1 INTRODUCTION

The fast-aging African turquoise killifish (Nothobranchius furzeri) has arisen as a novel vertebrate biogerontology model as it recapitulates many traits of human aging (Reuter et al., 2018). Despite a broad characterization of aging hallmarks in the killifish, its potential to regrow central nervous system (CNS) axons, and in particular how this capacity alters with age, remains largely unexplored. As a teleost fish, the killifish has an astonishing regenerative ability in its adult CNS, with a high de novo neurogenic ability and capability to regrow axons, rebuild circuits, and restore function after injury (Fleisch et al., 2011; Rasmussen & Sagasti, 2017; Zupanc & Sîrbulescu, 2012). However, aging has been shown to affect nervous system functionality, vulnerability to trauma and disease, as well as its regenerative potential, both in non-vertebrates and vertebrates (Vanhunsel et al., 2020). Identifying the molecules and signaling pathways contributing to this flawed regeneration in an aged setting is therefore of great interest in neuroregenerative research. We propose the killifish CNS, and its age-dependent repair capacity, as a promising model system to study the mechanisms that underlie functional neural circuit restoration, thereby contributing to new insights on how to stimulate regeneration in the (old) mammalian CNS.

The visual projection is a well-studied model system to investigate axonal regeneration in the CNS, in mammals as well as fish. The retinofugal projection, bringing visual information from the retina to the brain, consists of the axons of a single retinal cell type, namely the retinal ganglion cells (RGCs), which bundle together to form the optic nerve. In teleosts, RGC axons completely cross over at the optic chiasm and predominately innervate the contralateral optic tectum of the fish brain. The optic tectum of adult fish is considered homologous to the superior colliculus in mammals and is a multilayered structure that can be divided in three main zones: the superficial and central zones, where tectal afferents terminate, and the periventricular gray zone (PGZ), containing the majority of the tectal neuronal cell bodies (Bally-Cuif & Vernier, 2010; Neuhauss, 2010; Nevin et al., 2010). Restoration of this optic projection after optic nerve injury is an example of the remarkable intrinsic ability of teleost fish to repair axonal damage in the adult CNS. Indeed, upon crush or even transection of the optic nerve, RGCs of adult fish are capable of regrowing their axons, which then navigate to and reinnervate their target cells in the optic tectum to finally restore vision, proving that the visual system of teleosts forms a perfect model system for investigating the different aspects of axonal regrowth (Bollaerts et al., 2018). According to zebrafish and goldfish literature, the repair process of optic nerve regeneration can be subdivided into five phases: (1) the injury response period, occurring at <1 day postinjury (dpi), which is characterized by shrinkage of RGC dendrites; (2) the axonal outgrowth initiation phase, wherein distal axons of the injured RGCs degenerate and new RGC axons prepare for neurite sprouting, marked by the maximal upregulation of growth-associated genes in the RGCs; (3) the axon elongation and target reinnervation period, in which RGC axons extend toward their target cells in the brain; (4) the synaptogenesis phase, wherein synapses are repaired and retinotopy is restored, and in which several vision-dependent reflexes (optokinetic response, dorsal light reflex, and optomotor response) are restored; and (5) a final period involving long-term synaptic rearrangements and recovery of complex behaviors (Becker & Becker, 2014; Beckers et al., 2019; Bhumika et al., 2015; Diekmann et al., 2015; Kato et al., 1999, 2013; Lemmens et al., 2016; McCurley & Callard, 2010; Van houcke et al., 2017; Zou et al., 2013).

Adopting the optic nerve crush (ONC) model in female killifish, we aim to further unravel the impact of aging on axonal regeneration in the visual system. In this study, after a detailed characterization of the model in young adult fish, we mapped the different phases of the optic nerve repair process in three additional age groups: middle-aged, old, and very old fish. Depending on the age, repair of axonal damage seems to be affected in different phases of the regenerative process. While middle-aged fish show a delay in optic nerve regeneration and only partially recover their vision following ONC, old and very old fish do not restore vision at all. Indeed, our data reveal a striking impairment in RGC axonal regrowth in older fish, starting with a delay in axon outgrowth due to a low intrinsic ability to grow axons, and next failing in the subsequent phases of target neuron reinnervation and synaptogenesis. Besides the poor intrinsic growth capacity of older RGCs, we suggest that also a non-supportive extrinsic environment, actuated by abnormally reactivated micro- and macroglial cells, contributes to the observed age-related loss of functional repair in female killifish. Our results shed light on the underlying aging processes affecting the regenerative potential and thereby contribute to the search for effective neuroregenerative therapies in the (aged) mammalian CNS.

