The generation of neurons is a highly conserved cellular and molecular mechanism across metazoans (Figure 1). Even in distant species such as the cnidarian Nematostella, neural progenitors are induced within an epithelium by BMP inhibition (Bier and De Robertis, 2015) and express SoxB family genes (e.g., Sox2 in vertebrates) and Notch receptors, among others (Hartenstein and Stollewerk, 2015; Kelava et al., 2015). Their daughter cells undergo epithelial to mesenchymal transition, downregulating SoxB and Notch while activating Notch ligands and proneural genes (Rentzsch et al., 2017; 2019). This transition commits them to a neuronal fate, defined by the onset of genes supporting neuronal communication. This program interacts with transcription factors differentially expressed along the body axis, generating region specific neuronal subtypes (Figure 1; Jessell, 2000; Puelles et al., 2013; Sachkova et al., 2025). Interestingly, gene modules for neuronal communication preceded the emergence of neurons, raising the intriguing possibility that their parallel acqusition in different body regions led to the appearance of region-specific neuron types from the very dawn of the neuronal era (Arendt, 2020; 2021; Najle et al., 2023; Sachkova et al., 2025). Neuronal progenitors may likewise have had heterogeneous evolutionary origins, deploying shared regulatory programs alongside spatial patterning mechanisms. Throughout evolution, neurons further diversified in a hierarchy of neuron families, still relying on spatial patterning to generate their variation (Arendt et al., 2019). At the final division, immature neurons inherit from their progenitors a commitment to a region-specific fate through transcriptional codes (Jessell, 2000; Oberst et al., 2019a). Modern neuronal taxonomy approaches, based on transcriptional and electro-morphological features, have shown that, within neuronal families with clear distinct embryonic origin, neurons can further diversify along a seemingly continuous spectrum (Scala et al., 2021). To what extent this further diversification depends on contextual factors rather than intrinsic programs inherited by their progenitors, remains to be established. Notably, positional context can influence morphology and function also independently of transcriptional identity (Hecker et al., 2025; Shainer et al., 2025; Zaremba et al., 2025).

As neuronal specialization increased, so did their reliance on support cells providing a distinct form of “maternal care” ensuring metabolic supply, maintenance of the extracellular milieu, ion and water homeostasis, neurotransmitter clearance, synaptic remodeling, physical compartmentalization from body fluids, and responses to brain lesions (Hartline, 2011; Freeman and Rowitch, 2013; Verkhratsky and Nedergaard, 2018; Sheloukhova and Watanabe, 2024). These support cells are known as glial cells and express specific gene modules that emerged in bilaterians and have remained relatively conserved throughout evolution (Morizet et al., 2024). In invertebrates, glial cells are terminally differentiated cells and do not express neuronal progenitor markers such as the Sox2 homologs (Haim and Rowitch, 2016; Simões and Rhiner, 2017; Egger, 2023; Morizet et al., 2024; Gujar and Wang, 2025). In vertebrates, by contrast, neurogenic and homeostatic functions converged in a single and highly versatile cell type: the astroglia (Freeman and Rowitch, 2013; Morizet et al., 2024). During development, these cells are the first to differentiate from neuroepithelial progenitors maintaining an epithelial organization while extending, on the basal side, a long radial process hence the name radial glia (RG; Rakic, 2003; Mori et al., 2005; Jurisch-Yaksi et al., 2020; Miranda-Negrón and García-Arrarás, 2022). Despite their overall similarity along the neuraxis, RG are regionalized into progenitor domains producing distinct neuron types (Figure 2). After neurogenesis, in non-mammalian vertebrates, the RG scaffold persists lifelong, serving both neurogenic and homeostatic roles (Ganz and Brand, 2016; Jurisch-Yaksi et al., 2020). In mammals, they transform into astrocytes, a closely related cell type that invades and tiles the parenchyma (Figure 2). While most astrocytes specialize in neuron–glia interactions, a restricted subset is retained as neural stem cells (NSCs) in two canonical niches, the ventricular-subventricular (V-SVZ) and subgranular zones (SGZ) (Kriegstein and Alvarez-Buylla, 2009; Bayraktar et al., 2014). This led to the hypothesis that in mammals homeostatic and neurogenic functions became separated again into two sister cell types (Morizet et al., 2024). However, we and others demonstrated that at least in some brain regions, parenchymal astrocytes retain a latent neurogenic potential, and in specific conditions can generate neurons outside canonical neurogenic niches (Magnusson et al., 2014; Nato et al., 2015; Péron and Berninger, 2015). This indicates that the similarities between astrocytes and non-mammalian RG cells are wider than previously thought and the adult brain parenchyma can be permissive for neuronal progenitor maintenance and activity.

