Radial glial cells (RGCs) were initially described around the mid-late 1800s in several mammalian species by Bentivoglio and Mazzarello (1999). Using the Golgi impregnation technique, these and other investigators reported that RGCs were found in the developing nervous system (NS) at different embryonic and neonatal stages (Figure 1). Ever since then, RGCs have been associated with: (1) characteristic morphological features, (2) their occurrence in nervous tissue, specifically in the central nervous system (CNS), and (3) their transient presence during embryonic development. Over one century later, multiple studies have confirmed the original findings and have extended them to incorporate the molecular basis of cellular phenotypes and their functions. RGCs, or cells with similar properties termed radial glia-like cells (RGLCs), have been documented in adults of various species, where functions, additional to their embryological role, have been described. In this review, we focus on the presence of RGLC’s in adult organisms and their roles in NS regeneration, we highlight the work of our group in studying the RGLCs in echinoderms and their prominent role during regeneration of the radial nerve cord (RNC).

Nervous system regeneration, or the lack of it, remains one of the least understood and most diverse areas of study in the field of regeneration biology. Many species can regenerate their NS components, though the cellular and molecular machinery they employ to attain their regeneration capacities can differ. Hence, understanding the fundamental biological processes by which regeneration-competent species achieve regeneration has been the field’s holy grail for quite some time. Given the diverse nature of regeneration, understanding these processes will greatly benefit from a comparative/evolutionary analysis using a highly diverse subset of species.

NS regeneration is defined as the process by which damaged nervous tissue undergoes regrowth or renewal, restoring the system’s morphology and physiological functions. At the cellular level, NS regeneration can be divided into two independent processes: axonogenesis and neurogenesis. Axonogenesis is the process by which the axon of a pre-existing neuron regrows after suffering an injury, re-establishing the original connection. In vertebrates, axon regeneration is commonly seen in the peripheral nervous system (PNS); however, the same phenomenon is highly limited or fails to occur in the CNS. Axonogenesis does not fall within the scope of this review. For those interested in axonogenesis, we recommend some publications on the topic (Chisholm et al., 2016; Wahane et al., 2019).

Our focus is mainly on injury-induced neurogenesis and the mechanisms by which adult organisms activate neurogenesis in response to injury. Similar to the processes of cellular homeostasis or regeneration in other organs or tissues, new nerve cells originate from: (1) a reservoir of stem cells that can proliferate and differentiate to maintain tissue physiology or restore and repair an injury site or (2) by dedifferentiation, where a cell population reverts to a less-differentiated state by losing some of their differentiated properties (Tanaka, 2016; Yao and Wang, 2020; Walker and Echeverri, 2022). These cells can eventually divide and produce replacement cells for the organ or tissue in question. (3) A third mechanism has been proposed where tissues can generate new cells via trans-differentiation. In this case, similar to dedifferentiation, the cells lose their specific phenotypic characteristics and then acquire the phenotype of a different cell type without transiting through a “less differentiated stage” (Walker and Echeverri, 2022).

Regeneration is a dynamic process that can vary significantly from one species to another. A common mechanistic link to achieve NS regeneration has not been elucidated and, in fact, there might not be one common mechanism, but mechanisms might differ according to the animal groups. For example, protostomes with extensive regenerating abilities, such as Hydra or Planaria, are known to have pluripotent stem cells that can give rise to the NS cells and other cell types (Cebrià et al., 2002; Agata and Umesono, 2008). Crayfish, a well-studied crustacean, possess neurogenic niches with primary precursor neural progenitors (Sullivan and Beltz, 2005; Ventura et al., 2019). These cells can divide into secondary precursors that migrate to the olfactory structures and later differentiate into olfactory interneurons or projection neurons. The primary precursors cannot self-renew and depend on a re-supply of circulating hemocytes to replenish the precursor cell stock (Chaves da Silva et al., 2013; Benton et al., 2014). A comprehensive review on crustacean and insect adult neurogenesis and their regenerative mechanism can be found in Simões and Rhiner (2017).

