Axonal regeneration in zebrafish spinal cord

Abstract In the present review we discuss two interrelated events—axonal damage and repair—known to occur after spinal cord injury (SCI) in the zebrafish. Adult zebrafish are capable of regenerating axonal tracts and can restore full functionality after SCI. Unlike fish, axon regeneration in the adult mammalian central nervous system is extremely limited. As a consequence of an injury there is very little repair of disengaged axons and therefore functional deficit persists after SCI in adult mammals. In contrast, peripheral nervous system axons readily regenerate following injury and hence allow functional recovery both in mammals and fish. A better mechanistic understanding of these three scenarios could provide a more comprehensive insight into the success or failure of axonal regeneration after SCI. This review summarizes the present understanding of the cellular and molecular basis of axonal regeneration, in both the peripheral nervous system and the central nervous system, and large scale gene expression analysis is used to focus on different events during regeneration. The discovery and identification of genes involved in zebrafish spinal cord regeneration and subsequent functional experimentation will provide more insight into the endogenous mechanism of myelination and remyelination. Furthermore, precise knowledge of the mechanism underlying the extraordinary axonal regeneration process in zebrafish will also allow us to unravel the potential therapeutic strategies to be implemented for enhancing regrowth and remyelination of axons in mammals.


INTRODUCTION
Among the vertebrates, teleost fish and urodele amphibians have a remarkable capacity to regenerate their axons after spinal cord injury (SCI) (Becker & Becker, 2014;Clarke, Alexander, & Holder, 1988;Hui, Dutta, & Ghosh, 2010;Hui, Monaghan, Voss, & Ghosh, 2013). In contrast to mammals, zebrafish serve as an important model because two efficient programs of neural regeneration exist in adult fish cord, namely neurite outgrowth and remyelination. The extent of axonal regeneration has been evaluated by using various injury paradigms, such as transection or crush injury in zebrafish spinal cord (Becker, Wulliman, Becker, Bernhardt, & Schrachner, 1997;Hui et al., 2010).
However, the lack of long distance axonal regeneration in the mammalian central nervous system (CNS) has been ascribed to inadequate capability of intrinsic growth of neurons and the creation of extrinsic inhibitory milieu after injury (Huebner & Strittmatter, 2009).
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. The regenerative ability of CNS axons differs between mammals and fish, although some of the molecular and cellular pathways underlying axonal regeneration are similar. The remarkable differences in regenerative capacity between the CNS and peripheral nervous system (PNS) are due to differences in the intrinsic capabilities of the injured neurons and differential environmental cues. Thus, in this review, we intend to dissect out the similarities and differences in the cellular and molecular mechanisms that regulate axonal regeneration between the CNS and PNS in both fish and mammals since further insight could be pivotal for inducing successful axonal repair and targeting innervations in higher organisms.
It is important to take into consideration that zebrafish represent a powerful experimental model for studying many neurogenetic disorders because the zebrafish genome has been sequenced and annotated and most zebrafish genes are highly conserved in mammals with a zebrafish ortholog identified for 70% of human genes (Howe et al., 2013). Comparative analysis of neuroanatomy between zebrafish and human central and peripheral nervous systems revealed a similar structural organization, including cell types and neurons. Thus by taking advantage of new tools available for genetic manipulation such as genome editing, high-throughput DNA/RNA sequencing, in vivo imaging, etc. most of human physiology and pathologies can be modeled in zebrafish (Babin, Goizet, & Raldua, 2014;Howe et al., 2013;Kabashi, Brustein, Champagne, & Drapeau, 2011). As some common molecular mechanisms after SCI such as microglial/macrophage response and faithful recapitulation of the myelinating program exist during regeneration and genes identified in the fish are conserved in higher vertebrates (D'Rozario, Monk, & Petersen, 2017), zebrafish offer a promising tool for translational research.