In non-mammalian vertebrates RG persists after development, but progressively reduces proliferation, transforming the VZ from a pseudostratified to a single-layered epithelium. As neurons accumulate and mature, RG develop fine processes resembling astrocytic leaflets and increase neuron contacts (Figure 2A; Jurisch-Yaksi et al., 2020). This morphological maturation is accompanied by a progressive shift from neurogenic to homeostatic functions (Diaz Verdugo et al., 2019; Mu et al., 2019; Raj et al., 2020; Morizet et al., 2024). In zebrafish, transcriptional differences between RG in different domains are more pronounced during embryonic stages, possibly reflecting the active role of patterning factors and selector genes in defining regional identity. These differences are gradually attenuated at the end of development, while RG converges toward a shared functional state, with only residual regional specificities (Raj et al., 2020; Mitic et al., 2024; Morizet et al., 2024). Of note, in adults, regional transcriptional differences are largely lost upon neurogenic activation, despite the generation of distinct neuronal types (Lange et al., 2020), implying that non-transcriptional mechanisms, such as chromatin modifications, may preserve positional information in adult astroglia.

In birds, RG persistence is accompanied by the emergence of parenchymal astrocytes, morphologically and transcriptionally related cells that detach from the ventricle and migrate into the parenchyma (Falcone, 2022; Ciani et al., 2024; Morizet et al., 2024). These cells are rare in reptiles and likely evolved independently in birds and mammals as an adaptation to increased brain complexity, though their functions in birds remain largely unexplored.

In mammals, during development RG progressively acquire astrocytic features such as increased branching and vascular contacts, paralleling neuronal maturation (Schmechel and Rakic, 1979a; 1979b; Takahashi et al., 1990; Mori et al., 2005). Around mid-gestation, a gliogenic switch generates a first wave of astroblasts migrating into the parenchyma, followed by a second wave at the end of neurogenesis (around P0–P3 in mice), when residual RG retract their processes, detach from the ventricle, and transform into astrocytes (Gressens et al., 1992; Ge et al., 2012; Clavreul et al., 2022), while the ventricular layer becomes lined by ependymal cells (Bayraktar et al., 2014; Redmond et al., 2019). In the mouse neocortex only about 20% of the EMX1+ RG generate astrocytes, while the others are consumed in neurogenic differentiation (Gao et al., 2014; Lin et al., 2021; Shen et al., 2021). In parallel, by late stages of neurogenesis a population of Olig2+ progenitors, called the Tri-IPCs, contributes oligodendrocytes, olfactory bulb-like interneurons and astrocytes in the cortex (Zhang et al., 2020; Wang et al., 2025). Following RG scaffold disassembly, immature astrocytes proliferate locally and tile the parenchyma in a near-uniform lattice (Ge et al., 2012; Clavreul et al., 2019; Endo et al., 2022). Clonal analyses show stochastic dispersion, which however is constrained within their domain of origin (Hochstim et al., 2008; Tsai et al., 2012); (Figure 2). Neither age, injury, nor the loss of neighboring astrocytic populations causes astrocytes to cross these boundaries (Tsai et al., 2012), suggesting that their regional identity may be intrinsically determined. The overall subdivision of these domains is highly conserved in vertebrates (Puelles and Medina, 2002; Puelles et al., 2013; Luzzati, 2015), suggesting conservation and/or co-evolution of regional astrocytes-neurons interaction (Figure 2). As in zebrafish, mammalian astrocytes converge towards similar gene expression profiles, consisting in a strong core of shared gene modules, with subtler region-specific adaptations (Morel et al., 2017; Lozzi et al., 2020; Endo et al., 2022; Kwon et al., 2025; Morel et al., 2017; Lozzi et al., 2020; Endo et al., 2022). Notably, similarities are higher within major divisions of the brain, such as pallium versus subpallium, while sharp transcriptional discontinuities separate them, a pattern that mirrors neuronal regionalization and may reflect the evolutionary history of these domains (Endo et al., 2022).

