While autism is a spectrum of disorders, it’s clear that the more significant cases involve physical differences in the brain’s nerve cells. Several studies have reported an excess in connections among neurons in the brains of people with autism. But when does this happen? Changes in neural connections are key components of learning and memory, and they can happen at any point in life; major reorganizations in connectivity occur from before birth up to the late teens.

Anecdotal reports of autism’s symptoms often suggest an onset between one and two years old. But a new study places the critical point extremely early in embryo development—at a point before there are any mature nerve cells whatsoever.

A series of challenges

Figuring out how autism starts is complicated. To begin with, it’s a spectrum that might include more than one disorder. You also can’t know in advance who’s going to develop it, so you can only look at it retrospectively, after the problems are apparent. Finally, the human brain is simply not something you can ethically do invasive experiments on.

The new work relies on techniques that weren’t available just a few decades ago. We now know how to take skin cells and convert them to stem cells. We’re able to direct stem cells to develop along the lineages that contribute to brain development. And we can structure that development in three dimensions to produce a miniature version of the mature tissue, termed an organoid. Combined, these approaches allow us to study the development of autism using nothing more than a small skin sample from autistic individuals.

For the new research, a large international team obtained skin cells from eight autistic people and five controls. These were converted into stem cells and then induced to develop along a pathway that leads to brain-like neurons. This pathway includes an intermediate step, called a neural stem cell, in which the cells are committed to developing as nerve cells but haven’t adopted a mature, specialized nerve cell identity (mature cells belong to distinct populations, like serotonin-producing dentate gyrus cells, etc.). As had been seen in past studies, the mature nerve cells derived from autistic individuals created very complex patterns of branching axons compared to control cells.

At five different time points during the development of these cells, the researchers separated out the nerve cells or nerve-cells-to-be. Then they obtained all the RNA from the cells, which provides a window into gene activity. Next, the researchers performed a computational analysis to identify groups of genes that were active at specific steps. This identified three distinct groups of genes (which they termed “modules”) that defined distinct stages of the developmental process. You can think of these stages as pre-neuron, neural stem cell, and maturing neuron.

Accelerated development

When these modules were compared in cells from autistic individuals and controls, there weren’t many differences in the two that marked later stages of development. The earliest active module, however, appeared to be active on an accelerated schedule in the cells that came from autistic individuals. In other words, while normal cells might reach a given stage of gene activity at day four, those from autistic patients might reach that at day two. This accelerated pace was also apparent in the physical changes the cells undergo as they mature.

The earliest two modules also contain a number of genes that had previously been identified as enhancing the risk of autism. And expression of some of these genes at early stages in the process could mimic the progression of autism, accelerating the developmental process.

The timing of all of this suggested to the authors that the problems in these autistic individuals came from the process of forming neural stem cells. This sets the stage for problems in everything that comes after it.

To test this idea, the authors came up with a clever solution. People have identified a way to bypass the neural stem cell stage of the process and force stem cells to develop directly into neurons. (Surprisingly, all this takes is the expression of a single gene.) If the specification of neural stem cells is where things go wrong, then skipping it entirely might rescue the problems. And, in fact, it does. The complexity of neural branching was similar in the experimental and control cells when neurons were generated using this approach.

We haven’t “solved” autism

It’s important to emphasize that this research doesn’t mean we’ve “solved” autism in any way. The participants in this study were selected as having a single symptom that clearly placed them on the autism spectrum; it’s not clear whether these results will apply to those who are on the spectrum due to other symptoms. And there’s a big difference between knowing something goes wrong during neural stem cell generation and knowing what, exactly, has gone wrong. So there’s still a lot of work to do here.

But the results do indicate that, at least in some individuals with autism, problems start extremely early. In humans, neural stem cells are specified before three weeks into the pregnancy—a point when many people aren’t even aware or certain they’re pregnant. Depending on how general this is, that may mean that interventions at the earliest stages of autism—either by directly addressing the problem or by limiting any environmental influences that promote autism—is pretty unlikely.

While this is an impressive body of work on its own, what’s really striking is how it puts together so many techniques that are relatively recent developments. These include the use of stem cells to study diseases that are otherwise difficult to address experimentally, the ability to do large-scale RNA sequencing, and the algorithms that let us analyze this data—all are relatively recent developments. Biology is filled with incremental developments, and it’s only when you stop to consider what had to happen before research like this was even possible that the rate of progress can be appreciated.

Nature Neuroscience, 2019. DOI: 10.1038/s41593-018-0295-x  (About DOIs).

https://arstechnica.com/?p=1438149