In most vertebrates, CNS neuroregenerative capacities have been associated with the presence of RGLCs. These cells are well-known to repopulate and replenish injured tissues. Thus, organisms that retain RGLCs in their adult stages are associated with higher CNS regenerative abilities (Alvarez-Buylla et al., 1987; Malatesta et al., 2000; Malatesta and Götz, 2013; Hartfuss et al., 2001; Jurisch-Yaksi et al., 2020). For example, RGCs in rodents normally disappear soon after birth. In neonatal rodents whose brains have been injured, RGCs persist for a more extended period, suggesting that after neonatal brain injury, RGCs serve as scaffolds for neuroblasts to migrate and populate the lesioned site (Jinnou et al., 2018). Several studies have addressed the concept of proliferation and neurogenesis in adult rodents’ SVZ and SGZ after inducing focal ischemia or traumatic brain injury (Kernie and Parent, 2010). Following these injuries, progenitor cells on both niches respond by proliferating. Whether the cells generated by RGLC survive, migrate, and incorporate into CNS circuits remain controversial. Studies have determined that, in mammals, most of the cells generated in response to cerebral cortex injury do not survive and those that do are unable to successfully migrate to the lession site and differentiate (Chang et al., 2016). However, cell extrinsic factors have been shown to increase cell survivability and/or migration. For example, SVZ cell exposure to fibroblast growth factor (FGF2) increased cell survivability while inhibition of ADAM17 metalloprotease increased not only survival but also migration to the injured primary motor cortex (Chang et al., 2016; Geribaldi-Doldán et al., 2018).

While these studies expose the neurogenetic response to injury of adult RGLCs and their potential for facilitating CNS repair, these capacities in mammals are highly limited. Thus, the reason to focus on alternative model systems where the role of RGLCs in CNS regeneration can be explored extensively. Here we provide information on three of them, two vertebrate animals, the axolotl, the zebrafish, and an invertebrate, the sea cucumber.

The spinal cord is a component of the CNS whose regenerative capacities have been intensively probed. While the formation of a glial scar hinders spinal cord regeneration in mammals, in some amphibians and fishes, morphological and functional recovery is possible. In the two best studied species, axolotl and zebrafish, new neurons are formed by so called “ependymal cells” surrounding the central spinal canal (Chernoff et al., 2003; Reimer et al., 2008). These ependymal cells clearly fulfill our description of RGLCs. First, they are derived from RGCs during development, Second, they retain their radial morphology and third they express some of the RG markers (GFAP and Sox2) (Hui et al., 2015b). In addition, several markers associated with RGCs are expressed by the proliferating cells following spinal cord injury (Hui et al., 2015a). After injury, these cells have been shown to retract their radial processes, proliferate and differentiate to form new spinal cord neurons (Reimer et al., 2008; Ribeiro et al., 2017; Cura Costa et al., 2021; Klatt Shaw et al., 2021; Walker and Echeverri, 2022). In fact, some researchers have suggested that these RGLCs undergo dedifferentiation during regeneration, as can be determined by the changes in cell morphology, the loss of the radial extensions, and the differential expression of genes associated with neural stem cell properties (Walder et al., 2003; Ribeiro et al., 2017; Walker and Echeverri, 2022).

It is important to contrast these RGLCs/ependymal cells of zebrafish and axolotl to those in adult mammals. In the latter, the ependymal cells are cuboidal and multiciliated and usually do not divide after differentiation or in some cases, give rise only to glia (Spassky et al., 2005; Meletis et al., 2008). Moreover, transcriptomic studies have shown that ependymal cells in mice are transcriptionally different from neural stem cells and are not induced to divide by brain injury (Shah et al., 2018).