In mammals, neurogenesis is highly active during early life, reflecting a key conserved role in postnatal brain development. In adulthood, it persists at much lower levels and varies substantially across species, possibly contributing to long-term circuit plasticity in specific contexts (Amrein et al., 2011; Barker et al., 2011; Bonfanti et al., 2011; Kempermann, 2012; Sailor et al., 2017; Charvet and Finlay, 2018; Kempermann et al., 2018; Sorrells et al., 2021). Whether this decline reflects NSC exhaustion, deep quiescence or an unfavorable niche environment has been the subject of intense debate (Kalamakis et al., 2019; Harris et al., 2021; Carvajal Ibañez et al., 2023). The NSCs in the V-SVZ and SGZ are mostly quiescent, and only 0.15%–0.3% of them undergo activation each day, ∼30-fold less than adult zebrafish NSCs (Calzolari et al., 2015; Ziebell et al., 2018). Clonal analyses and intravital imaging in the V-SVZ and SGZ suggested that most NSCs undergo a few rounds of divisions, generating multiple waves of fast-expanding progeny before exhaustion (Calzolari et al., 2015; Pilz et al., 2018) or terminal astrocyte differentiation (Encinas et al., 2011). In parallel to this “disposable NSC population”, other studies in the SGZ identified cells with longer self-renewal capacities (Bonaguidi et al., 2011; Bottes et al., 2021), possibly acting as the zebrafish “reservoir pool” (Dray et al., 2021b). Alternatively, NSC behavior may distribute over a fluid spectrum of self-renewal potential. As clonal analyses possess intrinsic limitations in the discrimination of self-renewal strategies (Parigini and Greulich, 2020), more extensive and protracted observations, akin to those done in the zebrafish would be required to resolve this issue.

In the V-SVZ, a combination of symmetric self-renewing divisions (20%) and symmetric consuming divisions (80%) gradually reduces the population of NSCs contacting the ventricle, also known as Type B1 cells (Obernier et al., 2018). This reduction however, is counterbalanced by the increase of basal NSCs or Type B2 cells, originally thought to represent non-neurogenic SVZ astrocytes but recently shown to also have neurogenic potential (Cebrian-Silla et al., 2025). A comprehensive model of mouse V-SVZ NSC dynamics has been proposed by Basak et al. based on long-term lineage tracing, clonal analyses, and mathematical modeling (Basak et al., 2018). Similar to classic stem cell systems, NSCs divide symmetrically and the daughter cells stochastically choose between returning to quiescence (maintenance) or progressing into differentiation (exhaustion). The probability of taking either fate correlates with the amount of neighboring NSCs, so that reducing numbers sustains maintenance while crowding stimulates exhaustion (Basak et al., 2018). In line with this model, the reduced NSC density observed in the aged niche may increase the probability of self-renewing asymmetric divisions in which one of the daughter cells return to quiescence (Bast et al., 2018; Harris et al., 2021; Wu et al., 2023; Cebrian-Silla et al., 2025), further contributing to the lifelong maintenance of neurogenesis. As for the zebrafish pallium the number and relative position of NSCs in long term clones is remarkably stable and does not show any neutral drift (Basak et al., 2018). The V-SVZ may thus be viewed as a mosaic of “restricted niches”, maintained by local factors, whose nature however remains to be established. This model may also apply to other embryonic and adult niches in vertebrates, supporting the organization in “radial units”.

Overall, these findings reveal remarkable conservation in NSC spatio-temporal dynamics between non-mammalian and mammalian neurogenic astroglial progenitors.