AXONAL INJURY RESPONSE IN MAMMALIAN CNS AND PNS
Adult mammalian CNS axons demonstrate very little capacity to regenerate injured axons. The initial injury impact destroys many neurons, glia, and endothelial and meningeal cells, and the loss is further intensified by subsequent secondary degenerative response. The loss of astrocytes leads to abnormal ionic homeostasis whereas oligodendrocyte loss contributes to poor myelination and impaired axonal activity (Grossman, Rosenberg, & Wrathall, 2001). Following an injury the distal part of severed axons, that had lost contact with the neuronal cell bodies, degenerates. Occasionally the proximal segments survive and grow short sprouts but fail to regenerate and re-innervate appropriate targets (Ramon & Cajal, 1928;Thuret, Moon, & Gage, 2006), despite the attempt of the injured axons to navigate through the lesion environment. Upon injury, axons exhibit dystrophic growth cones identified as sterile clubs by Ramon and Cajal (1928) and later confirmed by others (Ertürk, Hellal, Enes, & Bradke, 2007;Tom, Steinmetz, Miller, Doller, & Silver, 2004). The formation of dystrophic end balls at the ends of lesioned axons is believed to be the reason for the inability for axonal regrowth. Morphologically these swollen entities contain disrupted cytoskeleton and accumulated organelles that persist for months to years after SCI (Hill, 2017;Hill, Beattie, & Bresnahan, 2001;Ruschel et al., 2015). Injured axons of the PNS can generate a new motile growth cone within hours of injury whereas injured CNS axons retract and form a retraction bulb. The formation of motile-growthcone-like structures also refers to a key difference between regeneration competent and incompetent axons. Further advancement of our understanding of retraction bulb formation compared to growth cone formation and growth cone collapse at different stages after injury is pivotal to recognizing the outcome and complexity of axonal injury response.
The landmark experiments by Aguayo, David and Bray (1981) showed that some injured CNS axons retain a limited capacity for regrowth and can regrow over a long distance in a permissive environment of sciatic nerve graft but cannot reintegrate into the CNS. Moreover, sprouting or regeneration without establishment of actual synaptic target innervations does not have any functional significance. Later, others showed that myelin from peripheral nerve is growth permissive whereas myelin from the CNS strongly inhibits nerve growth (Caroni & Schwab, 1988). It is also known that adult PNS neurons retain regenerative ability after injury and both sensory and motor axons can regenerate over a long distance. Thus substantial anatomical regeneration leads to functional recovery (Abe & Cavalli, 2008). Dorsal root ganglia (DRG) are unique as their axons bifurcate to innervate peripheral targets like skin and muscles whereas the central branch supplies sensory information to the CNS. While the peripheral branch of DRG can regenerate following injury, the central branches are unable to do the same.
Thus, the CNS and PNS respond differently to injury. This striking difference in regenerative ability is due to intrinsic properties of injured PNS and CNS neurons as well as to the differential extracellular environment of PNS and CNS axons. The difference in response to injury also resides in the glial population which may elicit either a pro-or antiregenerative response (Lutz & Barres, 2014). A deeper understanding of these cellular and molecular mechanisms highlights the differential ability of CNS and PNS axons and would allow us to devise future therapeutic strategies to induce effective axonal regeneration in the CNS.

AXONAL INJURY IN FISH CNS LEADS TO FUNCTIONAL RECOVERY
Adult zebrafish have the capability of robust axonal regeneration following spinal cord and optic nerve injuries. Severed axons can regrow following injury and as a consequence functional restoration can take place, which is quite remarkable. After injury, loss of function reflects the level of injury, i.e., fishes are paralyzed caudal to the lesion site but regain their normal swimming behavior within 4−6 weeks ( Figure 1) (Becker et al., 2004;Dias, Yang, Ogai, Becker, & Becker, 2012;Hui et al., 2010;van Raamsdonk, Maslam, de Jong, Smit-Onei, & Velzing, 1998). Regeneration of some axons after SCI is robust, e.g., severed axons of brainstem neurons can project across a significant distance-approximately 3.5 mm beyond the lesion site (Becker et al., 1997). But for others, axonal regrowth to a large extent is variable, e.g., Mauthner neurons exhibit poor axonal regrowth, although it can be augmented by changing intracellular signaling by adenosine 3 ′ ,5 ′ -cyclic monophosphate (cAMP) (Bhatt, Otto, Depoister, & Fetcho, 2004). Similarly, dorsal root axons and ascending axons of intraspinal neurons do not show any significant regrowth (Becker, Leiberoth, Becker, & Schachner, 2005;Becker et al., 1998), whereas descending mono-aminergic axons are capable of axonal regrowth for only a few micrometers into the distal stump. However, the volume of white matter renewal is not 100% compared to the original uninjured cord (Hui et al., 2010) and most of the regenerating axons extend through gray matter rather than through white matter, perhaps taking support of macrophages and microglia (Becker & Becker, 2001).

