Müller Glia maintain their regenerative potential despite degeneration in the aged zebrafish retina

Abstract Ageing is a significant risk factor for degeneration of the retina. Müller glia cells (MG) are key for neuronal regeneration, so harnessing the regenerative capacity of MG in the retina offers great promise for the treatment of age‐associated blinding conditions. Yet, the impact of ageing on MG regenerative capacity is unclear. Here, we show that the zebrafish retina undergoes telomerase‐independent, age‐related neurodegeneration but that this is insufficient to stimulate MG proliferation and regeneration. Instead, age‐related neurodegeneration is accompanied by MG morphological aberrations and loss of vision. Mechanistically, yes‐associated protein (Yap), part of the Hippo signalling, has been shown to be critical for the regenerative response in the damaged retina, and we show that Yap expression levels decline with ageing. Despite this, morphologically and molecularly altered aged MG retain the capacity to regenerate neurons after acute light damage, therefore, highlighting key differences in the MG response to high‐intensity acute damage versus chronic neuronal loss in the zebrafish retina.


| INTRODUC TI ON
The physiology and structure of the healthy human eye is known to degrade with age. Hallmarks of the ageing human retina include tissue thinning, neuronal loss, especially in the macula, and reduced visual function (Arden & Jacobson, 1978;Marshall, 1987;Salvi et al., 2006;Skalka, 1980;Slataper, 1950;Weale, 1975). Accordingly, ageing is a significant risk factor for degeneration and disease in the retina, such as age-related macular degeneration or primary open angle glaucoma (Al-Ubaidi et al., 2013;Marshall, 1987). Retinal degeneration is often a process that plays out over months or years, whereby neurons gradually die leading to dysfunction (Marshall, 1987;Salvi et al., 2006). Regenerative-based therapies, such as stimulating endogenous Müller Glia (MG) to regenerate dying neurons, offer a great promise for the treatment of various types of degenerative blinding conditions, including those associated with ageing.
Thus, determining the effects of ageing on the regenerative potential of the retina is an important consideration for the efficacy of such potential treatments going forward.
Müller Glia are the principal cells tasked with supporting and maintaining the retina throughout life. They provide retinal neurons with a myriad of supportive functions, including trophic support, neurotransmitter recycling and energy metabolism (Reichenbach & Bringmann, 2013). While these functions are critical for healthy retinal function, MG also have a prominent role during disease and after neuronal insult. In the mammalian retina, neuronal damage results in a MG gliotic response, involving the up-regulation of stress proteins, proliferation and morphology changes (e.g. hypertrophy) (Bringmann & Wiedemann, 2012). This response is thought to be neuroprotective initially, but may ultimately culminate in dysfunction and death via loss of metabolic support or tissue integrity (Bringmann et al., 2009). While the mammalian retina lacks significant regenerative capacity (Jadhav et al., 2009;Ooto et al., 2004), it has been shown that re-introduction of key molecular signals into MG can stimulate the genesis of new neurons in mice after acute damage (Jorstad et al., 2017(Jorstad et al., , 2020Karl et al., 2008), raising hopes that identifying the molecular mechanisms driving regeneration may provide therapeutic targets that we can exploit in the future. One such example is Hippo/Yap signalling, which has been identified as a signalling pathway that is altered after neuron degeneration and injury within MG cells (Hamon et al., 2017(Hamon et al., , 2019Rueda et al., 2019).
Accordingly, intensive efforts have been made to uncover the molecular mechanisms regulating these endogenous regenerative responses in the retinas of "lower vertebrates", such as fish and amphibians, where MG cells have the capacity to regenerate the retina after neuronal damage (Goldman, 2014). However, even though retinal regeneration is largely studied in these models, such as in the zebrafish (Fausett & Goldman, 2006;Fischer et al., 2002;Lenkowski & Raymond, 2014;Raymond et al., 2006;Thummel et al., 2008;Wan & Goldman, 2016), mechanisms are generally investigated in young adults, and regeneration is stimulated by inducing a high-intensity acute damage injury, which is likely to differ from the slowly accumulating chronic damage associated with natural ageing. In the zebrafish retina, after high-intensity acute damage, such as intense light or toxins (Lenkowski & Raymond, 2014;Thomas et al., 2012), MG undergo a brief reactive gliosis-like phase that transitions into a regenerative response: a de-differentiation and proliferation of the MG progenitor cell to specifically generate new neurons and restore vision (Eastlake et al., 2017;Thomas et al., 2016;Thummel et al., 2008). In contrast to high-intensity acute injury, damage to the retina caused by degenerative disease in ageing will manifest itself over much longer timescales (wildtype zebrafish can live up to c.43 months in the lab (Carneiro et al., 2016)) and results in the gradual accumulation of DNA damage and cell death (Carneiro et al., 2016;Henriques et al., 2013;Salvi et al., 2006). As such, chronic degeneration in ageing may significantly differ from injury models, especially in its capacity to induce a regenerative response from glia.
While multiple acute injuries do not appear to reduce the overall regenerative response of MG (Thomas et al., 2012), the effects of natural chronic age-related damage on the MG regenerative potential has not been determined.
The zebrafish offers a suitable model for studying regeneration in the context of ageing, as it displays key human-like hallmarks of ageing such as shorter telomeres and associated DNA damage (Carneiro et al., 2016). Accordingly, the well-established tert −/− mutant zebrafish have impaired regeneration, both in steady-state in tissues such as the gut (Anchelin et al., 2013;Henriques et al., 2013) and in response to injury, in the heart (Bednarek et al., 2015).
The tert −/− model therefore offers the possibility of identifying telomerase-dependent and telomerase-independent molecular drivers of regeneration in a context of accelerated ageing. Moreover, the use of premature ageing models have provided key molecular insights into the mechanisms driving the ageing process and can contribute to the identification of therapeutic targets for age-associated dysfunction and disease (Kubben & Misteli, 2017).
In this study, we used the naturally aged wildtype (WT) zebrafish and the prematurely aged tert −/− as models to study the impact of ageing and telomerase function on the maintenance of neuronal structure and regenerative capacity of MG cells. We started by confirming that naturally aged retinas display known hallmarks of retinal ageing, including tissue thinning, accumulation of DNA damage and neuronal loss (Van Houcke et al., 2019), which we show are accompanied by significant histopathological abnormalities in the retina. However, we now show that most of this damage is not accelerated in the absence of telomerase (tert −/− mutant), suggesting these to be largely telomerase-independent.