The main hallmarks of neurogenic niches, such as anatomical confinement, planarity and apico-basal polarity are largely lost in the 3D lattice of parenchymal astrocytes, raising the question of how their neurogenic potential may have adapted to this new environment. In the QA model, clonal analyses showed that neurogenic foci originate from the local expansion of individual astrocytes (Fogli et al., 2024). These foci persist for about 10 days, during which they grow and mature while continuously producing post-mitotic neuroblasts (Figure 3B). As each TAP clone completes its cycle, a new one emerges, maintaining a steady-state turnover. Astrocytes undergo only a brief activation that seeds neurogenic clones, after which in most cases they re-enter quiescence maintaining close contact with their daughter cells. According to the model of symmetric division and competition for a restricted niche (Basak et al., 2018), this high maintenance rate may result from lower astrocyte density in the striatum, increasing the likelihood of niche re-rentry.

By using the neurogenic foci as a proxy for the location of astrocyte activation events over the last 10 days, we demonstrated that these events are randomly distributed, with only a minor inhibitory effect from nearby clusters, closely matching the behaviour of zebrafish pallial RGs (Dray et al., 2021a). The activation rate was remarkably similar in the V-SVZ and in the striatal neurogenic area, at about 0.2%–0.4% astrocytes/day (Calzolari et al., 2015; Fogli et al., 2024), suggesting a similar prevalence of neurogenic progenitors in these regions, thus potentially encompassing the entire striatal astrocyte population.

Thus, despite the different tissue organization, when activated, parenchymal neurogenesis unfolds through the same cellular mechanisms and spatio-temporal dynamics as in planar neurogenic niches across evolution.

Mammalian neurogenic production relies more heavily than other vertebrates on TAPs, particularly in late developmental phases and in adult life. This may compensate for the need to boost both neuronal output and homeostatic support, despite their intrinsic trade-off. While TAPs are a crucial step in regulating the neuronal outcome, their behaviour is poorly studied. During development, TAPs typically follow deterministic division patterns, performing a few symmetric divisions before differentiating (Lin et al., 2021). Adult TAPs were initially thought to retain this rigid logic: in V-SVZ they divide about three times (Ponti et al., 2013; Calzolari et al., 2015). However, intermediate progenitor cell activity can be modulated by different neurogenic stimuli, as clearly shown in the DG (Kronenberg et al., 2003). More recently, clonal analyses (Bast et al., 2018) and live imaging (Pilz et al., 2018) revealed that most V-SVZ and SGZ TAPs divide symmetrically, while ∼25% divide asymmetrically, displaying highly variable division patterns. Although TAP expansion is lower in the SGZ, and closer to zebrafish dynamics, these results imply that TAP behavior is less deterministic than previously assumed (Götz, 2018). In the QA lesion, the TAPs clones had highly variable size and cellular composition. Mathematical modelling revealed that this heterogeneity is driven by a stochastic division process acting in parallel to an accelerating differentiation propensity, that introduces a deterministic factor constraining their expansion and ultimately leading to their exhaustion (Fogli et al., 2024). The high heterogeneity among striatal neurogenic foci may thus underlie a locally regulated control of neuronal output, and exploring the underlying cellular and molecular mechanisms could provide valuable insights into how parenchymal neurogenesis is modulated.

Overall, at least in the striatum and likely also in the neocortex, the main difference between canonical and parenchymal neurogenic niches lies not in the presence of quiescent NSCs, but in how their activity is regulated. Canonical niches provide protected and permissive environments that preserve progenitor stemness, maintain a quiescent pool for lifelong neurogenesis, and coordinate NSC activation by integrating multiple physiological and pathological cues. These mechanisms have been extensively reviewed elsewhere (Fuentealba et al., 2012; Lim and Alvarez-Buylla, 2014; Vicidomini et al., 2019; Llorente et al., 2022; Chaker et al., 2024; Foley et al., 2024). In adult canonical niches NSCs are mostly quiescent, but their activity can be stimulated by several external cues (Kempermann et al., 2015; Song et al., 2016; Obernier and Alvarez-Buylla, 2019; Caron et al., 2022; Chaker et al., 2024). Experience-dependent cues, such as physical exercise, environmental enrichment or learning can increase SGZ neurogenesis by multiple pathways including the release of BDNF, modulation of local GABAergic tone, and excitatory inputs onto NSCs (van Praag et al., 2000; 2002; Catavero et al., 2018; Liu and Nusslock, 2018; Li et al., 2023). Some stimuli can elicit localized effects such as nutritional state or pregnancy that activate specific V-SVZ domains to generate specific OB subtypes (Paul et al., 2017; Chaker et al., 2023). By contrast, in all known vertebrate niches examined so far, brain lesions induce a strong activation of both quiescent and primed NSCs (Arvidsson et al., 2002; Kernie and Parent, 2010; Jurisch-Yaksi et al., 2020; Morizet et al., 2025). In mammals, these responses are partly mediated by inflammatory cytokines (IL-6, CNTF, TNFα, Interferons; Llorens-Bobadilla et al., 2015; Belenguer et al., 2021).