NEUROTRAUMA AND INFLAMMATORY RESPONSE
Traumatic injury to the nervous system triggers an inflammatory reaction. Inflammation in the CNS and PNS differs from each other. Based on various studies we compared the inflammatory response between

Microglia/macrophage mediated response after mammalian SCI
The inflammatory response after CNS injury is primarily due to activation of microglia and recruitment of peripherally derived macrophages which are known to contribute to secondary degenerative response within the mammalian CNS (Fitch & Silver, 2008). Both these cells respond to injury in different proportions depending on the type and extent of injury and may produce various cytotoxic to trophic factors which can play a deleterious as well as a beneficial role in CNS tissue. Resident CNS cells upregulate pro-inflammatory cytokines and chemokines within the first hour after injury which may cause neuronal death and destruction in the vicinity of the injury and thus may play a crucial role in disconnecting existing neuronal connections (Banati, Gehrmann, Schubert, & Kreuzberg, 1993;Giulian, Chen, Ingeman, George, & Noponen, 1989). Microglia become activated within 1 day after injury and play a role in phagocytosis of tissue debris, thus becoming involved in scavenging, resisting infections, and restoration of tissue homeostasis (Hanisch & Kettenmann, 2007;Jin & Yamashita, 2016). The role of microglia in clearing cell debris is amplified in CNS injury, although their phagocytic capacity may be limited compared to blood-borne macrophages (Neumann, Kotter, & Franklin, 2008). After CNS injury microglia may produce anti-inflammatory cytokines such as interleukin 4 (IL-4), IL-10 and transforming growth factor (TGF-) having a neuroprotective role (Hanisch & Kettenmann, 2007;Jin & Yamashita, 2016;Streit, 2005).
The inflammatory response due to microglial production of cytokines or chemokines recruits peripheral immune cells such as monocytes, neutrophils, dendritic cells, and T-lymphocytes at the injured site in different phases of injury. Most importantly, in mammalian SCI, hematogenous macrophages and microglia persist indefinitely at the injury site (Donnelly & Popovich, 2008) whereas depletion of macrophages improves recovery and augments reparative macrophage phenotypes increasing axon growth and motor activity (Popovich et al., 1999;Schwartz & Yoles, 2006). Moreover, delayed and limited recruitment of macrophages into the lesion site and the persistent presence of myelin debris create a non-conducive environment for axon growth leading to CNS regeneration failure (George & Griffin, 1994;Perry, Brown, & Gordon, 1987;Schwab, Kapfhammer, & Bandtlow, 1993).
Macrophages phagocytose debris and secrete neurotrophic factors and play a direct role in axonal retractions and axonal dieback which occur extensively following injury. Microglial response after injury also contributes to scar formation (Dibaj et al., 2010;Silver, Schwab, & Popovich, 2015). Macrophages engulf axonal fragments at the vicinity of the injured site where inflammation is most intense . Activated microglia/macrophages release matrix metalloproteinase-9 (MMP-9) and are implicated in blood−brain barrier disruption, neutrophil invasion, and secondary myelin degradation.
Blood derived macrophages rather than resident microglia are responsible for prolonged dieback of injured axons (Evans et al., 2014), which is initiated by MMP activity (Horn, Busch, Hawthorne, Van Rooijen, & Silver, 2008). Acute, transient upregulation of MMP-9 and delayed but persistent upregulation of MMP-2 has been reported in SCI and spinal nerve injury (Noble, Donovan, Igarashi, Goussev, & Werb, 2002;Verslegers, Lemmens, Van Hove, & Moons, 2013). MMP-9 and MMP-2, in particular, are predominantly recognized as key players in clearing the path for axons to regrow by breaking chondroitin sulfate proteoglycan (CSPG) scar tissue. An in vitro study also revealed that activated macrophages cause marked axonal retraction which can be inhibited by functionally blocking MMP-9, but not MMP-2 (Busch, Horn, Silver, & Silver, 2009). Macrophages are exposed to the milieu of injured CNS and differentiate into a functionally distinct subset of cells with differential effect on neuronal survival and axonal regrowth (Kigerl et al., 2009). M1 type macrophages responsible for interferon- (IFN-) and Toll-like receptor signaling secrete pro-inflammatory cytokines which augment inflammation and removal of debris and cause axonal dieback (Horn et al., 2008;Kigerl et al., 2009), whereas M2 type macrophages are alternatively activated, cause cell proliferation and migration, release growth factors such as neurotrophins, enhance oligodendrocyte progenitor cell (OPC) differentiation, remyelination, and axonal regrowth, and reduce axonal dieback. So M2 macrophages initiate anti-inflammatory responses (Kigerl et al., 2009;Miron et al., 2013) and can overcome axon growth inhibition by CSPG and myelin (Kigerl et al., 2009). Mammalian SCI response begins with an early pro-inflammatory response. The M1 polarized macrophages persist long after injury exerting neurotoxic effects leading to chronic inflammation and impaired axonal regeneration (Beck et al., 2010;Kigerl et al., 2009).