K E Y W O R D S
ageing, degeneration, Müller glia, proliferation, regeneration, retina, telomerase, Zebrafish Interestingly, the accumulation of chronic damage with ageing is not sufficient to stimulate MG to proliferate or replace lost neurons. Instead, ageing leads to morphological and molecular aberrations in MG and loss of vision, recapitulating hallmarks of human retinal degeneration with advancing age. Moreover, we show that MG cells express Yap throughout lifespan, a key molecule involved in the regenerative response (Hamon et al., 2019;Rueda et al., 2019), but that this expression decreases with advancing age. Despite reduced Yap levels, aged MG cells are still capable of regenerating neurons in old zebrafish, after acute light injury.
We, therefore, identify differences in the MG response to highintensity acute damage versus chronic damage, an important consideration for potential therapies aiming to stimulate endogenous regenerative mechanisms to treat human retinal disease.

| The aged zebrafish retina displays neurodegeneration independently of telomerase
Similarly to humans, the zebrafish retina consists of three nuclear layers separated by two synaptic plexiform layers (Easter & Nicola, 1996). The nuclear layers consist in the outer nuclear layer (ONL), containing photoreceptors; the inner nuclear layer (INL) containing bipolar cells (BCs), amacrine cells (ACs), horizontal cells (HCs) and MG; and the ganglion cell layer (GCL) mainly containing retinal ganglion cells (RGCs) (see diagram in Figure 1a). Further, RGCs, ACs and BCs are neurons that come together to make connections in the major synaptic neuropil, called the inner plexiform layer (IPL; Figure 1a). The zebrafish central retina is considered fully specified by 73 h post-fertilisation (hpf) (Easter & Nicola, 1996) and continually grows at the periphery throughout life due to proliferation of the ciliary marginal zone (CMZ) (Easter & Nicola, 1996). Thus, the central region of the adult retina can be considered the "oldest", as it is largely generated during early developmental stages, and where there is very little proliferation to generate new neurons (Fu et al., 2018;Julian et al., 1998), with the exception of rods (Johns & Fernald, 1981;Julian et al., 1998;Nelson et al., 2008;Otteson et al., 2001). In contrast, the peripheral region of the retina is continually expanding at the margins due with the newest born cells being found adjacent to the CMZ It has recently been shown that the ageing WT zebrafish retina shows degeneration phenotypes with advancing age (Fu et al., 2018;Van Houcke et al., 2019). Telomerase is known to be important for proliferation and regeneration in the zebrafish (Anchelin et al., 2013;Bednarek et al., 2015;Carneiro et al., 2016;Henriques et al., 2013); however, apart from its described role in the human retinal pigment epithelium (RPE) (Drigeard Desgarnier et al., 2016), its specific role in the ageing retina remains unclear. Here, we characterised key known hallmarks of human retina ageing (Eriksson & Alm, 2009), including retina thinning, morphology and histopathological alterations. We show that the WT and tert −/− zebrafish central retina progressively thins with ageing ( Figure 1b, b',c,c',c''), consistent with previous studies (Eastlake et al., 2017;Thomas et al., 2016). This retinal thinning is specific to the central retina, as the peripheral retina maintains its thickness over-time (Figure 1c',c'' and Figure S1). Thinning of the central retina is accompanied by several morphological and histopathological alterations, including cellular spatial rearrangement, cellular atrophy and the development of vacuoles in the RPE (Figure 1c-c'' and Figure S2). However, apart from specific alterations in the RPE, we show that this occurs largely independently of telomerase, as the telomerase-deficient model tert −/− displays retinal thinning and most morphological and histopathological alterations at the same rate as the WT (Figure 1b',c').
