Mechanisms underpinning spontaneous spinal cord regeneration

Abstract

Vertebrates exhibit a range of regenerative capacities following spinal cord injury. At one end of the spectrum are chief regenerators, including teleost fish and urodele amphibians. At the other end, most mammalian species exhibit limited repair and multicellular complications following spinal cord injury. Pro-regenerative immune, glial and neuronal injury responses underlie innate spinal cord repair in highly regenerative vertebrates. In many instances, fundamental mechanisms of spinal cord repair represent ancestral neuroprotection mechanisms that are conserved but become overwhelmed by anti-regenerative effects in mammals. Reflecting recent advances in the field, we review how fine-tuned immune responses, pro-regenerative glial cell reactivity and multimodal neuronal repair direct innate spinal cord repair.

Summary:

This Review gives an overview of how pro-regenerative immune, glial and neuronal strategies are deployed for recovery from spinal cord injury, which is limited to select vertebrate species.

Introduction

The basic form and functions of the spinal cord are conserved across all vertebrates. Yet while common developmental principles guide intricate spinal cord morphogenesis during vertebrate embryogenesis, adult vertebrate species elicit widely varied responses following spinal cord injury (SCI). In most mammalian species, including humans, SCI causes permanent loss of sensory, motor and systemic functions, posing a global medical and socioeconomic burden (Ahuja et al., 2017). Curiously, vertebrates such as teleost fish and urodele amphibians – and even some mammalian species – display a remarkable ability to tolerate damaging injuries and regain lost functions (Zottoli et al., 1994; Becker et al., 1997; Hanslik et al., 2019; Thygesen et al., 2019; Streeter et al., 2020).

The unique ability of some vertebrates to mend a lesioned spinal cord have fascinated scientists for centuries. As neurons are the ultimate effectors of spinal cord function, earlier regeneration studies were largely focused on neuronal repair, mirroring neuron-centered approaches in the neuroscience field. Fueled by considerable understanding of developmental axon growth and neurogenesis, the molecular and cellular mechanisms of axon regrowth and neuronal regeneration in regenerative species were the first to be dissected (Bastmeyer et al., 1991; Becker et al., 1998; Hui et al., 2013). Over the past decade, renewed appreciation for the roles of immunoglial cells, including astrocytes and microglia, has expanded SCI research into new directions (Mokalled et al., 2016; Cavone et al., 2021; de Sena-Tomás et al., 2024; Shaw et al., 2024 preprint). In addition to stimulating immune and glial regeneration research, recent technological advances, including single-cell transcriptomics and imaging tools, have also invigorated studies of neuronal repair (Huang et al., 2021; Pedroni et al., 2024; Saraswathy et al., 2024).

Differential immune, glial and neuronal injury responses underlie disparate injury outcomes between highly and poorly regenerative vertebrates (Fig. 1). In non-regenerative mammals, additional cell types, such as fibroblasts and pericytes, exhibit exaggerated injury responses that impact extracellular matrix deposition and scarring outcomes (Dias et al., 2018; Saxena et al., 2019; Kolb et al., 2023). Mechanisms that differentially promote or inhibit repair have been extensively reviewed and will still be covered in this Review. However, one underappreciated feat of tissue regeneration is the ability of regenerative vertebrates to deploy mechanisms that are often conserved but overwhelmed by anti-regenerative complications in mammals. This Review aims to highlight how fine-tuned immune responses, pro-regenerative glial cell reactivity and multimodal neuronal repair can support spinal cord regeneration in the absence of intrinsic and extrinsic inhibition. Rather than portraying regenerative vertebrates as peculiar exceptions with superior regenerative competence, we present multiple lines of evidence that fundamental mechanisms of spinal cord regeneration are conserved and could be deployed to promote neural repair in mammals. This Review focuses on recent advances that have been made in zebrafish over the past decade and directs readers to reviews covering other regenerative species or prior zebrafish literature.

Fig. 1. Mechanisms underlying differential regenerative competence after SCI in vertebrates.

Top: Regenerative species, such as zebrafish and axolotls, exhibit transient immune responses, astroglial bridging and multiple modes of neuronal repair. Bottom: Mammals exhibit some pro-regenerative responses, such as acute immune cell infiltration, astrocytic wound healing and residual neuronal plasticity (inner circle), but these cellular processes are overshadowed by multicellular complications (highlighted in gray) that impede regeneration. Schematics depict lesioned spinal cord tissues in highly regenerative and less-regenerative vertebrates, with immune cells (blue), astroglial cells (magenta) and axons (green).

Regenerative capacity is unevenly distributed among vertebrates

Tissue regeneration is manifested in different mechanisms and efficiencies among vertebrates, including mammals. Relative to other tissues, the central nervous system (CNS) is particularly resistant to undergoing repair following injury or disease (Zheng and Tuszynski, 2023). While the general assumption is that an animal’s regenerative competence inversely correlates with its position and complexity in the phylogenetic tree, gains and losses of regenerative capacity do not adhere to a clear principle and are not confined to specific phylogenetic nodes in evolution (Fig. 2). In fact, nature presents only a few exceptions for which tissue regeneration bestows an unambiguous evolutionary advantage. For instance, adult lizards retain an evolved defense mechanism that allows them to self-amputate their tails and regenerate the entire structure, including spinal cord, following caudal autonomy (Alibardi, 2022). Beyond such rare examples of adaptive regeneration, not all fishes are capable of spinal cord repair, whereas some mammalian species can achieve scarless healing following SCI.

Fig. 2. Distribution of regenerative capacity among vertebrate models of SCI.

For each species, elevated (green) or limited (magenta) regenerative competence is shown throughout the lifespan. Regenerative capacity is generally elevated at younger life stages across the phylogenetic tree. However, unlike less-regenerative vertebrates, regenerative species, such as zebrafish, retain the ability to mend spinal cord lesions in adult stages. Frogs benefit from a temporally restricted regenerative period, although this ability is lost post-metamorphosis. Spiny mice represent an exceptional case among mammals, maintaining life-long scar-free tissue repair.