Macrophages/microglia, after zebrafish SCI, exhibit controlled inflammatory response and augment rapid removal of myelin debris
Several chemokines and cytokines are upregulated after SCI in zebrafish (Hui et al., 2014), and among these IL-4r, interferon-1 (IFN-1), and transforming growth factor beta 1 (TGF -1) are also upregulated in mammalian CNS injury. Similar to other vertebrates, zebrafish microglia express typical vertebrate macrophage genes. Furthermore the expression of many transcriptional regulators, immune pathogen receptors, and pruning associated genes, which are also found in mammals, suggests functional conservation between mammals and fish (Oosterhof et al., 2017).
The secretion of MMPs can be seen in both mammalian and fish CNS injury. Dynamic expression of four specific MMPs (MMP-2, MMP-9, MMP-13a, and MMP-14) during different phases of retinotectal regeneration has been reported in zebrafish (Lemmens et al., 2016;McCurley & Callard, 2010). Our microarray analysis data demonstrate upregulation of several MMP genes in different phases after SCI in adult zebrafish. In the absence of glial scar in zebrafish CNS injury, we do not see upregulation of MMP2. Most importantly, both MMP-9 and MMP-13 are commonly expressed after SCI and optic nerve crush injury, probably indicating their implication in injury response and axonal regrowth (Table 1) (Hui et al., 2014;McCurley & Callard, 2010).
The spatiotemporal activation and distribution of microglia and macrophages after SCI in zebrafish differ from mammals and we observe early activation of microglia and infiltration of blood-borne macrophages (2−3 days post injury) in the wound site (Hui et al., 2010).
In mammals macrophages persist at the injury site for a long time after SCI (42 days post injury in rodents, 12 months post injury in humans) (Fleming et al., 2006;Kigerl, McGaughy, & Popovich, 2006), whereas in adult zebrafish cord depletion of blood-borne macrophages was observed 10 days post injury (Hui et al., 2010). Functionally different macrophage subsets also exist in zebrafish (Nguyen-Chi et al., 2015), the expression of a higher number of M2 type macrophage genes being upregulated very early after SCI whereas expression of only two M1 type macrophage genes with very low fold change was observed (Table 1) (Hui et al., 2014). So far all the circumstantial evidence puts forward a hypothesis that, in zebrafish SCI, there is an initial brief proinflammatory state followed by an anti-inflammatory response as M2 polarized macrophages persist and predominate. So there is a general bias towards an anti-inflammatory state as M2 macrophages are not neurotoxic and hence lack of chronic inflammation promotes axonal regeneration after SCI. Further experimental analysis and functional validation is still required to fully uncover the role of different subsets of macrophages in controlling inflammation after SCI in zebrafish.

Some microglial responses after SCI in mammals and zebrafish and in PNS injury are common
There are some obvious similarities in the molecular signals emanating from macrophages and microglial cells that exist after SCI in mammals and fish. A recent study showed that blockage of lysophosphatidic acid, an important mediator of inflammation, improves outcome after SCI in both zebrafish and mouse models (Goldshmit et al., 2012).
Several inflammatory stimuli such as lipopolysaccharide and zymosan can promote the M1 type macrophage phenotype and augment a regenerative response in both peripheral and central axons (Boivin et al., 2007;Gensel et al., 2009;Silver et al., 2015;Yin et al., 2003).
Microglia are thought to be the first cell type reacting to CNS injury in both mouse (Bollaerts,  although there may be temporal variation. The microglial response after injury in the CNS in mammals is mediated by ATP and receptor P2Y12 that serves as an attractant to reach the injury site (Davalos et al., 2005;Haynes et al., 2006). We confirmed the presence of activated microglia in early injured cord ( Figure 2). In zebrafish also, ATP and P2Y12 purinergic receptor are required for microglial migration after CNS injury, which highlights a conserved role of these molecules by which microglia can sense neuronal damage (Sieger et al., 2012).
Even in the absence of mechanical injury, when a single motor neuron was selectively ablated in zebrafish larval spinal cord, microglia became rapidly activated within 30 min as the dying neuron sent an "eat me" signal and subsequently became phagocytosed by microglia at the lesion site (Morsch et al., 2015). In PNS, ATP secretion from Schwann cell related lysosomal exocytosis during Wallerian degeneration has also been reported (Jung et al., 2014).

