouabain [32–35]	GCL, INL or ONL
	NMDA [39, 41, 85]	GCL
	MNU [40, 208]	Photoreceptors
	6-OHDA [209–211]	Dopaminergic neurons
	Tunicamycin [212]	Photoreceptors
	CoCl2 [52–54]	Photoreceptors, ganglion cells
	Light-induced
	[56, 57, 85, 213, 214]	Photoreceptors
	Genetic ablation
	Nitroreductase/metronidazole [215–218]	Photoreceptors, bipolar cells
	Retinitis pigmentosa [219, 220]	Photoreceptors
	Diabetic retinopathy [221–223]	RGCs
	Age-related macular degeneration [224–226]	RPE and photoreceptors
	Glaucoma [227, 228]	RGCs
Amphibians	Mechanical
	Surgical removal [69, 70]	All the retinal cells
	Chemical
	Tunicamycin [46]	Photoreceptors
	Genetic ablation
	Nitroreductase/metronidazole [229, 230]	Photoreceptors
	Retinitis pigmentosa [231–233]	Photoreceptors
Birds	Mechanical
	Surgical removal [73–75, 160] Chemical	All the retinal cells
	NMDA [80, 89, 234]	GCL
Mammals	Mechanical
	Retinal detachment [27, 28]	Photoreceptors
	Chemical
	NMDA [37, 38, 40, 42, 90]	GCL
	MNU [40, 235]	Photoreceptors
	Tunicamycin [48]	Photoreceptors
	Ouabain [236]	INL
	Light-induced
	[237, 238]	Photoreceptors
	Genetic ablation
	Retinitis pigmentosa [239, 240]	Photoreceptors
	Diabetic retinopathy [241, 242]	RGCs
	Age-related macular degeneration [243, 244]	RPE and photoreceptors
	Glaucoma [245–248]	RGCs

Mechanical injury is one of the feasible methods to study whole retina regeneration that ensures uniform damage to all retinal layers [10]. It involves surgical procedures, including a poke, small incisions or even the removal of a small retinal patch. In poke or stab wound injury with a 30G needle, the eyeball is tilted with the help of forceps and stabbed with a needle so that it penetrates all the retinal layers to the vitreous [11]. Another injury method involves retinal detachment with subretinal saline or hyaluronic acid injections through small scleral incisions [12, 13]. Retina regeneration studies also involve the removal of a small retinal flap, including all retinal layers involving transscleral injuries with the help of a micro-knife and the removal of a small rectangular retinal patch [14, 15]. In general, mechanical damage is a frequent mode of injury to animal tissue. Repairing a mechanical injury through regeneration or wound healing could have been reflected in the natural selection and evolutionary advantage to the animal’s survival.

The chemical injury involves injecting chemical moieties such as ouabain, N-methyl D-aspartate (NMDA), 6-hydroxy dopamine (6-OHDA), tunicamycin, N-methyl-N-nitrosourea (MNU) or hypoxia inducing factors. These chemicals are injected in varying doses depending on the retinal layer to be targeted and hence can be used to mimic various eye diseases. Ouabain is a cardiac glycoside that inhibits the Na+/K+ pump by binding to the K+ binding site of ATPase, thus affecting the resting potential of nerves and disrupting the retinal metabolism [16]. Varying concentrations of ouabain have a differential effect on retinal layers [17, 18], whole retina [19], ganglion cell layer, inner nuclear layer [20] and photoreceptor cells [17]. NMDA causes neuronal cell death by overexciting synapses due to increased NMDA receptor-mediated cation influx [21]. NMDA has been used to cause retinal injury by inducing ganglion cell death, activating glial cells to cope with damage [22–27]. The most accessible dopaminergic neurons of the vertebrate retina are dopaminergic amacrine cells. Retinal dopamine has multiple roles in vision, like adapting light/dark retinal circuits and influencing trophic processes [28]. The 6-OHDA (a hydroxylated analog of natural dopamine) is one of the most common neurotoxins that induce rapid death of dopaminergic neurons. In the solution, 6-OHDA converts to quinone, which stimulates the production of free radicals. For using relatively high doses of 6-OHDA, sodium ascorbate in the solution prevents the free radical formation and hence non-specific damage [25, 44]. Tunicamycin is an antibiotic inhibiting N-glycosylation of asparagine-linked oligosaccharides [45]. Rod photoreceptors contain the visual pigment rhodopsin, a membrane protein with two asparagine-linked glycoproteins. Tunicamycin treatment inhibits glycosylation of opsin, leading to disruption of the rod outer segment membrane assembly [31] and shortening of rod outer segments and hence photoreceptor-specific damage [47, 48]. MNU is an alkylating agent causing DNA damage. It causes photoreceptor damage by apoptosis-induced cell death [49]. CoCl₂ affects hemoglobin by preventing iron inclusion in heme and affecting oxygen carriage, leading to the production of hypoxia-inducible factors [50]. CoCl₂ also inhibits proteasomal degradation of hypoxia-inducing factors, thereby promoting hypoxia-mediated injury [36]. Intravitreal injection of CoCl₂ is used to mediate hypoxic injury and has been used to damage different retinal layers as photoreceptors [52, 53] or ganglion cell layers [54]. Chemical modes of retinal injury enable us to understand and emphasize the regenerative mechanisms adopted by an organism in response to various cellular insults initiated by chemical toxins.

Light is necessary for vision, but constant long-term exposures or high-intensity light can damage the photoreceptor layer [55]. The electromagnetic spectrum in the range of 400–1400 nm, which is allowed to pass through the retina, is considered the retinal hazard zone [55]. Light-induced damage follows the disruption of the standard light and dark cycle to a long dark cycle followed by exposure to high-intensity visible light [56] or ultraviolet light [57]. The light-induced damage is with tungsten halogen lamps, metal halide lamps and fiber optics [25]. Photothermal, photochemical and photomechanical mechanisms can mediate light-induced damage. Photothermal damage happens by the transfer of radiant energy in the form of photons to the retinal tissue, which increases intramolecular collision and hence dissipation of energy as heat. Irreversible thermal damage to the retina occurs after the ambient temperature is raised by at least 10°C. Depending on the extent of damage, cells may undergo apoptosis, necrosis or immediate cell death [55]. This kind of injury occurs in laser light photocoagulation and optical coherence tomography-guided laser injuries [25]. Photochemical damage is caused by oxygen free radicals. Light in the high-energy portion of the visible spectrum interacts with chromophore molecules like photoreceptors, flavoproteins, heme proteins, melanosomes and lipofuscin present in the retina and retinal pigment epithelium (RPE). These chromophores then release energy to oxygen, generating singlet oxygen species. These free radicals lead to the oxidation of polyunsaturated fatty acids and protein moieties that break down membranous structures. Retinal photoreceptors, especially outer segments, possess large amounts of membranes and are hence more susceptible to free radical-mediated damage [25, 55]. Rapid introduction of energy into melanosomes generates mechanical compressive and tensile forces, causing photomechanical damage. These photomechanical forces result in micro cavitation bubbles, lethal to RPE and other cells [25, 55]. Photo-ablation ensures the damage to photoreceptors, the functional cell types of vision. Photoreceptors when selectively damaged in the retina respond via rod progenitors and Müller glia depending on the extent of damage to the retina.

