Investigating Connectivity

Report from NYAS symposium
Carl Sherman
April 20, 2016
Slides of dendrites growing

As mice learned a new task, repeated imaging of the same neurons over one-day intervals showed remodeling of synapses, with new branches (dendritic spines) forming and others eliminated during training. Image courtesy of Xu et al

The brain is a dynamic mass of interconnected circuits, exquisitely engineered to maintain balance and stability in the midst of constant change. Unraveling its structural and functional connectivity in health and disease has been a dominant theme in modern neuroscience, and a recent symposium at the New York Academy of Sciences offered some insight into what we’ve learned.

The primary focus of the afternoon was the synapse, the point of contact between neurons, with particular emphasis on dendritic spines, key players in excitatory neurotransmission.

“Why do we look at spines so much?” asked David Sulzer, professor of neurology, psychiatry, and pharmacology at Columbia University. “They’re easy to see.” For visualizing other aspects of neurotransmission, such as presynaptic activity, “our techniques are not so good, and we’re missing a lot of interesting information if we ignore them.”

That said, dendritic spines provide a useful index of change in connectivity, as was evident in Bruce McEwen’s discussion of stress.

“We now know that dendrites can grow or shrink with experiences good and bad,” said McEwen, professor of neuroscience at Rockefeller University and a member of the Dana Alliance for Brain Initiatives. “With repeated stress, apical dendrites of the CA3 region of the hippocampus shrink in a reversible manner. Neurogenesis is reduced by certain stressors, enhanced by enriched environment and exercise.”

The hibernating European hamster offers “one of the most dramatic example of [such] plasticity,” he said. When its day length is artificially shortened, “the atrophy of apical dendrites occurs within hours, and can be reversed equally rapidly when you wake the animal up.” It appears that de- and repolymerization of the active cytoskeleton underlies these quick structural changes, he said.

Generally, stress-related corticosteroid secretion triggers excessive glutamine release, which shrinks dendrites over time. Other mediators, growth factors, cell surface and adhesion molecules, and neurotransmitter systems, are also involved, McEwen said.

“A stressful experience causes alterations in neural architecture, and in behavior,” he said. Although effects were first studied and are particularly evident in the hippocampus, they are distributed widely. When animals are stressed, dendrites shrink in the prefrontal cortex (PFC), leading to their expansion in the basolateral amygdala and orbitofrontal cortex, with associated increased anxiety and vigilance, and their shrinkage in the medial amygdala, which may explain reduced social interaction. The system normally returns to its pre-stress state when danger passes.

In a human study, stress in medical students preparing for exams was associated with reduced cognitive flexibility and functional connectivity within the PFC. The alterations promptly reversed after a vacation.

Such resilience apparently declines with age and in people with mood and anxiety disorders. Neurogenesis stimulated by regular exercise or intense learning experiences can redress some of these losses, however, as can treatment with antidepressant drugs.

Altogether, stress effects on brain structure and cognitive function can be best described as “an inverted U-shaped dose-response curve,” McEwen said. “Acutely, adrenal steroids and excitatory amino acids enhance synaptic function and improve some aspects of memory. More intense or prolonged stressors have the opposite effect.”

How dendrites learn

Wenbiao Gan, professor of neuroscience and physiology at NYU Medical Center, detailed the microanatomy of memory. “Synaptic connections are highly plastic,” he said. “These changes are important for development, adaptation to the environment, and learning.” At the same time, stability in these circuits is essential to retain information over time.

“To understand the basis of learning and memory, we developed a way to directly image dendritic spines in the mouse cortex,” he said.

Using two-photon microscopy, Gan and colleagues observed the same neurons over hours, days and months. “Most spines were remarkably stable,” he said: In the adult cortex, some 70 percent of those seen at 5 months remained 19 months later. At the same time, a substantial number of new spines were formed, some long-lasting.

Gan described experiments in which young mice were trained to adjust their gait to follow accelerations of a motorized device. The rate of spine formation doubled after a half-hour of training for the first few days, but slackened off afterward. “It’s when you learn something new that you form new connections,” he said. Over succeeding weeks, most newly-formed spines disappeared; 16 months later, only a small percentage survived.

With this process—initial creation of many new spines, followed by selection and stabilization— “the brain can keep and integrate new connections. A small number suffice to modulate circuit function without disrupting the connectivity behind storage of previous memories,” Gan said.

Dendritic growth related to learning isn’t uniform, he observed, but limited to subsets of a neuron’s branches. Training in a different task spurs spine formation in a different branch.

Another microscopy study suggested sleep’s role in the dendritic changes underlying  learning. “Without sleep, branch-specific formation of new spines was severely reduced,” Gan said. Apparently, the neurons engaged by the motor task are reactivated during non-REM sleep.

A second training period didn’t promote spine formation as much as sleep. “It’s better to sleep after learning than to keep studying,” he concluded.

Awry in autism

Two other talks suggested that autism involves abnormal connectivity in the brain. Bernardo Sabatini, professor of neurobiology at Harvard University, focused on the basal ganglia, a midbrain region that links reward to motor activity and whose dysfunctions are apparently involved in disorders as diverse as Parkinson’s and Huntington’s disease, obsessive compulsive disorder, and drug addiction.

“There is a case building in the literature that the region makes a contribution to autism” as well, Sabatini said. The core features of autism—disturbed language development, aberrant eye contact, and repetitive movements—“all can be mapped onto the basal ganglia.”

He described research suggesting that synaptic connections between the cortex and basal ganglia develop in response to input from the cortex, and that modifying cortical activity in very young animals can influence their number and strength.

Disruptions in this process early in development might account for the repetitive movements (e.g. hand-flapping, rocking) common in autism, which apparently involve recurrent signals cycling through the basal ganglia. Sabatini described research focusing on a gene expressing a protein that regulates synapse formation, linking its abnormalities to hypergrooming, a behavioral abnormality in mice that resembles such stereotypical behaviors. Deletions or truncations in the part of the human genome containing the equivalent gene are associated with rare forms of autism, he said.

David Sulzer summarized research linking autistic spectrum disorders to a deficit in synaptic pruning. Normally, synapses proliferate profusely in infancy and are cut back by late adolescence: A post-mortem study found that the length and number of dendritic spines decreased by 50 percent in normal brains between childhood and the teen years, but just 15 percent in autistic brains.

Macroautophagy, by which unwanted cell parts are degraded by cellular machinery, apparently plays a role in pruning. Sulzer said that in animal models, activating a protein, mTOR, that inhibits macroautophagy resulted in abnormally dense synapses and autistic-type behaviors. The brains of autistic patients similarly have increased levels of mTOR, and a surplus of damaged cellular components, such as mitochondria, that are normally eliminated by macroautophagy.

A number of genetic mutations linked to autism apparently increase mTOR activity, he said.  When animals genetically engineered to overexpress mTOR were given a drug to inhibit the protein in adolescence, it reactivated macroautophagy and normalized autism-like behaviors.

Although the specific drug may not be appropriate, “this is very promising clinically,” said Sulzer, because the finding suggests that “during the process of active pruning [several years after autism symptoms typically appear] it may not be too late to treat these children.”