4
Title
Axonal regeneration in zebrafish
Authors
Thomas Becker, Catherina G. Becker
Centre for Neuroregeneration, School of Biomedical Sciences, The Chancellor’s Building, 49 Little France Crescent, Edinburgh EH16 4SB, UK
For correspondence: ,
Abstract
In contrast to mammals, fish and amphibia functionally regenerate axons in the central nervous system (CNS). The strengths of the zebrafish model, i.e. transgenics and mutant availability, ease of gene expression analysis and manipulation and optical transparency of larvae lend themselves to the analysis of successful axonal regeneration. Analyses in larval and adult zebrafish suggest a high intrinsic capacity for axon regrowth, yet signalling pathways employed in axonal growth and pathfinding are similar to those in mammals. However, the lesioned CNS environment in zebrafish shows remarkably little scarring or expression of inhibitory molecules and regenerating axons use molecular cues in the environment to successfully navigate to their targets. Future zebrafish research, including screening techniques, will complete our picture of the mechanisms behind successful CNS axon regeneration in this vertebrate model organism.
Introduction
The high capacity of fish and amphibians to regenerate organs, appendages and CNS structures has long been recognized [1]. However, it is still unresolved why amniotes, and in particular mammals, have lost the capacity for CNS regeneration during evolution. There is hope that mechanisms of successful regeneration can be gleaned from studying regenerating model vertebrates, such as zebrafish, and translating knowledge to analyse non-regenerating species. The zebrafish combines experimental and genetic accessibility with an enormous regenerative capacity and is, therefore, advancing to be a widely used vertebrate model of successful CNS regeneration. In the zebrafish CNS, entire neuronal populations regenerate from progenitor cells, which has recently been expertly reviewed elsewhere [2,3]. Here we focus on the events related to injury of axons in the CNS and regrowth from those neurons that have been axotomized.
Extent of axon regrowth
In mammals, axon regrowth in the CNS is extremely limited and some types of axotomized neurons, such as retinal ganglion cells, even perish [4]. In contrast, anamniotes, including zebrafish, have an astonishing capacity to successfully regrow long-range projection axons over distances that are much greater than when these axons first made connections during development. For example, after a crush or complete transection lesion of the optic nerve, retinal ganglion cells (RGCs) survive [5,6] and their axons regrow to faithfully and topographically re-innervate nine termination fields in the adult brain [7]. Regenerating RGC axons reach the tecum opticum, the largest termination field, by 8 days post-lesion (dpl). During the entire regeneration process only a few projection errors are made. A few axons erroneously grow ipsi-laterally into the brain, and some optic axons from the dorsal retina grow erroneously through the dorsal brachium of the optic tract and vice versa [7]. However, these pathfinding errors do not lead to apparent errors in target innervation, as retinotopy of the projection is reestablished by 42 dpl in a pattern that is indistinguishable from unlesioned animals.
After a complete transection of the spinal cord, severed axons of brainstem neurons with spinal projections cross into the distal part of the spinal cord by 14 dpl and project at least 3.5 mm beyond the initial transection site by 42 dpl [8,9]. Interestingly, regenerating axons re-route through the central grey matter, rather than growing through the peripheral white matter, thus navigating novel pathways [10]. Remarkably, not all severed axons regenerate equally well. Some brain nuclei with descending axons show poor axon regrowth [11]. This includes the individually identifiable paired Mauthner neurons, unique large brainstem neurons in aquatic vertebrates. Moreover, descending monoaminergic axons manage to cross the lesion site, but penetrate only a few micrometers into the distal spinal cord [12]. Severed dorsal root axons and ascending axons of intraspinal neurons also show little detectable regrowth [13]. Thus the regrowth capacity for a number of axon types varies and is well established in the adult zebrafish, providing a model to study differences in the regenerative success of CNS axons.
Axon regeneration is functional
Despite minor targeting errors in the regenerated optic projection and imperfect regeneration of spinal axons, functional recovery is spectacular. After optic nerve injury the optokinetic and optomotor responses are recovered by 14 and 28-35 dpl, respectively, matching axon regrowth and re-establishment of retinotopy. More complex visually guided behaviours, such as two fish chasing each other, take longer to recover (3 months), perhaps related to long-term synaptic rearrangements [5,14].
After spinal cord transection, fish are completely paralyzed caudal to the lesion site, but within 42 dpl, most regain swimming activity and the ability to maintain their position in a water flow, similar to uninjured control animals [15,16]. Creating a physical barrier to axon regrowth in the spinal cord prevents recovery [9] and re-transecting the spinal cord abolishes it [12], providing evidence that regrowth of axons is indispensible for recovery. Hence, functional recovery after axonal regeneration in the optic projection and spinal cord are robust and can be assessed by a number of quantitative assays.