With the advent of research in transcriptomics, researchers have come up with the association of several genes with retinal disorders. This knowledge has been implied in transgenics and genome editing techniques including TALENS, CRISPR-Cas and morpholino-based gene silencing to mimic the disease in model organisms. Various animal models including fish, amphibians and mammals have been developed to mimic retinal diseases like retinitis pigmentosa, diabetic retinopathy, macular degeneration and glaucoma. Transgenic approaches involving nitroreductase/metronidazole have been used in zebrafish to cause cell-specific ablation (reviewed in [25]). The retinal damage paradigms used in different animal models are given in Table 1. Despite the differences in the mode of injury and the species used, the fundamental principle of regeneration remains similar. In all the cases, there is a necessity to form progenitors from pre-existing cells before the differentiation into various retinal cell types.

REGENERATION MECHANISMS

Regeneration relies on replacing damaged or lost cells that happen through different mechanisms, including dedifferentiation, trans-differentiation or reprogramming [43]. Dedifferentiation involves changing terminally differentiated cells to a less differentiated state. These less differentiated cells within their lineage can then divide and replace the lost cells in their vicinity by changing into their final differentiated stage. In trans-differentiation, the already differentiated cells change lineage to form another cell type. To replace the damaged or lost tissue, existing differentiated cells either first dedifferentiate and then differentiate to a new cell lineage or directly change to form new cells. In reprogramming, a fully differentiated cell reverts to its pluripotent stage. In response to injury or damage, fully differentiated cells start dedifferentiation, attaining stem cell-like characteristics which then proliferate and differentiate into different cell types, restoring the lost tissue (Fig. 3). Efficient retina regeneration ensues in cold-blooded vertebrates like fish, urodele amphibians, frogs and embryonic stages of rodents and birds. In these animals, different cellular sources contribute to retina regeneration. These cells, including neural stem cells at the retinal periphery (ciliary marginal zone, CMZ), Müller glia, rod progenitors and RPE, adopt any of the above modes for regeneration to pursue (Table 2).

Figure 3.

An overview of three different retina regeneration mechanisms followed in the vertebrates. (a) The transdifferentiation of RPE in embryonic stages of amphibians and chicks after whole retina removal. The RPE regenerates the whole retina by losing its pigmentation, changing morphology and proliferating to differentiate into different neuronal retinal cell types. In anuran and urodelian amphibians the RPE needs the presence of a choroid as a support to form the whole retina. In contrast, a portion of neural tissue (retina) ensures retina regeneration in chick embryos. Still, it is of reverse polarity, i.e. the photoreceptor layer faces the vitreous humor, and ganglion cells are toward the vascular layer (choroid). (b) The post-retinal damage reprogramming of Müller glia in zebrafish, chick and mice involves sensing damage by glial cells, which then dedifferentiate to form multipotent stem cells. The dedifferentiated Müller glia divide and migrate to different retinal layers by interkinetic migration, where they differentiate to form lost retinal cell types and restore the vision. The regeneration by Müller glia reprogramming is inherently successful and complete in zebrafish, while in chicks, not all proliferated Müller glia form neuronal cell types. Mice Müller glia are reluctant to reprogram, but in recent studies, the genetic manipulations and/or inhibition of HDACs have enabled their Müller glia to regenerate whole retinal tissue. (c) The CMZ containing constantly proliferating cells at the retinal periphery contributes to retinal growth in adult zebrafish and amphibians throughout their lives. In chick retina, CMZ contributes to neuronal cell types but only amacrine and bipolar cells, up to one-month post-hatching. CMZ is also known to contribute to adult mice retina growth by forming only ganglion cells. In response to light-mediated damage, the rod progenitors, with their origin either in the outer nuclear layer or from progenitors in the inner nuclear layer, contribute to the restoration of lost photoreceptors in the zebrafish retina.

Table 2.

The retina regeneration mechanisms adopted at different developmental stages of mammalian and non-mammalian vertebrates enlisting the retinal cell types involved and the status of regeneration in them

Animal	Animal age and retinal progenitors involved	Mechanism	Regeneration status
Fish	Adult
	CMZ [60]	Asymmetric division along radial axis	RPE regeneration
	RPE [71, 225]	Proliferation and differentiation	Continuous Retinal growth
	Muller glia [78, 79, 86]	Reprogramming	Complete and functional retina regeneration RPE regeneration
	Embryo
Amphibians	Embryonic
	[249, 250]	Developmental regrowth	Functional eye regrowth which declines with age
	Tadpole
	CMZ [59]	Proliferation and differentiation	Retinal growth at margins
	Muller glia [233, 251]	Reprogramming	More efficient in aged animal
Birds	Embryonic
	Adult
	CMZ [66]	Proliferation and differentiation	Whole retina regeneration
	RPE [69, 70]	Transdifferentiation	Neural retina and lens regeneration
	Muller glia [251]	Reprogramming	More efficient in aged animal
Birds	Embryonic
	RPE [73, 74, 76]	Transdifferentiation with FGF2 or lin28 and a part of neural retina	Retina of reverse polarity
	Post-hatch
	CMZ [252]	Proliferation and differentiation	Amacrine and bipolar cells
	CMZ [253]	With growth factors (FGF2 and insulin)	Amacrine, bipolar and ganglion cells
	Muller glia [80, 88, 89]	Reprogramming (NMDA, FGF2, CNTF)
Mammals	Embryonic
	CMZ [65]	Proliferation, translocation and differentiation	RGCs generation
	Post-natal
	Muller glia [193]	Reprogramming (forced Ascl1 expression)	Amacrine, bipolar and photoreceptor cells
	Adult
	Muller glia [37, 42, 89]	Reprogramming (NMDA, FGF, insulin, retinoic acid)	Few bipolar, rod and amacrine cells
	Muller glia [90]	Reprogramming (forced Ascl1 expression with HDAC inhibitor)	Functional retina regeneration with more bipolar cells