These regulatory mechanisms are mediated by multiple cellular components, including blood vessels, microglia, neuronal afferents, and astrocytes, all of which are also present throughout the parenchyma. In non-mammalian vertebrates, RG cells engage with these elements directly within the parenchyma, and this environment likely represents an integral component of their niche (Jurisch-Yaksi et al., 2020). In mammals, the transformation of RG into astrocytes has profoundly modified the anatomical relationship of individual cells, further parcellating these ancestral niches. These new microenvironments, however, still regulate the delicate balance between quiescence and activation of astroglial stem cell potential. Accordingly, key pathways involved in NSC quiescence and maintenance, including Notch (Imayoshi et al., 2008; Engler et al., 2017; Lampada and Taylor, 2023; Morizet et al., 2025), BMP (Lim et al., 2000), β1-integrin (Porcheri et al., 2014), S1P (Codega et al., 2014), and GABA (Liu et al., 2005; Song et al., 2012; 2016) are strongly active in parenchymal astrocytes (Lim et al., 2000; Robel et al., 2011; Magnusson et al., 2014; Nagai et al., 2019; Singh et al., 2022). In line with the developmental switch between neurogenic and homeostatic states, these pathways also regulate astrocyte morphogenesis (Meyers and Kessler, 2017; Cheng et al., 2023; Guo et al., 2023; Gonzalez and Reinberg, 2025). The adult parenchyma can thus be considered as a quiescent neurogenic niche, subdivided into a mosaic of multiple sub-niches.

Unlike in canonical niches, abrogation of these pathways is not sufficient to activate neuronal production from parenchymal astrocytes (Lim et al., 2000; Porcheri et al., 2014). However the underlying changes in astrocyte states have been investigated in detail only for the Notch signalling (Magnusson et al., 2020; Zamboni et al., 2020). Notch pathway is a highly conserved regulator of NSC activity whose abrogation induces neurogenic activation across metazoans, from cnidarians (Richards and Rentzsch, 2015) to vertebrates (Rothenaigner et al., 2011; Shimojo et al., 2008; Imayoshi et al., 2010). Deletion of Rbpj, a central Notch effector, drives striatal and cortical astrocytes into a shallow quiescent or “primed” state, which occasionally progresses to overt neurogenesis in the medial striatum and medial cortex (Magnusson et al., 2014; 2020; Zamboni et al., 2020). Brain lesions can independently induce similar primed states in cortical and striatal astrocytes (Zamboni et al., 2020; Kremer et al., 2024) and at least in the striatum it involves methylome changes that closely resemble those observed during V-SVZ astrocyte activation. These observations suggest that, as in canonical niches, parenchymal astrocytes may transit to a primed state and only subsequently to actual neurogenic state. However, in some conditions, such as stab injury or after Notch abrogation, this primed state does not progress into neurogenesis (Buffo et al., 2008; Magnusson et al., 2020; Zamboni et al., 2020), suggesting that reawakening of neurogenic competence and its expression may be regulated by distinct factors. The molecular factors controlling these states in physiologic and pathologic conditions remain to be established.