To determine which type of neuronal loss was underpinning retinal thinning, we used immunohistochemistry with several molecular markers specific for different neuronal populations from young to old ages, in the presence and absence of telomerase (tert −/− ). As  (Lo, 2019) and we show that long double-cones in the zebrafish retina undergo structural changes (short and/or disorganised) with ageing, shown by ID4 staining (Figure 2h,h'). To confirm the photoreceptor outer segment disorganisation, we stained rod and cone outer segments with the zpr3 antibody ( Figure S3a) and red/green double cones with the zpr1 antibody ( Figure S3b).
Zpr3 staining clearly shows the shortened and misaligned outer segments in the WT aged retina, but is not accelerated in tert −/− ( Figure   S3a (Fu et al., 2018). These defects are also not accelerated in the tert −/− at the ages tested, suggesting these are likely telomerase-independent phenomena. | 5 of 20 MARTINS eT Al.

| MG do not proliferate in response to chronic retinal neurodegeneration with ageing
It is well-established that MG in the zebrafish retina proliferate and regenerate lost neurons after high-intensity acute damage (Jadhav et al., 2009;Jorstad et al., 2017;Ooto et al., 2004). Although the progressive age-related retinal neurodegeneration would suggest a lack of regenerative response by MG, it remained unclear whether this degeneration was being counteracted, at least to some degree, by proliferation in the aged retina. To test this, we used a 5-ethynyl-2′-deoxyuridine (EdU) pulse-chase strategy to identify any cell divisions in the retina, from young to old ages, which would be indicative of potential regeneration. To characterise the steadystate regenerative capacity of the central and peripheral CMZ until old age, we carried out a 3-day pulse of EdU followed by 0-and 30- ing that there is no compensatory proliferation from the CMZ in response to neuronal loss with ageing. Finally, the fact that there are very low and decreasing levels of proliferation in wildtype ageing could explain why removing telomerase (tert −/− ) has no further significant effect. As proliferation of MG is the primary source of neurons after acute injury in fish (Bernardos et al., 2007;Fausett & Goldman, 2006), we wanted to determine whether there could be any small compensatory proliferation over-time, within MG cells specifically.
To do so, we co-labelled cells with the MG specific marker glutamine synthetase (GS) and counted the GS-positive; EdU-positive cells. what has been reported to occur in response to high-intensity acute damage (Jadhav et al., 2009;Jorstad et al., 2017;Ooto et al., 2004), our data show that age-related chronic cell death does not trigger proliferation of MG cells. Finally, rods can originate from rod-specific progenitors into adulthood (Johns & Fernald, 1981;Julian et al., 1998), which are found in the ONL and are derived from MG that slowly divide in the WT retina (Bernardos et al., 2007;Nelson et al., 2008;Otteson et al., 2001;Raymond et al., 2006). To test whether there is any compensatory proliferation with ageing occurring specifically from rod-specific progenitors found scattered throughout the ONL of the retina (Johns & Fernald, 1981;Stenkamp, 2015), we