Recent single-cell transcriptomics have identified distinct precursor populations that generate neuronal and glial cells during zebrafish embryonic development (Scott et al., 2021). Likewise, single cell transcriptomics have shown several populations of RGLCs to be present in the developing and adult axolotl brain, and one of these appears to be activated by injury (Wei et al., 2022). Thus, we expect that future experiments will provide crucial information on the ependymal populations in the adult zebrafish and axolotl and the transitions they undergo during spinal cord regeneration.

Zebrafish RGLCs in different niches can generate new neurons under normal homeostatic conditions and in response to injury (Barbosa et al., 2004; Dray et al., 2015). However, there are differences between homeostatic neuron production and their formation in response to injury. In the former, newborn neurons remain below the progenitor zone while in the latter, newborn neurons migrate to populate injured brain regions (Barbosa et al., 2004; Dray et al., 2015). During normal conditions, only type II zebrafish RGLCs are in the active proliferative state (Dray et al., 2015). Their proliferation modes have been shown to be symmetric and asymmetric, providing self-renewing and progenitor-generating properties (Barbosa et al., 2004, Barbosa and Ninkovic, 2016; Rothenaigner et al., 2011; Dray et al., 2015). Most of the RGLC proliferation capacities are enhanced in response to injury. For example, using an engineered neurotoxic amyloid-β42 (Aβ42)-dependent zebrafish, which represents degenerative neurotoxicity similar to that observed in Alzheimer’s disease, researchers found that some RGLCs responded to this injury by extensive proliferation and enhanced neurogenesis (Bhattarai et al., 2016). One unique difference between homeostatic- and injury-generated neurons is that the latter have enhanced migratory capacities, rendering them able to restore damaged tissue (Barbosa et al., 2015; Barbosa and Ninkovic, 2016).

Zebrafish also display an impressive ability to regenerate their retina. The zebrafish retina, typical of vertebrate retinas, is composed of three cellular layers, six principal neuron population types, and four types of glial cells (Fadool and Dowling, 2008). One of the glial cells, the Müller glia, stands out in that it plays a unique role in retinal regeneration (Figure 6). The Müller glia apico-basal morphology enables structural and metabolic support for the retinal neuron population (Reichenbach and Bringmann, 2013; Lenkowski and Raymond, 2014). Müller glia are considered mature, differentiated cells. Under normal retinal physiology, they contribute to retinal synaptic activity by recycling GABA and glutamate neurotransmitters. Moreover, these cells support neurons by regulation of extracellular volume and molecular composition, transporting metabolic waste, and providing neurons with nutrients (Lenkowski and Raymond, 2014; Gao et al., 2021).

Müller glia have been proclaimed as being the retinal RGLC, mainly because of their radial morphology but also because they share some of the RGC markers, such as GS, GFAP, GLAST, and VIM (Linser and Moscona, 1979; Malchow et al., 1989; Lenkowski and Raymond, 2014; Gao et al., 2021). In response to injury, Müller glia play a pivotal role in retina regeneration. This process has been extensively studied at both cellular and molecular levels. It has been documented that these RGLCs respond to injury in the following manner: (i) Müller glia sense injury in the retina, (ii) The cells dedifferentiate or reprogram, (iii) Dedifferentiated Müller glia divide asymmetrically, producing one Müller glia cell and one Müller glia-derived progenitor cell, (iv) The progenitor cells divide and populate the injury site creating neurogenic clusters.

Transgenic zebrafish lines have provided crucial data to acknowledge Müller glia as the neural stem cell source for retinal regeneration (Bernardos and Raymond, 2006; Fimbel et al., 2007; Thummel et al., 2008). A study assessing their proliferation capacity by BrdU incorporation or PCNA immunodetection determined, that although there are proliferating Müller glia in the uninjured retina, most of the cells are in a quiescent state that is regulated by Notch signaling (Bernardos et al., 2007). As a response to injury, Notch expression decreases, facilitating Müller glia proliferation (Ruddy and Morshead, 2018; Sahu et al., 2021). Similar to other RGLCs, the Müller glia are also a source of neurons, as demonstrated by using transgenic reporter lines showing that proliferating Müller glia gave rise to rod progenitors (Bernardos et al., 2007).