Throughout evolution, brain lesions not only strongly stimulate NSCs activation, but also induce profound changes in homeostatic functions of glial cells activating context- and time-dependent states, also called reactive states, that limit damage and support repair (Götz et al., 2015; Sofroniew, 2020; Escartin et al., 2021; Clayton and Liddelow, 2025). Given the dual nature of vertebrate astroglia, and the further elaboration of astrocyte functions and heterogeneity in mammals, a non-trivial and still mostly unresolved question is how these multiple functions and states are regulated and if they can coexist only at the population or also at the individual cell level. In both rodents and humans, the reawakening of astrocyte neurogenic competence, measured with the neurosphere assay, was found associated only with invasive injuries, where the type of astrocytes reactivity include cells that proliferate and transiently re-acquire more immature features (Buffo et al., 2008; Sirko et al., 2013; 2023). In both stroke and QA model, further neurogenic activation of primed astrocytes is typically delayed relative to the peak of reactivity, possibly associated with a more reparative phase, and at least in the QA was independent from early proliferation or expression of C3, a marker of a reactive astrocyte subtype (Fogli et al., 2024). This indicates that the neurogenic program is compatible with multiple reactive sub-states, within the context of the sub-acute phase of an invasive injury.

The tight temporal control of the striatal neurogenic response is paralleled by an equally tight spatial control. Indeed, although the neurogenic potential is widespread in the striatal parenchyma, its activation is typically focal (Figure 4). Following QA-induced lesions, the neurogenic region aligns with the lesion border, mostly to its rostral and ventro-medial part (Fogli et al., 2024). In contrast, in mouse models of progressive degeneration, neurogenic activation is shifted toward more lateral striatal domains (Luzzati et al., 2011), remains confined dorso-medially in normal rabbits (Luzzati et al., 2006), and ventro-laterally in guinea pigs at weaning (Luzzati et al., 2014). Such localized activation generates neurons interacting with specific neuronal networks, raising the question of whether these same circuits may contribute to neurogenic activation. In canonical niches NSCs can be directly modulated by neurotransmitters and are highly sensitive to neural activity (Song et al., 2012; Paez-Gonzalez et al., 2014; Káradóttir and Kuo, 2018). The evolutionary transformation of RG into astrocytes resulted in more localized interactions with specific neuronal populations, allowing greater flexibility in neuron–astroglial networks. While specializing in local circuit-specific interaction, mammalian astrocytes may have thus also refined the circuit-specific regulation of their neurogenic capacity.

Altogether, these findings indicate that similarly to neurogenic niches, parenchymal neurogenic activation may pass through multiple sequential steps, each potentially responding to specific signals. While the spatial organization of different neurogenic niches and their activatory stimuli remains to be established, the stability of astrocyte units provide a framework for highly tunable spatio-temporal dynamics of adult neurogenesis in mammals.

The RG fate potential is shaped by spatial and temporal patterning. Morphogen gradients (e.g., BMP, Shh, Wnt) subdivide the VZ in domains of committed progenitors generating specific neuronal subtypes (Jessell, 2000; Puelles et al., 2000; Appan et al., 2023). In the vertebrate telencephalon, pallial RG generate glutamatergic neurons, while sub-pallial RG (e.g., MGE, LGE, CGE, PoA) produce GABAergic neurons (Marin et al., 2000; Cobos et al., 2001; Flames et al., 2007; Luzzati, 2015; Quintana-Urzainqui et al., 2015; Schmitz et al., 2022). RG can also change its potential over time. In the cortex, RG cells sequentially generate deep-layer neurons, upper-layer neurons and then glia (Oberst et al., 2019a; Lin et al., 2021). This sequence is driven at least in part by intrinsic mechanisms, as shown by clonal cultures and was initially thought to be irreversible (Frantz and McConnell, 1996; Shen et al., 2006; Gaspard et al., 2008). However, recent work shows that late RG can resume a previous neurogenic state in response to earlier-stage environmental signals, revealing temporal plasticity (Oberst et al., 2019b). A similar capacity to resume developmental neurogenic sequences has been proposed to support regeneration after lesion in some non-mammalian vertebrate models (Ohnmacht et al., 2016; Zheng et al., 2022; Foucault et al., 2024).

