Notes on: The development and evolution of inhibitory neurons in primate cerebrum
A paper written by me, my mentor Alex Pollen, et al., vaguely about what makes us human :)
Read the real paper at: https://doi.org/10.1038/s41586-022-04510-w
TL;DR
- We used single cell RNA sequencing to look at all the different types of inhibitory neurons that exist at 6 points during rhesus monkey brain development, and compared them to similar public mouse developing brain data
- We find that as soon as newborn inhibitory neurons are produced by dividing stem cells, they stop dividing and choose one of about 16 “initial class” categories based on where they are born.
- These initial classes undergo a process of maturation and eventually take on new identities as the “terminal neuron classes” of the adult brain.
- We use RNA velocity to predict which initial classes become which terminal classes in the adult brain.
- We show that TAC3+ Striatal Interneurons that only exist in the brains of primates are already distinct from the original striatal interneuron class during development.
- We show that initial classes of neurons that come from the LGE, which usually migrate straight to the olfactory bulb, also migrate into the cortex and striatum in larger numbers in primates than in mice.
- We find 3 pairs of mature twin neuron types, separated after birth, with one sister in the olfactory bulb and the other sister in a different region of the brain
- By following the redistribution of these types of neurons, we reveal the identity of a common deep white matter inhibitory neuron and a rare striatum laureatum neuron.
Cool coincidences (see the end of the article for a full discussion):
- The latest born neurons in the brain, known to play a role in maintaining plasticity in the olfactory bulb, are redirected into the latest maturing region of the human brain, the white matter of the prefrontal cortex.
- In addition, the most reduced region of the primate brain (the olfactory bulb) contributes more neurons to the most expanded region of the primate brain (the cerebral cortex). Here’s the “reduce and reuse” model we propose that could explain these coincidences:
Primate decreased reliance on smell means they have a smaller olfactory bulb. Meanwhile, Expanded cortex for thinking and generally being human means expanded regions where OB neurons are born. Perhaps the excess olfactory neurons without a home in the bulb go to the cortex to be “reused” in white matter!
Is this more than just a coincidence? Only future research will tell!
Overview of the Paper
or:
How to stumble your way into a string of awesome discoveries
Every neuroscientist has an argument for why the type of neuron that they study is the most interesting. One could argue that cortical excitatory neurons are the most interesting because the larger cortex is the single most defining aspect of the human brain, but one would be WRONG because inhibitory neurons are clearly the most fascinating neuron type. First of all, they are crucial for intelligence, providing balance to excitatory neurons to prevent the brain from sliding into mass excitation like epilepsy. They also are the key to the computational processing units of the brain, being the primary mechanism by which logic gates can be encoded in neural circuits. Beyond their function, their genesis is as exciting as the first moments of a sea turtle’s life: while most neurons of the brain are born right where they will live for the rest of their life, cerebral inhibitory neurons of each type are born in a specific zone, from which they migrate great distances to the many places they’re needed. With these complexities, there are many things we don’t know about these processes in mice, and the developing brains of primates have been studied even less.
We initially assumed that we would find exactly the same neurons in the mouse and macaque brains, and so we set out to quantitatively compare and contrast what features made the two species’ excitatory and inhibitory neurons different. Given how different rodent and primate brains end up being, we cast a wide net hoping to find differences during development that might snowball into the massive adult differences. We therefore performed single cell RNA sequencing on a number of regions from 6 different timepoints during rhesus monkey brain development, and collected many different public mouse datasets to compare primate to rodent neurons. I bioinformatically separated the inhibitory neurons from the excitatory neurons using a simple cutoff for cells that expressed well known marker genes for inhibitory neurons (GAD and DLX genes). Once I had a nice clean set of inhibitory neurons, I found that there was no magic way to group the cells into meaningful clusters. Neurons, it seems, are like children in that they are more like each other than any one might be to their adult self. What I ended up doing was making more groups of neurons than I thought there were, then merging groups by hand by going through the types of genes that defined each from the other groups.
In doing this process, I found that the data appeared to support a model as follows: Inhibitory neurons in the brain come from divisions of progenitors in the ganglionic eminences, and as soon as they stop dividing they immediately adopt one of a discrete number of “initial class” identities. These initial classes differentiate along continuous trajectories and appear to eventually choose a crystalized terminal class that is a subclass of their initial class.