2 RESULTS 2.1 Detailed characterization of the axonal regeneration process in young adult killifish subjected to ONC

To characterize optic nerve regrowth in the young adult female killifish, we evaluated the different regenerative phases following ONC in 6-weeks-old young adult fish. We first assessed the initiation of axonal regrowth by determining retinal mRNA levels of growth-associated protein 43 (gap43) and tubulin alpha 1a chain (tuba1a), both validated markers for the intrinsic axonal outgrowth potential of RGCs (Beckers et al., 2019; Kato et al., 2013; McCurley & Callard, 2010; Van houcke et al., 2017), at 0, 3, 7, and 14 dpi. Gap43 (Figure 1a) and tuba1a (Figure 1b) mRNA levels were significantly upregulated at 3 dpi, followed by a decline in expression to baseline values by 7 dpi. Killifish RGCs thus transiently express these growth-associated markers to initiate axonal regrowth (Table 1).

Time course of optic nerve regeneration following optic nerve crush injury in young adult killifish. (a, b) RT-qPCR for gap43 (a) and tuba1a (b) demonstrates a significant increase in expression levels in the retina at 3 dpi. n = 3. (c) Illustration of the retrograde biocytin tracing method, explained in detail in the Experimental Procedures section. (d) Counting the number of DAPI+ cells in the GCL upon ONC reveals a pronounced cellular loss between 2 and 7 dpi (black line). The number of biocytin−/DAPI+ cells is comparable in injured and control retinas, except at 2 dpi when this cell population not only contains displaced amacrine cells but also RGCs of which the axons have not reached the site of tracer placement yet (blue line). The number of biocytin+/DAPI+ cells rises during the course of regeneration, to reach a plateau at 7 dpi (green line). n = 7–10. (e) Illustration of the anterograde biocytin tracing method, explained in detail in the Experimental Procedures section. (f) Microscopic image of a biocytin-labeled coronal brain section of an uncrushed control fish, revealing black RGC axons innervating the tectum. Scale bar = 200 µm. (g) Representative images depicting tectal reinnervation at several time points following ONC. While no biocytin-labeled axons are detected in the tectum at 2 dpi, regenerating RGC axons start entering the ventral and dorsal parts of the optic tectum around 4 dpi (arrowheads). From 14 dpi onwards, the biocytin labeling matches that of uninjured fish. Scale bar = 200 µm. (h) Schematic illustration of the synaptic layers in the optic tectum of uncrushed control fish after immunolabeling for synaptotagmin reveals the presence of pre-synaptic vesicles (red dots) in the SO, SFGS, SGC, and S/S layers of the killifish optic tectum. (i) Following ONC, labeling of pre-synaptic vesicles starts to decrease in the SFGS layer and is almost completely absent at 7 dpi. At 14 dpi, synaptic labeling increases again, and by 21 dpi, synaptotagmin signals in the SFGS are comparable to that of uninjured fish. All data show mean ± SEM, means with a different letter are significantly different (One-way ANOVA for panels a and b, Two-Way ANOVA for panel d), see Table S1 for exact p-values. DAPI, 4′,6-diamidino-2-phenylindole; dpi, days postinjury; gap43, growth-associated protein 43; GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; NFL, nerve fiber layer; ONC, optic nerve crush; ONL, outer nuclear layer; OPL, outer plexiform layer; PGZ, periventricular gray zone; PRL, photoreceptor segment layer; RGC, retinal ganglion cell; RPE, retinal pigment epithelium; RT-qPCR, real-time quantitative polymerase chain reaction; S/S, projection zone between SAC and PGZ; SAC, stratum album centrale; SFGS, stratum fibrosum et griseum superficiale; SGC, stratum griseum centrale; SM, stratum marginale; SO, stratum opticum; tuba1a, tubulin alpha 1a chain TABLE 1. Overview of the phases of the regenerative process following optic nerve injury in three different fish models of young adult age, that is, Danio rerio, Carassius auratus, and Nothobranchius furzeri Phases of the regenerative process following optic nerve injury D. rerio C. auratus N. furzeri Injury response <1 dpi Outgrowth initiation 1–7 dpi 1–5 dpi <7 dpi Synthesis RNA/proteins for neurite sprouting 1–7 dpi <7 dpi Number of regenerating RGCs in retina starts increasing 2 dpi 2 dpi Axon elongation and target reinnervation 5–30 dpi 14–40 dpi 4–14 dpi Axons appear in the optic tectum 5–7 dpi 14–20 dpi 4 dpi Complete reinnervation of the optic tectum 10–30 dpi 30–40 dpi 14 dpi Synaptic repair and restoration retinotopy 14–80 dpi Complete reformation of synapses / OKR 14 dpi 14–21 dpi DLR 18 dpi 21 dpi Re-establishment retinotopy / OMR 25–63 dpi Long-term synaptic rearrangements and recovery complex behaviors 60–120 dpi 150–180 dpi Chasing and/or shoaling behavior 60–100 dpi 150–180 dpi References

