This is a technology site, and we don’t normally write about medical tech. But since I’m interested in brain-computer interfaces, I was struck by science writer Ed Yong’s article on "model human brains" last month in National Geographic. They’re jelly-bean-sized neural lumps that include rudimentary retinas and cortices, layered similarly to normal human brains, recently grown in an Austrian lab. The trick was to take human skin cells, chemically rewind them into stem cells, then coax these to become neurons. I thought what any reasonable person would: This is it. This is how we end up making those brain plug human-machine interfaces in The Matrix.
As Neuroskeptic speculated on the Discover magazine blog, I too wondered about the ethics of growing human brains in laboratories. What if the thing that develops from human progenitor cells is eventually able to complete the program encoded in its genes? What if we end up with Terminators? This process is cut short in the current versions, because the brains have no circulatory system, no blood to bring oxygen and nutrients to the cells in the center of the blob, so they develop malformed with all the sophisticated parts on the outside, "like a car with its engine on the roof" as one author of the paper put it. But some individual brain parts develop much like a normal embryo's, up to the ninth or tenth week.
All my heady sci-fi visions were scuttled by some grounded brain experts, but I learned some fascinating stuff. Here’s everything you ever wanted to know about “mini-brains.”
Matt Weber, University of Pennsylvania, Post-Doctoral Fellow, Psychology Department: One thing to keep in mind is that these little guys are developed to the equivalent of nine weeks old. That limits what you can do with them because they aren't responsive to the outside world, so their development is insulated from the organizing influence of sensory input. And this influence kicks in quite early in development: Fetuses move their limbs, they taste what their mothers eat, they learn their parents' voices. So the ability to generalize from anything we observe in these disembodied brains to even a 20-week-old fetus might be quite limited.
The upside is that we might be able to separate genetic contributions to development from experiential ones. But that assumes we can reconcile ourselves to the ethics of developing a disembodied brain to a stage where, if it were embodied, it could respond to the world. That's a deep rabbit hole, and I won't venture in except to note that "responding to the world" is a pretty low bar for meaningful mental life: Vegetative patients still have brainstem-mediated pupillary responses to light, and retinas detached from a brain not only respond to light, but actually can learn about regularities in visual displays. Which isn't to say there might not be good reasons to resist developing a disembodied brain that far--it's just that it's a complicated question.
Ania Dabrowski, University of Michigan Medical School, MD-PhD Doctoral Candidate, Neuroscience:
This paper is a developmental neurobiologist's wet dream. Deeming this a "mini-brain" is a bit of a stretch, however it's important to put this into the context of how we've been studying human neural development. The most common model for studying these questions in humans are mice. Because they're mammals, and they're kind of similar to us. But a mouse is not a human, and we don't know how so (kind of weird thing to say, right?). It seems obvious that there are differences between how human and mouse brains develop, but our resources to study this are limited!
In comes mini-brain. Suddenly, you can take human cells, and see how they form rudimentary brains. And you can manipulate human genes, and see how they are relevant to cortex development. Sure, it's not a real human brain. But it's also not a mouse brain.
Mini-brain is a tool. On its own, it's just a bunch of cells in a vat, not even close to Stanslaw Lem's conscious brain in a jar. But mini-brain is a powerful tool to answer specific scientific questions. And there are ultimately two scientific questions we're all interested in answering: One: How does the human brain work, and two: What's going wrong in various neuropsychiatric diseases and how can we fix it?
Stephanie Bielas, University of Michigan, Assistant Professor of Human Genetics:
There are many very common diseases that a lot of people are attracted to studying--Alzheimer's, Autism, Parkinson's, schizophrenia. I think that stem cell models [like mini-brain] will be invaluable, but at this particular juncture it can be difficult.
This paper is one example of microcephaly [small-brain disorder, a genetic condition that Bielas' lab studies, and which the mini-brain experiment modeled] as a model for stem cells, which is something I'm working on. It's not like it's a terribly unique idea to them. There's been a couple of other papers that have shown similar things, including one last year in Nature Neuroscience that did almost exactly the same experiment, to a T. I think that the only thing that differentiated this paper is a section of one of their, whatever, "organoids," happened to move off that looked like an olfactory bulb. So, it was a small break. But otherwise, [mini-brains] are very much like neural rosettes [what the rest of the field calls 3-D balls of neurons in culture, which aren't new]. I haven't compared the two papers word for word, but what's different? Maybe just talent at manipulating the press.
What you're developing in a dish, they're not organoids. It's not like you're developing small brains. I've always heard them called "neural rosettes." They have similarities to brains, but you have to know exactly what you want to look at, because they aren't just developing small brains in a dish.
Let's take schizophrenia: What cells are being affected in schizophrenia? I don't think anyone really knows or for sure knows the genes that regulate them or if it's a transmitter issue or some other network issue. You can get mature [neural] networks in a dish. Of course, they're not going to be in a brain, but there will be synapses with connections and everything. But: number one, it is exceedingly challenging to actually grow these things in a uniform way so that you can test drugs, but even if you could: What is the phenotype you are testing? You can't say "Are they having a hallucination? Are they having abnormal behavior?," which is the read-out at this point [for diagnosing schizophrenia]. It's not like we know "Okay, the cells get schizophrenia if we give them this particular drug; this cell has that response, that's schizophrenia." We don't know that!
Sam Wang, Princeton University, Associate Professor of Molecular Biology and Princeton Neuroscience Institute:
First, to rain on the parade fairly hard: The researchers were able to recapitulate a few aspects of neocortex development, namely a layered structure. And that's about it. So it's more like they created the layered version of one brain region. Kind of. No brainstem, no dopamine, acetylcholine, serotonin, opioids, neuropeptides [the brain transmitters, released by the brainstem, which drive emotion and motivation, and respond to mind drugs like antidepressants, ritalin, LSD, or cocaine], no cerebellum [the motor-control region which Wang's lab specializes in], no white matter [the myelin-sheathed high-speed axonal power-cords which speed signals across the brain]... Um, in what way is this a mini-brain? What would I do with it? Try it on toast points, maybe.
But seriously, what would I do? Maybe test if long-distance axons can really pull parts of the neocortex together to make folds. That's an intriguing idea about where brain folding patterns come from.
So even this toughest cynic (who, full disclosure, taught my first neurobiology course in college) concedes there are exciting questions to be answered with the new cultured brains. But rest easy--it’s not what it sounds like. Thanks are due all the scientists who contributed their points of view to this article, for showing both the prospects and limitations of farming "brains" in vats.
[Image: Flickr user Ekkehard Streit]