Extrinsic determinants of axon regrowth
In mammals, lack of axon regrowth is brought about in part by a hostile growth environment in which astrocytes and other cells form the glial scar containing growth-inhibitory extracellular matrix (ECM) components, such as chondroitin sulfate proteoglycans (CSPGs). In addition, myelin and myelin debris from degenerating fibers contain growth-inhibitory molecules [17]. After a lesion of the optic nerve in zebrafish, there is no evidence of a CSPG expressing glial scar [18]. In the spinal cord, ependymo-radial glial cells, which also have astroglial functions, even form bridges that re-connect the severed spinal cord. In transgenic larvae, expressing green fluorescent protein under the glial fibrillary acidic protein promoter in ependymo-radial glial cells, rejoining of the spinal and glial bridges can be observed by time-lapse microscopy [8]. These bridges have been suggested to guide or support axon growth across the lesion site [8]. Ependymal progenitor cells in the mammalian spinal cord, which are similar to ependymo-radial glial cells in zebrafish, generate scar cells [19]. Remarkably, in zebrafish, but not in mammals, ependymo-radial cells generate neuronal cell types after a lesion [20,21], perhaps in lieu of scar cells, which might be one of the reasons for the absence of a detectable glial scar in zebrafish.
Myelin-associated inhibitory molecules, such as NogoA/RNT4 [22] and MAG/siglec-4 [23] do exist in the zebrafish CNS. However, at least for zebrafish NogoA it has been shown that the NogoA-specific N-terminal inhibitory domain is missing, and the other protein domain that is inhibitory in mammals (Nogo66) fails to elicit growth cone collapse of regenerating axons in vitro [22]. Moreover, zebrafish oligodendrocytes, in contrast to mammalian oligodendrocytes, increase expression of recognition molecules that may promote axon growth, such as contactins [24,25], P0 [26] and L1-related molecules [27], after a CNS lesion.
Receptors for inhibitory molecules, such as Nogo66 and CSPG receptor NgR [28] and the CSPG receptors RPTP-sigma and LAR [29] are expressed in the zebrafish CNS and at least for NgR there is evidence for expression on regenerating axons [22]. However, the exact spatio-temporal regulation of receptor expression during axon regeneration needs further investigation.
Similar to mammals, there is also a strong activation of macrophages/microglial cells after an optic nerve or spinal lesion in zebrafish [10,26,30,31]. Lysophosphatidic acid induced boosting of the immune response after spinal injury negatively impacted neurite growth in a recent study [30]. However, the exact mechanisms how the immune response influences axonal regrowth need to be determined. Overall, the cellular and molecular composition of the adult lesioned CNS in zebrafish presents an environment that is presumably more conducive to axon regrowth than the CNS environment in mammals.
Neuron-intrinsic factors regulating axon regrowth
In general, severed CNS axons in zebrafish have a high capacity for regrowth. For example, in retinal explant culture, RGC axons grow much more vigorously than mammalian counterparts and they upregulate a number of well known regeneration/growth-associated molecules, such as GAP-43 [32], L1-related proteins [33] and alpha-1 tubulin [34]. Similar to mammalian neurons with non-regenerating axons, L1-related proteins and GAP-43 are not upregulated in those zebrafish brain nuclei that show a low regenerative capacity [11]. The functional importance of these genes has been demonstrated in vivo or explant culture by inserting a gelfoam pledget soaked with anti-sense morpholino oligonucleotide into either spinal or optic nerve lesion site. The morpholino is retrogradely and selectively transported to the somata of axotomized neurons where it suppresses expression of target genes for weeks, long enough to show effects on regeneration [33,34].
While the above-mentioned genes are also expressed during developmental axon growth, there is evidence that they are differently regulated during development and adult regeneration. Transgenic reporter lines, using different fragments of the regulatory sequences of GAP-43 or alpha-1 tubulin to drive reporter gene expression, have demonstrated the presence of regeneration-specific mechanisms of gene regulation [32,35]. Indeed, expression profiling of retinal ganglion cells undergoing axon regrowth revealed that some genes are uniquely expressed during regeneration, such as the transcription factor KLF6/7, which in turn regulates expression of alpha-1 tubulin [34,36,37]. The powerful combination of expression profiling with morpholino knock-down has led to the identification of a number of additional genes that may play specific roles in regeneration [38 and citations therein].