McCurley and Callard (2010)

Kato et al. (2013)

Zou et al. (2013)

Becker and Becker (2014)

Bhumika et al. (2015)

Diekmann et al. (2015)

Lemmens et al. (2016)

Van houcke et al. (2017)

Beckers et al. (2019)

Kato et al. (1999)

Kato et al. (2013)

Current study Note Given values are approximate. Abbreviations: DLR, dorsal light reflex; dpi, days postinjury; OKR, optokinetic response; OMR, optomotor response; ONI, optic nerve injury; RGC, retinal ganglion cell

Outgrowth initiation was further investigated using retrograde biocytin tracing in 6-weeks-old fish, followed by analysis on midsagittal retinal sections (Figure 1c). In baseline conditions, this retrograde tracing results in labeling of all RGCs in the retinal ganglion cell layer (GCL). Analyzing the proportion of biocytin+/4′,6-diamidino-2-phenylindole (DAPI)+ and biocytin−/DAPI+ cells revealed that, similar as reported in other fish species (Mack et al., 2004), 51.2 ± 3.7% of the cells in the GCL are RGCs that take up the dye (biocytin+/DAPI+) and 48.8 ± 3.9% of the cells are displaced amacrine cells (biocytin−/DAPI+) (Figure 1d). Upon ONC, however, cell numbers and cell type proportions changed. Counting the total number of DAPI+ cells in the GCL of retinal midsagittal sections demonstrated a pronounced cell loss between 2 and 7 dpi that resulted in a 39.3 ± 3.4% reduction in total cell number by 21 dpi (Figure 1d). To confirm the occurrence of cell death following ONC, the number of activated-caspase-3+ cells in the GCL was quantified. The number of apoptotic cells tended to increase in the injured retinas when compared to uninjured ones, with a maximum number at 7 dpi that decreased again by 14 dpi (0.3 ± 0.2 apoptotic cells in the GCL per midsagittal retinal section in uninjured fish, 3.1 ± 1.0 cells at 7 dpi, 0.9 ± 0.1 cells at 14 dpi), indicating that apoptosis indeed occurs in young adult killifish subjected to ONC (see also Figure 3c). Next, to disentangle whether the observed cell loss in the GCL resulted from RGC loss only, or from loss of both RGCs and displaced amacrine cells, the proportion of biocytin−/DAPI+ and biocytin+/DAPI+ cells was evaluated in retinas of nerve crushed fish. At 2 dpi, the number of biocytin−/DAPI+ cells exceeded that of uninjured fish as this cell population not only contained displaced amacrine cells, but also RGCs of which the axons did not reach the site of tracer placement yet. At all later time points post-ONC, the number of biocytin−/DAPI+ cells remained constant and comparable to baseline values, indicating that the ~40% loss of cells in the GCL is not attributable to loss of displaced amacrine cells (Figure 1d). The number of biocytin+/DAPI+ RGCs, however, showed a permanent decrease as a result of ONC. First, at 2 dpi, their number was significantly reduced compared to uninjured numbers as not all RGC axons reached the site of tracer placement yet (with 11.8 ± 1.2% of the RGCs, relative to those traced in the uninjured retina, arriving at the site of biocytin tracer placement). The number of biocytin+/DAPI+ RGCs then increased and reached a plateau at 7 dpi. However, with 51.4 ± 5.6% of the total number of RGCs labeled at 7 dpi, it never reapproached uninjured levels over the course of regeneration (Figure 1d). Strikingly, these data indicate that, of the total RGC population, ~40% is lost following ONC, ~50% is able to regenerate, and the remaining ~10% seems to stay alive in the retina without regrowing an axon.

We next evaluated the timelines of the axon elongation and tectal reinnervation phase. Target reinnervation following ONC was studied via anterograde biocytin tracing, followed by analysis of axonal density levels within the RGC axon innervation area on midcoronal optic tectum sections (Figure 1e). In uninjured control fish, RGC axons could be visualized in the stratum fibrosum et griseum superficiale (SFGS) and stratum opticum (SO) of the tectum (Figure 1f). No biocytin-labeled RGC axons were detected at 2 dpi, showing that they did not reach the tectum yet. Axons started re-entering the dorsal and ventral parts of the optic tectum by 4 dpi. At 7 dpi, reinnervation of the tectal layers was still ongoing and analysis of axonal density levels revealed that 90.6 ± 2.0% of the tectal reinnervation area was reinnervated, as compared to uninjured fish. Reinnervation was completed at 14 dpi, with 98.49 ± 0.9082% of the tectal area reinnervated (Figure 1g). In young adult females subjected to ONC, it thus takes about 14 days for the optic tectum to be completely reinnervated by RGC axons (Table 1).