We took great care to go through each type we detected and to go through the literature to give it a proper annotation. The neurons of the brain have been studied in depth for more than a hundred years, so every reasonably common type of neuron is guaranteed to have been seen by someone, at some point. By looking through the literature and hundreds of immunostains we did, we gave each distinct class of newborn neurons a standard name. We also made a taxonomy of all the initial classes of neurons that are born, and what terminal classes they appear to turn into as they mature. By doing this, we were able to see the full diversity that exists during brain development, and also ask if there was anything there that shouldn’t have been.
The evolutionary birth of a neuron
As Fenna Krienen nicely articulated in her 2020 study (https://doi.org/10.1038/s41586-020-2781-z), there are a number of ways that cell types can evolve to change the way tissues function. First of all, a cell type can become more or less abundant. Secondly, gene regulation can be changed in neurons that changes the properties of a cell type. Thirdly, a cell type can change its location, which may or may not mean it is doing the same job in a different context. Lastly and most dramatically, entirely new cell types can come into existence (great review here https://doi.org/10.1038/nrg.2016.127). In the same study, she highlighted that in adult primate brains, the only example of a novel cell type (defined by transcriptome) was the TAC3+ striatal interneuron. We were very excited to find that during development, we could easily distinguish a TAC3+ initial class from the ancestral class of striatal interneurons. We found that even in development, the two classes of primate striatal interneuron (MGE_CRABP1/TAC3 and MGE_CRABP1/MAF) have gene expression differences that would affect how the two classes function, like TAC3+ cells expressing receptors for acetylcholine. Because TAC3+ striatal interneurons only exist in primates, it is difficult to ethically study them in living tissue. This developmental snapshot can help future researchers figure out how these cells may be derived from stem cells in a dish so that they can be studied in more detail.
Neurons where they don’t belong?
To return to our original focus, once we had enumerated the initial classes of inhibitory neurons, we looked at the region from which the different classes was derived. The general rule of thumb for where neurons born in the ganglionic eminences is as follows: CGE neurons go to the cortex, MGE neurons go to the cortex and basal nuclei, LGE neurons go to the basal nuclei and olfactory bulb, but NOT the cortex. Imagine our surprise (read: confusion) when nearly all of our frontal lobe cortical samples contained MGE and CGE neurons, along with a mystery inhibitory neuron that expressed MEIS2. Had we sampled the macaque olfactory bulb at the time, I think this would have been immediately obvious, but since we didn’t we played a little game of “where’s Cell-do” with the cells, since to do scRNAseq, you blow up the tissue and don’t know where in your chunk of tissue you dissociated a cell came from or what it looks like.
From our immunostains, we eventually determined that the mystery neurons were coming from the very edge of the LGE (the dLGE). Because this is generally known to be the place where many olfactory bulb neurons come from, this led us to compare our data to the public mouse olfactory bulb data to see if our mystery neurons matched the profile of any olfactory bulb neurons.
When the data showed a clear correspondence between the LGE types and olfactory bulb granule cells (LGE_MEIS2/PAX6), the whole story started to unfold. In the stain below, you can clearly see that large numbers of neurons from the same place diverge, with one group traveling the highway of cells to the olfactory bulb (RMS) with other cells, and others follow a previously unknown route straight into the prefrontal and anterior cingulate cortical white matter (Arc-ACC).
It was especially exciting because the neurons streaming into the cortex correspond to possibly the most studied cells of the olfactory bulb, the adult born “granule cells”. In almost all parts of the brain, the neurons a mouse is born with are essentially all the only neurons they will ever have. However, the olfactory bulb is one of two places where neurons continue to be replaced throughout life (the other is the hippocampus). We know that in humans, these olfactory bulb neurons continue migrating into childhood. It will be very interesting to see if these neurons continue to migrate into the cortex as well!
Another interesting thing that we stumbled upon while looking for these olfactory bulb-like cells in the cortex was that it seemed the granule cells weren’t the only olfactory bulb neurons to have strayed from the path. Starting from the granule cells, we followed the thread–if one type of neuron from the olfactory bulb had found a new home, perhaps there were others like it. Indeed there were. In addition to a correspondence between olfactory bulb FOXP2+ periglomerular cells, superficial white matter inhibitory neurons and eccentric spiny neurons (LGE_FOXP2/TSHZ1) (it’s a long story), we also found perhaps the second-most famous olfactory bulb neurons, the dopaminergic periglomerular neurons, in an unexpected place.