Zebrafish CNS injury exhibits a response similar to PNS injury
Previous investigations suggest that one of the reasons for failure of axonal regeneration after mammalian CNS injury is contributed by inefficient myelin clearance, whereas in PNS an efficient myelin clearance during Wallarian degeneration by Schwann cells and macrophages augments regeneration (David & Lacroix, 2003;Neumann et al., 2008;Vargas & Barees, 2007). In support of this concept, we observe that zebrafish macrophages are also involved in debris

INTRINSIC CONTROL OF AXON GROWTH AFTER INJURY
A major limiting factor for successful axon regeneration of adult mammalian CNS is the poor intrinsic property of injured neurons. In contrast, following axotomy in PNS, a vast collection of regeneration asso- The response could be elevated and sustained probably because of upregulation of several neuron intrinsic growth associated factors such as GAP-43 and L1 related molecules and alpha tubulin (Becker et al., 2004;Kusik, Hammond, & Udvadia, 2010;Veldman et al., 2010).
Axonal regeneration in both zebrafish CNS and PNS decline with age (Becker et al., 1997(Becker et al., , 1998Graciarena, Dambly-Chaudière, & Ghysen, 2014) and low regenerative capacity of these axons is often related to a failure to upregulate GAPs (Rasmussen & Sagasti, 2017). Some of the RAGs identified in fish are also capable of axonal regeneration in mammals. Furthermore, our array analysis data suggest that several RAGs associated with regeneration of PNS are upregulated after SCI in zebrafish (Hui et al., 2014). Specific intrinsic molecular differences contribute to differential axonal growth response in PNS and CNS injury, and exploitation or manipulation of specific PNS related RAGs could promote limited CNS regrowth after injury. Thus one of the future strategies to induce regeneration in mammalian CNS would be manipulation of RAGs following injury.
CNS axons are capable of regenerating over long distances within a permissive environment

EXTRINSIC FACTORS-CNS MYELIN ASSOCIATED INHIBITORS AND THEIR ROLE IN AXONAL REGROWTH
Following injury, the PNS axons in higher vertebrates are capable of regeneration whereas CNS axons fail to regenerate. This difference in regeneration capacity is not only because of differential intrinsic properties of CNS and PNS neurons but also due to the respective environments. The extrinsic factors include extracellular matrix (ECM), trophic factors, chemorepulsive guidance cues, and myelin associated lipids and proteins. Accumulation of myelin breakdown products released from severed axons and the formation of inhibitory glial scar at the injury epicenter lead to a chemical and physical barrier that perturbs axonal growth and regeneration (Busch & Silver, 2007;Fawcett, 2006;Schwab, 2004). The major extrinsic barriers to axonal regrowth in injured CNS are several growth inhibitory and repulsive factors expressed by different glial cells such as glial progenitors, oligodendrocytes, astrocytes, and microglia (Silver et al., 2015). Regenerating axons of both fish and mammalian neurons are prevented by mammalian oligodendrocytes and myelin, although axons can grow in the presence of fish oligodendrocytes (Bandtlow, Zachleder, & Schwab, 1990;Bastmeyer, Beckmann, Schwab, & Stuermer, 1991;Fawcett, Rokos, & Bakst, 1989;Schwartz et al., 1985;Wanner et al., 1995). This experimental evidence indicates that the factor(s) in the environment of mammalian CNS and PNS may account for a differential axonal regeneration response

Adult CNS myelin is inhibitory for neurite outgrowth in mammals
CNS myelin is found to be the primary source of inhibition as immobilized CNS myelin, but not PNS myelin, is the one that inhibits axonal outgrowth (Schwab & Thoenen, 1985). The three isoforms of Nogo (Nogo A, B, and C) belonging to the reticulon family (RTN 4) of membrane proteins are present in mammalian CNS. Among these, Nogo A is highly expressed in oligodendrocytes in mammalian CNS after injury (Huber, Weinmann, Brösamle, Oertle, & Schwab, 2002).
Other CNS myelin associated factors that can strongly inhibit axon outgrowth in vitro include myelin associated glycoprotein (

Involvement of multiple ligands and multiple receptors in axonal growth inhibition
Amongst several ligands MAG is synthesized in oligodendrocytes and Schwann cells, but it is a relatively minor constituent of both CNS and PNS myelin (Quarles, 2007). Although MAG is responsible for maintenance of myelinated axons, it has been widely used as an inhibitory substrate for neurite outgrowth assay in postnatal and adult neurons.
It serves dual functions depending on age and type of neuron, i.e., promotes neurite outgrowth in young and inhibits the same in adult neurons. Furthermore, genetic deletion studies revealed that MAG may have opposing roles-inhibitory on some neurons (reduced sprouting of corticospinal tract axons) while promoting on others (enhanced serotonergic axon sprouting) even in adult CNS. All the genetic deletion analysis as well as in vitro and in vivo experimental evidence suggests that MAG may take a divergent role in disease and injury. It may promote axonal growth and protect axons from further degeneration, contrary to its well publicized role in axonal growth inhibition.
OMgp, a glycosyl phosphatidyl inositol (GPI) linked protein containing a leucine rich repeat (LRR) domain, is widely distributed in the CNS and causes growth cone collapse in many neuronal populations. It has also been demonstrated that this protein is enriched in the membranes of oligodendrocyte-like cells around the nodes of Ranvier, hence damaging collateral sprouting (Huang et al., 2005).