Differentiation

Dedifferentiation in the strict sense is not reported during retina regeneration. Here, the consistently growing population of stem cells residing in the CMZ contributes to either retinal growth or regeneration by differentiating into different retinal cell types. In amphibians and teleost fish, the retina grows proportionately to eye or body growth. This continuous growth is due to stem cells residing throughout their life in CMZ. In frogs, most of the retina forms during the tadpole stage from CMZ, with only marginal growth had happened during embryonic development [59]. The stem cells residing in CMZ also contribute to zebrafish retinal growth by dividing asymmetrically in the radial axis and adding concentric rings of new cells [60]. The newly added cells differentiate to form different neural retinal cell types in the proposed sequence from ganglion cells first to cones, horizontal cells, amacrine cells, rods and bipolar cells. After the neurogenesis phase is over, the remaining cells differentiate into Müller glia, which are the last to form and have functions similar to that of astrocytes in the CNS [61]. Chicks’ eyes continue to grow for about a month after hatching with retinal development by cells within the CMZ. However, unlike amphibians and fish, only amacrine and bipolar cells are added to the chick retina under normal physiological conditions without any external factor supplementation. The inability of stem cells in chick CMZ to generate all types of neurons may not be intrinsic to progenitors. Still, it could be due to local factors, which could be overcome by exogenous regulatory molecules like insulin, FGF2, etc. [62, 63]. Retinal stem cells are also identified in adult mice CMZ [64] and shown to form ganglion cells by traveling laterally from CMZ to the neural retina in embryonic stages (E10-15) [65]. CMZ is not the primary source of stem cells after injury in regeneration-competent animals, including fish and amphibia. However, a study in Xenopus tropicalis has shown that CMZ participates in complete retina regeneration after total retina removal [51]. Thus, the differentiation of CMZ as a mode of regeneration is often limited in its efficacy except in Xenopus.

Transdifferentiation

The embryonic stages of anuran amphibians and avian embryos can regenerate their retina even after its complete removal, and this regenerative capability is retained even up to adulthood in some urodelian amphibians [67]. In these animals, retina regeneration occurs through RPE transdifferentiation. The RPE restores the damaged retina by dedifferentiating into proliferative neurogenic progenitors [68]. Dedifferentiating RPE loses its pigmentation, changes morphology, detaches from the basement membrane [68] and starts expressing progenitor markers such as Klf4, Sox2, Pax6 and c-Myc [69]. Transdifferentiation requires interaction between the connective tissue and RPE. In the newt, the choroid acts as a connective tissue while in the embryonic chick it is a fragment of the neural retina [67]. Whole retina regeneration is also evident in post-metamorphic Xenopus laevis by transdifferentiation of RPE and differentiation of stem cells in CMZ [70]. In adult fish, RPE can regenerate itself after its ablation but does not contribute to retina regeneration [71, 72]. Notably, the avian embryos do not regenerate the retina spontaneously from RPE, but growth factor treatments such as fibroblast growth factor (FGF) [58] or lin28, an RNA binding protein that is important pluripotency inducing factor involved in reprogramming [74], are known to induce RPE transdifferentiation in them. Furthermore, unlike amphibians, embryonic chicks regenerate a nonfunctional retina of reverse polarity (i.e. photoreceptor layer faces the vitreous) and RPE-derived cells do not transform into RPE itself [75, 76]. Though in vitro studies have shown FGF2 to transdifferentiate early differentiated embryonic rat RPE to neural retina up to a certain stage [77], mammalian RPE cells in vivo are incompetent to reprogram into neural retina. These studies suggest that transdifferentiation as a means of retina regeneration is ineffective in non-amphibian vertebrates.

Reprogramming

Reprogramming of Müller glia is the potential source of retina regeneration in zebrafish [78, 79], chicks [80] and mammals [37, 42]. Müller glia are the non-neuronal cells with nuclei in the inner nuclear layer and cytoplasmic extensions spanning all the retinal layers. The unique morphology of Müller glia enables them to sense the damage in any of the retinal layers induced by either light, chemicals, cell-specific genetic ablation or mechanical injury [81]. Soon after the tissue damage, the chromatin and transcriptome of Müller glia change to allow them to dedifferentiate [81–83]. This reprogrammed Müller glia enters the cell cycle and starts expressing early neural progenitor markers [61, 84, 85]. In 2006, the use of 1016α1T: GFP transgenic fish, in which adult zebrafish tubulin1α promoter drives GFP expression in injury responsive MGPCs, established Müller glia as the cellular source of fruitful retina regeneration [79]. Since then, several studies have been there to elucidate molecular mechanisms of zebrafish Müller glia reprogramming [9, 61, 81, 86, 87]. The very first evidence of Müller glia as a source of regenerated neuronal cells is from the chick retina [80]. In response to retinal damage, Chick Müller glia enter the proliferative state and express proneural genes [82]. The reprogrammed chick Müller glia are sustained for a long time, expressing undifferentiated cell markers [80, 88], but even after attaining progenitor-like characteristics, the majority of them do not produce neurons probably due to the non-attainment of mature retinal progenitor cell identity [61]. Reprogramming mammalian Müller glia needs either growth factor supplementation, epigenetic modification, transgenic approaches or a combination of them [37, 89, 90]. Without these manipulations, mouse Müller glia does respond to retinal damage, by migrating and expressing progenitor and cell cycle-specific markers but does not enter the cell cycle [91]. The inherent inability of mouse Müller glia to reprogram could be due to epigenetic factors including DNA methylation [92], histone modifications [90] or cell cycle inhibition [91]. In general, the reprogramming of the Müller glia through genetic and epigenetic mechanisms allows the smooth switching into a proliferating population of progenitors, which gives rise to major retinal cell types during retina regeneration.

In the zebrafish retina, continuous growth is supported by stem cells in CMZ and by the continued addition of rod photoreceptors [93]. These rod progenitors arise from a population of slowly dividing cells in the INL. These cells form radially elongated neurogenic clusters and migrate from INL to the ONL, generating rod precursors [94]. Even in the uninjured zebrafish retina, Müller glia proliferates at a low rate expressing the multipotent progenitor marker, Pax6. Müller glia-derived progenitors, which migrate to ONL express Crx (cone-rod homeobox), are the retinal progenitors generating rod photoreceptor lineage [78]. These rod progenitor cells, which originate in INL, divide and migrate to ONL, proliferate further before differentiating and are known to replenish the rod photoreceptors [95]. In the case of photoreceptor-specific ablation, when a retinal injury is chronic or not strong enough to activate Müller glia, the rod progenitors respond to damage [96, 97]. In the diabetic model of zebrafish with pdx1 homozygous mutation rod and cone cells regenerate from slowly dividing neurod: GFP expressing progenitors with their origin in ONL [98]. The pancreatic and duodenal homeobox 1 (pdx1) mutant fish represents diabetic features with reduced beta cells and insulin levels along with elevated glucose levels. In adult zebrafish, the cells destined to become photoreceptors express Neurod [99]. Moreover, the photoreceptors originate without Müller glial activation, which suggests the involvement of rod progenitors in the replenishment of both rod and cone cells [98, 100].