Among the closest living group to jawed vertebrates are the jawless lamprey fish, which undergo spontaneous repair following repeated spinal cord injuries (Hanslik et al., 2019). Benefiting rom a strategic phylogenetic position between vertebrates and invertebrates, lampreys have greatly advanced our understanding of intrinsic and extrinsic regeneration mechanisms (reviewed by Maxson Jones and Morgan, 2023). Noting that functional recovery is achieved despite imperfect cellular regeneration, lamprey research pioneered the notion of compensatory plasticity after SCI (Hibbard, 1963). Below, we highlight vertebrates such as zebrafish and axolotls as champions of spinal cord repair, the usefulness of studying the regenerative and non-regenerative stages in frogs, and the curious example of the spiny mouse as a mammalian model of successful spinal cord repair. Questions related to the evolution of spinal cord regenerative capacity are fascinating but not amenable to empirical studies. If and why regenerative competence was lost in some species will continue to captivate scientists for years to come.

Elevated regenerative competence in teleost fish and urodele amphibians

Teleosts deployed for SCI research included goldfish before zebrafish became a primary model for teleost regeneration (Zottoli et al., 1994; Becker et al., 1997; Rasmussen and Sagasti, 2017). Zebrafish (Danio rerio) are tropical freshwater fish and an excellent model of successful CNS regeneration. In addition to their elevated regenerative capacity, zebrafish SCI research is rapidly evolving due to their amenability to genetic manipulations. Fish SCI models include complete transection or crush injuries in adults, and mechanical transection or laser ablation in larvae (Becker et al., 1997; Hu et al., 2018; Bremer et al., 2019; Burris et al., 2021; Alper and Dorsky, 2022; Jensen et al., 2023). Benefiting from shorter experimental timelines, advanced imaging tools and amenability to electrophysiological studies, larval zebrafish research is widespread. However, limitations of the larval SCI models include an underdeveloped immune system, delayed microglial maturation and baseline expression of growth factors and regeneration-permissive molecules (Davidson and Zon, 2004; Mokalled et al., 2016; Hui et al., 2017). Although technically more challenging, adult zebrafish have a fully developed innate and adaptive immune system (Davidson and Zon, 2004; Stachura and Traver, 2011), exhibit pro-regenerative glial cell reactivity in the absence of developmental guidance cues (Reyes and Mokalled, 2024), and activate regenerative programs despite eliciting limited growth in adulthood compared to other teleosts (Biga and Goetz, 2006).

Similar to teleost fish, axolotls are urodele amphibians that stand out for their renowned ability to reverse the anatomical and functional deficits of SCI. Experimental systems include tail amputation, spinal cord transection or crush injuries (Tazaki et al., 2017; Thygesen et al., 2019; Walker et al., 2023). Axolotls remain in neoteny, do not undergo metamorphosis and do not develop adaptive immunity (Menger et al., 2010). Experimentally induced metamorphosis in axolotls reduces the rate and fidelity of regenerative responses but does not abrogate functional recovery, linking metamorphosis to lower regenerative capacity (Monaghan et al., 2014). Recent genetic advances in axolotls have enabled cell fate mapping. For instance, recent lineage tracing showed that Shh⁺ floor plate cells exclusively contribute to the regenerating floor plate, indicating lineage restriction during regeneration (Arbanas et al., 2024). Molecular studies revealed species-specific roles for the transcription factor Junbb and microRNA miR200a during innate spinal cord repair (Sabin et al., 2019; Walker et al., 2022). Tail and spinal cord regeneration in axolotls have been reviewed by Tazaki et al. (2017).

Regenerative competence is temporally restricted in anuran frogs

Tadpole regenerative capacity is restricted to a limited time window: after tail development is concluded and before the onset of feeding (Filoni and Bosco, 1981). As tadpoles, Xenopus laevis undergo epimorphic regeneration of multiple tissues, including spinal cord, following tail amputation (Cannata et al., 2001). The presence of regeneration-competent and -incompetent phases stimulated intra-species comparisons into the molecular drivers and inhibitors of regeneration. Moreover, inter-species comparative studies between X. laevis and Xenopus tropicalis were prompted by X. tropicalis showing a milder refractory period than X. laevis (Wang and Shi, 2021; Williams et al., 2021). Xenopus studies championed the hypothesis that cold-blooded animals express regenerative genes that are lost in warm-blooded vertebrates. Supporting this hypothesis, genes encoding thioredoxin domain-containing protein Ag1, small GTPases Ras-dva1/2 or cold-blooded animal-specific wound epithelium receptor protein C-answer were lost during the evolution of birds and mammals and are required for tail regeneration in tadpoles (Ivanova et al., 2013, 2018; Tereshina et al., 2014; Korotkova et al., 2019). Despite these drastic examples of complete gene loss, the prevailing model is that injury induces comparable gene sets across species and that nuanced gene regulatory changes alter the timing or specificity of gene expression to impact regeneration. Advances in Xenopus tail and spinal cord regeneration were recently reviewed by Phipps et al. (2020).