In a following step, the synaptogenesis phase was studied using immunostainings for synaptotagmin (Znp-1 antibody) (Beckers et al., 2019; Fox & Sanes, 2007; Zappa Villar et al., 2021), a calcium sensor protein present in the membrane of pre-synaptic vesicles, on optic tectum sections at several time points after ONC. Pre-synaptic vesicles were clearly visible in the SO, SFGS, stratum griseum centrale (SGC), and S/S (projection zone between the stratum album centrale (SAC) and PGZ) tectal layers of uninjured fish (Figure 1h), but disappeared following ONC, particularly in the SFGS, which is the tectal layer to which most RGC axons project in teleost fish (Bally-Cuif & Vernier, 2010; Neuhauss, 2010; Nevin et al., 2010). This degradation of synapses started from 2 dpi onwards in the young adult killifish and continued up to 7 dpi, at which almost no synapses could be observed in the SFGS. At 14 dpi, we noted an injury-induced synaptogenesis, and synaptic numbers approached pre-injury levels by 21 dpi (Figure 1i). Thus, injury-induced synaptogenesis in 6-weeks-old killifish occurs in the second and third week post-ONC (Table 1).

To evaluate functional recovery after ONC, two different behavioral tests were used. First, an optokinetic response test was performed to assess visual acuity of fish subjected to a bilateral ONC. This test revealed that immediately after ONC, eyesight was completely lost. However, over time, the young adult fish regained their primary vision. Visual acuity was retrieved from 4 dpi onwards and reached baseline values by 14 dpi (Figure 2a). Next, we evaluated the dorsal light reflex. Upon unilateral nerve damage, fish have been demonstrated to alter their swimming position and swim oblique toward the side of the uninjured eye. During the course of regeneration, this oblique swimming position gradually returns to baseline levels (Diekmann et al., 2015). Similar as for the optokinetic response, young adult killifish lost primary vision in their left eye directly after ONC, as they started to swim oblique already at 1 dpi when compared to uninjured fish. Tilting of their frontal body axis increased, reaching a maximum at 7 dpi, and then normalized with progressing axonal regeneration. Indeed, starting from 21 dpi, body axis tilting was no longer significantly different when compared to baseline levels (Figure 2b). Altogether, primary vision has thus completely recovered by 21 dpi in young adult female killifish (Table 1).

Recovery of primary vision following optic nerve crush injury is impaired in aged killifish. (a) Determining the maximal spatial frequency provoking an optokinetic response at different time points postbilateral ONC discloses that young adult killifish start reshowing this reflex from 4 dpi onwards and full recovery is achieved by 14 dpi. In 12-weeks-old fish, the response returns from 7 dpi on, yet maximal spatial frequency values never approach uninjured values. Both 18-weeks- and 24-weeks-old fish do not regain the optokinetic response after ONC. n = 6–10, except for 6-weeks-old fish at 10 dpi (n = 5) and for 24- weeks-old fish at 18 dpi (n = 4). (b) Assessing the degree of tilting at several time points after unilateral ONC reveals almost complete restoration of the dorsal light reflex from 21 dpi onwards in 6-weeks-old killifish. Fish of the older age groups, that is, 12-weeks-, 18-weeks-, and 24-weeks-old fish, never regain this reflex and remain to swim in a tilted position. n = 5–9, except for 18-weeks-old fish at 0 dpi (n = 4), and 24-weeks-old fish at 55 dpi (n = 3) and 65 dpi (n = 1). All data are represented as mean ± SEM, means with a different letter are significantly different (Two-Way ANOVA), see Table S1 for exact p-values. dpi, days postinjury; ONC, optic nerve crush 2.2 Impaired functional recovery following ONC in aged killifish

In a next set of experiments, we aimed at evaluating whether older fish recover from optic nerve damage equally well as young adult killifish. Visual recovery was determined using the optokinetic response test in middle-aged (12 weeks), old (18 weeks), and very old (24 weeks) fish at various time points following ONC (Figure 2a). Middle-aged fish showed a delay in functional recovery, regaining visual acuity from 7 dpi in comparison with 6-weeks-old fish that already displayed the first signs of recovery at 4 dpi. However, and in contrast to young fish, the optokinetic reflex was never restored to uninjured values at the measured time points, suggesting an age-related impairment in the regeneration process. Strikingly, the older age groups (18 weeks and 24 weeks) never showed any recovery of this primary reflex, not even after more than 8 weeks, indicating that these older fish did not succeed in restoring the retinotectal circuit after ONC.