IS REGENERATION SIMILAR TO EMBRYONIC DEVELOPMENT OR CANCER?

Tissue regeneration often shares hallmarks of embryonic development, despite the differences. Several recent studies reveal that various gene regulatory networks active during tissue regeneration are similar to that of embryonic development (reviewed in [101]). During embryonic development, the retina grows as an outgrowth of the forebrain, and different neuronal cell types form from a multipotent progenitor cell. These multipotent progenitors can undergo multiple rounds of cell division, generating different retinal cell types in a rough sequence. A ‘clock’ controls this sequence of generating different cell types during embryonic development with the transition of progenitors from making early cell types to those generated later. A set of gene regulatory events and transcription factors control this transition, driving the acquisition of neural fate and then sequential generation of different cell types. After a retinal injury, this set of events needs to be restarted for restoring different retinal cell types. The Müller glia and RPE, in addition to their immense division capability, also retain the ability to ‘dedifferentiate’ and ‘transdifferentiate’ into progenitor cells. These progenitors resemble those present during embryonic development stages; hence, responding cells seem to follow ‘winding the clock back’ during retina regeneration. In amphibians, the RPE-derived progenitors regenerate the retinal cell types in the same order as during development [102]. Chicks Müller glia proliferate after a retinal injury, but only a few differentiate into neuronal cell types consisting mainly of interneurons [80, 88, 103]. Proliferating Müller glia in chick retina express some progenitor cell markers as Ascl1, Chx10, Pax6, Klf4 and cFos but do not turn on the Otx2, a feature of photoreceptor cells [80, 104, 105]. In zebrafish, after an acute injury, Müller glia-derived progenitors can restore all the retinal cell types. The order of generating new neurons is similar to that of other vertebrates [78, 79]. The early expression of Atoh7 marks the developing ganglion cells as the first cells generated during zebrafish development [106]. During retina regeneration, Müller glia in fish express progenitor-cell associated transcription factors like Pax6, Atoh7, Islet1 and Otx2 [34, 81, 85], and the temporal order of proneural transcription factors is the same as during development [61]. It suggests the ability of fish Müller glia to ‘wind back’ their molecular clock to that of a developmental progenitor state [85]. One of the master regulators behind zebrafish retina regeneration seems to be Lin28a. Soon after the retinal injury, Müller glia starts expressing Lin28, repressing the microRNA (miRNA)-let-7. The Lin28-mediated suppression of let-7 allows Müller glia specific expression of pluripotency factors and efficient retina regeneration (Fig. 4) [107]. The role of Lin28 is conserved across the diverse genera, and it regulates cell fates during development by regulating miRNAs [108–110]. A recent study also shows the transcriptional similarity between the fish Müller glia-derived progenitors and embryonic progenitors [82]. Despite the differences in the cellular niche, the damaged retina often recapitulates various developmental biology events during a regenerative response.

Figure 4.

The diagram depicts the set of molecular events leading to Müller glia reprogramming and retina regeneration focusing on zebrafish. Any retinal damage induces the activation of Müller glia (MG) from the resting stage in the intact retina. A number of factors lead to dedifferentiation of activated MG attributed mainly to the increase in Lin28 expression, which degrades let-7 miRNA. This series of events is accompanied by a hike in pro-inflammatory cytokines and pluripotency inducing factors (PIFs), which lead to activation of many signaling pathways and MG proliferation. A decline follows the proliferative phase in expression levels of PIFs with simultaneously increased expression of neuronal markers culminating in the generation of different retinal cell types with MG cells as the last to form.

The epithelial to mesenchymal transition (EMT) is an essential cellular phase during cancer. The onset of cancer involves EMT that enables the rapid proliferation and subsequent migration of the cancerous cells to various parts of the affected organism. The opposite mesenchymal to epithelial transition (MET) contributes to the establishment of a new cellular phenotype. Similar steps are needed even during retina regeneration. There are several gene expression events occurring in the zebrafish Müller glia similar to MET, which contribute to the reprogramming and induction of progenitor cells. The cellular reprogramming into pluripotent stem cells is also physiologically related to MET. Several studies on MET during reprogramming have proven that SNAI1 (SNAIL), which is a facilitator of EMT, causes a reduction in reprogramming efficiency both in human and mouse cells [111]. During zebrafish retina regeneration, we see a decline in Snail family genes suggesting the existence of a MET-like situation to induce progenitor cells [112]. Moreover, other findings also support that EMT not only provides cell motility but is also capable of inducing stem-cell properties alongside preventing cellular apoptosis and senescence [113–115]. During zebrafish retina regeneration, the progenitor cell proliferation seen after an injury is dependent on TGF-β signaling that contributes to the downregulation of E-cadherin [112]. The zinc-finger transcription factors ZEB1 and ZEB2 induce EMT by suppressing E-cadherin and hence reorganization of epithelial cells to become migratory mesenchymal cells, a characteristic of cancer metastasis. Zeb1 and Zeb2, are directly controlled by TGF-β signaling during zebrafish retina regeneration leading to the proliferation of the retinal progenitors [112]. Thus, Müller glia reprogramming is initiated with EMT followed by MET before switching into proliferating progenitors that differentiate into retinal cell types. Such observations during retina regeneration draw a parallel with embryonic development and cancer.

IMMUNE RESPONSE IN RETINA REGENERATION

After the tissue injury, damaged cells release damage-associated molecular patterns (DAMPs), which are recognized by pattern recognition receptors. DAMPs, including purine metabolites such as ATP or uric acid, heat shock proteins and high mobility group 1 protein, are sequestered intracellularly in intact cells [116]. But upon the damage, injured cells release these DAMPs extracellularly, activating both classical and non-classical pattern recognition receptors, including toll-like receptors, Nod-like receptors, etc. These events lead to the induction of downstream signaling pathways and the production of pro-inflammatory signals (Fig. 5). These signals recruit resident and circulating immune cells to the injury site and promote chemical mediators’ production and release (reviewed in [117]). The innate immune response mediated instant and acute inflammation helps restore tissue homeostasis and activate healing. But dysregulation of inflammation can lead to chronic inflammation, which is harmful to injured tissue and does not culminate in regeneration [118]. In regenerating tissues, inflammatory signals enhance chromatin remodeling for access to DNA by reprogramming factors and promote proliferation in reprogrammed cells [119, 120]. Upon retinal injury, the mammalian Müller glia undergoes reactive gliosis. The reactive gliosis seems to be neuroprotective initially, but it eventually leads to neuronal cell loss and scarring [116]. In zebrafish also, retinal injury induces an inflammatory response, but instead of the glial scar, it pursues Müller glia proliferation and replacement of lost retinal neurons. A recent study with transcriptomic profiling of chick, zebrafish and mice retina suggests mice Müller glia quiescence due to nuclear factor I (Nfia/b/x) and Sox5. The mice Müller glia express these transcription factors at the resting stage, which are downregulated soon after retinal damage but get upregulated to match the resting Müller glia at later stages [82]. These NFI factors maintain mice Müller glia quiescence, preventing transition to a progenitor-like state and neurogenesis during development as well as post-retinal damage [82, 121]. Injury-induced inflammation also plays a vital role in postnatal chick retina regeneration, where it activates the reprogramming of Müller glia. Still, persistent activation of the NF-κB pathway inhibits the proliferation of progenitors [122]. The complement component C3a alone induces embryonic chick retina regeneration via STAT3 activation and induction of pro-inflammatory cytokines, TNF-α, IL-6 and IL-8 [123].