In non-mammalian vertebrates, multiple RG domains remain neurogenic throughout life, preserving the embryonic regionalization and commitment (Furlan et al., 2017; Raj et al., 2020). This enables the generation of region-specific neurons and, in some cases, successful brain regeneration (Alunni and Bally-Cuif, 2016; Joven and Simon, 2018; Lust and Tanaka, 2019; Jurisch-Yaksi et al., 2020), as in teleost fishes (e.g., zebrafish and medaka) and urodeles (e.g., axolotl and salamanders) whereas in anurans this capacity is lost upon metamorphosis (Endo et al., 2007; Berg et al., 2011; Kyritsis et al., 2012). In teleost fish, in physiological conditions, newly generated neurons remain mostly confined within their domain of origin and are supposed to acquire region specific identities, although their fate has been verified only in some regions, mainly in the pallium and subpallium (Adolf et al., 2006; Hinsch and Zupanc, 2007; Ganz and Brand, 2016; Lange et al., 2020). In reptiles, besides the olfactory bulb and the medial cortex (homologous to the DG) new neurons are also generated in various telencephalic regions, including the striatum and amygdala (Lopez-Garcia et al., 1988; Font et al., 2001; Delgado-Gonzalez et al., 2011; Grandel and Brand, 2013). In birds, the identity and integration of adult-born neurons have been characterized in greater detail. Within the dorsal pallium, RG cells give rise to different subtypes of glutamatergic projection neurons that integrate into song-related circuits (Nottebohm, 2004; Scott and Lois, 2007). Conversely, subpallial RG cells generate both interneurons that migrate to pallial regions and striatal projection neurons (Scott and Lois, 2007; Kosubek-Langer et al., 2017).

In mammals, SGZ progenitors retain the embryonic commitment of their domain of origin, generating glutamatergic granule neurons throughout life (Bond et al., 2021). In contrast, V-SVZ progenitors, despite originating from multiple pallial and subpallial embryonic domains (Figures 2A,B), converge to produce OB interneurons (Merkle et al., 2007; Dodson et al., 2015; Fuentealba et al., 2015). These interneurons belong to the LGE-MEIS2/PAX6 neuronal class that, during embryonic development, originates mainly from the dorsal lateral ganglionic eminence (LGE; Stenman et al., 2003; Schmitz et al., 2022). With the establishment of the postnatal V-SVZ, this neuron class starts to be produced also by progenitors derived from other embryonic domains (Kriegstein and Alvarez-Buylla, 2009; Bayraktar et al., 2014). In adults, the embryonic regionalization is preserved but repurposed to generate different OB interneuron subtypes (Merkle et al., 2007; 2014). New neurons differentiate but only about half of them integrate into mature circuits (Kempermann et al., 2004; Deshpande et al., 2013; Akers et al., 2014; Sailor et al., 2017; Denoth-Lippuner and Jessberger, 2021). Some progenitors are activated on demand to generate specific OB interneuron subtypes only in specific conditions (Paul et al., 2017; Chaker et al., 2023; 2024).

Heterotopic transplantations indicated that subtype identity depends on the intrinsic commitment of astroglial progenitors (Merkle et al., 2007). Interestingly, these fate choices are only weakly and incompletely discriminated at the transcriptional level, both in the progenitors and in neuroblasts (Tepe et al., 2018; Cebrian-Silla et al., 2021) suggesting similar differentiation trajectories. When the switch to OB interneuron commitment is established remains unclear, but it likely coincides with the gliogenic switch, when the RG begins to generate astrocytes and oligodendrocyte precursors (Fuentealba et al., 2015; Zhang et al., 2020; Lee et al., 2024; Wang et al., 2025). The convergence of diverse progenitor domains toward a single interneuron class likely represents a specific evolutionary adaptation of the mammalian telencephalon, possibly related to the evolution of astrocytes.