| Zebrafish vision declines with ageing, independently of telomerase
Loss of visual acuity and contrast sensitivity with advancing age in humans has been well documented (Marshall, 1987;Salvi et al., 2006). Here, we report that the presence of retinal neuron loss and tissue thinning in the aged retina is largely independently of telomerase (tert). Therefore, to determine whether the molecular and structural changes we observe have a pathological consequence on the retinal function, and whether this is accelerated in the absence of telomerase, we tested visual acuity in WT and tert −/− zebrafish, from young to old ages. Visual testing in the zebrafish has been used to screen for mutants with defects in retinal development and function (Brockerhoff et al., 1995;Gross et al., 2005;Lessieur et al., 2019;Neuhauss et al., 1999), but were yet to be tested in the context of ageing. The optokinetic response (OKR) is used to measure innate visual responses and provides readout of visual acuity (Brockerhoff et al., 1995;Tappeiner et al., 2012)  unlike what is described in other tissues (Carneiro et al., 2016). We, therefore, focused on characterising the morphological and molecular changes underpinning the lack of MG regenerative responses in the ageing WT retina from here onwards.

| MG become morphologically altered with ageing
MG respond to damage or injury in most, if not all, retinal degenerative diseases (Bringmann et al., 2009). In the mammalian retina alterations in MG morphology is a hallmark of MG response after retinal damage (Bringmann & Wiedemann, 2012), but also commonly observed in the aged retina (Telegina et al., 2018). The characteristic mammalian response is gliosis, whereby MG morphology becomes aberrant, accompanied by changes in expression and localisation of structural proteins like glial fibrillary acidic protein (Gfap) (Bringmann et al., 2009;Reichenbach & Bringmann, 2013). As we do not observe a MG regenerative response to neurodegeneration, but we do observe vision loss in naturally aged zebrafish, we wondered whether zebrafish MG were undergoing an ageing-associated damage response similar to mammals. Here, we labelled MG using the well-described GS antibody and characterised MG morphology and organisation using the markers Gfap, Vimentin and cellular retinaldehyde-binding protein (Cralbp) (Garcia & Vecino, 2003;Verardo et al., 2008). Although the number of MG cells is maintained in the central retina throughout the zebrafish lifespan ( Figure S5a

| Müller glia express reduced levels of Yap in old age
As we observe morphological alterations in MG cells and a lack of MG proliferation in response to age-associated chronic damage, we next asked whether aged MG have altered molecular expression of Yap, a key molecule in the retinal regenerative process (Hamon et al., 2019). Specifically, when Hippo signalling is activated, LATS1/2 kinases phosphorylate the transcriptional co-factor Yap, leading to its degradation. Importantly, Yap expression is required for MG regeneration (Hamon et al., 2019) and its expression goes up in reactive MG cells after acute damage (Rueda et al., 2019). We therefore tested whether Yap expression was affected in ageing and in MG specifically. Here we show, by immunohistochemistry, that Yap is expressed in MG until old age (Figure 5a-c), where it is thought to be required for the maintenance of MG proliferative capacity (Rueda et al., 2019). Nevertheless, overall expression levels in the eye are reduced, as assessed by western blot (Figure 5d,d', and Figure S6).
As such, MG retain the expression of a critical molecular player required for the regenerative process after damage, but its levels are significantly reduced. Given this, it was still unclear whether the F I G U R E 2 Zebrafish retina undergoes neurodegeneration with ageing, independently of telomerase. The central retina immunolabelled with HuC/D and PKC (amacrine in magenta and bipolar cells in green, respectively), in both WT and tert −/− , in (a) young (5 months)  naturally aged fish could regenerate their retina in response to a high-intensity acute injury.