Figure 5.

Post-retinal damage induction of immune response leads to successful retina regeneration. Any sort of retinal damage induces cell death as a result of which the dying cells release DAMPs, including heat shock proteins (HSPs), adenosine triphosphate (ATP) and uric acid. These DAMPs then induce the immune cells to release pro-inflammatory cytokines such as IL-6, IL-8 and TNF-α. The cytokine surge culminates in the induction of various signaling pathways resulting in the reprogramming of responding cells and retina regeneration.

Microglia are the resident macrophages acting as immune cells in the nervous system, which sense and respond to insult by producing cytokines. After the retinal injury, microglia play a crucial role in modulating Müller glia response, and microglial absence hinders chick or zebrafish Müller glia proliferation [87, 124]. The response of microglia can be pro- or anti-inflammatory depending upon their polarization or the duration up to which they stay active. During zebrafish retina regeneration, suppression of immune response before retinal damage delays, but the post-injury application of immune suppressants accelerates Müller glia/photoreceptor regeneration. The induction of inflammatory response is necessary for zebrafish retina regeneration, but it needs a timely resolution to allow successful progression [125]. The microglia/macrophage-mediated inflammation is a key regulator of mTOR in the Müller glia enabling mTOR-mediated retina regeneration [126]. Pharmacological and genetic ablation reveals the mTOR pathway as an integral part of zebrafish RPE regeneration also, where it recruits macrophages/microglia to the injury site that in turn further activates the mTOR pathway [127]. The microglia are also crucial for zebrafish RPE regeneration, where RPE expresses cytokines including interleukin34 (il34) as a leukocyte recruitment factor [128]. In contrast to this, the microglia in the mouse retina are known to induce inflammatory genes reducing neurogenesis [129]. The knowledge of the immune response during retina regeneration in various vertebrates reveals a crucial involvement of various immunological molecules that govern a smooth regenerative response.

MOLECULAR BASIS OF RETINA REGENERATION

The retinal injury induces Müller glia and efficient retina regeneration in cold-blooded animals like teleost fish but not so in mammals. Different research groups in varying parts of the world have uncovered the molecular mechanisms underlying the regenerative capability of vertebrates. Several developmental signaling and other pathways are known to induce Müller glia reprogramming or maintain their quiescent state (Figs 4 and 5).

Cytokines

Retinal damage causes cell death and the release of pro-inflammatory cytokines. In the Xenopus eye, retinal removal induces Il-1β and Tnf-α, which upregulate mmp9 and mmp18 essential for RPE transdifferentiation [134]. In the photoreceptor regeneration using transgenic zebrafish, TNF-α accounts for the transition of Müller glia from non-proliferative gliosis to a regenerative state [131]. Upon retina damage, dying retinal neurons produce Tnf-α, which is necessary for Müller glia proliferation in zebrafish retina regeneration [135]. After neuronal damage, the dying Müller glia release cytokines and inflammatory Mmp9. Mmp9 is a matrix metalloproteinase, which plays a crucial role in inflammation by cleaving precursor or mature cytokines, making them active or rendering them inactive. Mmp9 plays an essential role in resolving an inflammatory response, as its mutants express higher levels of pro-inflammatory cytokines such as TNF-α [132]. The importance of Mmp9 is also supported by the induction of mmp9 soon after damage and its strong upregulation in MGPCs [133]. During photoreceptor regeneration, Mmp9 plays a vital role in cone photoreceptors’ survival [132]. In zebrafish, the photoreceptor damage upregulates mmp9 in dividing Müller glia and photoreceptor progenitors. The expression of mmp9 is also induced by TNF-α, even in intact retinae. Insulin, IGF-1, HB-EGF and cytokines synergize to stimulate Müller glia reprogramming even in the uninjured zebrafish retina through a set of signaling pathways [136]. The leptin and Il-6 family cytokines promote zebrafish retina regeneration post-injury by Jak/Stat signaling [137]. The midkine cytokine is necessary for G1 to S phase transition in zebrafish. Its loss results in reactive gliosis of Müller glia, which may be one of the reasons for unsuccessful mammalian retina regeneration [130]. While the persistent release of cytokines is regulated through various cell signaling events, their necessity is proven to be a hallmark of retina regeneration across multiple model systems.

TGF-β signaling

TGF-β signaling governs many biological processes that have an anti-proliferative role in various biological systems. In zebrafish, TGF-β3 promote retina regeneration via a canonical pathway involving mycb and junb gene family activation [138], while other study reports its inhibitory role probably via a non-canonical path [139]. We have found the biphasic role of the TGF-β1 signaling pathway during zebrafish retina regeneration, where it regulates regeneration-associated genes and miRNAs for successful Müller glia reprogramming, proliferation and then differentiation to maintain the retinal homeostasis [112]. TGF-β signaling plays a neuronal cell death protective role during mice’s retina development [140], while in adult mice, its absence in microglia leads to neuroinflammation and retinal degeneration [141]. Post-retinal damage, TGFβ/Notch signaling axis, reprograms mouse Müller glia to epithelial lineage and glial scar formation [142]. TGF-β1 and TGF-β2 activate non-canonical p38MAPK signaling in mice leading to gliosis. It is interesting to note that the anti-proliferative TGF-β signaling in mammals, which often turns pro-proliferative during the cancerous conditions, is pro-proliferative during zebrafish retina regeneration.