Interestingly, not all genes that are upregulated in neurons with regenerating axons promote axon growth. For example, socs3, a strong neuron-intrinsic inhibitor of axon regeneration in mammals, is upregulated in retina ganglion cells after an optic nerve lesion in zebrafish, and attenuates axonal regeneration [39]. This suggests that the molecular injury response in zebrafish neurons may be more similar to that in mammals than previously thought.
Studying zebrafish neurons that do not regenerate axons and in which upregulation of regeneration-associated genes fails, might be particularly instructive. For example, the Mauthner neuron’s regenerative capacity is limited already in larvae, which makes it possible to directly observe effects of manipulations that augment axon regrowth, as in the PNS (see Box 1). Like in mammalian neurons, increasing the levels of cAMP in the axotomized Mauthner neuron leads to increased and directional axon regrowth, resulting in recovery of function [40]. Moreover, the small size of the zebrafish larvae makes it ideally suited for drug screening efforts and indeed, paradigms are being established for semi-automated high-throughput laser-sectioning of the Mauthner axon, greatly facilitating screens that aim to identify factors to promote regeneration [41].
Guidance of regrowing axons
It is perhaps the most astonishing property of adult axon regeneration in zebrafish that axons reach their targets over long distances to make functional reconnections. Indeed, it has been noted in mammalian systems that even when regrowth of CNS axons is experimentally induced, axons frequently fail to navigate correctly [42,43]. It could be hypothesized that regenerating axons in the zebrafish CNS simply retrace their former pathways along degenerating tracts due to physical constraints, similar to the PNS [44]. The observation that most regenerating axons in the spinal cord re-route to the gray matter during regeneration and not through the denervated white matter does not support this idea [10]. Moreover, we tested this hypothesis experimentally in a mutant of the robo2 recognition molecule. This mutation leads to the random and variable appearance of ectopic tracts during development [45]. If degenerating tracts guided optic axons, these should faithfully be re-used by regenerating axons. This was, however, not the case, refuting the hypothesis that degenerating tracts present the predominant guidance cue to regenerating axons [46].
What guides regenerating adult axons? CSPGs and other inhibitory molecules have functions in developmental axon guidance, by repelling axons from areas that are not to be innervated. Regenerating zebrafish optic axons are sensitive to axon-repellent/inhibitory guidance molecules. For example, these axons do not penetrate a substrate boundary of axon-repellent ECM molecules, such as tenascin-R and CSPGs in vitro [18,47]. Indeed, in vivo, tenascin-R surrounds the optic projection, consistent with repellent axon guidance. Tenascin-R and CSPGs are particularly strongly expressed in the posterior pretectal nucleus, a diencephalic nucleus that is engulfed by optic axons, but does not receive primary visual input. Enzymatic removal of CSPGs in vivo allows optic axons to partially invade this nucleus, suggesting that repellent ECM molecules guide regenerating optic axons around this nucleus [18]. Adult zebrafish also retain graded expression of axon repellent ephrins in the tectum [7], which is important for correct retinotopic mapping of optic axons during development. The presence of guidance cues in the adult optic system in zebrafish may be related to the continuous addition of retinal ganglion cells in adults, which need to find their way to the tectum. These cues are available also to regenerating axons. In adult rats, in which no new axons are added to the optic projection, ephrin gradients and appropriate receptor gradients in retinal ganglion cells are down-regulated. However, this is reversed upon a lesion of the adult optic nerve, indicating that the ephrin guidance system might also be available in the lesioned CNS of mammals [48].
Conclusion
Successful CNS regeneration is a process for which a number of intrinsic and extrinsic factors have to interact to allow functional reconnections. Analyses of axon regrowth in the CNS of zebrafish show high intrinsic axon growth capacity, minimal scar formation, low expression of growth inhibitors and guidance of regenerating axons by molecular cues. It is striking that in zebrafish, environmental and intrinsic factors are so well co-ordinated to allow for axonal regeneration, whereas in mammals the opposite appears to be the case. However, intrinsic and extrinsic factors might be functionally connected, such that changing only a few parameters affects both axons and environment. For example, stabilizing microtubuli in the lesioned spinal cord of mammals using Taxol improves both intrinsic axon regrowth and environmental scarring [49]. Given the unique array of genetic, optic and screening tools available in the zebrafish model, we expect zebrafish to contribute to elucidating the molecular mechanisms underpinning axon regrowth and navigation leading to functional reconnections in the vertebrate CNS.
Acknowledgements
We thank Drs. David Lyons and Dirk Sieger for critical reading.
TEXTBOX 1
The peripheral nervous system (PNS) of larvae is highly accessible to study cell-cell interactions after laser-microsurgical lesions.