Generation of dominant negative NgR mutation and antibody to
NgR suggests that both are capable of blocking neurite outgrowth by MAG (Domeniconi et al., 2002).

GLIAL SCAR FORMATION AFTER MAMMALIAN SCI
Apart from degenerating myelin, astrocytic scar formation has been regarded as an important component of axonal growth inhibition after injury in the mammalian CNS. The glial scar is a complex of ECM and cell types that form a dense structure, which may serve to protect adjacent tissue but may also be a major impediment to regenerating axons. Various previous reports correlate the failure of axonal regrowth to the presence of mature astrocytes, astrocytic scars, and CSPG produced by astrocytes (Liuzzi & Lasek, 1987;. However, recently Anderson et al. (2016) challenged the dogma that the glial scar is the major impediment to axonal regrowth and functional recovery. The astrocyte response to injury is termed reactive gliosis which is characterized by cellular hypertrophy, changes in gene expression, and cellular proliferation (Sofroniew, 2005(Sofroniew, , 2009).
While studying the molecular mechanism underlying reactive gliosis and its effect on astrocyte function, it has been suggested that astrogliosis may result in both beneficial and detrimental effects on axonal growth depending on its time course and dynamic features (Bush et al., 1999;Faulkner et al., 2004;Wanner et al., 2013). Some of these responses may have beneficial properties causing reduction of inflammation and cellular degeneration, while a particular population of astrocytes may support axonal regrowth (Faulkner et al., 2004). The astrocytic scar or glial scar is made up of two distinct components: (a) the lesion penumbra with hypertrophic astrocytes and (b) the core of the lesion composed of NG2 + oligodendrocyte precursors, PGDFR + fibroblasts, and activated infiltrating macrophages and pericytes. Dystrophic axons within the lesion remain in close vicinity and make stable contact to NG2 + glia (Busch et al., 2010;McTigue, Tripathi, & Wei, 2006). Thus the presence of segregated populations of different cells within structurally layered glial scars is indicative of the existence of specific chemorepulsive/attractive mechanisms (Cregg et al., 2014). NG2 + glia initially provide a supportive substance which helps to prevent axonal dieback due to the presence of inflammatory cells. Later, the interaction causes entanglement of dystrophic axons to NG2 glia. These synaptic specializations of dystrophic axons and NG2 + glia are responsible for hindering axonal growth. Although OPC proliferation is inhibited by NG2, its protein scaffold can also be degraded by MMP-9, which in turn can facilitate remyelination.

CSPG mediated growth inhibition
Astrocytes are considered to be the major cell type in glial scars and produce different classes of proteoglycans, such as heparin sulfate proteoglycan, dermatan sulfate proteoglycan, keratin sulfate proteoglycan, and CSPG (Johnson-Green et al., 1991;. Expression of CSPG increased markedly after CNS injury, in the vicinity of a disrupted blood−brain barrier. Evidence suggests that the inhibitory activity of CSPGs depends on the GAG components as these studies demonstrated that treatment with chondroitinase ABC removes sulfated GAG chain from CSPG and hence abolishes inhibition . Chondroitinase treatment not only enhances axonal regeneration and functional recovery after SCI, but promotes collateral sprouting and generation of new synapses (Alilain, Horn, Hu, Dick, & Silver, 2011;Bradbury et al., 2002). Although it is clear that CNS myelin and glial scars both inhibit axonal regeneration, their relative importance in vivo is not very convincing as there are conflicting reports suggesting that some overlap and spatiotemporal difference exists in regulation.
Several general mechanisms are implicated for CSPG-mediated growth inhibition, such as masking of neuronal integrin interaction with growth promoting ECM, e.g., laminin, NCAM, etc., facilitation of inhibitory effects of Sema5a, and limiting calcium availability to neurons by binding extracellular Ca + and by binding with CSPG receptors. Several CSPG receptors like LAR phosphatase, PTP , Ngr1, and Ngr3 (Lutz & Barres, 2014) have been reported. The intracellular consequences of CSPG mediated growth inhibition involve Rho activation, AKT inactivation, and calcium related signals like PKC (Yiu & He, 2006). Interestingly, the presence of both the growth promoting and the growth inhibitory components in the ECM and CSPG might act by interacting with growth promoting substrates. Conversely, exposure of CSPG can also enhance dystrophic growth cone formation in injured sensory axons, suggesting that CSPG mediated inhibition could affect both cytoskeletal and membrane elements of the growth cone structure (Ramon & Cajal, 1928;Tom et al., 2004).