Pluripotency factors

After the initial reprogramming events, Müller glia attain stem cell-like characteristics with induction of pluripotency factors [107]. We have explored the role of oct4 during zebrafish retina regeneration, where initial pan-retinal expression of oct4 regulates Müller glia reprogramming via regulation of regeneration-associated genes, E-cadherin and micro-RNAs. Oct4 also plays an essential role in cell cycle exit by suppressing pro-proliferative genes in collaboration with Hdac1 [143]. The morpholino-mediated knockdown of sox2, klf4 and nanog, too, impair the zebrafish retina regeneration (unpublished data). We have shown Myca/b to be an early induced pluripotency factor that regulates zebrafish Müller glia reprogramming via activation of ascl1a and lin28 [144]. The Sox2 is a crucial player in zebrafish Müller glia reprogramming and proliferation, where it activates ascl1a and lin28a, and its loss affects cone photoreceptor regeneration [145]. In medaka fish, Sox2 and other factors maintain Rx2 expression in CMZ, regulating stem cell fate toward neural retina or retina pigment epithelium [146]; alongside this post-retinal damage, sustained Sox2 expression allows a regenerative response similar to zebrafish [147]. In chicks, Sox2 reprograms RPE differentiation toward retinal neurons [148]. The conditional expression of Ascl1 and Lin28 in uninjured zebrafish and mice retina stimulates sparse Müller glia proliferation. The simultaneous inhibition of Notch signaling enhances neuronal regeneration in the former but not in the latter [149]. In rats, subretinal injections of OCT4 reprogrammed human pluripotent stem cells cause RPE generation [150]. The spontaneous induction of pluripotency-inducing factors in the injured retina or RPE paves the way for switching into a proliferating population of progenitors during retina regeneration.

mTOR pathway

Retinal damage induces the release and production of various factors, activating several signaling pathways. Stab wound injury in zebrafish induces mTOR signaling and is essential for Müller glia dedifferentiation and proliferation forming MGPCs by activating regeneration-associated factors like lin28a, ascl1a, cytokines and cell-cycle regulators [126]. Further, the downregulation of the tumor suppressor Pten that activates the Akt–mTOR pathway in the zebrafish retina is also pivotal in Müller glia reprogramming [152]. Retinal damage triggers mTOR signaling in activated Müller glia and its inhibition impairs MGPCs proliferation even in EGF2 treated retinae. Inhibition of the mTOR pathway also surpassed the MGPC-promoting effects of glucocorticoid, sonic hedgehog (Shh) and wnt signaling in chick retina [151]. These observations support the view that the mTOR pathway is important for successful retina regeneration.

Hippo signaling

Hippo signaling is an important developmental pathway regulating eye growth through YAP and TAZ. Yap is required for the Müller glia to respond to an injury by regulating their cell cycle re-entry and progenitor cell formation, leading to the differentiation of new photoreceptors in zebrafish [156]. Yap suppresses photoreceptor differentiation during zebrafish development by suppressing Rx1-mediated activation of photoreceptor genes [157]. Knockdown of yap1 impairs Müller glia proliferation and neurogenesis after photoablation in the zebrafish retina [82]. Upon retinal damage, YAP gets upregulated in mice and Xenopus Müller glia, and its conditional deletion inhibits cell-cycle gene expression and hence reactive gliosis in the former [154]. In postnatal mice, TAZ overexpression compensates for the loss of YAP. In adult mice-retina, YAP expression pertains to Müller glia. Its loss is not compensable by TAZ, leading to cone degeneration and Müller glia dysfunction in aged animals [153]. The hippo pathway in ‘on-state’ inhibits YAP nuclear translocation and blocks mammalian retina regeneration [155]. Müller glia-specific deletion of its component genes and transgenic overexpression of hippo unresponsive YAP leads to Müller glia reprogramming to a progenitor-like state [155]. In mice, the YAP-EGFR axis plays an essential role in the exit of Müller glia quiescence and proliferation in response to injury [154]. The YAP overactivation in the mouse Müller glia induces their reprogramming into highly proliferative cells, a feature necessary in all retinal regeneration events [154]. The developmental pathway of Hippo signaling in its ‘off-state’ is crucial to the retina regeneration in different vertebrates.

Shh signaling

Shh is another developmental pathway whose dysregulation is known to cause anomalies such as cyclopia. The Shh pathway is very important in several species studied in the retina regeneration context. In the developing retina of zebrafish and Xenopus, the Hedgehog pathway shortens the G1 and G2 phases, thus speeding up and early exit of the cell cycle [158]. Early Shh activation induces gliosis, proliferation and neuroprotection and its continued activation facilitates amacrine and ganglion cell differentiation in injured zebrafish retina [159]. In zebrafish, the blockade of Shh signaling completely abolishes the regenerative potential, and this pathway activates regeneration-associated genes to facilitate zebrafish retina regeneration [133]. Overexpression of recombinant Shh also activates proneural gene Ascl1 in both zebrafish and mice [133]. In chick retina, Shh overexpression alone induces proliferation from the anterior margin of the eye without external FGF2 [160]. Activation of the Shh pathway with its agonist purmorphamine induces transdifferentiation of Müller glia into rod photoreceptors in rat retina [161]. Shh also protects ganglion cells from chronic hypertension in adult rats [162]. Taken together, these observations suggest the crucial roles of yet another developmental pathway, Shh signaling, during retina regeneration.

Wnt/beta-catenin signaling

Wnt signaling is essential in embryonic development, stem cell maintenance, tissue repair and cancer [163]. During zebrafish retina regeneration, Wnt/β-catenin signaling plays a crucial role, where inhibition of GSK-3β to stabilize the β-catenin alone induces Müller glia dedifferentiation and retinal cell types formation even in the uninjured retina [164]. After retinal damage, Wnt signaling is stimulated via an early pan-retinal induction of Insm1a that suppresses pan-retinal dkk (a negative regulator of wnt signaling), ascl1a, and its expression to restrict Ascl1a-Insm1a-Dkk axis back to the site of injury until the regeneration is completed [165]. Wnt and BMP signaling reprogram the neural retina into RPE in the chick retina [166]. Activation of Wnt/β-catenin signaling promotes the reprogramming of Müller glia to precursor state, proliferation and neural regeneration in mice [167, 168] via the Lin28/let-7 pathway even in the uninjured retina [169]. Mice Müller glia after β-catenin gene transfer also reprograms to form rod photoreceptors on rod specific gene transfer [170]. These observations emphasize the pivotal roles played by Wnt signaling in many vertebrate models of retina regeneration.

MAP/ERK pathway

The MAP/ERK pathway is very important for cells to respond to external signals and subsequent cell proliferation. The MAP/ERK pathway is essential in many biological events from developmental biology, cell division, tissue repair and cancer. Several external signals, including EGF2, IGF-1, BDNF, cytokines, neurotrophins and DNA damage, activate the MAPK pathway. During zebrafish retina regeneration, ERK signaling gets activated in Müller glia by induction of multiple growth factors, including HB-EGF, IGF1, FGF2 and insulin. This pathway then acts synergistically with other signaling pathways to promote transdifferentiation and Müller glia reprogramming (reviewed in [171]). Soon after a retinectomy, ERK signaling in the adult newt gets activated with p-ERK nuclear translocation and loose cell contact, causing β-catenin nuclear translocation. Extracellular FGF2 keeps ERK activated, which along with β-catenin and HB-FGF signaling, leads to cell-cycle re-entry, transdifferentiation and proliferation of RPE. EGF2 administration in X. laevis and chicks activate the ERK pathway to induce pax6 expression and hence RPE reprogramming (reviewed in [171]). In the injured chick retina, the Müller glia showed accumulation of p-ERK1/2, and p-CREB proteins, along with transient expression of cFos and Egr1, an indication of active MAPK-signaling [172]. In mice retina, the p38 MAPK pathway provides LIF (leukemia inhibitory factor) dependent neuroprotection after light-induced degeneration [173]. Altogether, the MAP/ERK pathway is one of the important signaling events soon after retinal damage observed in both mammals and non-mammalian animals.

