Brains are bewilderingly complicated systems of connections between neurons. Mapping those connections is an important step in understanding how brains work. Scientists have recently completed the most ambitious effort yet to construct such a map: a complete document of every neuron and every connection in the brain of an adult fruit fly.
The research represents the first such map for an animal that can walk and see, and the first complete map of the brain of an adult animal. It traces each and every one of the 139,255 neurons in the brain of Drosophila melanogaster, along with the 50 million connections between them, and is by far the largest and most detailed ever produced. A similar map has been created for the larva of the same species, but that brain is much smaller, with only around 3,000 neurons. The brain of an adult also has to handle far more information and behaviors.
The map is described in two papers published October 2 in Nature. It is the result of a collaboration between a team of 287 researchers at 76 institutions around the world, and used over 100 TB of data.
Co-authors Phillip Schlegel, Sven Dorkenwald, Sebastian Seung, Gregory Jeffris, Davi Bock and Mala Murthy spoke to Popular Science about the landmark achievement.
There’s obviously an immense difference in size, and the fly’s brain includes structures like the mushroom body that we don’t have—but at the level of how neurons connect and how “wiring” is formed, how similar is a fruit fly’s brain to ours?
Philipp Schlegel: At the level of neuron-to-neuron connections, our brains and those of insects are extremely similar, which is why Drosophila is such a great model system for studying how brains work. That said, there are of course some differences and I often find myself more intrigued by instances of divergence than similarity.
Sebastian Seung: The mushroom body is a great example. It is true that this olfactory structure doesn’t exist in our brains. However, our piriform cortex is thought to be analogous to the insect mushroom body in its connectivity. (This analogy can be extended to the rest of the olfactory system.) Because fly and human genomes are similar, we have long known that fly and human brains must have similarities at the molecular level. We have been slower to realize that there are also similarities at the circuit level, revealed by examining patterns of connectivity.
[One question is] why there are circuit similarities, when the evolutionary ancestor of flies and humans is so ancient. Perhaps the similarities were produced by convergent evolution. At the molecular level, the insect and human olfactory systems seem very different; the genes for olfactory receptors are different. But the circuits may have ended up analogous because they have to solve the same computational problem.
Video: 3D rendering of all ~140k neurons in the fruit fly brain. Credit: Data source: FlyWire.ai; Rendering by Philipp Schlegel (University of Cambridge/MRC LMB).
Davi Bock: In both the fruit fly brain and ours, large networks of neurons somehow combine to process information, drive behavior, and store and recall memory. In both brains, neurons fire action potentials, use common neurotransmitters, etc. Both brains are characterized by massively recurrent connectivity; and both brains show signatures of interesting network structure we would very much like to understand in greater detail.
So while there are differences, the central question of how to organize large networks of neurons to process, store, and recall information will almost inevitably have common underpinnings across species. Figuring this out is a hard problem, and the adult fly seems to be a happy medium between “much simpler than human” and “still really interesting” in terms of information processing and behavioral capabilities.
To the extent that there’s such a thing as a “typical” neuron, how does a typical neuron in a fly’s brain compare to a human’s? Are they the same size? Do they have similar numbers of dendrites and synaptic terminals? Do they form similar numbers of connections?
PS: It’s non-trivial to define a “typical” neuron, as evident by the amount of cell types (over 8,000) we found in the fly’s comparatively small brain. In the fly’s visual system, for example, neurons are on average about 0.6mm “long” (i.e. the combined length of all the neuron’s branches) and have about 270 inputs and 500 outputs. Greg already mentions this below but saying that mammalian neurons are about 10x larger than fly neurons is probably not far from the truth.
In mammalian brains individual synapses are typically one-to-one, i.e. they form a connection between exactly two neurons. In contrast, insect synapses are typically one-to-many (“polyadic”), meaning that they connect multiple different neurons. Why that might be is up for speculation but perhaps it has something to do with insect brains trying to pack as much connectivity (and hence compute power) as possible into a very tiny smaller brain.
SS: A human neuron is generally much larger. A human neuron can extend from one side of the brain to the other, or between the brain and spinal cord. A giraffe or whale neuron can be larger still.