The spiny mouse is an exception to limited mammalian regeneration

Mouse species within the genus Acomys, also known as spiny mice, are members of the Muridae family, which includes the widely used experimental Mus and Rattus models. Spiny mice sparked research interests for their unique ability to sustain scar-free healing of the skin and ear (Seifert et al., 2012; Gawriluk et al., 2016; Matias Santos et al., 2016). Acomys were subsequently shown capable of robust regenerative competence after SCI (Streeter et al., 2020; Nogueira-Rodrigues et al., 2022). Whereas Mus axons arrest at the lesion border, Acomys axons are not deterred from entering the lesion after thoracic spinal cord transection. By 12 weeks, corticospinal axons show bilateral sprouting, collateral arborization and partial conduction of compound action potentials across the lesion (Nogueira-Rodrigues et al., 2022). Functionally, Acomys mice regain bladder function and establish some weight bearing and plantar support capacity (Nogueira-Rodrigues et al., 2022). Decreased inflammation, enhanced expression of stem cell and axon guidance genes and pro-regenerative astrocyte reactivity underlie enhanced cellular bridging and functional repair in Acomys compared to Mus (Streeter et al., 2020). Reduced fibrosis is a common characteristic of Acomys responses to various injuries (Saxena et al., 2019). In the spinal cord, Acomys deposit less collagen IV and heparan sulfate proteoglycan and increase deposition of keratin sulfate proteoglycans (KSPGs) (Streeter et al., 2020; Nogueira-Rodrigues et al., 2022). Supporting the hypothesis that increased KSPG biosynthesis provides a pro-regenerative matrix environment, expressing a key KSPG biosynthetic enzyme (β3gnt7) is sufficient to promote neurite outgrowth in vitro (Nogueira-Rodrigues et al., 2022). Why and how Acomys evolved to replace lesion scarring with a scarless tissue bridge is a fascinating observation and open question. Acomys spinal cord regeneration was recently reviewed by Maden and Varholick (2020).

Transient and finely tuned immune responses contribute to spinal cord repair

The first obstacle to regeneration is the requirement to re-enact developmental processes within the hostile environment of adult tissue lesions (Fig. 3). Shortly after mammalian SCI, reactive cell-derived or debris-derived molecules that inhibit regeneration overwhelm the extracellular milieu in and around the lesioned cord (Li et al., 2004; Cafferty et al., 2010; Lee et al., 2010b; Schwab and Strittmatter, 2014; Chio et al., 2022). Attempts to clear cellular debris begin with the phagocytic activity of lesion-adjacent microglia. In addition to their own phagocytic activity, activated microglia secrete pro-inflammatory cytokines that recruit infiltrating immune cells (Kigerl et al., 2009; Hakim et al., 2021). Distinct populations of innate and adaptive immune cells are sequentially recruited to the lesion (Popovich et al., 1996, 1997; Ankeny et al., 2006; Beck et al., 2010).

Fig. 3. Pro-regenerative immune, glial and neuronal responses postSCI in zebrafish.

Schematics represent longitudinal sections of spinal cord tissues in adult zebrafish. (A) In the absence of SCI, neurons and glia are organized around ependymal progenitor cells (gray) lining the central canal (white).

(B) In acute SCI, zebrafish activate a transient, pro-regenerative immune response characterized by the infiltration of immune cells, predominantly microglia and macrophages, into the lesion site. Lesion-adjacent neurons survive the injury and adopt a plasticity signature, while ependymal progenitors proliferate in response to injury.

(C) Chronic SCI is marked by neurogenesis, axonal regrowth and glial regeneration. Glial progenitors (magenta-gray) in the ventral domain undergo epithelial-to-mesenchymal transition (EMT) and contribute to astroglial bridging across the lesion. Neurogenic progenitors (green-gray) generate newborn neurons (green) and axons regrow across the lesion.

Immune cell activation plays dual roles after spinal cord injury

As the first line of defense against neural damage, the immune system dictates the cascade of regenerative events following SCI. After human SCI, immune activation initiates acutely but then persists for years following injury. This has been perceived as inefficient, exaggerated or both (Means and Anderson, 1983; Carlson et al., 1998; Fleming et al., 2006). Consistent with the emergence of overt anti-regenerative inflammation, treatment with anti-inflammatory drugs or depletion of microglia and macrophages improved functional recovery after mouse SCI (Popovich et al., 1999). Sustained, chronic inflammation is a chief contributor to secondary neurotoxicity and the recruitment of scar-depositing fibroblasts to the lesion (Dorrier et al., 2021). Conversely, targeted microglial and macrophage ablation in acute SCI revealed a requirement for early immune activation in mice (Shechter et al., 2009; Greenhalgh et al., 2018; Uderhardt et al., 2019; Brennan et al., 2022). These studies underscore the challenges of discerning specific immune functions in mammals, highlighting the need to improve our understanding of the temporal impact of specific immune populations on regenerative outcomes.

Transient and reversible activation of multiple immune populations in zebrafish

Regenerative vertebrates present a valuable model to untangle pro-regenerative immune functions in the absence of inflammation. The zebrafish immune system comprises innate myeloid cells and adaptive lymphoid cells that are conserved across vertebrates (Davidson and Zon, 2004; Stachura and Traver, 2011). The CNS is also privileged with tissue-resident microglia that elicit a phagocytic response after injury or disease (Wittamer et al., 2011; Nguyen-Chi et al., 2015; Krasemann et al., 2017; Villani et al., 2019). Spinal cord transection in adult zebrafish triggers transient and reversible immune activation, peaking 3 days post-injury (dpi) and significantly cleared by 14 dpi (Shaw et al., 2024 preprint). Neutrophils, blood-derived macrophages and tissue-resident microglia, followed by T cells, serially infiltrate the lesion (Shaw et al., 2024 preprint). Importantly, elevated leukocytes are known to persist in chronic SCI in mouse and humans, whereas immune SCI responses are transient in zebrafish (Gadani et al., 2015; Gillespie and Ruitenberg, 2022; de Sena-Tomás et al., 2024; Shaw et al., 2024 preprint). Establishing an immune requirement for spinal cord repair, blocking immune activation prolongs inflammation and reduces regeneration (Tsarouchas et al., 2018; de Sena-Tomás et al., 2024; Shaw et al., 2024 preprint). Below, we review the identities, activation profiles and roles of major immune subtypes after SCI.

