Around 320 million years ago the first animal able to reproduce on land, known as an amniote, emerged from the sea. All reptiles, birds, and mammals evolved from this common ancestor, so researchers who study brain evolution look for commonalities and differences in the brains of these species that might help them deduce which regions and cells are conserved (and therefore “ancient”), and which might be evolutionary innovations specific to the species.

Traditional approaches involve comparing brain regions in terms of connectivity patterns, cell architecture, gene expression, and developmental origin. This suffices for anatomical regions, leading to the idea that the brain consists of shared, and thus “old”, and specific, or “new” regions. For instance, it is widely believed that the mammalian cortex, which houses higher brain functions, is new, whereas sub-cortical regions are usually ancient and shared. “A lot of claims are made that this brain region in mammals is what leads to their special capabilities,” says David Hain, a graduate student at the Max Planck Institute in Frankfurt, Germany. “But it’s actually not well-known which building blocks are ancient, and which are mammalian novelties.” Furthermore, traditional approaches do not resolve cell types, so it was not known if this model holds at the level of individual cell types within regions.

Here be dragons

A package of four papers, published this September in Science, tackles this question for some lesser-studied species – lizards and salamanders – using single-cell RNA sequencing (scRNA-seq) technology. Hain and colleagues generated a whole-brain cell-type atlas of the Australian bearded dragon, a reptile. Jamie Woych, of Columbia University, New York, US, and colleagues, profiled the developing and adult salamander telencephalon (a region which, in mammals, contains the cortex). Katharina Lust, of Vienna Biocentre, Vienna, Austria, and colleagues, studied the telencephalon of the axolotl, a salamander, both during development and in its capacity to regenerate. Xiaoyu Wei, of the Beijing Genomics Institute, China, and colleagues, also studied axolotl brain regeneration.

Single-cell sequencing was critical to all four studies. “It allows the study of thousands to hundreds of thousands of cells at the same time,” says Hain. “So it enables us to get a global picture of the brain.” Previous approaches have used marker genes to look for analogous cells across species, but cells may differ in that gene, yet still be very similar and derive from the same ancestor. “Previously, people have missed some of these similarities, because they were focusing only on a subset of genes,” Hain says. Sequencing entire transcriptomes avoid this problem.

Hain et al. used scRNA-seq to profile 285,483 cells from the bearded dragon brain and identified 233 distinct neuron types. They integrated this data with publicly available mouse data and found lizard and mouse neurons that roughly correspond in terms of location and neurotransmitters. The cell types exhibited distinctive combinations of transcription factors important for development, and genes involved in neuronal connectivity and signaling. This suggests the existence of broad, conserved types inherited from amniotes, defined by both developmental origin, and their role in neural circuits.

A mosaic of old and new cells

Hain was also motivated by the observation that brain regions do not work in isolation. “If evolution changes one brain area, there’s probably pressure on other areas that are highly connected to it to change as well,” he says. Likewise, newer regions evolved from older regions, and so may still contain “old” cells. When the team inspected the cells with finer-grained analyses, this is indeed what they found. Some cells could be mapped to analogous types between species, but others could not. This held for all brain regions they studied, indicating that almost all brain regions are a mosaic of both conserved and new cell types. This was particularly pronounced in a region called the thalamus, which showed mostly conserved types in the medial (middle) region, and highly divergent types in lateral areas. The lateral thalamus projects to higher sensory regions, so this likely reflects the fact that those connected regions diverged extensively during evolution.

Finding a mixture of old and new cell types across the brain is at odds with any simple notion of old and new brain regions, but it tallies with the other studies. All three focused on the salamander telencephalon, and they all found highly conserved types of inhibitory neurons (cells that reduce the firing of neurons around them) in this region. Conversely, all four studies found that excitatory cells (which send outputs that increase neural firing in other regions) were far less conserved, suggesting this broad class are typically much more specialised.

Grow New Brains

Lust and colleagues used single-nucleus RNA sequencing (snRNA-seq) and a technique called single-nucleus assay for transposase-accessible chromatin (snATAC-seq, which assesses where DNA is accessible for transcription), to catalogue cell types in the axolotl telencephalon, and cutting-edge spatial transcriptomic techniques to locate cell populations in the brain. The team found excitatory neurons similar to mouse and turtle cells in three regions: the hippocampus (a critical memory region), the dorsal cortex, and the olfactory cortex. They then showed that the olfactory cortex-like neurons receive input from the olfactory bulb, reinforcing the conclusion that these are conserved cell types involved in processing smell.

To study the axolotl’s capacity to regenerate its brain, the team used a technique called Div-Seq, which combines snRNA-seq with a method that labels dividing cells with a marker incorporated into their DNA. Of particular interest was a type of cell called ependymoglia. These are the predominant type of glia (non-neuronal, “support” cells) in the salamander, and have similarities to both a kind of epithelial cell in mice, and progenitor (stem) cells. They are responsible for generating neurons during development, maintenance of normal states (homeostasis), and during regeneration. “We wanted to understand how brain regeneration works at the level of cells and molecular pathways,” Lust says. “And how similar is it to regular homeostatic neurogenesis?”

The researchers excised a 1mm cube from the olfactory cortex-like region, then labelled cells at multiple time points during regeneration, from two to 26 days after injury, and sequenced their RNA at points from one to 12 weeks afterwards. “By doing this we could really understand the transcriptome profile of the new cells made during regeneration,” Lust says. Cells proliferated and formed new neurons, leading to restoration of all previously existing cell types. Precisely how the process knows what cell types to generate is unclear. “This is something we have to look at in the future,” Lust says.

The process was mostly the same as homeostatic neurogenesis, with one key difference. The researchers identified a transcriptional state in ependymoglia, specific to injury, in which gene expression related to wound healing and cell migration was elevated, early in the regeneration process. “Early on, these cells probably need to migrate to close the wound before they regenerate new neurons,” Lust says. “This is something they don’t do in the normal homeostatic brain.” Wei and colleagues also saw this distinct response to injury. Finally, the team used tracing methods to show that input connections were reestablished, suggesting some restoration of function. “We haven’t done functional studies, to see if these neurons are activated in response to smells,” Lust says. “But we see that there is again input from the olfactory bulb to the regenerated region, which probably leads to recovery.” Establishing the extent to which regenerated neurons are fully integrated into functional circuits is another direction Lust plans to pursue in the future.