Glucocorticoid receptor signaling

The glucocorticoid receptor signaling (GCR) is connected with anti-inflammatory responses, and its agonists are used to treat inflammatory eye diseases. In the zebrafish retina, activation of the GCR pathway before rod photoreceptor ablation delays Müller glia proliferation and replacement of lost rod cells. Its activation after damage enhanced the photoreceptor regeneration, probably by rapid immune response resolution [125]. Intact chick eyes have GCR expression in CMZ, and upon acute retinal injury, it gets upregulated in Müller glia. The activation of the GCR pathway is anti-proliferative by inhibiting the FGF2/MAPK pathway, while its inhibition, in the injured chick retina, promotes Müller glia proliferation and neuronal differentiation [174]. In mice retina, activation of the GCR pathway prevents light-induced photoreceptor death [175] and induces retinal stem cell proliferation in CMZ [176]. These observations suggest the potential of GCR in adapting to different cell physiology and is an essential contributory pathway during retina regeneration.

Notch signaling

Notch signaling is known to maintain Müller glia quiescence in the uninjured retina. After retinal damage, Notch signaling keeps the injury-responsive cells at the site of injury and regulates the threshold and proliferation of injury-responsive Müller glia [177]. In the undamaged zebrafish retina, inhibition of Notch signaling along with TNFα activation reprograms Müller glia generating different retinal neurons similar to regenerating retina [178]. Inhibition of Notch signaling and forced expression of Ascl1a and Lin28a stimulate widespread multipotent Müller glia progenitors in intact zebrafish retina but not in mice [149]. Notch signaling along with Tgfb3 inhibits zebrafish retina regeneration [139]. Similarly, Fgf8 mediated suppression of Notch signaling stimulates Müller glia proliferation in the young fish retina, while increasing Notch signaling and suppressing Müller glia proliferation in the older fish [179]. Notch signaling mediates the early steps of RPE transdifferentiation in the newt retina, where Notch-1 expressing cells first appear in the regenerating retina [180]. In the chick retina persistently active Notch signaling is inhibitory to neuronal regeneration, while initial activation is essential for Müller glia dedifferentiation; for efficient neuronal cell differentiation, it needs to be suppressed at later time points [103]. LIM/homeobox transcription factor (Lhx2, a transcriptional activator) mediated regulation of Notch signaling is essential for Müller glia specification and differentiation in mice retina [181]. In the rat model of retinal degeneration, intravitreal injection of olfactory ensheathing cells preserves the visual function by suppressing Notch signaling and inhibiting Müller glia activation associated with gliosis [182]. These diverse observations of various vertebrate models suggest that Notch signaling plays a unique and stage-dependent role during retina regeneration.

Jak/Stat and other pathways

The Jak/Stat pathway and its stringent regulation are crucial during embryonic development and diseases such as cancer. Soon after retinal injury in zebrafish, damage-induced cytokines activate the Jak/Stat signaling culminating in Müller glia reprogramming by Ascl1a regulation and successful neuronal regeneration [137]. The NMDA-induced retina damage or growth factor administration activates Jak/Stat signaling in chick Müller glia facilitating proliferation but inhibiting neural differentiation [88]. In mice retina, TNFα activates Müller glia proliferation and inflammatory response instead of neural regeneration via Jak/Stat and MAPK pathways [183]. The inability of mice Müller glia neural regeneration is due to STAT directed binding of Ascl1 to inappropriate targets like Id1 and Id3, which keeps the cells in a progenitor-like state and hence prevents differentiation [184]. Thus, the stringent regulation of STAT and subsequent gene activations are essential for the normal and effective regenerative response in the damaged retina.

Several studies demonstrate that many developmentally essential pathways are active and necessary for efficient retina regeneration. It is tempting to speculate that retina regeneration employs several developmental biology pathways because of the recapitulation of embryonic development during retina regeneration. It is also important to note that several of these pathways contribute to the onset and progression of cancer, which also parallels retina regeneration. However, it is essential to note that despite many developmental and cancer biology pathways being turned on during retina regeneration, the tissue efficiently controls the cell proliferation through apoptosis and avoids shifting into cancerous conditions in normally regenerating organisms such as fishes and frogs.

EPIGENETIC ASPECTS OF RETINA REGENERATION

Retina regeneration involves responding cells transition from a fully differentiated quiescent state to stem cell-like features. Many genetic and epigenetic events happen during reprogramming, allowing chromatin accessibility and gene transcription [101]. The epigenetic regulations involving DNA methylation, histone modifications and non-coding RNAs enable the switching of differentiated cells to pluripotent-like states. The changes in genes encoding chromatin-modifying proteins are associated with various congenital retinal malformations (reviewed in [185]). DNA demethylation contributes to Müller glia reprogramming in zebrafish, but consistent hypomethylation affects progenitor cell proliferation, migration and differentiation. The changes in methylation patterns were revealed by reduced representation bisulfite sequencing. Indicating that active DNA demethylation occurs soon after damage during zebrafish retina regeneration, allowing Müller glia reprogramming, and then methylation patterns are regained to enable migration and differentiation [83]. The RPE of the chick (embryonic stages 23–25) retina undergoes dynamic changes in bivalent histone marks (H3K27me3/H3K4me3) and DNA demethylation during reprogramming. In the chick retina, tet methylcytosine dioxygenase (TET3) facilitates DNA demethylation and RPE reprogramming, even in the absence of external growth factors [186].

Moreover, promoters of regeneration-associated genes were already hypomethylated in the resting Müller glia of the zebrafish retina, which stayed the same in the progenitor cells [83]. In contrast to zebrafish, the Müller glia in medaka is less responsive to injury forming only photoreceptors. After retinal damage, Sox2 gets downregulated (silenced) in Medaka fish, and upon restoration of Sox2, medaka Müller glia can behave like that of zebrafish [145, 147]. Like the zebrafish retina, the epigenome of mice Müller glia resembles that of late retinal progenitor cells with hypomethylation of regeneration-associated genes [187]. Though mice Müller glia transcriptome changes post damage, it fails to stimulate cell cycle and retinogenic factors to the states observed in early retinal progenitor cells [188]. Soon after the injury, the promoter of Oct4 is hypomethylated but then returns to methylated status as that of quiescent Müller glia with concomitant changes in DNA methyltransferase (dnmt3b) levels [92]. Since Oct4 plays a vital role during zebrafish retina regeneration, silencing of Oct4 after mice retinal damage may account for unsuccessful renewal in the latter [143].