Neutrophils

Neutrophils are the first immune cell population to infiltrate the lesioned spinal cord in regenerative and non-regenerative vertebrates (Taoka et al., 1997; Tsarouchas et al., 2018; de Sena-Tomás et al., 2024; Shaw et al., 2024 preprint). Neutrophils phagocytose cellular debris, recruit microglia and macrophages, and tune inflammatory events in spinal cord lesions (Neirinckx et al., 2014). In zebrafish larvae subjected to SCI, neutrophil infiltration peaks 4–6 h post-injury and is resolved to pre-injury levels by 72 h (de Sena-Tomás et al., 2024). Using transgenic lines in which neutrophils express the photoconvertible protein Kaede, lesion-associated neutrophils were shown to reverse migrate from the lesion to distal anatomical locations without clear organ preference (de Sena-Tomás et al., 2024). The regenerative functions of neutrophils require Cxcr1/2 activation (de Sena-Tomás et al., 2024). Intriguingly, genetic manipulations of rac2, which alters neutrophil mobility and blocks their recruitment to the lesion, had no impact on cellular or functional regeneration, neutrophil clearance or the recruitment of microglia and macrophages (de Sena-Tomás et al., 2024). Thus, spinal cord regeneration can occur in the absence of neutrophils, ostensibly via compensatory immune functions from other cell populations (de Sena-Tomás et al., 2024). Conversely, accelerating neutrophil clearance by administering a Cxcr4 antagonist accelerated progenitor cell proliferation, motor neuron regeneration and cellular bridging across the lesion (de Sena-Tomás et al., 2024). Molecularly, Cxcr4 inhibition induced pro-inflammatory cytokines such as il-1b and tnfa, and activated pro-inflammatory tnfa⁺ mpeg⁺ microglia/macrophages, which were shown to promote progenitor cell proliferation and regeneration (Tsarouchas et al., 2018; de Sena-Tomás et al., 2024).

Microglia and macrophages

Supporting a central regenerative role for macrophages after SCI, irf8 mutant zebrafish larvae lack macrophages and exhibit regeneration defects reminiscent of broad immune inhibition by dexamethasone (Tsarouchas et al., 2018). A subset of lesion-adjacent macrophages express tnfa, promoting progenitor cell proliferation and neuronal differentiation via AP1/Hdac1 signaling (Cavone et al., 2021). By contrast, colony-stimulating factor 1 receptor (csf1ra/b) mutant zebrafish larvae, in which microglial differentiation is impaired, did not exhibit altered axon regeneration (Tsarouchas et al., 2018). These results established a requirement for macrophages after SCI, and suggested microglia are dispensable for larval spinal cord regeneration. However, csf1ra/b mutants show increased macrophage numbers (Tsarouchas et al., 2018), which could compensate for the loss of microglia in this model. Moreover, developmental migration of microglia into the CNS and lymphocyte maturation extend for weeks after zebrafish larval development (Willett et al., 1997; Davidson and Zon, 2004; Xu et al., 2015; Hui et al., 2017). This timeline of immune cell development underscores the need to validate immune cell requirement and functions in adult animals. In adult zebrafish, depletion of mpeg1.1⁺ microglia and macrophages between 1 and 14 dpi diminished cellular bridging and the recovery of swim function (Shaw et al., 2024 preprint). Notably, allowing mpeg1.1⁺ cells to repopulate the spinal cord at 14 dpi was not sufficient to direct regeneration between 14 and 28 dpi. These results indicated that mpeg1.1⁺ microglia and macrophages are acutely required for adult spinal cord repair.

Regulatory T cells

Regulatory T cells (T_regs) comprise a subset of CD4⁺ T cells, mediate antigen recognition and play important roles in immune suppression (Workman et al., 2009; Josefowicz et al., 2012). After injury, T_regs exert regenerative functions by repressing proinflammatory chemokine expression and dampening innate and adaptive inflammation (reviewed by Gupta et al., 2021). Using a forkhead box P3a (foxp3a)-driven zebrafish reporter line, T_regs were shown to emerge in the juvenile zebrafish thymus (Hui et al., 2017). In adult zebrafish, peripheral and lesion-associated foxp3a⁺ T_regs peak at 7 dpi in close association with proliferating progenitors and newly born neurons (Hui et al., 2017). Molecularly, spinal T_regs express general T_reg activation genes such as nr4a1 and amphiregulin, in addition to CNS-specific growth factors, including neurotrophin 3 (ntf3) (Hui et al., 2017). Conditional ablation of foxp3a⁺ T_regs results in impaired cellular bridging, disorganized rostral and caudal axon sprouting and diminished swim recovery (Hui et al., 2017). Establishing T_reg-derived Ntf3 as a crucial regulator of progenitor cell proliferation, administration of recombinant human NTF3 was sufficient to rescue the number of Sox2⁺ Pcna⁺ cells in T_reg-depleted animals at 7 dpi (Hui et al., 2017). These findings show the pivotal role played by adaptive immune cells in progenitor cell activation and spinal cord regeneration.

Efficient phagocytosis and lack of chronic inflammation

Adult zebrafish achieve efficient debris clearance and innate repair in the absence of chronic inflammation (Shaw et al., 2024 preprint). Cross-species comparisons between zebrafish and mammals support a model in which zebrafish immune cells elicit superior phagocytic capacity, and identified transcription and immune response regulator (tcim) as a macrophage-enriched zebrafish gene (Shaw et al., 2024 preprint). Genetic deletion of zebrafish tcima and tcimb impedes phagocytic capacity and regeneration, reprogramming leukocytes into a pro-inflammatory transcriptional state similar to that of mice. Conversely, genetic overexpression of human TCIM accelerates debris clearance and regeneration by reprogramming myeloid precursors into activated phagocytes (Shaw et al., 2024 preprint). Manipulations that impact phagocytosis, such as tcim mutagenesis, often leads to excessive accumulation of lipid droplets (Shaw et al., 2024 preprint). These zebrafish observations link lipid processing to phagocytic and regenerative capacities. Similar to zebrafish mutations that impair phagocytosis, murine lipid-laden myeloid cells that fail to process and recycle lipids transition into a hyper-inflammatory, foamy phenotype (Wang et al., 2015). Conversely, lipid breakdown and fatty acid oxidation enhance macrophage differentiation, phagocytosis and microglial neuroprotection in mice (Jacquel et al., 2012; Zhang et al., 2012; Leng et al., 2022). These studies and others emphasize the value of deploying highly regenerative vertebrates to dissect and promote neuroprotective immune mechanisms after SCI. We propose that regenerative vertebrates will inspire approaches to accelerate debris clearance in acute SCI, and to promote immune quiescence in sub-acute to chronic SCI in humans.