Among the events that involve Müller glia in the regeneration process, we would like to highlight the dedifferentiation response since cell dedifferentiation stands out as a process by which RGLCs in various organisms appear to contribute to NS regeneration. Müller glia dedifferentiation has been assessed by the re-expression of retinal progenitor neurogenic markers, such as BLBP, ALCAMA, Rx1, six3b, and Pax6 (Fimbel et al., 2007; Thummel et al., 2008; Nagashima et al., 2013). In addition, Sox2 role has been functionally assessed in the regenerating zebrafish retina, demonstrating it to be an essential factor regulating Müller glia proliferation and reprogramming (Gorsuch et al., 2017; Elsaeidi et al., 2018). Even though progenitor marker re-expression is observed, it has been suggested that Müller glia cells only partially dedifferentiate, given that mature Müller glial cell markers persist at a lower level during dedifferentiation. Nonetheless, dedifferentiation enables Müller glia cells to divide asymmetrically producing a neural progenitor cell that proliferates and builds neurogenic clusters that will differentiate into the retina neural cell types (Bernardos et al., 2007).

Conserved functions of the Müller glia mediating ocular regeneration have also been documented in other vertebrates (Fischer and Bongini, 2010; Goldman, 2014). In the early 2000s, Fischer and Reh (2001) demonstrated that, in chickens, following retinal injury by injection of N-methyl-D-aspartate (NMDA), Müller glia cells re-entered the cell cycle and could differentiate into retinal neurons as determined by co-labeling of BrdU and retinal markers. The available data point to Müller glia cells as the essential cell for retinal repair (Fischer and Reh, 2003).

Radial glia-like cells also play a major role in the regeneration of the echinoderm NS following transection of the RNC. Immediately after transection, the injured edges of the RNCs begin to swell, nerve fibers become distorted, and programmed cell death occurs. This initial phase is characterized by extracellular matrix (ECM) deposition at the wound site and elongation of the RNC ectoneural component (Mashanov et al., 2008, 2014). At the cellular level, evidence using electron and fluorescence microscopy strongly suggests that the RGLCs dedifferentiate and provide the precursors for the new nervous tissue (Figure 7; San Miguel-Ruiz et al., 2009; Mashanov et al., 2010, 2013). Two observations have been key to outline the RGLCs role: First, the degradation of the RGLC radial cytoskeletal fibers during regeneration and second, BrdU experiments demonstrating the dedifferentiated RGLC proliferative capacities. The morphological changes are key features consistent with other known dedifferentiating cells. Hence, by losing their complex radial extensions, they transition into a less specialized state and proliferate. The dedifferentiated cells also migrate toward the injury gap and participate in one of the main events in RNC regeneration: the reconnection of the transected stumps. During this phase, both RNC edges from each side of the transected side close together until they reconnect and restore the RNC. This is thought to be one of the most complex processes involving signaling molecules that control growth, migration, axial polarity, and cell-to-cell interactions (San Miguel-Ruiz et al., 2009; Mashanov et al., 2013).

Following RNC reconnection, dedifferentiated cells in the regenerative zone redifferentiate, which is reflected by the growth in the size of the “new” tissue between the stumps. Birth dating experiments have shown that dedifferentiated RGLCs cells play a role in this event by giving rise to both RGLCs and neuronal populations in the regenerated structure (Mashanov et al., 2013). What regulates RGLC redifferentiation to neuronal or non-neuronal lineage still needs to be studied. However, many researchers have suggested that this process is mainly intrinsically regulated, while still having extrinsic signaling molecules and/or cell-to-cell interaction that could influence the outcome (Mashanov et al., 2008, 2013; San Miguel-Ruiz et al., 2009). By the end of the regenerative response, cell proliferation and apoptosis return to basal levels, and cellular and anatomical organization of the “new” RNC segment resembles that of the non-transected cords.