| Müller glia retain the ability to regenerate after acute damage until old age
Studies on retinal regeneration typically focus on the young adult retina and have not taken old age into account for the regenerative response. There is evidence from ex vivo mouse explants that MG neurogenic competence in response to neuronal death is limited with advancing age (Loffler et al., 2015); however, this study was restricted to early post-embryonic stages and not more mature adult stages. Several high-intensity acute damage paradigms induce retinal neuron death and elicit a regenerative response in zebrafish (Bernardos et al., 2007;Cameron, 2000;Hanovice et al., 2019;Ranski et al., 2018). Repeated damage to the adult retina, specifically between 9-18 months of age, which is still considered young for a WT fish, results in MG continuously re-entering the cell cycle to proliferate and regenerate lost neurons, although they do show signs of chronic activation after repeated insults (Ranski et al., 2018). However, it remains unclear whether MG retain their ability to proliferate and regenerate the retina throughout life, and whether altered MG, such as those often occurring in retinal disease or old age retain proliferative capacity to replace damaged neurons. Since we show that MG retain Yap expression into old ages, albeit at reduced levels, it remained possible that aged zebrafish retinas could still regenerate in response to high-intensity acute damage. To test this, we used the light-damage model where aged albino zebrafish, at three stages of their lifespan, were treated with light to elicit photoreceptor damage. To detect increased proliferation of MG in response to damage, a 3-day pulse of bromodeoxyuridine (BrdU) was performed, followed by a 28-day chase period (Figure 6a Figure S7). Together, these results suggest that the regenerative response remains intact throughout zebrafish lifespan and the molecular and morphological alterations in aged MG does not impact their regenerative capacity in old age.

| DISCUSS ION
Using a combination of cell labelling strategies throughout adulthood into old age, we show that the zebrafish retina retains its potential to regenerate in response to acute damage into old age. Therefore, the lack of compensatory proliferation in response to chronic, ageassociated cell death in the ageing zebrafish retina is not due to a loss of capacity of MG to proliferate per se, but likely due to the absence or insufficient levels of the required stimuli for MG proliferation and regeneration.

| Telomerase-dependent and telomeraseindependent hallmarks of zebrafish retinal ageing
Previous work has suggested that telomerase and telomere length are important for human retinal health. In particular, the RPE has been reported to have shorter telomeres than the neural retina, and it accumulates senescence over time (Drigeard Desgarnier et al., 2016). Accordingly, reactivation of telomerase has been described to ameliorate symptoms of age-associated macular disease (Dow & Harley, 2016;Rowe-Rendleman & Glickman, 2004) and telomerase activators are currently in clinical trials (e.g. NCT02530255).
However, retinal homeostasis requires more than RPE maintenance.
It requires a steady-state level of proliferation of different cell types F I G U R E 3 Aged zebrafish retina does not show signs of regeneration in response to spontaneous cell death and neuronal loss. (a) Schematic figure of the experimental design: 3-day pulse of EdU, by IP injection, followed by 0-or 30-day chase. We anticipate that in a healthy young fish the retina has some cells proliferating in the CMZ, which over-time will replace older cells in the central retina. In the case of injury, we anticipate that there will be elevated levels of proliferation in peripheral and central retina to replace the dead cells. involved in multiple aspects of retina function. As in other proliferative tissues (Anchelin et al., 2013;Carneiro et al., 2016;Henriques et al., 2013), it would be expected that telomerase levels would influence zebrafish retina homeostasis. We, therefore, hypothesised that the zebrafish retina would degenerate in a telomerase-dependent manner, leading to vision loss. However, our data suggest that, apart from aberrations in the RPE, most age-related changes described here in the naturally aged zebrafish retina, including vision loss, are largely telomerase-independent, since depletion of telomerase (tert −/− ) does not accelerate or exacerbate any of these phenotypes.
Telomere dysfunction is known to affect mostly highly proliferative tissues (reviewed in Henriques and Ferreira, (2012)) and what our data show is that, in the region most affected by ageing phenotypes, the central retina, there is very little proliferation to start with, potentially explaining why telomerase depletion has no further significant effect.