Pro-regenerative glial cell reactivity directs scarless spinal cord repair

As the largest and one of the most abundant cell types in the CNS, astrocytes play crucial roles in the development and maintenance of neuronal circuits (Takano et al., 2006; Attwell et al., 2010; Paixão and Klein, 2010; Freeman and Rowitch, 2013; Blanco-Suárez et al., 2017). Astrocytes are highly sensitive to CNS perturbations, exhibiting a spectrum of reactivity states and functional changes, referred to as ‘reactive astrogliosis’ or ‘astrocyte reactivity’ (Zamanian et al., 2012). Although glial cell injury responses have been extensively studied in mammals, the complexities of these responses in mammals and our understanding of how astrocytes respond to SCI in highly regenerative vertebrates are relatively limited.

Reactive astrocytes play dual roles after mammalian spinal cord injury

Mammalian SCI is marked by scarring. The ‘glial scar’ is a historically coined term that depicts the detrimental effects of reactive astrocytes on neuronal survival and repair (Ramón y Cajal et al., 1991; Sofroniew, 2018; Yang et al., 2020). However, as spinal lesions attract multiple cell types in addition to glial cells, the SCI scar has been recently categorized into a fibrotic core bordered by a glial compartment. The fibrotic scar consists of infiltrating immune cells, fibroblasts and pericytes (Göritz et al., 2011; Soderblom et al., 2013; Zhu et al., 2015), whereas the surrounding glial scar comprises astrocytes, microglia and oligodendrocyte progenitor cells (Anderson et al., 2016; Hara et al., 2017; Wanner et al., 2013). Concomitantly, the concept that the glial scar is both detrimental and beneficial for SCI repair has emerged (Pekny et al., 2014; Sofroniew, 2018; Tran et al., 2022; Shafqat et al., 2023). Highly regenerative vertebrates exhibit pro-regenerative glial cell reactivity with minimal detrimental effects, which presents a unique opportunity to dissect the transcriptional, morphological and functional features of pro-regenerative astrocyte reactivity.

Astrocyte reactivity is a conserved defense mechanism across vertebrates

Under physiological conditions, astrocytes provide trophic and metabolic support for neurons, regulate the development and maintenance of synapses, and control spinal blood flow, fluid availability and ion homeostasis in the CNS (Takano et al., 2006; Attwell et al., 2010; Paixão and Klein, 2010; Blanco-Suárez et al., 2017). A conserved, central role for astrocytes is to protect the CNS against external injuries. In acute SCI, astrocytes accumulate around the lesion in zebrafish and mammals (Auguste et al., 2007; Dawley et al., 2012; Goldshmit et al., 2012). In mammals, reactive astrocytes border the fibrotic scar, limit the expansion of inflammation and protect spared neural tissues (Yuan and He, 2013; Sofroniew, 2018). In addition to their wound-healing functions, astrocytes help restore tight junctions and repair the blood–spinal cord barrier (Pitter et al., 2014). Astrocytes clear excess glutamate and were reported to phagocytose dead cells in vitro and in vivo (Lööv et al., 2012; Morizawa et al., 2017). Consequently, depleting reactive astrocytes in mice leads to enlarged lesions, increased inflammation, barrier failure and worsened injury outcomes (Faulkner et al., 2004; Myer et al., 2006; Okada et al., 2006; Sofroniew, 2009).

Reactive astrocytes have detrimental effects in mammals

Despite their acute beneficial roles, mammalian astrocytes form a physical and molecular barrier that obstructs neural repair at later stages (Tran et al., 2022). They increase the expression of anti-regenerative molecules, including intermediate filaments, cytokines, toxic amino acids and extracellular matrix molecules (Silver and Miller, 2004; Massey et al., 2006, 2008; Galtrey and Fawcett, 2007; Andrews et al., 2012; Hara et al., 2017). These molecules trigger neurotoxicity in addition to amplifying cyclical reactivity in surrounding astrocytes (Li et al., 2020). One of the most prominent astrocyte-derived inhibitory molecules is chondroitin sulfate proteoglycan (CSPG), which inhibits axon regrowth, neurogenesis, oligodendrocyte progenitor cell migration, and remyelination (Smith-Thomas et al., 1995; Bradbury et al., 2002; Barkho et al., 2006; Lee et al., 2010a; Siebert and Osterhout, 2011; Siebert et al., 2011; Wang et al., 2011; Hammond et al., 2014). Thus, chronic astrogliosis generates a microenvironment that impedes neuronal repair, hinders remyelination and curbs regeneration.

Bipolar astrocyte morphology correlates with pro-regenerative effects

Bipolar radial glial morphology correlates with pro-regenerative effects across vertebrates. In mammals, astrocytes with bipolar features support neuronal migration and axonal growth in the developing CNS (Joosten and Gribnau, 1989; Vaccarino et al., 2007). After SCI, elongated astrocytes are associated with regeneration-permissive functions in regenerative and non-regenerative species (White et al., 2008; Filous et al., 2010; Goldshmit et al., 2012; Zukor et al., 2013; Mokalled et al., 2016; Shen et al., 2025). Early observations of glial bridging in regenerative species were made in lampreys in 1959. At the time, K. Maron noted the formation of a glial-derived bridge of ependymal cells at the lesion site starting at 5 dpi (Maron, 1959). Subsequent zebrafish studies established glial cells as a central component of cellular regrowth across the lesion (Goldshmit et al., 2012; Mokalled et al., 2016). In 2012, Goldshmit et al. characterized bipolar glial cells that express the general astrocyte marker Gfap in the regenerating tissue at the lesion (Goldshmit et al., 2012). These Gfap⁺ bridging glia were later identified as astroglial cells that connect the transected spinal cord (Klatt Shaw et al., 2021) (Fig. 3). Genetic lineage tracing and cell ablation showed that bridging astroglia emerge from a confined niche of glial progenitors and are required for functional recovery (Zhou et al., 2023). Intriguingly, glial bridging is required for axon regrowth in adult zebrafish and older larvae but not in younger larvae at 3 days post-fertilization (Wehner et al., 2017; Walker et al., 2025), suggesting the existence of developmental axon guidance cues that can substitute for bridging in larval zebrafish.