Much of what is known of RNC regeneration in holothurians has been provided by histological analyses, including immunohistochemistry, in situ RNA hybridizations, and RNA sequencing analysis. The transcriptomics data analyses have provided a list of differentially expressed genes that are possible candidates for genes controlling or modulating the regeneration response (Mashanov et al., 2014). For example, ECM-related genes were among the most differentially expressed transcripts in all regeneration stages. Many other genes have been targeted as possible candidates and even transposable elements have been suggested to be associated with RNC regeneration (Mashanov et al., 2012).

Specific genes have been probed to modulate particular cellular events, for example, the process of RGLC dedifferentiation. The first candidate genes to be evaluated as promoters of RGLC dedifferentiation were the Yamanaka pluripotency factors Oct1/2/11, c-Myc, Sox 2, and Klf1/2/4 homologs (Mashanov et al., 2015b). The study demonstrated that (i) all four factors were expressed during early and late stages of RNC regeneration, (ii) Myc was significantly overexpressed during all the stages of regeneration, suggesting that it played an important role in the regeneration process (iii) Oct1/2/11 and SoxB1 (sox 1 homolog) were significantly downregulated at the late stages of RNC regeneration (Mashanov et al., 2015b). These findings were further explored by evaluating the role of Myc via RNA knockdown in regenerating RNC. The results demonstrated that Myc regulated programmed cell death and RGLC cell dedifferentiation, two major events known to occur during the initial regeneration response (Mashanov et al., 2015c). Therefore, it has been assumed that the in vivo de-differentiation observed in the RGLCs is more akin to a partial reprogramming that partially dedifferentiates cells to a precursor-like phenotype than to the total reprogramming that produces pluripotential stem cells.

Two recent advances open the possibility for new findings on the regeneration of the echinoderm CNS and the possible role of the RGLCs. The first is the development of an in vitro RNC explant preparation (Quesada-Díaz et al., 2021). Most, if not all, of the previously performed studies were done using the holothurian RNC complex, that in addition to the RNC is composed of longitudinal muscle bands, water vascular canal, and the body wall. This has made it almost impossible to directly target the RNC without affecting the surrounding tissues. Using an isolated RNC explant that can be cultured provides a unique opportunity to study cellular events associated with dedifferentiation and regeneration without the influence of adjacent tissues (Quesada-Díaz et al., 2021). Moreover, it has been documented that in in vitro cultures, RGLCs within the RNC continue to exhibit the same morphological changes shown to occur during in vivo RGC dedifferentiation. Hence, this method provides a tool to study the dedifferentiation process while excluding other cellular events in the surrounding tissues. This culture system is presently being used to provide a transcriptomic profile of regenerating RNC explants.

The second advancement with a significant impact on RNC regeneration studies is the sequencing of the H. glaberrima genome. This was published as a draft genome, providing a tool that can benefit both the regenerative and developmental biology fields (Medina-Feliciano et al., 2021). The genome serves as a database to identify potential genes and molecules that are key to the regeneration process. In addition, for researchers working with H. glaberrima that have obtained transcriptomic data from regenerating intestine and nerve complex at different time points, the genome provides a reference to the identification of the differentially-expressed genes. The impacts of these results will reach their full potential when a reference genome becomes available.