| Chronic vs. high-intensity acute damage in MG responses: a "tipping point" required to stimulate regeneration in ageing?
Regeneration studies so far have relied on several damage paradigms, including phototoxic (Ranski et al., 2018) and ouabaininduced lesions (Mitchell et al., 2019), which induce rapid cell death post-insult. While high-intensity acute damage models are suitable to explore the cellular and molecular mechanisms underpinning the regenerative potential of the retina, they do not test whether this regenerative response is also occurring with natural ageing in the retina. Our results show that zebrafish develop vision loss with ageing and that this is accompanied by retinal neurodegeneration and MG morphological and molecular aberrations, rather than proliferation and replacement of dying neurons. This is counterintuitive to what would be expected based on the well-known regenerative response to neuronal death by MG. Nevertheless, we show that aged zebrafish retinas maintain their regenerative response and capacity after acute injury. This study, therefore, shows that there are critical differences between the MG regenerative response to the neuronal loss induced by chronic damage in ageing and after high-intensity acute injury.
"Natural ageing" and associated neuronal loss play out over months or even years and we now show that this does not elicit the same regenerative response as high-intensity acute damage. This may be because the slow degeneration is not producing a strong enough signal to induce MG to undergo the cellular and molecular processes of regeneration. It has been shown that the level of cell death can induce a differential response of MG cells. Whilst a large amount of rod death causes a regenerative response, small amounts do not (Lessieur et al., 2019;Montgomery et al., 2010;Thomas et al., 2012). Furthermore, there may be distinct differences in signals released after apoptosis or necrosis in ageing vs damage models (Bergmann & Steller, 2010;Nelson et al., 2013). In support of this concept, recent work suggests that there may be key signalling differences underpinning the difference between a "regenerative" or a "reparative" response to injury (Conedera et al., 2020). Thus, we propose that a molecular signal, or expression changes of such signal, will regulate the "tipping point" required to elicit a MG regenerative response in ageing.
Research in the context of acute damage paradigms in the zebrafish retina has uncovered many of the molecular mechanisms regulating this regenerative response (reviewed in Goldman, 2014;Julian et al., 1998). For instance, proliferation appears to be a key mechanism for the initiation of the regenerative response as blocking proliferation after damage in the zebrafish retina results in MG alterations, and not regeneration (Thummel et al., 2008), similarly to humans.
Moreover, age-associated degeneration of retinal neurons and their synapses, as we observed to occur in this study, may result in a loss of neurotransmitter release, such as gamma-aminobutyric acid (GABA), which has been shown to facilitate the initiation of MG proliferation (Rao et al., 2017). Thus, dysregulation of neurotransmitters upon neurodegeneration could inhibit the key molecular pathways regulating regeneration. Alternatively, the initial inflammatory response has also been shown to be determinant for the repair process. Upon light-induced retinal damage, overexpression and subsequent release of tumour necrosis factor (TNF) by apoptotic photoreceptors seems to induce MG proliferation (Iribarne et al., 2019;Thomas et al., 2016).
After high-intensity acute damage, there is also an increased number of microglia in the retina (Mitchell et al., 2018(Mitchell et al., , 2019 and they have been shown to be essential for retinal regeneration (Conedera et al., 2019;White et al., 2017). In contrast, we show that microglia numbers do not increase with ageing in the zebrafish retina, and therefore, may compromise the regenerative response. Nonetheless, it is important to consider that the available immunohistochemistry techniques to identify microglia numbers have a few limitations. Namely, there may be a spike in microglia number in the aged retina that is rapidly resolved and missed at our time points or microglia may undergo F I G U R E 4 Zebrafish vision declines with ageing, which is accompanied by signs of glial morphological alterations. (a) OKR assay was performed by immobilising the fish in between soft sponges, inside a petri dish containing water, placed in the centre of a rotation chamber. The walls of the rotation chamber had 0.8 mm-thick black and white stripes and the chamber was maintained at a constant velocity of 12 rpm throughout the experiment. signal, and the level required for inducing the response, will be critical to stimulate regeneration in the human retina. It is very promising that mammalian MG can be molecularly manipulated to increase their regenerative potential after damage (Jorstad et al., 2017). However, for these therapies to be truly transformative for human blinding conditions this regenerative capacity must be paired with the appropriate "damage" stimuli since using acute light damage or toxins in humans is unlikely to be therapeutically feasible.