Manipulations that either induce or correlate with the emergence of bipolar astrocytes have been shown to improve axon regrowth and regeneration across species (White et al., 2008; Goldshmit et al., 2012; Zukor et al., 2013; Mokalled et al., 2016; Shen et al., 2025). Mouse studies have implicated Pten, TGFα and Wnt signaling in pro-regenerative astrocyte reactivity (White et al., 2008; Zukor et al., 2013; Shen et al., 2025). Advances in transcriptional and mutagenesis tools in zebrafish revealed a gamut of additional factors, including fgf, ctgf, yap1/taz (wwtr1), bach1, egr1 and junb (Goldshmit et al., 2012; Mokalled et al., 2016; Klatt Shaw and Mokalled, 2021; Klatt Shaw et al., 2021). Here, we discuss the roles of some of these pathways.

Fgf

Fibroblast growth factor (Fgf ) signaling is a pro-regenerative mechanism in the glia of zebrafish and mice (Goldshmit et al., 2012; Kasai et al., 2014). In zebrafish, multiple components of Fgf signaling, including fgf8a, fgf3, spry4, pea3 (etv4) and erm (etv5b), are upregulated in Gfap⁺ glial progenitors at 3 dpi (Goldshmit et al., 2012). By 14 dpi, fgf3 and spry4 transcripts become restricted to neurons, whereas fgf8a, pea3 and erm maintain their glial expression (Reimer et al., 2009; Goldshmit et al., 2012). The glial bridging functions of Fgf were established using a small molecule inhibitor of the pathway, dominant-negative Fgf receptor manipulations, sprouty mutants and recombinant Fgf8 (Goldshmit et al., 2012). In these experiments, pharmacological and genetic loss-of-function approaches reduced the proliferation and migration of Gfap⁺ cells into the lesion and compromised axonal bridging. Conversely, Fgf activation using spry4 mutants and intraperitoneal Fgf8 injections increased glial and axonal bridging at the lesion (Goldshmit et al., 2012). Following mouse SCI, exogenous application of Fgf2 lowered the levels of TNFα and CSPG, decreased inflammation and improved motor function (Goldshmit et al., 2014). Importantly, mouse astrocytes treated with Fgf2 adopt a bipolar morphology that is thought to facilitate axonal growth (Goldshmit et al., 2014). These studies correlate the emergence of bipolar astroglia cells with pro-regenerative effects across vertebrates.

Ctgf

Connective tissue growth factor (Ctgf), also known as Ccn2, is a matricellular molecule that elicits a range of cellular responses, including cell proliferation and differentiation (Grotendorst and Duncan, 2005). Zebrafish studies identified ctgfa (ccn2a) as a central glial bridging factor and described the emergence of ctgfa⁺ gfap⁺ bridging glia after SCI (Mokalled et al., 2016; Zhou et al., 2023). ctgfa transcripts are upregulated in glial progenitors and the nascent cellular bridge in the regenerating spinal cord (Mokalled et al., 2016). The glial requirements of ctgf during bridging and of glial bridging during spinal cord regeneration were established using genetic mutants and cell-ablation tools (Mokalled et al., 2016; Zhou et al., 2023). ctgfa mutants exhibit reduced swim endurance and fail to extend glial or axonal projections through the lesion (Mokalled et al., 2016). Similar phenotypes were observed following depletion of ctgfa⁺ cells (Zhou et al., 2023). Conversely, genetic and pharmacological Ctgf gain of function enhance glial bridging and functional repair. Intriguingly, murine CYR61, also known as CCN1, was recently implicated in the reactivity of lesion-remote astrocytes after thoracic hemisection (McCallum et al., 2024 preprint). These studies suggest that CCN family members, including CTGF and CYR61, have conserved pro-regenerative roles in glial cell reactivity.

Epithelial-to-mesenchymal transition

Transcriptional and genetic evidence indicate that epithelial-to-mesenchymal transition (EMT) reprograms glial progenitors into pro-regenerative glia (Klatt Shaw et al., 2021; Zhou et al., 2023). EMT is often linked to stem cell activation and tissue plasticity after injury (Jessen and Arthur-Farraj, 2019; Wilson et al., 2020). gfap⁺ ctgfa⁺ bridging glia progenitors displaying hallmarks of EMT, including twist1a expression and Yap1 activation, express mesenchymal markers and downregulate epithelial markers (Klatt Shaw et al., 2021). Phenocopying Fgf and Ctgfa loss of function, CRISPR mutagenesis showed that egr1, junbb and yap1/taz are required for cellular bridging and functional recovery (Klatt Shaw et al., 2021). Furthermore, Twist1 overexpression promotes glial bridging and functional recovery (Klatt Shaw et al., 2021). As EMT gene expression is more pronounced in zebrafish astroglia compared to mammalian astrocytes (Klatt Shaw et al., 2021), whether EMT can confer pro-regenerative functions to mammalian astrocytes remains to be tested.

As disparate as glial cell reactivity may appear between regenerative and non-regenerative vertebrates, the studies reviewed here provide multiple lines of evidence supporting a conserved neuroprotective role for astroglial cells after SCI. We propose that acute glial cell reactivity is an ancestral regenerative function across vertebrates, but that chronic astrogliosis is specific to mammalian SCI. While mammalian research has predominantly focused on dampening the detrimental effects of astrogliosis, regenerative models such as the zebrafish could inform complementary strategies to promote beneficial glial cell reactivity and wound healing.