ABSENCE OF ASTROGLIOSIS IN REGENERATING ADULT ZEBRAFISH CNS
In adult zebrafish, the most important aspect of axonal regeneration is that the regenerated axons reach the appropriate targets over long distances and can make re-innervations, a phenomenon that does not happen in the case of mammals. It has been hypothesized that regenerating axons could trace their original pathways along the degenerating tracts as happens in PNS (Graciarena et al., 2014). It has been proposed that regenerating axons re-route through gray matter during regeneration which does not match the previous hypothesis (Becker & Becker, 2001). So what would be the guidance cues  (Hui et al., 2015).
There is an obvious lack of a permanent glial scar and formation of a growth permissive glial bridge that may lead to augmentation of axonal regeneration in adult zebrafish CNS (Baumgart et al., 2012;Goldshmit et al., 2012;Hui et al., 2010). Strand et al. (2016) (Vitalo, Sîrbulescu, Ilieş, & Zupanc, 2016). Application of Fibroblast Growth Factor-2 (FGF2) after mammalian SCI generates a radial-glia-like or pro-regenerative glial-progenitor-like state and thus FGF2 mediated response leads to a change in glial morphogenesis and attenuated scar formation in both mammalian and zebrafish SCI (Goldshmit et al., 2014).
The composition of CNS ECM is different from the PNS. The dense glial scar and ECM network act as a physical and molecular barrier to axon regeneration in injured CNS. The major ECM molecules in the CNS include a huge amount of glycosaminoglycan hyaluronan and the glycoproteins tenascin-C and thrombospondin whereas PNS ECM includes laminins, collagen, and heparin sulfate proteoglycan. The mechanism of glial bridge formation seems to be conserved between fish and mammals, since application of FGF2 improves recovery after mammalian SCI and promotes bridge formation in zebrafish (Goldshmit et al., 2014). The glial bridging behavior also resembles Schwann cell bridging in mammalian PNS. Schwann cells secrete a basal lamina rich in growth promoting ECM, which is crucial to the ability of these cells to myelinate (Bunge et al., 1990). Upregulation of pro-regenerative ECM molecules like laminin−integrin interaction could trigger PI3K activation, Akt signaling, and cytoskeletal rearrangements, all these signals play important role in PNS regeneration (Chen, Yu, & Strickland, 2007), although in intact peripheral nerve Schwann cells express CSPG and after injury CSPG expression is upregulated (Höke et al., 2006). Basal lamina tubes play a vital role in shielding axons from CSPGs inside endonurium. Furthermore, upregulation of MMP-2 and MMP-9 in the distal stump of injured PNS relieves CSPG inhibition by degrading the same and favoring a more regeneration permissive environment for axonal growth (Ferguson & Muir, 2000). Similar to Schwann cells, astrocytes may upregulate ECM molecules like fibronectin and laminin after injury, but modest upregulation of pro-regenerative CNS ECM is overshadowed by a huge upregulation of CSPGs which are inhibitory to axonal regrowth (McKeon, Höke, & Silver, 1991 in CNS. After injury in the CNS, a complex ECM environment is generated in the wound site that may be inhibitory for axon regrowth. The CNS glia and ECM are known to contribute to inhibitory scars whereas PNS glia affect post injury ECM favoring axon growth. In zebrafish spinal cord, Wnt signaling is activated after SCI that controls fibroblast and ECM deposition. Wehner et al. (2017) demonstrated that Wnt/b catenin signaling controls collagen XII deposition and promotes axonal regeneration after SCI. Similarly, a fibrous scar was generated after SCI in goldfish but, despite the fibrous scar, regenerating axons can enter and pass the lesion aided by the glial process (Takeda, Atobe, Kadota, Goris, & Funakoshi, 2015).