| MG response to chronic damage in ageing
The regenerative capacity of vertebrate tissues tends to decrease after repeated injury and as animals advance in age (Ranski et al., 2018). In the context of the central nervous system, this is aggravated by gliosis, a reactive change in glial cells in response to damage. In the mammalian retina, MG cells undergo gliosis in response to damage and in many retinal degenerative diseases (Bringmann et al., 2009). However, in the zebrafish retina MG cells undergo the initial reactive gliotic response after acute damage but quickly shift to the regenerative pathway. Here, we show that while we observe morphological and molecular alterations in MG cells in response to agerelated neurodegeneration, MG in the aged retina retain their ability to regenerate in response to high-intensity acute damage of photoreceptors. This provides evidence that regenerative mechanisms in tissues that have been damaged or are degenerating remains possible. Importantly, the fact that MG are still capable of proliferating in response to a high-intensity acute injury in old animals, shows that the lack of regeneration in the context of ageing is not due to an intrinsic inability of MG to proliferate per se, suggesting that it is the stimuli itself that is different/insufficient in the context of ageing. Mechanistically, in contrast to what has been described in high-intensity acute damage models, where Yap levels go up (Rueda et al., 2019), we show that overall Yap levels decrease in the chronic damage setting of ageing. It is, therefore, a likely possibility that at least one of the key stimuli missing may be one that is responsible for switching off Hippo signalling pathway, preventing Yap degradation.

| CON CLUS IONS
Our work demonstrates that, in the context of age-induced neuronal degeneration, zebrafish MG undergo degeneration, rather than regeneration. This resembles what occurs in the aged human retina as well as in many human retinal degenerative diseases. Importantly,
The telomerase mutant line tert AB/hu3430 was generated by Nethyl-nitrosourea mutagenesis (Utrecht University, Netherlands; Wienholds, 2004). It has a T→A point mutation in the tert gene and is available at the ZFIN repository, ZFIN ID: ZDB-GENO-100412-50, from ZIRC. The fish used in this study are direct descendants of the ones used previously (Carneiro et al., 2016;Henriques et al., 2013), by which point it had been subsequently outcrossed five times with WT AB for clearing of potential background mutations derived from the random ENU mutagenesis from which this line was originated.
The tert hu3430/hu3430 homozygous mutant is referred to in the paper as tert −/− and was obtained by incrossing our tert AB/hu3430 strain.
In order to study age-related phenotypes in the zebrafish retina, in this study, we use an age >30 months old fish for what we consider old in WT (in the last 25-30% of their lifespan), and we consider the tert −/− old fish at the equivalent age (>12 months) that corresponds to the last 25-30% of their lifespan, approximately. In specific, "Old" was defined as the age at which the majority of the fish present age-associated phenotypes, such as cachexia, loss of body mass and curvature of the spine. These phenotypes develop close to the time of death and are observed at >30 months of age in WT and at >12 months in tert −/− (Carneiro et al., 2016;Henriques et al., 2013).
In addition, we used adult albino zebrafish for retinal regeneration studies (described in detail below). Importantly, none of the animals included in this study displayed visible morphological alterations in the eyes (e.g. cataracts).

| Optokinetic response (OKR) assay
Fish were anaesthetised in 4% tricaine methanesulfonate (MS-222; Covetrus, pharmaceutical grade) and placed in a small bed-like structure made of sponge, located inside a small petri dish containing fish water. During the experiment, fish were maintained still by the strategic use of needles that sustained the sponges close to the fish so that the fish could not move. The petri dish was then placed inside a rotation chamber with black and white striped walls (8 mm-thick stripes). After fish recovered from anaesthesia, the trial began, and the walls of the chamber started rotating at 12 rotations per min (rpm; for 1 min to the left side followed by 1 min to the right side).
Eye movements were recorded using a digital camera throughout the experiment. After the experiment, the number of eye rotations per minute was measured by video observation and manually counting. The counting was performed blindly by two independent researchers. In the end, the results were normalised for the WT young from the same day/batch, in order to control for different days of experiments.

| Intense light-damage paradigm with BrdU incorporation
A photolytic damage model in adult albino zebrafish was utilised to destroy rod and cone photoreceptors and elicit a regenerative response (Thomas et al., 2012). Briefly, adult albino zebrafish were dark-adapted for 10 days prior to a 30 min exposure to ~100,000 lux from a broadband light source. Next, fish were exposed to ~10,000 lux of light from four, 250 W halogen lamps for 24 h.
Following 24 h of light treatment, fish were transferred to a 1 L solution containing 0.66 g of NaCl, 0.1 g Neutral Regulator (SeaChem Laboratories, Inc.), and 1.5 g BrdU (5mM; B5002; Sigma-Aldrich) for 48 h. This timeframe for BrdU incubation was based on previous studies in order to label MG cell-cycle re-entry (Garcia & Vecino, 2003). Following a 48-h incubation in BrdU, the fish were split into two groups: one group was euthanised by an overdose
Briefly, fish were injected with 5 μl of 10 mM EdU diluted in dimethyl sulfoxide (DSMO), by intraperitoneal (IP) injection, for 3 consecutive days (3-day pulse). In order to differentiate proliferating cells and low-proliferative EdU-retaining cells, the fish were separated into two groups: 0-day chase and 30-day chase groups. The fish from the first group were culled 2 h 30 min after the last injection of Edu, whereas the fish from the second group were culled 30 days after the last injection of EdU. After culling, whole fish were processed for paraffin-embedded sections as described above. In order to detect EdU labelling in paraffin-embedded sections, the slides were deparaffinised, hydrated, underwent antigen retrieval, were The incubation in the secondary antibody was performed as previously described.