Intersecting modes of neuronal plasticity and regeneration enable neural repair

Lesioned spinal neurons encounter a multitude of environmental and molecular challenges that are context, injury and species dependent. As discussed above, extrinsic immunoglial effects can be anti- or pro-regenerative. In addition to integrating extrinsic cues, neurons undergo intrinsic molecular changes that impact their ability to regenerate. Recent single-cell transcriptomics identified a small population of injury-surviving spinal neurons that expressed regeneration-associated genes in the mouse spinal cord and were linked to spontaneous plasticity (Matson et al., 2022). Despite this signature of spontaneous plasticity, neurotoxicity and intrinsic inhibition of axon regeneration prevail in less-regenerative vertebrates. These inhibitory effects were recently reviewed by Zheng and Tuszynski (2023). Intriguingly, lesion-adjacent zebrafish neurons that activate regeneration-associated gene expression in acute SCI are associated with increased survival and plasticity (Saraswathy et al., 2024). Innate neuronal repair in regenerative vertebrates is achieved by cooperate survival, plasticity and regeneration mechanisms (Saraswathy et al., 2024) (Fig. 3). Below, we focus our discussion on the multiple modes by which neurons meet the challenges of SCI in zebrafish.

Neuronal plasticity and survival

Plasticity refers to spontaneous reorganization of spared neural circuitry via axon sprouting, not to be confused with regeneration in which a transected axon regrows to its original target site. Neuroplasticity offers a promising therapeutic avenue that does not require de novo cellular or axonal regeneration and is spontaneously detected after mammalian SCI (Tendolkar and Mokalled, 2026). Zebrafish studies have revealed that plasticity precedes cellular regeneration and is required for functional recovery following SCI (Vandestadt et al., 2021; Pedroni et al., 2024; Saraswathy et al., 2024). Similar to lampreys, zebrafish achieve significant functional recovery before axon regrowth and neurogenesis are completed (Hibbard, 1963; Reimer et al., 2008; Kuscha et al., 2012; Pedroni et al., 2024; Saraswathy et al., 2024). In larval zebrafish, caudal calcium propagation and muscle function are restored as early as 2 dpi, when the majority of lesion-associated neurons are not newborn (Vandestadt et al., 2021). Similarly, in adult zebrafish, 5-ethynyl-2’-deoxyuridine (EdU) and 5′-bromo-2′-deoxyuridine incorporation assays showed little neurogenesis in isl1⁺ or hb9 (mnx1)⁺ neurons during the first 7 days after injury, despite some recovery in swim function (Reimer et al., 2008; Vandestadt et al., 2021). Imaging of neuronally expressed elavl3: GCaMP6s along with EdU labeling indicated that initial circuit repair relies on the recruitment of pre-existing post-mitotic neurons, rather than neurogenesis (Vandestadt et al., 2021). These findings were supported by recent single-cell transcriptomics revealing a transient population of injury-responsive neurons (iNeurons) that acquire a neuroblast-like transcriptional signature between 3 and 7 dpi (Saraswathy et al., 2024). Similar to the early migrating neurons described by Vandestadt et al. (2021), the majority of iNeurons were morphologically mature and did not incorporate EdU after SCI. Ruling out the notion that mature neurons dedifferentiate after SCI, Vandestadt et al. (2021) showed that early migrating neurons did not express the neural stem cell markers nestin or sox2. However, in-depth molecular characterization by single-cell RNA sequencing supported a model in which mature neurons acquire an immature neuroblast-like cell identity, even if they do not express sox2. By tracking transgenic reporter lines for motor neurons and interneurons, multiple neuronal subtypes were thought to survive SCI and elicited an iNeuron transcriptional signature (Vandestadt et al., 2021; Saraswathy et al., 2024). These studies support a model in which injury-surviving neurons revert to a neuroblast-like stage and are immediate responders to SCI in larval and adult zebrafish.

iNeurons express known regeneration-associated genes during acute to sub-acute SCI (Saraswathy et al., 2024). Using CellChat analysis, cell–cell signaling outgoing from iNeurons and received by other cell types is predicted to account for the majority of all neuronal signaling at 7 dpi. Independent lines of evidence corroborated vesicular trafficking and endocytosis in acute neuronal plasticity after SCI (Vandestadt et al., 2021; Hilton et al., 2022; Saraswathy et al., 2024). Applying inhibitors of clathrin-dependent and -independent endocytosis impaired neuron recruitment to the lesion, establishing a requirement for endocytosis in neuronal plasticity (Vandestadt et al., 2021). Moreover, CRISPR mutagenesis showed that synaptotagmin XI (syt11a/b) and vesicle-associated membrane protein 4 (vamp4) are crucial for functional recovery despite being dispensable for cellular regeneration across the lesion. Syt11 is a non-canonical SNARE that inhibits spontaneous neurotransmission and bulk endocytosis, whereas Vamp4 promotes calcium-dependent neurotransmitter release and bulk endocytosis. These findings are consistent with recent mammalian studies showing that inhibition of the presynaptic release machinery is important for axon regrowth (Hilton et al., 2022). Together, these studies highlight endocytic vesicular trafficking as a putatively essential mechanism that underlies spontaneous plasticity and functional recovery after SCI.

Insights into the mechanisms that activate neuronal survival and plasticity after SCI are only beginning to emerge. Despite experiencing a transient surge in glutamatergic drive, which results in glutamate toxicity in mammals, zebrafish motor neurons employ calcium-binding proteins to buffer excess calcium and tolerate SCI damage (Pedroni et al., 2024). Improved calcium buffering is concomitant with a transient glutamate-mediated upregulation in gap junctional connexins (Cx35 and Cx36) and an increase in bidirectional electrical coupling between motor neurons (Pedroni et al., 2024). A model has been proposed in which double-stranded RNA (dsRNA) molecules secreted by damaged cells activate neuronal toll-like receptors (TLRs), and that dsRNA-TLR activation is an early injury response that stimulates neuronal plasticity (Vandestadt et al., 2021). Supporting this model, treating zebrafish larvae with a global RNase or dsRNA-specific RNaseIII, but not DNases, inhibited the recruitment of isl1⁺ neurons into the lesion and impaired functional recovery (Vandestadt et al., 2021). Similar phenotypes were obtained upon pharmacological or genetic loss of function of TLR3 or TLR22. Conversely, the application of the dsRNA-mimicking ligand and TLR3 activator Poly(I:C) increased isl1⁺ neuron migration and functional recovery, without impacting neurogenesis. Importantly, Poly(I:C) treatment was sufficient to rescue RNaseA-induced phenotypes, showing that TLR activation is a major driver of neuronal plasticity and SCI repair (Vandestadt et al., 2021).