How does a fly’s neurochemistry compare to a human’s? Do we see all the neurotransmitters in a fly’s brain that we observe in a human brain, or only some of them? Do they play the same roles in both brains? And are there neurotransmitters in the fly’s brain that aren’t observed in the human brain?
PS: Flies use the same neurotransmitters (Dopamine, GABA, Acetylcholine, etc.) as us.
SS: The major neurotransmitters are the same, but there are differences in how they work. For example, glutamate is mostly excitatory in our brains, but usually inhibitory in the fly brain. There are also similarities though. For example, dopamine does seem to be important for “reward learning” in both fly and human brains.
Ultimately, are we looking at a brain that works in a manner that’s fundamentally similar to a human brain but smaller and easier to map? Or are there significant differences?
SS: This is a glass half-full or half-empty question. There are both similarities and differences. We now know that fly circuits for olfaction, vision, and navigation have architectural similarities with mammalian circuits for the same functions. What I mean is that analogous connectivity motifs exist in these circuits, much as similar architectural motifs might be found in different buildings.
The fly connectome is helping neuroscientists achieve the first really deep understanding of the function of any brain. At this point in time, any brain that we can truly understand helps us to understand all brains.
Why is this particular species of fruit fly so well-studied? What makes it an attractive subject?
SS: Drosophila melanogaster has been used as a model organism in biology for over a century, so it was naturally adopted by neuroscientists too. That being said, connectomes of other insect species are on the horizon: ants, mosquitoes, bees, etc. Comparing these connectomes and relating to the rich diversity of insect behaviors will be an exciting and fun area of research.
Mala Murthy: Flies are an important model system for neuroscience since their brains solve many of the same problems we do, and since they are capable of sophisticated behaviors like the execution of walking and flying, learning and memory behaviors, navigation, feeding, and even social interactions. My lab studies how the brain mediates social communication, and we discovered that flies continually use feedback cues, like sights and sounds, from a social partner to decide what action to produce at each moment in time—they can even make different choices in different contexts. This kind of complex behavior involves much of the brain, necessitating a complete map of all connections in the brain to solve it.
The papers discuss the idea of a “snowflake”, and the question of how representative any individual brain can be of a species. Is the question here that while the structure of a brain (its “circuitry”, to use an electronic metaphor) is similar across a species, the actual connections within that structure (the “wiring”) can be different? E.g. neuron x and neuron y could be connected physically in two brains, but in one brain the connection could be inactive while in another it could be active?
PS: That is an interesting question. In connectomics we tend to say a connection between two neurons is “strong” when it is made of many individual synapses. Based on our comparison between the FlyWire dataset and a previous partial brain map, strong connections are almost always present in both brains. You (and in fact many other neuroscientists) are now rightfully asking whether that “structural” strength of a connection necessarily translates into “functional” strength.
The short answer is: most likely yes. The slightly longer answer is: we don’t know for sure yet but as far as I am aware, functional connections reported in past physiological and behavioral studies also turn out to be structurally strong in the connectome. That said: this remains a big question which the neuroscience community will have to address in the years to come.
This raises the question of whether there’s some universality in “wiring”, too—i.e whether there are some neurons that are always connected, or “hard-wired”.
MM: There are indeed similarities in brain organization and it will be exciting to understand how these relate to function.
The fruit fly has about 10^5 neurons, and a mouse brain has about 10^8 neurons. How does the number of connections between neurons scale in comparison to the number of neurons? Does a mouse’s brain have 1000x the connections between neurons than a fly’s, or many more?
Gregory Jeffris: Actually, the number of synapses per neuron probably differs by no more than a factor of 10, perhaps partly because fly synapses are smaller than mammalian synapses and because of their [polyadic nature]. There may also be more individual synapses between strong partner neurons in the fly connectome (the champion connections between two neurons are in the thousands of individual synapses).
Finally, could you talk a little bit about exactly how having such a detailed map of the fly’s brain will allow for further research?
PS: Yes, absolutely. This brain map establishes a baseline against which future experimental connectomes can be compared. For example, one could train a fly to dislike an odor that would otherwise be appealing to the animal, then generate a map of that fly’s brain and compare it against FlyWire to see what changed on the circuit level. While EEG doesn’t work in flies (too small), overlaying a connectome with other modalities such as calcium imaging or electrophysiological recordings is something that is actively worked on across many research groups.
