Organoids, Autism, Mice, and Men

Modeling neurodevelopmental disorders in a dish.

Posted Oct 17, 2020

Anyone who follows progress in the biological sciences at a distance might have spotted the recent emergence of a new word: organoids. Those with a science-fiction bent might particularly enjoy the variant, ‘cerebral organoids,’ that now appears, not in a late-night horror movie, but in conventional neuroscience journals. This sounds like some brain-in-a-bottle super-consciousness come to take over the world, but in fact, it is much more interesting than that.

Until recently, what we understood about risk factors for developmental disorders such as autism and schizophrenia came from studies of the mouse. The disciplines of epidemiology and genetics reveal how some toxins and drugs are associated with neurodevelopmental conditions; how these conditions are also linked to certain genes; and how some immune disorders predispose to a subsequent diagnosis. Associations are fine, but if you wanted to understand how those risks led to the emergence of the disorder, then you had to turn to an animal model, usually a mouse. Mice can be manipulated to express risk genes. They can be exposed to environmental hazards. And from those studies has emerged our understanding of the biology of risk.

Such mouse models can be interrogated in two distinct ways: behaviourally and developmentally. Remarkably, the first of these has proven to be quite robust. For example, the gene SHANK3 is one of the most strongly associated with autism. It underlies Phelan-McDermid syndrome, a syndromic form of autism caused by a deletion of the end of chromosome 22. This SHANK3 gene lies in this missing piece, and without the protein this gene encodes, nerve cells have more difficultly building synapses, the connections between them. Mice with SHANK3 mutations have a remarkable behavioural correspondence to autism. They react differently to other mice, they groom incessantly, and they make funny squeaks. With the eye of faith, you can convince yourself that these accord with the social deficits, repetitive behaviour, and language difficulties displayed by autistic children.  

Researchers argue interminably about the validity of this comparison, and whether it really gives credence to the developmental studies that follow. If the outcome of this mutation in mice is similar to autism, then presumably the developmental path taken by these mice reflects that pursued by autistic children. If this is true, then the biochemical consequences of carrying the mutated gene will be similar for the mouse and the human. Ergo, where the mouse goes wrong, there too goes the child.

This theory only works if you don’t look too closely. Both being mammals, mice and humans have broadly similar brains and brain development, but the differences are almost as profound. A mouse builds its cerebral cortex in a week; a human takes months. The mouse then takes another week to wire its cortical circuits. Humans take twenty years. Mice, it turns out, aren’t quite human.

Enter ‘cortical organoids.’ I have written elsewhere of the breakthrough in stem cell biology that has given easy access to human pluripotent stem cells. You can now take a simple skin biopsy and grow skin cells. Then you can use a genetic trick to turn these cells into stem cells that can in turn make any tissue type you desire, including brain. Initially, this meant growing the cells in a two-dimensional sheet in a culture dish: human brain cells but not organised brain tissue. But now, thanks to the work of Madeleine Lancaster [1] and others, we can actually grow small pieces of brain tissue—organoids.

The first cerebral organoids were pretty disorganised, nothing more than confused aggregates of brain tissues. But as accuracy and reproducibility has improved, they’ve quickly become the model of choice. They reproduce on a small scale many aspects of human brain development, previously inaccessible to researchers.

Coupled to this is a second ground-breaking technology development: single-cell transcriptomics. If you want to understand as fully as possible what a cell is doing at any particular point in time, the best single measure is to ask: what genes is it expressing? This can reveal what physiological activities the cell is capable of, and crucially during development, it can tell you what that cell is likely to do next. The problem has been that transcriptomics—this measure of total gene expression—was only possible on large groups of cells. So in the developing brain, for example, one could measure what a few thousand or million cells were expressing jointly, but not what each individual cell was expressing. This posed a severe limitation, particularly in the brain where two neighbouring cells are frequently very different. Now, however, it is possible to dissociate a piece of tissue or an organoid, then analyse quantitatively every single gene expressed by thousands of cells that comprise that tissue. 

Just how powerful this combination of technologies can be is shown in a new study [2] by Themasap Kahn and colleagues. They have coupled organoids and single-cell transcriptomics with neurophysiological analysis to study another syndromic form of autism. DiGeorge Syndrome is caused by a different genetic deletion involving the unfortunate chromosome 22. Rather than the terminal portion of the chromosome going missing, as happens with Phelan-McDermid Syndrome, this deletion takes out a stretch in the middle of the chromosome containing perhaps 30 or 50 genes. The consequences are profound: heart problems, characteristic facial features, and cleft palate. The neurobiological consequences are widely varied: autism or schizophrenia, frequently with associated epilepsy.

The first thing to note about this new study is simply its power. The research team has managed to assemble 43 stem cell lines from 15 different DiGeorge Syndrome individuals. Until now, a study that had three to five lines was considered robust. Most studies of this genetic anomaly in mice would have used a single mutated mouse line, albeit with multiple individual mice. At a time when science is increasingly worried about reproducibility, this added power is exceedingly welcome. Noticeably, this tour-de-force is a collaboration between multiple high-powered Institutions including Stanford, UCLA, and various laboratories across the USA and beyond.

So what have they discovered? The first thing to notice is that the DiGeorge cells differentiate pretty well. There is no gross distortion in their ability to build neural tissue. This is reassuring. Whatever goes wrong in these cells, it is relatively subtle, and may well be reversible. Following that, the study reveals an electrophysiological fault in nerve cells derived from the DiGeorge lines in comparison with lines from neurotypical individuals. It’s a relatively subtle difference: the nerve cells can still fire a stimulus, but the threshold for firing is decreased, making these cells more spontaneously active than controls. The authors speculate that this might help explain the increased risk of epilepsy in DiGeorge individuals. The problem seems to lie with calcium channels, neuronal membrane components that play a crucial role in modulating the electrical activity of nerve cells. This isn’t the first study to implicate neuronal ion channels, indeed it is a familiar theme emerging in autism studies. We’ve known for some time, for example, that the gene, CACNA1G, which also encodes a calcium channel also underlies autism in some individuals. And this is one of several such genes now identified.

This discovery raises the prospect of future therapies for DiGeorge sufferers and adds to the developmental profile that is emerging for neurodevelopmental disorders more generally. But there was a further interesting finding that might cause a few eyebrows to raise. When the authors analysed the genes whose expression was altered in the DiGeorge lines, they found that they overlapped substantially with those reported earlier in a similar human stem cell model of 22q11 deletion. Reassuringly, this confirmed the reproducibility of this approach. There was, however, no overlap between this gene set and the set identified using a mouse model of 22q11 deletion. This raises the horrible possibility that whatever the similarities between the mouse and human response to this deletion, it does not involve the same genes. If that is the case, it may well not involve the same biological mechanisms. Human cerebral organoids may have arrived on the scene none too soon.


1. Lancaster MA, Renner M, Martin C-A, Wenzel D, Bicknell LS, Hurles ME, et al. (2013) Cerebral organoids model human brain development and microcephaly. Nature 501: 373-379

2. Khan TA, Revah O, Gordon A, Yoon S-J, Krawisz AK, Goold C, et al. (2020) Neuronal defects in a human cellular model of 22q11.2 deletion syndrome. Nat Med: doi: 10.1038/s41591-020-1043-9