Thus, innate spinal cord repair requires more than axon regrowth and de novo neurogenesis to achieve functional recovery. Recent advances in transcriptomic, electrophysiological and imaging tools are finally providing molecular clues regarding why or how regenerative species can regain function prior to cellular regeneration. The discovery of an acute neuroplasticity signature during innate spinal cord repair unlocks new prospects to leverage the zebrafish model for neuronal plasticity research, in addition to regeneration.

Neuronal and axonal regeneration

Neuronal regeneration and adult neurogenesis are similar processes involving de novo regeneration of neuronal cells from adult stem cells. In regenerative vertebrates, potent adult progenitors retaining radial glial features replenish lesion-associated neurons (Reimer et al., 2008; Briona et al., 2015; Ribeiro et al., 2017; Donato and Vickaryous, 2022). Newly differentiated neurons populate the regenerating tissue, as pre-existing neurons regrow axons across lesioned tissues (Reimer et al., 2008, 2013; Barreiro-Iglesias et al., 2015). Single-cell transcriptomics has shown that tightly regulated waves of excitatory neuron regeneration precede slower, continuous neurogenesis of inhibitory neurons (Saraswathy et al., 2024). Consequently, excitatory-inhibitory (E/I) ratios are marked by early imbalance towards an excitatory phenotype at 7 dpi, and restoration of baseline E/I activity by 42 dpi (Saraswathy et al., 2024). As altered E/I transmission leads to severe behavioral deficiencies in humans (Isaacson and Scanziani, 2011), how zebrafish progenitors ensure faithful and balanced neurogenesis over the extended timeline of spinal cord regeneration remains unknown.

Axon regeneration refers to axonal regrowth from injured neurons. Mammalian axon regeneration is challenged by intrinsic neuron inhibition (Goldberg et al., 2002; Filbin, 2003; Park et al., 2008) in addition to the extrinsic immune and glial factors discussed above. Conversely, early axon-tracing experiments in adult zebrafish showed that brainstem neurons are capable of long-distance axon regrowth, extending over 3.5 mm beyond the lesion site (Becker et al., 1997). Although brain-derived projections show slight differences in their regrowth capacities, both dopaminergic and serotonergic axon terminals are spontaneously established (Kuscha et al., 2012). Even reticulospinal Mauthner axons, which were thought to represent a unique example of failed regeneration in zebrafish, were subsequently found to be capable of regeneration following two-photon laser-mediated transection (Becker et al., 1997; Hu et al., 2018). A requirement for axon regrowth during innate spinal cord repair was uncovered by creating a physical barrier to axon regrowth at the lesion site and through transection of regenerating axons after complete regeneration (Becker et al., 2004; Kuscha et al., 2012). As neurogenesis and axonal regeneration have been appreciated for decades, we direct readers to recent reviews covering the classic literature related to these topics (Ghosh and Hui, 2018; Tsata and Wehner, 2021).

In more recent studies, transiently overproduced serotonergic neurons were shown to correlate with the regrowth of long spinal projections and functional circuit recovery (Kuscha et al., 2012; Huang et al., 2021). As serotonergic input is both brain and spinal derived, dual spinal cord transections were used to study spinal serotonergic effects independently of brain-derived projections. Depletion of tph2-expressing serotonergic neurons in this two-cut SCI model showed that lesion-associated serotonergic interneurons are required for axon regrowth and functional recovery (Huang et al., 2021). Serotonergic neurons proximal to the lesion have distinct electrophysiological and transcriptional signatures compared to their lesion-remote counterparts (Huang et al., 2021; Saraswathy et al., 2024). Mechanistically, Huang et al. proposed a model in which expression of the serotonergic receptor htr1b in glutamatergic excitatory neurons may facilitate glutamatergic axon regrowth in response to localized serotonergic signaling. In support of this model, glutamatergic axon regrowth and functional recovery are compromised in htr1b mutants (Huang et al., 2021).

Together, these studies uncover an orchestrated series of regenerative events that direct innate neuronal repair following SCI. Crucial to this process is the cooperative regulation of neuronal plasticity, axon regrowth and neurogenesis. We expect future studies will reveal intricate multidimensional regulatory mechanisms that underlie these processes at the cellular, tissue and temporal levels.

Concluding remarks

Despite recent advances highlighted in this Review and beyond, the regeneration field holds significantly more open questions than answers. Reflecting current interests in the field, our Review focuses on acute mechanisms of immune activation, astroglial wound healing and neuronal plasticity. However, our understanding of the signaling mechanisms that guide regenerative patterning, remyelination and functional recovery remain limited. When manifested, SCI repair is largely stem cell derived (Reimer et al., 2008; Donato and Vickaryous, 2022; Maxson Jones and Morgan, 2023; Arbanas et al., 2024). While lineage-tracing experiments in zebrafish and axolotls are beginning to unravel how specific niches of progenitor cells contribute to spontaneous regeneration (Zhou et al., 2023; Arbanas et al., 2024), comprehensive dissection of stem cell niches and the mechanisms that retain stem cell potency in adult animals remain unknown. Finally, the evolutionary principles of tissue regeneration and how observations in regeneration-competent animals would translate to mammals continue to attract regeneration researchers into the field. We propose that a first step to sufficiently address these overarching questions is to acknowledge that regenerative capacity depicts ancestral injury responses that are conserved across species. Rather than picturing regenerative capacity in magic terms that may seem unattainable in mammals, in-depth studies of similarities and differences are more likely to yield a realistic understanding of regenerative mechanisms that are either masked or missing in mammals. We expect this approach, along with continued tool development and cross-species comparisons, will provide the roadmap to understanding and promoting spinal cord repair.