Throughout developmental biology, and indeed biology as a whole, structure and function are intimately linked. This relationship is elegantly exemplified by neural stem cells of the developing cerebral cortex. These cells, which are known as radial glia, are multipotent: in addition to generating neurons, they generate astrocytes, oligodendrocyte precursor cells and, later, ependymal cells. In mice, radial glia fibers extend from the ventricular surface of the cortex all the way out to the pial surface, creating a physical scaffold along which newborn neurons can migrate radially, from the ventricular zone (VZ) to the cortical plate. By contrast, the human cortex is massively expanded and more morphologically diverse. Outer radial glia (oRG, also known as basal radial glia) occupy the evolutionarily expanded outer subventricular zone (OSVZ) and lack physical contacts with the ventricular surface but retain radial fibers that extend to the pial surface. Additionally, truncated radial glia (tRG) are found in the VZ with processes contacting the ventricular but not the pial surface (deAzevedo et al., 2003, Nowakowski et al., 2016). Although oRG cells have been identified in small proportions in mice (Wang et al., 2011), tRG have not been observed in mice or other lissencephalic species studied to date, making it challenging to elucidate their function. Moreover, RG subtypes in humans are not just morphologically and spatially distinct, but also transcriptomically separable (Nowakowski et al., 2016), hinting at a functional relevance of the scaffold split. Indeed, tRG and oRG cells may generate morphologically and molecularly distinct subtypes of astrocytes (Allen et al., 2022), but the broader role of tRG cells in cortical development, especially late in gestation, is unknown. A recent preprint by Bilgic et al. (2022 preprint) address these questions surrounding tRG using the ferret as an model system.
Ferrets are gyrencephalic small mammals in the order Carnivora, making this study the first to describe non-primate tRG. Using immunohistochemistry for the tRG marker CRYAB, Bilgic et al. reveal that ferret tRG, like human tRG, emerge during peak neurogenesis. In the ferret, this is around embryonic day (E) 34, a few days before birth. The tRG pool grows until postnatal day (P) 14, which spans the majority of gliogenesis, then diminishes by P35. Next, using single cell RNA sequencing of ferret cortices spanning these ages, the authors reveal that, as in the human, ferret tRG are transcriptomically distinct from oRG. The authors then compare ferret radial glia subtypes to previously published single cell data from human samples (Nowakowski et al., 2017). They show that both tRG and oRG from ferrets are transcriptomically similar to those found in humans. However, tRG only make up about 5% of the radial glia identified in ferrets and many express the ventricular radial glia (vRG) and oRG marker HOPX. Some vRG remain even at late stages in the ferret, reinforcing the idea that a complete scaffold split does not occur in ferrets as it does in humans mid-way through neurogenesis (Reillo et al., 2011; Nowakowski et al., 2016). Together, these comparative studies establish the presence of tRG in ferrets and open the door for similar studies in other species.
Next, the authors ask the question: how are tRG generated in the ferret? To address this, they use in utero electroporation of a plasmid that expresses GFP under the control of a Hes5 promoter along with the thymidine analog EdU, followed by live imaging of slice cultures. They identify two different modes for generating tRG. In all cells observed, however, CRYAB+ cells are generated as progeny of cells that already lack a basal process and divide asymmetrically, suggesting that tRG are, perhaps surprisingly, not generated directly from proliferative divisions of vRG. Live imaging is challenging to do at scale, but more examples would strengthen support for this model of tRG generation. As the authors point out, it is unclear whether the ‘mother’ cells they observe are already CRYAB+ tRG, as staining is performed post-fixation. Further characterization using this ferret slice culture system could reveal which factors contribute to the dramatic tissue-wide decision to split the neurogenic niche.
Finally, the authors leverage the unique access to late gliogenesis afforded by the ferret model system to address the function of tRG. Specifically, they ask: what cell types do tRG generate in the cortex? As the tRG population is greatest during late neurogenesis through gliogenesis, the assumption is that tRG are capable of generating both neurons and glial cells. Interestingly, tRG pools diminish around the time when post-mitotic multiciliated cells that line the ventricles in adults (known as ependymal cells) begin to emerge. Given this location and timing, as well as the radial glial origin of ependymal cells in mice (Spassky et al., 2005), it is tempting to wonder whether tRG give rise to ependymal cells in ferrets and humans. To address this, the authors begin with co-expression analysis. They find that the proportion of CRYAB+ cells that co-express FOXJ1, a canonical ependymal cell marker, increases steadily during the early postnatal period, suggesting a steady transition from tRG to mature ependymal identity of cells lining the ventricles. Using pseudotime trajectory analysis from their single cell data, they also find that the cluster containing most tRG (NPC3) also contains cells that express FOXJ1, as well as AQP4 – an astrocyte marker. Although this provides strong evidence for tRG as the origin of ependymal cells in the ferret based on timing, location and pseudotime analysis, demonstration of a bona fide lineage relationship between tRG and ependymal cells will require future investigation.
Overall, this study demonstrates that the developing ferret cortex contains an increased diversity of radial glia, similar to that observed in humans. The investigators also demonstrate that tRG give rise to ependymal cells, highlighting that the specification of tRG and oRG may represent a lineage splitting event during cortical development that generates two divergent stem cell populations. Further studies are needed to determine what other developmental roles may be attributed to tRG cells, and what adult brain cell types may be derived from tRG cells specifically. The study by Bilgic et al. also highlights that the ferret may serve as an animal model for addressing some of these questions experimentally. Finally, we know that oRG-like cells exist in the mouse in small proportions – are there also tRG-like cells that simply have not been identified because they do not express the canonical tRG marker CRYAB? Perhaps mice and men are not so different after all.