DEMYELINATION AND REMYELINATION AFTER INJURY
During CNS injury axons are severed and cells in the white matter may die. Demyelination of axons following injury results in severe loss of function and has grave consequences in both the CNS and PNS. So, remyelination proves to be a crucial step to achieve amphibians, axonal regeneration has been studied in both these scenarios in fish. While studying the myelination and remyelination process in the zebrafish nervous system it has been observed that the fundamental structure and composition of myelin and the underlying molecular mechanism controlling myelination are conserved between fish and mammals (Preston & Macklin, 2015). OPCs generate myelin forming oligodendrocytes continuously in adult rodent brain (Rivers et al., 2008). Similarly, the number of oligodendrocyte lineage cells increases and axon myelination occurs continuously in post embryonic fish cord by Myelin Basic Protein (MBP) oligodendrocytes (Jung et al., 2010;Park, Shin, Roberts, & Appel, 2007). Unlike mammalian CNS myelin, fish myelin and oligodendrocytes do not inhibit axon regrowth (Bastmeyer et al., 1991). Following CNS injury in zebrafish, the expression of growth promoting cell surface protein is linked to axonal regeneration. Myelin protein zero in mammalian Schwann cells is known to promote axonal growth whereas mpz genes encoding zebrafish ortholog Po and contactin 1a are strongly upregulated in oligodendrocytes in regenerating white matter tracts after injury (Schweitzer et al., 2003;Schweitzer et al., 2007). It has also been observed that separate cis-regulatory elements mediate mpz expression in myelinating oligodendrocytes and its transcriptional induction by axonal damage. Thus different mechanisms exist during myelination and maintenance of CNS myelin in uninjured and injured zebrafish CNS (Bai, Parris, & Burton, 2014). Interestingly the anatomical characteristic of zebrafish cord refers to the fact that axons of ventral and dorsolateral cord are highly myelinated, whereas other regions of the white matter harbor branches of radial glia along with non-myelinated axons. This may indicate that unmyelinated axon bundles and radial processes may provide a more permissive cellular environment for axonal regeneration after injury in adult zebrafish cord.
The regenerative capacity of axons is dependent on and supported by Schwann cells-myelin forming glias of the PNS. As studied in a variety of paradigms, injury to the PNS would cause Wallarian degeneration, followed by neurite regeneration and remyelination (Chen et al., 2007;Scheib & Höke, 2013 to play a role after SCI in mammals (Giger et al., 2010;Harel & Strittmatter, 2006;Hollis, 2016;Jacobi, Schmalz, & Bareyre, 2014). It has been suggested that inhibitory and chemo-repulsive axon guidance molecules are likely to play an important role in synaptic stabilization and limitation of neuronal plasticity in adult life. Significantly, after SCI, several members of the slit and semaphorin family, syn-CAM, neuroligin, and ephrin B are expressed in spinal neurons, for e.g., slit 1-3 and semaphorin 7a and neuroligin-1 are expressed in propriospinal neurons and a few glycinergic interneurons in spinal cord (Jacobi et al., 2014). Wnt signaling in adult CNS injury also contributes to brain and spinal cord circuitry (Onishi et al., 2014). The reappearance of developmental cues after injury limits both descending and ascending motor and sensory axons. In particular Wnt-PCP signaling plays an important role in axon guidance. Ryk mediates Wnt repulsion by inhibiting PCP signaling. The role of Wnt in directing peripheral axon growth has not been studied in its entire gamut.

Future perspectives
The similarities and differences of molecular and cellular mechanisms of adult zebrafish CNS regeneration and mammalian PNS regeneration give us insights to better understand how a permissive niche can be created to achieve successful axonal regeneration in the mammalian CNS and to adopt successful therapeutic strategies. The abundant data discussed above provide a strong basis to pursue research on zebrafish, with a goal to induce regeneration in mammals including humans. To consider some of the future therapeutic strategies a better understanding of the differential immune response after injury is required. Zebrafish could serve as an ideal model to uncover the beneficial immune response and maintenance mechanism for successful regeneration. To induce axon regeneration, understanding glia present in the axon environment and their response to injury is crucial.
A deeper understanding of the mechanism employed by PNS glia could shed more light on the regeneration permissive environment.
Successful axonal regrowth in zebrafish CNS relies on high intrinsic capacity, absence of a glial scar, and appropriate axon guidance molecules. Future studies can be facilitated by large scale gene expression analysis to identify novel RAGs, axonal guidance molecules, etc.
along with efficient functional analysis by adapting reverse genetic techniques paving the road for rapid identification of the molecular pathways involved in axonal regeneration. A continued focus on comprehensive dissection of the molecular mechanisms of endogenous capacity of axonal regeneration in zebrafish could be maximally exploited to achieve functional recovery after CNS injury.
A significant focus of SCI research has been directed towards the mammalian model, which is regeneration incompetent. There has been much progress to understanding the cellular and molecular basis of regeneration failure in mammalian CNS, yet successful repair strategies cannot be adopted from mammalian SCI. The injury response in mammals proves to be complex and dynamic. The failure and success of regeneration depend on several cellular events and are controlled by many event-specific molecules and factors. However, much of the therapeutic strategy to induce regeneration, particularly in spinal cord, is actually based on specific molecules which may have a role in controlling any specific events. In order to improve therapeutic intervention after SCI, a combinatorial approach should be taken. Manipulation of a single event, molecule, or a group of molecules may not improve the clinical outcome. Serious thought has to be given to a more comprehensive approach. In such a scenario a translational approach by studying a model organism to enhance CNS regeneration is invaluable and needs further attention.

CONFLICT OF INTEREST
No conflict of interest.