We also determined the dorsal light reflex in 12-weeks-, 18-weeks-, and 24-weeks-old fish (Figure 2b). Notably, normalization of body axis tilting could not be detected in any of these age groups. They reached their maximum recovery between 14 and 21 dpi, but, even after 65 days, remained to swim in a tilted position when compared to uninjured controls of the same age group. Middle-aged, old, and very old fish thus did not regain this reflex following ONC. All in all, our findings show a clear impairment of functional recovery following optic nerve damage in female killifish of older age.

2.3 Age-associated alterations in the intrinsic growth potential of RGC axons after ONC

As a diminished intrinsic growth potential of RGCs might underlie the observed age-related defect in functional recovery, retinal mRNA expression levels of the growth-associated genes gap43 and tuba1a were investigated in older killifish. Similar to 6-weeks-old fish, real-time quantitative polymerase chain reaction (RT-qPCR) on retinal tissues demonstrated an upregulation of gap43 mRNA in 12-weeks-, 18-weeks-, and 24-weeks-old fish following ONC. However, its expression level at 3 dpi was reduced when compared to young adult fish, suggesting a diminished ability to initiate axon regrowth. Notably, as gap43 mRNA levels were still elevated at 7 dpi, the outgrowth response seems to be delayed as well (Figure 3a). Additionally, tuba1a mRNA expression was lower or even absent in older fish subjected to ONC in comparison with 6-weeks-old fish (Figure 3b). These data clearly indicate that aging impacts the intrinsic ability of RGCs to initiate axon outgrowth.

Aging impacts optic nerve regeneration already early in the regenerative process. (a, b) mRNA expression studies reveal reduced gap43 (a) and tuba1a (b) levels at 3 dpi, respectively, in all older age groups and in 18-weeks- and 24-weeks-old killifish, as compared to young adult killifish. n = 3. (c) Quantification of activated-caspase-3 immunopositive apoptotic cells in the GCL of fish subjected to ONC discloses that significantly more cells undergo cell death at 7 dpi in the older age groups when compared to 6-weeks-old fish. n = 3–5. (d) Representative images of retinal sections of 24-weeks-old fish after immunostaining for activated-caspase-3 show an increased number of apoptotic cells in the GCL at 7 dpi, predominantly in the central retina. Scale bar = 100 µm. Boxed area is magnified in (d’). (e) Following ONC, the total cell number (DAPI+) in the GCL decreases in all older age groups. n = 6–10. (f) Analysis of the proportion of biocytin−/DAPI+ and biocytin+/DAPI+ cells in the GCL using retrograde biocytin tracing reveals that, while the number of biocytin−/DAPI+ cells remains rather constant, the number of biocytin+/DAPI+ cells fails to reach uninjured levels during the regenerative process. n = 6–10. (g) Representative images of the retrograde biocytin experiment depict a lower number of RGCs outgrowing their axons past the crush site (biocytin+/DAPI+) in 24-weeks-old fish from 3 dpi on. Scale bar = 50 µm. (h) Quantification of the percentage of biocytin+/DAPI+ RGCs (relative to the total number of biocytin+/DAPI+ cells in the GCL of uninjured control fish of the same age group) reveals a clear reduction in the number of regenerating RGCs in all older fish during the early phases of the regenerative process. n = 6–10. All values are shown as mean ± SEM, statistical difference is indicated using different letters (Two-way ANOVA), see Table S1 for exact p-values. Note that the data from 6-weeks-old fish are identical to those depicted in Figure 1. DAPI, 4′,6-diamidino-2-phenylindole; dpi, days postinjury; GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; ONC, optic nerve crush; ONL, outer nuclear layer; OPL, outer plexiform layer; RGC, retinal ganglion cell 2.4 Occurrence of apoptosis and reduced number of regenerating RGCs following ONC in older killifish

To investigate whether ONC also results in apoptosis of RGCs in the older age groups, we quantified the number of activated-caspase-3+ cells in the GCL of 12-weeks-, 18-weeks-, and 24-weeks-old fish and demonstrated that the number of dying cells in the GCL at 7 dpi was significantly higher in comparison with 6-weeks-old fish, with an increased number still visible at 14 dpi (Figure 3c). Interestingly, apoptotic cells were mainly found in the central retina of these aged fish, where the older, presumably more vulnerable, cells are located (Figure 3d, magnification in d’).

