Abstract

Narayanan “Bobby” Kasthuri combines the power of electron microscopy and supercomputing to construct connectomes—comprehensive wiring diagrams of the neurons and synapses in the brain. In 2015, Kasthuri joined Argonne National Laboratory, becoming the first neuroscientist to be hired at the U.S. Department of Energy Laboratory. Kasthuri is also an assistant professor in the Department of Neurobiology at The University of Chicago.
In this interview, Kasthuri discusses how his team uses connectomics to examine what makes the human brain unique and the differences in the brains of those with mental illnesses—and explains how he plans to apply these insights to design better treatments for patients.
Although in neuroscience, we often talk about neural circuits, no one has actually ever seen a circuit—maybe with the exception of one animal called Caenorhabditis elegans, for which we have mapped out every neuron. For most circuits, no one's asked: How many neurons are in this circuit versus that circuit? Are there any neurons in common? That was because mapping brains at the level of neuronal connections was a hard technical problem.
I realized that if we wanted to have a top-down physical understanding of how mental illnesses work, then we needed a map of the brain, just like we have a map of the genome. With the map of the genome, it's much easier to find your mutation and map it onto this reference atlas. The analogy is, if you have a map with the wiring diagram of a healthy brain, you can then compare it to brains from individuals with mental illnesses or other diseases.
The second thing is more of a philosophical difference. What I'm interested in is slightly tangential to other projects. I think getting the wiring diagram of a brain is awesome, but that it is kind of useless unless you're comparing it to something. In some animals, like mammals, we know that the brain rearranges as a result of that animal's experience. So we may get a wiring diagram that's only unique to that animal.
I've always been of the mindset that what we really want to do is to compare across brains. We're interested in what makes the human brain so special. I love the fact that we've done a fly brain. I love the fact that we're going to do a fish brain someday, and then maybe a mouse brain. Those are all great systems to study, but they really are poor analogies of the human brain.
The other thing I noticed in kids broadly—including me when I was a kid, I assume—is that there are tons of mental illnesses in kids that we just refuse to diagnose. Let me give you an example. It turns out that my daughter, who's in sixth grade now, had an imaginary friend when she was in first grade. When I talked to my mom about it, she said, “Oh, yeah, you had an imaginary friend, too.” At the time, I was just out of medical school, and I took a sheet that you're supposed to use to interview schizophrenics, and I used it to interview my daughter. And I realized it wasn't a joke to her. Her imaginary friend was real, and she could really hear those voices.
There are a bunch of behaviors that are perfectly acceptable in children but not acceptable in adults. I have a suspicion that some mental illnesses are these Peter Pan syndromes—brains that just refuse to grow up. And what we do is we keep the behaviors of childhood into adulthood when it's inappropriate. That would imply that we keep the wiring diagram of childhood into adulthood as well.
Someday, I hope we'll be able to do the wiring diagram of a baby human brain and an adult human brain, and compare that to the brain of an adult with schizophrenia. My suspicion is that the adult schizophrenic wiring diagram will more match the baby's wiring diagram than it will the adult's. Right now we're working on that with animal models—but addressing that in humans is the ultimate goal.
Then we do statistics on it to ask questions such as, “On average, how many connections did a baby neuron make versus an adult neuron?”
In terms of personalized medicine, there is a separate project that we've been working on. There are lots of moments when people take biopsies from humans to make predictions about a disease, to make predictions about treatments, etc. A specific project we're working on is with people who have Parkinson's disease.
One of the main therapies for Parkinson's is something called deep brain stimulation, where you put an electrode into the patient that stimulates the affected brain region, enabling their movement to be restored. It's a kind of remarkable treatment—but for some reason it fails in some patients, and it's not clear why.
With a neurosurgeon here, we've been taking brain biopsies from the patients they put deep brain stimulating electrodes into, and we've been trying to analyze the synapses in that volume. The idea is, if all the synapses are gone, it doesn't really matter how much you stimulate the brain. That region is no longer connected to any other region, so your stimulation isn't going to have much effect. If we knew how many synapses were there, we think we can predict the patient's outcome with something like deep brain stimulation. We think the same is true for epilepsy. Again, these are early days, but this is how I imagine pushing this into direct help for patients.