The dopamine-producing neurons of the olfactory bulb are well-studied, as there are only a few small places in the brain containing neurons that produce dopamine, a key neurotransmitter for pleasure/reward, motor and a number of other systems in the brain. As the scRNAseq hinted that there might be TH expressed in our mystery cells (the TH gene codes for a key enzyme that makes dopamine), we were searching for these cells in the cortex by immunostaining. We were lucky that I always scan the entire brain slice in all colors with the microscope, and were able to catch a few highly distinct neurons clustered around the edge of the striatum. While we don’t yet know if these cells actually produce dopamine, from what we can tell their profile matches that of the olfactory bulb dopaminergic neurons. Since these cells were not previously described, we decided it would be helpful for finding and studying them in the future if they had a name. Based on the rough, wreath-like shape one might imagine them to form along the external striatum of the whole brain, we called them striatum laureatum neurons, Latin for neurons which crown the striatum with laurels. Not really knowing what they do yet, perhaps it’s premature to give them such an honorable name. I, for one, hope they live up to it!
Why have these cells not been found before?
In many ways, they have. In 2003, Luca Bonfanti’s group observed chains of migrating neurons in the cortex in rabbits and a number of other species(https://doi.org/10.1073/pnas.1735482100), while in 1987, a study observed a density of cells expressing a protein in the dopamine pathway ringing the striatum (https://doi.org/10.1016/0304-3940(87)90298-9). Had these groups had the power of scRNAseq to identify all cell types simultaneously using scRNAseq, I have no doubt they would have quickly made the connection to match the identities of the cells in the cortex to those in the olfactory bulb. On the other hand, in the past 5 years, there has been a lot of work using the same techniques to try to put together a “parts list” for the brain. The idea here is that if you want to understand how a machine works, you first need to know all the different pieces it is made of. The problem with this approach is that while the mouse brain is roughly the size and shape of an almond, the human brain is enormous, and the vast majority of neurons of the cerebral cortex are on the surface. These neurons float above an ocean of white matter, the internal area of the cortex mostly consisting of axons running from neurons in the cortex to other parts of the brain that have relatively few neurons. It’s similar to the problem of counting all the species in the ocean. You can learn a lot looking at the easily accessible coast and surface waters. However, once you go below these rich, obvious areas, things get much farther apart and you need to survey a much larger area to find the diversity. Thus, we’re especially unaware of cells that are scattered sparsely or live in border regions in the primate brain, the hiding spots which wouldn’t be the first place you shine the flashlight while searching through the darkness.
[Speculating Wildly About] What Our Discoveries Mean
In this paper, we have tried to contrast two different mechanisms of cell type evolution. Firstly, we provide new insights about the duplication and divergence of the single ancestral class of striatal interneurons into two distinct classes in primates. This opens many new questions about how it is genetically programmed and how a new neuron type is integrated into existing circuits.
The neurons of the olfactory bulb appear to be particularly affected by primate brain reorganization. They represent the latest born neurons in the brain, which are known to play a role in maintaining plasticity in the olfactory bulb, AND they are redirected into the latest maturing region of the human brain, the white matter of the prefrontal cortex. In addition, the olfactory bulb is the most diminished region of the primate brain, and it is about 2% of the mouse brain by volume in contrast to about 0.01% of the volume of the human brain(https://en.wikipedia.org/wiki/Mouse_brain#cite_note-9)). Coincidentally it is contributing more neurons to the most expanded region of the primate brain (the cerebral cortex). The “reduce and reuse” hypothesis that we’re proposing and that I am now working to test can be explained as follows:
Decreasing reliance on smell in primates led to evolution shrinking the olfactory bulb. In parallel, the cortex and the rest of the cerebrum was expanded by the increased fitness of smarter primates. Because the birthplace of the olfactory bulb neurons is attached and likely was scaled up with the cortex, excess olfactory bulb neurons are produced and are sent to the cortex where they can be “reused”. Of course it’s also possible that the causality could go the other direction, where the demands of increasing cortex size, which may already have been using “olfactory bulb” neurons in the white matter in some distant ancestor (there is some evidence that this might be true in animals like pigs, cows and dolphins), drove the shrinking of the olfactory bulb by directly reallocating its neurons. There are more experiments to do to prove it one way or the other, but I’m on the case!