Whether neuronal progenitors retaining embryonic commitments persist in a quiescent state in canonical niches or in the brain parenchyma is still unclear. In some mammalian species, low-level neurogenesis can occur outside the two canonical niches in physiological conditions, particularly in the striatum, where neurogenesis has been proposed to occur also in humans (Bédard et al., 2002; Luzzati et al., 2006; 2014; Ernst et al., 2014; for reviews see Bonfanti and Peretto, 2011; Feliciano et al., 2015; Jurkowski et al., 2020). Brain lesions can further stimulate parenchymal neurogenesis also in the cortex and striatum of rodents (Kernie and Parent, 2010; Magnusson and Frisén, 2016; Williamson et al., 2019). The newborn parenchymal neurons may arise from either the V-SVZ or local progenitors, but their identity and circuit integration remain poorly defined. Early studies proposed that these neurons could adopt appropriate region-specific fates, such as corticospinal neurons after focal cortical ablation (Chen et al., 2004) or medium spiny neurons, the striatal projection neurons, after stroke (Arvidsson et al., 2002), yet such claims have not been subsequently confirmed (Liu et al., 2009; Luzzati et al., 2011; Wei et al., 2011; Magnusson et al., 2014; Nato et al., 2025). Instead, a recurrent and highly consistent observation across independent studies and experimental models is that most newborn parenchymal neurons are short-lived (Gould et al., 2001; Arvidsson et al., 2002; Chen et al., 2004; Luzzati et al., 2006; 2011; Ohira et al., 2010). In line with the neurotrophic theory (Levi-Montalcini, 1987), the physiologic loss of newly generated neurons, reaching up to 50% in adult-born neurons and developing cortical interneurons, is commonly interpreted as a selection process, in which differentiating neurons compete for integration (Kempermann et al., 2004; Denoth-Lippuner and Jessberger, 2021). Similarly, the poor survival of newborn parenchymal neurons has been interpreted as a strong selection, caused by a non-permissive environment. However, transplanted embryonic precursors survive and integrate into the adult parenchyma (Fricker et al., 1999; Wernig et al., 2004) even in the very same environment in which locally generated cells die (Luzzati et al., 2011), suggesting that their survival is regulated by cell-intrinsic factors.

During development, immature neurons can play transient roles in circuit assembly that are unrelated to their functions in mature circuits (Cossart and Garel, 2022). Similarly, in adult niches, immature neurons are endowed with special plastic capacities that affect local circuits before the critical selection windows (Ge et al., 2007; Akers et al., 2014; Sailor et al., 2017). In this view, the initial number of neurons generated may meet the demand of transient plasticity, while selection may subsequently fine-tune this surplus to match the functional requirements of mature circuits. Accordingly, during development, the extent of neuronal selection can range from negligible (Sarma et al., 2011; Gao et al., 2014), to partial, as in the case of interneurons, up to neuron types such as Cajal–Retzius or subplate neurons that, despite integrating into circuits, are almost entirely eliminated by the end of development (Cocas et al., 2016; Riva et al., 2019; Molnár et al., 2020; Elorriaga et al., 2023). In adults, pregnancy induced OB interneurons have a transient life that is dependent on the presence of pups (Chaker et al., 2023).

To explore whether neurons generated by striatal astrocytes after QA lesion correspond to a transient type, we recently combined morphological, functional, connectivity and transcriptomic analyses (Nato et al., 2025). Despite their transient life, these lesion-induced neurons mature morphologically and functionally, integrating into circuits. Single-cell RNA-seq revealed that they do not belong to striatal neuron lineages but rather to the same LGE-MEIS2/PAX6 class as the OB interneurons generated in the V-SVZ. Re-analysis of neurons generated by striatal and cortical astrocytes after Notch abrogation (Magnusson et al., 2020; Zamboni et al., 2020) revealed a widespread commitment of telencephalic astrocytes towards this neuron class. Although this class was thought to contribute only to OB interneurons, in primates, they were recently shown to transiently populate the embryonic striatum and cortex (Schmitz et al., 2022; Wang et al., 2025). Spatial transcriptomics revealed that these cells are also present in the mouse cortex and striatum during embryonic and postnatal development (Nato et al., 2025). A more detailed analysis using Sp8, a specific marker of these cells in the striatum, showed that they are transiently populating the rostro medial striatum up to the weaning period. In this same developmental stage, similar transient Sp8+ cells are present in the lateral striatum of the guinea pig (Figure 4; Luzzati et al., 2014). Striatal LGE-MEIS2/PAX6 cells may thus represent a new developmental player reused in a context-specific manner to support plasticity.