Cell loss after nerve damage was also studied using retrograde biocytin tracing in 12-weeks-, 18-weeks-, and 24-weeks-old fish at several time points post-ONC. Similar as observed in young adult fish, the total DAPI+ number in the GCL declined after ONC, again resulting in a ~40% reduction in total cell number by 21 dpi (39.1 ± 3.2%, 42.9 ± 1.8%, and 45.6 ± 1.8% reduction in total cell number at 21 dpi in 12-weeks-, 18-weeks-, and 24-weeks-old fish, respectively) (Figure 3e). Counting the number of biocytin−/DAPI+ and biocytin+/DAPI+ cells revealed that ONC does not result in death of displaced amacrine cells but eliminates RGCs (Figure 3f). Together with the observed increase in the number of apoptotic cells, these results show that ONC induces RGC death in aged females, just like in young adults.

Using this retrograde tracing paradigm, we also looked more rigorously into age-associated differences in the number of RGCs that are regenerating, and thus whether the decreased intrinsic outgrowth potential of aged RGCs led to a reduced axonal outgrowth. As compared to young adult fish, the number of regenerating RGCs with axons extending up till the site of biocytin placement was visibly reduced at 3 and 4 dpi in 12-weeks-, 18-weeks-, and 24-weeks-old fish, indicating that axonal outgrowth is indeed affected in the early phases of regeneration (Figure 3g,h). By 21 dpi, 12-weeks- and 18-weeks-old fish seemed to have caught up with the young adults as they showed a similar number of axon-regrowing RGCs, suggestive for a delay in axonal regrowth, rather than a defect. This contrasts findings in 24-weeks-old killifish, in which the number of RGCs with regenerating axons tended to remain lower (with 29.0 ± 3.0% of the RGCs outgrowing an axon at 21 dpi). While axonal outgrowth thus seems to be decelerated in middle-aged and old killifish, a stronger hindrance of outgrowth might occur in very old fish (Figure 3h).

2.5 Reduced tectal reinnervation in aged killifish after ONC

Via anterograde biocytin tracing and morphometrical quantification on midcoronal tectal sections, we next investigated whether the impaired functional recovery in the older age groups is linked to a diminished RGC axon reinnervation of the optic tectum. At 7 dpi, RGC axonal density levels in the SO and SFGS tectal layers were significantly reduced in 12-weeks-, 18-weeks-, and 24-weeks-old fish as compared to 6-weeks-old fish (Figure 4a). Nevertheless, axonal density measurements reapproached uninjured levels at 21 dpi for 12-weeks-old fish. The axonal density levels in the optic tectum of 18-weeks- and 24-weeks-old fish, on the other hand, did not return to uninjured values at 21 dpi (Figure 4b). Axon elongation thus takes longer or might even be permanently impaired in these older females, eventually resulting in delayed or incomplete reinnervation of the optic tectum.

Tectal reinnervation and injury-induced synaptogenesis are impaired in older killifish. (a) Representative images of coronal optic tectum sections, stained for biocytin after anterograde biocytin tracing, visualize a gradual reduction of RGC axonal density levels (arrowheads) in the tectal reinnervation area at 7 dpi in 12-weeks-, 18-weeks-, and 24-weeks-old fish when compared to 6-weeks-old fish. Scale bar = 100 µm. (b) Quantification of axonal density levels, defined as the ratio of the biocytin+ area to the total area of RGC innervation in the optic tectum, shows a clear decrease in tectal reinnervation at 7 and 14 dpi in all older age groups. In middle-aged fish, tectal reinnervation at 21 dpi is comparable to that in uninjured age-matched controls. Axonal density levels in the optic tectum of 18-weeks- and 24-weeks-old fish never approach uninjured values. n = 4–9. (c) Quantification of the Znp-1+ area in the SFGS layer of the optic tectum shows a downregulation of the pre-synaptic signal at 2 dpi in 6-weeks-old fish, and at 7 dpi in 12-weeks-, 18-weeks-, and 24-weeks-old fish. While synaptotagmin expression in the SFGS increases again and is comparable to uninjured control levels at 21 dpi in young fish, it does not reapproach baseline levels at any timepoint after ONC in the older fish. n = 3–5. (d) Immunostainings for synaptotagmin on coronal brain sections reveal the presence of pre-synaptic vesicles in the killifish tectum, with synapses in the SFGS layer indicated with arrowheads. Note that the 6-weeks-old data are derived from the same data as those illustrated in Figure 1h,i. Scale bar = 100 µm. All data are represented as mean ± SEM, means with a different letter are significantly different (Two-Way ANOVA), see Table S1 for exact p-values. DAPI, 4′,6-diamidino-2-phenylindole; dpi, days postinjury; ONC, optic nerve crush; PGZ, periventricular gray zone; RGC, retinal ganglion cell; S/S, projection zone between SAC and PGZ; SAC, stratum album centrale; SFGS, stratum fibrosum et griseum superficiale; SGC, stratum griseum centrale; SO, stratum opticum 2.6 Impeded synaptic repair in the optic tectum following ONC in aged killifish