Chromatin modifying enzymes are emerging as important regulators of zebrafish retina regeneration. Histone methyltransferases (Dotl1, an H3K79 methyltransferase) and histone deacetylases (Hdac1, histone deacetylase 1) are essential for Müller glia dedifferentiation and proliferation in zebrafish [144, 189, 190]. The mammalian RPE reprogramming inefficiency is also due to the highly methylated promoters of photoreceptor-related genes and repressive chromatin marks on the non-photoreceptor neuronal genes [191]. The in vitro studies with mice Müller glia in dissociated cultures demonstrated reactivation of many genes and neuronal generation after Ascl1a overexpression. Chromatin immunoprecipitation analysis revealed a reduction in repressive (H3K27me3) while enhancement in activating (H3K27ac) histone marks at the reactivated genes [192]. The Ascl1a transgenic mice could reprogram their Müller glia, generating neurons (bipolar and amacrine cells) in the early stages (2 weeks old) but not in mature animals [193]. The DNase-seq analysis in adult mice revealed the loss of open chromatin near neuronal genes, and inhibition of histone deacetylases (HDACs), along with Ascl1a overexpression, could make them regenerate successfully [90]. Epigenome modifications are key to the regenerative response in an injured retina both for induction of progenitors and differentiation to various retinal cell types.

The roles of non-coding RNAs

The non-coding RNAs, including short non-coding RNAs (microRNAs), long non-coding RNAs and circular RNA, play an essential role in several cellular processes. They regulate cell proliferation, differentiation and apoptosis and are also significant players in retinal development and the initiation and progression of retinal diseases (reviewed in [194, 195]). miRNAs are small non-coding RNAs that regulate many biological processes by interfering with the translation of target proteins. The miRNA transcripts need processing by two RNAse III proteins, Drosha and Dicer [196]. Alterations in miRNA expression are associated with retinal diseases such as age-related macular degeneration and glaucoma. The importance of miRNAs during retinal development and regeneration was noticeable after dicer inactivation (reviewed in [197]). Dicer mutant zebrafish and X. laevis have defective retinal growth, including cell differentiation and retinal lamination [198, 199]. Though dicer deficient mice did not have evident defects at an early postnatal time, aged mice showed defects in light responsiveness and retinal structure [200] (reviewed in [197]). miRNAs are also essential during retina regeneration as Dicer knockdown after retinal damage impaired proliferation of zebrafish Müller glia [201]. The small RNA sequencing at different time points after light damage revealed some miRNAs to be upregulated (miR-142b, miR-146a, miR-7a, miR-27c and miR-31). The morpholino-mediated knockdown of these miRNAs proved they are essential for retina regeneration [201]. The levels of some miRNAs get downregulated post-retinal injury, suggesting their involvement in maintaining a quiescent state or inhibiting Müller glia reprogramming. A well-studied miRNA let-7 maintains Müller glia quiescence, which must get downregulated by Lin-28 for efficient retina regeneration [107, 133, 189]. Similar to let-7, the other two miRNAs must be downregulated post-retinal damage, i.e. miR-203 and miR-216a, to allow successful zebrafish retina regeneration. The miR-203 targets Pax6b, inhibiting the proliferation of progenitors, and miR-216a targets Dot1l, inhibiting dedifferentiation and proliferation of Müller glia [190, 202]. The role of other miRNAs, miR-143, miR-145, miR-200a and miR-200b, is known during zebrafish retina regeneration. Oct4 and TGF-β signaling regulate these miRNAs to culminate in the efficient and controlled retina regeneration [112, 143]. The manipulation of miRNAs also promotes mammalian retina regeneration in vitro (reviewed in [197]). These studies emphasize the importance and necessity of miRNA-mediated gene regulations for efficient retina regeneration.

ARE WE NEARING MAMMALIAN RETINA REGENERATION?

The increasing number of shreds of evidence from non-mammalian vertebrates suggests that retina regeneration is carried out through efficient orchestration of gene induction and repression in a stringent spatial and temporal fashion. In mammals, soon after injury, Ascl1 [193] and Oct4 [92] are induced similar to that found in the injured zebrafish retina [107, 143]. Several mammalian models also suggest the propensity of retinal regeneration akin to the non-mammalian models in a controlled scenario such as overexpression of Ascl1 [192]. However, despite the inefficiency of regeneration, this type of forced expression of certain master regulators, making the mammalian retina more inclined toward retinal regeneration, suggests epigenetic factors’ possible involvement. Further, the overexpression of Ascl1 along with repression of HDAC [90] or inhibitor of STAT-signaling [184] made the mammalian retina more congenial for regeneration. Apart from the epigenome-influencing factors, various micro RNAs such as miR-25, let-7 and miR-124 also play important roles in regenerative potential in the mammalian retina [203]. Unlike the fishes and amphibians, most other vertebrate models of retina regeneration have a terrestrial life where wound healing is often seen after an injury compared to functional restoration through regeneration. The prevalence of wound healing in land animals compared to their aquatic counterparts is not fully understood. However, it is believed that land animals perform a faster wound healing instead of slow regeneration, probably to avoid the risk of infection or the ability to lead a normal life despite having a compromised organ structure. It is also interesting to note that all vertebrates possess excellent regenerative capability during their embryonic stages where a complete or semi-aquatic environment is present. This ability to regenerate early during development could also be attributed to the less repressive epigenetic environment facilitating gene expression. Even in axolotls, after a forced metamorphosis, which now leads to a terrestrial habitat lacking the robust regenerative capability it had during its aquatic life [204]. The reduced regenerative potential in salamander after the metamorphosis could also be implied to the immunological barrier to regeneration [205]. It would be exciting to explore the regenerative potential of several obligate aquatic mammals. Taken together, it is tempting to speculate that mammals do possess the ability to regenerate as found in non-mammalian vertebrates provided ample opportunity in their genetic and epigenetic landscape.

FUNDING

P.S. acknowledges her support from the SERB NPDF (PDF/2019/001148) for postdoctoral fellowship. R.R. also acknowledges research funding from Science Education and Research Board SERB, DST, India (EMR/2017/001816), DBT India (BT/PR17912/MED/31/336/2016), STAR grant from DoE (STARS1/180) and support from IISER Mohali.

CONFLICT OF INTEREST

None declared.