Fifteen years ago, the proteins that Princeton neuroscientist Lisa Boulanger has staked her career on weren't even thought to exist in the brain. Known as major histocompatibility complex class I, or MHCI proteins, they are essential for an adaptive immune response. The thought at the time was that the brain was an area of the body where the immune system wasn't active. It simply wouldn't need MHCs.


An immunological surprise
As a postdoc at Harvard, Boulanger was studying how depth perception forms in the brain, and did an unbiased screen for the genes responsible. Unexpectedly, MHC genes popped up.

"We assumed it was a mistake," she said. "Because if you open any immunology textbook, the MHC chapter starts by saying it's found in most nucleated cells in the body except neurons."

But it wasn't a mistake. In the years that followed, it became clear that MHCs weren't just doing things in the brain with regard to vision, either. In the hippocampus, the brain's learning and memory center, MHCs alter the strength of communication among neurons. And MHCs help limit the number of synapses.

"If you have 10 synapses where you should have two, you'll have a big problem, potentially," Boulanger says. "Even though bigger brain, more synapses sounds like a great idea, it's actually not."

Boulanger wanted to figure out how MHCs were performing this critical job, and knew that yet another unusual suspect also controlled synapse density: the insulin receptor. In the rest of the body, those receptors help regulate the amount of sugar in the bloodstream. But in neurons, signaling through them increases the number of synapses.

Boulanger remembered decades-old studies that suggested that MHCs might affect insulin receptor signaling in liver cells and fat cells. It seemed like a long shot, but she wondered if MHCs were insulin receptors might connect to what MHC was doing in the brain.

A graduate student, Tracy Dixon-Salazar, first looked to see whether insulin signaling was normal in mice without MHC. Consistent with a connection, signaling in the mice was abnormally high.

A visual test
To find out for sure, postdoctoral fellow Carolyn Tyler used a drug to block insulin signaling in brain samples from either normal mice, or those without MHC. Then it was time to count synapses.

The most direct way, Boulanger says, is to physically count them. Princeton undergraduate Joseph Park spent dozens of hours snapping pictures of brain slices on an electron microscope. Zooming in 4,000 times closer than the naked eye can see, synapses come into focus and become countable.

"This is something that you see in textbooks when you're doing your training," says Tyler. "To see it in your own tissue on the scope, up close, it's really amazing."

Even most neuroscientists, she says, don't ever get a chance to personally see a synapse. To an untrained eye, the black and white images are difficult to decipher.

"If you're not used to it, you would look at this picture and say this is a really bad satellite image of a very crowded city," says Boulanger.

After mastering the skill of identification, Park saw that mice without MHC had about 20 percent more synapses than regular mice do. But in those same mice, the brain samples that had been treated with the drug were normal.

"When we fixed their insulin signaling using a drug, we fixed their synapse density," says Boulanger. "That tells us this is actually the way that MHC is changing the number of synapses in the developing brain."

The team published their results in the Journal of Neuroscience.

Manifold connection to disease
The findings might explain why inflammation — which increases MHC levels — might lead to insulin resistance and type 2 diabetes elsewhere in the body.

More troubling is the fact that MHC, even if in the brain, still can perform its immune duties, which include siting on the surface of cells and offering samples of what's inside to T cells searching for infection.

"You have one molecular machine that's moonlighting in these two places," says Boulanger. "You could have some wanted or unwanted interactions between those two functions of this one group of proteins."

Boulanger's results are also consistent with an emerging understanding of some types of autism as a failure of the brain to trim its many connections, says Manny DiCicco-Bloom, a neuroscientist and child neurologist at Rutgers Robert Wood Johnson Medical School.

DiCicco-Bloom says the same drug that Boulanger found so effective in fixing the brain connections in the mice without MHC is also used to treat a disorder called tuberous sclerosis. Caused by a mutation in one of two genes, it's often accompanied by autism. The drug, rapamycin, is used in short bursts to stave off tuberous growths in the heart, but it might also be doing more.

"We really probably can't sanction giving children with autism rapamycin," says DiCicco-Bloom. "But we can consider, and we have done a clinical trial with children who have tuberous sclerosis, to see whether it improves their social function."

Insight into our past
The clinical applications are intriguing, but Boulanger cautions that for now they are only speculative. On the other end, she's also thinking about what MHC's presence in the brain might mean in terms of evolution.

"Very primitive organisms have neurons and synapses, and the adaptive immune system is a relatively new thing," says Boulanger. "So what if [MHC] came from the brain and the immune system borrowed it?"

For Boulanger, that prospect offers an exciting possibility: that the secrets she's uncovered about immune proteins in the brain would return the favor, and help immunologists find the origins of our immune system.