Radial glial cells are found in all developing vertebrate CNS studied to date, and RGLCs are found in adult CNS components of vertebrates and invertebrates. The morphological data accumulated over decades, together with the avalanche of new molecular data provided by new sequencing methods is molding our view of these cells. In fact, we can describe the RG/RGLCs in terms of Ernst Haeckel dictum “Ontogeny recapitulates Phylogeny” where their evolution and their presence in different animal species parallels their embryological development. In this scenario, the RGCs that form in the early deuterostome embryo can be considered the precursors of both neurons and glia of the adult CNS. In invertebrate deuterostomes, these cells remain in the adult CNS, retaining morphological properties and particular molecular markers along with their competence to undergo neurogenesis. Thus, organisms where RGLCs are found are capable of robust CNS regeneration. A similar situation is observed in some amphibians and fishes, where the progeny of RGCs is retained as RGCLs and are responsible for both homeostatic and regenerative neurogenesis in adults. In contrast, in mammals, only certain areas of the adult CNS retain RGLCs. The cells in these niches might express specific markers along with more general RGLC markers, giving rise to particular subpopulations. Nonetheless, these RGLC subpopulations are able to undergo homeostatic neurogenesis and, after injury, these RGLCs are activated and appear to participate in some, albeit limited, CNS restoration. One last example will serve to strengthen the evolution/regeneration comparison. It is well documented that in most vertebrates, the capacity for CNS regeneration is lost during the developmental process. Nowhere is this better shown than in the Xenopus model where the ability to regenerate their transected spinal cord changes dramatically as the animal undergoes metamorphosis. In this respect, the work by the Larraín lab (Edwards-Faret et al., 2018, 2021) has correlated the regenerative capacity to the presence of RGLCs, expressing certain molecular markers and showing neurogenic capacity found in the regenerative stage but lost or greatly diminished in the non-regenerative stage.

In addition to the neurogenic regenerative potentials of RGLCs, there are also parallels in the processes by which RGLCs in different species or different regions of the CNS respond to injury. Specifically, changes observed in their radial morphology and, at the molecular level, in their gene expression profile point out toward a degree of dedifferentiation, where a mature cell reverts into a progenitor/stem cell phenotype as described by Walker and Echeverri (2022). Dedifferentiation allows cells to enter the cell cycle and give rise to new tissue cells. In this respect, it is interesting that activated astrocytes, following trauma to the adult mammalian brain, have also been shown to re-express markers of more immature precursors suggesting a degree of dedifferentiation (Zambusi et al., 2020; Lattke and Guillemot, 2022).

To fully understand the RGCs and RGLCs it becomes crucial to determine the lineage and evolutionary relationship of these cell types. Are all adult RGLCs remnants of the embryonic RGC population that persist in certain areas of the adult brain? Are the invertebrate RGLCs evolutionarily related to the vertebrate RGCs? What, if any, specific marker can be used to identify the RGLC and RGC subpopulations? Modern cell-typing techniques will be important in answering these and other questions. The use of gene expression markers will be decisive in determining the expression profile of the different cell types and the changes that might take place during developmental or functional stages. Equally important will be to determine the factors that induce the neurogenic activity in these cells and the process by which cells might dedifferentiate and/or enter the cell cycle.

The possibility exists that the neurogenic abilities of these cells can be manipulated to respond to traumatic injury so that they serve as proliferating neurogenic progenitors in favor of CNS repair. Apropos to this, dedifferentiation is also being explored as a potential therapy to regenerate the CNS. The idea is to be able to transform the state of the brain into a regeneration-promoting state by, for example, converting terminally differentiated astrocytes into a precursor type cell that could give rise to neurons (Yao and Wang, 2020; Zamboni et al., 2020; Robel et al., 2011; Lattke and Guillemot, 2022). Nonetheless, it is imperative to first lay a foundational understanding of RGC roles, their pathways of regulation, and possible differentiation stages. Key information could be obtained from comparative studies using non-traditional model animals that show extraordinary regeneration abilities. Therefore, we propose that the study of the RGLCs in the echinoderms, particularly in the sea cucumber H. glaberrima will provide important information to elucidate neurogenesis activation in adult animals.

YM-N and JG-A wrote part of the manuscript, contributed to the article, and approved the submitted version.