To study whether also defects in synaptic repair in the optic tectum lie at the basis of the impaired visual recovery in older fish, synaptic dynamics after ONC were investigated in 12-weeks-, 18-weeks-, and 24-weeks-old fish and compared to those in young adult animals. Immunostainings for synaptotagmin at 0, 2, 7, 14, and 21 dpi showed that degradation of RGC pre-synaptic vesicles seemed to start later in the older age groups, that is, at 7 dpi in fish with 12 weeks, 18 weeks, and 24 weeks of age instead of at 2 dpi in young adult fish. While new synaptic vesicles were already noticeable in the SFGS tectal layer of 6-weeks-old fish at 14 dpi, reappearance of pre-synaptic vesicles only became visible in 12-weeks-old fish from 21 dpi onwards. Strikingly, this was not the case in 18-weeks- and 24-weeks-old fish, where the reformation of synapses remained non-existent within the time frame studied (Figure 4c,d). These data indicate a clear effect of aging on injury-induced synaptogenesis, with a delayed or even absent synaptic repair in middle-aged and old/very old fish, respectively.

2.7 Contribution of an altered inflammatory response and glial reactivity to hindered regeneration

Besides the RGC intrinsic decline in axonal outgrowth, reinnervation, and synaptic restoration potential, we also determined age-related effects of extrinsic factors that might affect circuit repair following ONC. Previous data from our group disclosed an upregulation of pro-inflammatory cytokines and an increased presence of resident and/or blood-born immune cells in the retina and optic tectum of older killifish, indicative of an inflammaging status in the old fish CNS (Vanhunsel et al., 2021). Immunohistochemistry for the pan-leukocyte marker L-plastin, and subsequent quantification of the immunopositive area, indeed revealed an increase in microglia/leukocyte number in the retina (Figure 5a, Figure S1) and tectum (Figure 5b,c) of uninjured older killifish as compared to uninjured 6-weeks-old animals. Following optic nerve injury, an inflammatory response was induced in both tissues of all age groups. In 6-weeks-old fish, immune cell numbers were elevated at 2 dpi and returned to control values around 14 dpi, both in the retina (Figure 5a, Figure S1) and tectum (Figure 5b,c). In the older age groups, and especially in 18-weeks- and 24-weeks-old fish, this increase in immune cells lasted longer in the retina as well as the tectum and was more extensive in the retina of 24-weeks-old fish as opposed to young adults (Figure 5, Figure S1). Additionally, RT-qPCR for several pro-inflammatory cytokines was performed on total retinal and tectal samples. This revealed an overall increased and extended, albeit not always statistically significant, expression of interleukin (il)-1β, tumor necrosis factor (tnf), il-6, and il-8 in the retina and/or tectum of 24-weeks-old killifish in comparison with 6-weeks-old fish at various time points after injury (Figure S2). Next to increased transcriptional variability in the old age group, the absence of significant findings might be the result of using complete retinal/tectal lysates and thus a dilution effect, as only a small fraction of all cells in the retina and optic tectum is represented by inflammatory cells. All in all, our results do point toward a stronger and/or prolonged immune response in the retina and optic tectum of older fish subjected to ONC, which might affect the circuit repair process.

Immune cells respond differently to optic nerve crush injury in aged killifish. (a, b) Quantification of the L-plastin+ area in the retina (a) and optic tectum (b) reveals a transient rise in immune cell number in young adult killifish. The immune response following crush in killifish of older age is more extensive and/or prolonged. n = 3–5. (c) Representative pictures of L-plastin-stained tectal sections of young adult, middle-aged, old, and very old killifish reveal an increased number of immune cells in the uninjured aged fish tectum. When subjected to ONC, the immune response seems to be more prolonged in killifish of older age, which is most clear in the tectum of fish at 18 weeks and 24 weeks of age. Scale bar = 100 µm. All values represent mean ± SEM, means with a different letter are significantly different (Two-Way ANOVA), see Table S1 for exact p-values. DAPI, 4′,6-diamidino-2-phenylindole; dpi, days postinjury; ONC, optic nerve crush; PGZ, periventricular gray zone; S/S, projection zone between SAC and PGZ; SAC, stratum album centrale; SFGS, stratum fibrosum et griseum superficiale; SGC, stratum griseum centrale; SO, stratum opticum

Additionally, an involvement of radial glial cells was both qualitatively and quantitatively investigated via immunohistochemistry for vimentin, a well-known marker for reactive gliosis (Figure 6, Figure S3) (Escartin et al., 2021; Zamanian et al., 2012). As described previously (Vanhunsel et al., 2021), Müller glia, the radial glia of the retina, were found to upregulate vimentin expression in both their cell somas and fibers with increasing age. Upon nerve damage, no clear injury-induced glial response could be observed in young adult or aged killifish (Figure 6a, Figure S3), indicating that Müller glia reactivation likely does not contribute to the regenerative response following ONC in killifish nor to the observed age-associated regenerative impairment in old fish.