| Imaging and quantifications
Paraffin-embedded sections were imaged by epifluorescence microscopy, using a DeltaVision microscope with a 40× oil objective.
Cryosections were imaged by laser scanning confocal imaging, using a Leica SP5 microscope, Nikon A1 Confocal microscope or a Zeiss 900 Airyscan 2 confocal, with a 40× oil objective. In either paraffin or cryosections, multiple 0.2-0.6 μm thick z-stacks were acquired in order to capture the whole retina region. For each staining, a total of 4 images were taken per retina, 2 from central and 2 from peripheral retinal. The central retina was defined to be the centre point between opposing CMZs (a minimum of ~1000 μm from the periphery).
The peripheral retina was defined as the tissue directly adjacent to the CMZ.
In order to quantify the alteration in the staining patterns, a zprojection was generated using ImageJ (Rasband, W.S., ImageJ, U. S. National Institutes of Health, https://imagej.nih.gov/ij /, 1997-2018.) and three boxes of 100 × 100 μm were drawn in each field of view.
Retinal thickness was measured in DAPI stained paraffin sections, using the software ImageJ to measure the individual layers and overall layers thickness as indicated in the diagram in Figure 1a. For key protein staining, the total number of positive cells was manually counted for each label.
Ribeye A, 1D4, ZO1, ZPR3 and Gfap staining were an exception to this, for which a qualitative assessment was performed instead.
To do so, structural and morphological defects were identified as follows. For the outer segments of long double-cones (1D4) staining, as young WT retinas usually display long and organised outer red green cones, and these were counted manually and their length was measured using image J ( Figure S3).
Finally, raw images were used for quantification purposes. The images were then processed with Adobe Illustrator 21.0.2 for display purposes.

| Protein extraction, quantification and Western blotting
Adult fish were culled by overdose of MS-222, followed by confirmation of death. Dissected eyeballs were snap-frozen in liquid nitrogen and kept at −80°C until protein extraction, as follows.
Frozen tissue was homogenised in 100 μl of radioimmunoprecipi- Immunoblotting was performed as follows. Membranes were blocked using Intercept ® TRIS buffered saline (TBS) Blocking Buffer for 1 h at RT and incubated with target primary antibody (see Table 1 for details) in TBS Blocking Buffer 0.1% Tween-20 (Sigma-Aldrich, Missouri) ON at 4°C, with mild shaking. Secondary antibody detection was performed after washing the primary antibody with 1X TBS (50 mM Tris-Cl, pH 7.5, 150 mM NaCl (Sigma-Aldrich)) with 0.1% Tween-20. Primary antibody detection was performed using

| Statistical analysis
Statistics were performed using the GraphPad Prism v7.00. Normality was assessed by the Shapiro-Wilk test. For normally distributed data where most of the groups had a sample size of ≥5, unpaired t-test was used to compare 2 data points and one-way ANOVA followed by Bonferroni post hoc test was used to compare more than 2 data points. For non-normally distributed data and/or data containing less than 5 animals in most of the groups Mann-Whitney test and Kruskal-Wallis tests were used instead. Two-way ANOVA followed by Tukey's multiple comparisons was used in order to compare more than 2 data points in 2 different groups (genotypes). Chi-square was performed to analyse structure and morphological changes in the retina based on qualitative assessment, having taken into account the number of animals per group displaying defects versus not displaying defects. A critical value for significance of p < 0.05 was used throughout the analysis. There were no repeated measurements performed in this study.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets and source data generated during and/or analysed during the current study are available from the corresponding author upon reasonable request.