Radial glia in the old killifish optic tectum behave differently in response to optic nerve crush. (a) Quantification of the vimentin+ area in the IPL and INL of the retina reveals that ONC does not change the vimentin expression pattern in the retina of fish at any age. n = 3–4. (b) Quantification of the vimentin+ area in the SGC, SAC, and S/S layers of the optic tectum shows an age-related upregulation of basal vimentin staining in the glial fibers. Following ONC, an increase in vimentin immunoreactivity is observed in the fibers of young adult killifish, particularly at 7 dpi. Upregulation in the radial fibers post-ONC is more pronounced and prolonged in the older age groups. n = 3–4. (c) Representative images showing immunostaining for vimentin in the optic tectum of young adult, middle-aged, old, and very old killifish, which reveals a gliotic response in the fibers of radial glia (arrowheads) in all age groups subjected to ONC. Scale bar = 50 µm. All values represent mean ± SEM, means with a different letter are significantly different (Two-Way ANOVA), see Table S1 for exact p-values. DAPI, 4′,6-diamidino-2-phenylindole; dpi, days postinjury; INL, inner nuclear layer; IPL, inner plexiform layer; ONC, optic nerve crush; PGZ, periventricular gray zone; S/S, projection zone between SAC and PGZ; SAC, stratum album centrale; SFGS, stratum fibrosum et griseum superficiale; SGC, stratum griseum centrale; SO, stratum opticum

In the optic tectum of uninjured fish, cell bodies of quiescent radial glia are positioned right below the neuronal layer of the PGZ and adjacent to the tectal ventricle. Their individual cytoplasmic processes (fibers) project upwards, protrude in the superficial tectal layers, and terminate as end feet on the pia surface (Lindsey et al., 2019; Than-Trong & Bally-Cuif, 2015; Vanhunsel et al., 2021). Similarly, our findings show that in uninjured young female killifish, vimentin staining of diffuse glial end feet was clearly present in the SGC, SFGS, and SO layers of the optic tectum, and to a lesser extent in the radial fibers. Following ONC in 6-weeks-old fish, vimentin signal was upregulated in the radial fibers, albeit only at 7 dpi and thus for a short time. Damage to the optic nerve thus seems to elicit a transient and limited response of the radial glia in the young adult killifish tectum. In the older age groups, basal vimentin immunoreactivity was visibly upregulated in the radial fibers, indicative for the occurrence of reactive gliosis in the optic tectum of older fish, as previously observed (Vanhunsel et al., 2021). While the gliotic response of the glial fibers appeared to be transient in 6-weeks-old fish, it was prolonged and more pronounced in the older age groups (Figure 6b,c). All in all, these results indicate that radial glia within the tectum of aged female killifish respond differently to ONC, which might contribute to the observed defect in restoration of a functional circuit.

Strikingly, also at the crush site in the optic nerve, we observed signs of gliotic and inflammatory responses in older fish. Following ONC, glial fibrillary acidic protein (Gfap) expression by astrocyte-like cells surrounding the scar seemed upregulated in all age groups, indicative for reactive gliosis (Escartin et al., 2021; Zamanian et al., 2012). Additionally, at 7 dpi, a Gfap− area could be observed at the ONC injury site in all age groups, reflecting the presence of a glial scar (Figure 7a). Notably, in 12-weeks-, 18-weeks-, and 24-weeks-old fish, this Gfap-negative area was enlarged as compared to 6-weeks-old fish at 7 dpi (Figure 7b) and still present in 24-weeks-old fish at 35 dpi (Figure 7c), suggesting more severe damage of the nerve, increased scarring, and/or impaired repopulation by astroglia in older killifish during the repair process (Hilla et al., 2017; Liu et al., 2021; Qu & Jakobs, 2013). Despite reduced Gfap staining within the scar area, this region was not cell-free (Figure 7d). Rather, in line with findings in rodents (Liu et al., 2021; Qu & Jakobs, 2013), L-plastin+ cells, representing microglia and leukocytes, were observed at the crush site of 6-weeks-, 12-weeks-, 18-weeks-, and 24-weeks-old fish as compared to uninjured young adults. In comparison with 6-weeks-old injured fish, the number of these immune cells at the lesion site was clearly augmented in the older age groups (Figure 7e), suggestive for a stronger inflammatory response. Interestingly, Sirius Red staining revealed collagen deposition in the glial scar of 12-weeks-old fish at 7 dpi, which became more prominent at older age, with a clear