Research Area: Neurobiology

Muscles and bones strengthen with exercise and the immune system ‘learns’ from vaccines or infections, but none of those changes match the versatility and flexibility your central nervous system shows in adapting to the world. The brain is a model remodeler. If it weren’t, you wouldn’t have learned how to read this and you wouldn’t remember it anyway.

The brain’s ability to change its cells, their circuit connections, and even its broader architectures in response to experience and activity, for instance to learn new rules and store memories, is called “plasticity.” The phenomenon explains how the brand-new brain of an infant can emerge from a womb and make increasingly refined sense of whatever arbitrary world it encounters – ranging from tuning its visual perception in the early months to getting an A in eighth-grade French. Plasticity becomes subtler during adulthood, but it never stops. It occurs via so many different mechanisms and at so many different scales and rates, it’s… mind-bending.

Plasticity’s indispensable role in allowing the brain to incorporate experience has made understanding exactly how it works – and what the mental health ramifications are when it doesn’t – the inspiration and research focus of several Picower Institute professors (and hundreds of colleagues). This site uses  the term so often in reports on both fundamental neuroscience and on disorders such as autism, it seemed high time to provide a primer. So here goes.

Beginning in the 1980s and 1990s, advances in neuroanatomy, genetics, molecular biology and imaging made it possible to not only observe, but even experimentally manipulate mechanisms of how the brain changes at scales including the individual connections between neurons, called synapses; across groups of synapses on each neuron; and in whole neural circuits. The potential to discover tangible physical mechanisms of these changes proved irresistible to Picower Institute scientists such as Mark Bear, Troy Littleton, Elly Nedivi and Mriganka Sur.

Bear got hooked by experiments in which by temporarily covering one eye of a young animal, scientists could weaken the eye’s connections to the brain just as their visual circuitry was still developing. Such “monocular deprivation” produced profound changes in brain anatomy and neuronal electrical activity as neurons rewired circuits to support the unobstructed eye rather than the one with weakened activity. 

“There was this enormous effect of experience on the physiology of the brain and a very clear anatomical basis for that,” Bear said. “It was pretty exhilarating.”

Littleton became inspired during graduate and medical school by new ways to identify genes whose protein products formed the components of synapses. To understand how synapses work was to understand how neurons communicate and therefore how the brain functions.

“Once we were able to think about the proteins that are required to make the whole engine work, we could figure out how you might rev it up and down to encode changes in the way the system might be working to increase or decrease information flow as a function of behavioral change,” Littleton said.

Built to rebuild

So what is the lay of the land for plasticity? Start with a neuron. Though there are thousands of types, a typical neuron will extend a vine-like axon to forge synapses on the root-like dendrites of other neurons. These dendrites may host thousands of synapses. Whenever neurons connect, they form circuits that can relay information across the brain via electrical and chemical signals. Most synapses are meant to increase the electrical excitement of the receiving neuron so that it will eventually pass a signal along, but other synapses modulate that process by inhibiting activity.

Hundreds of proteins are involved in building and operating every synapse, both on the “pre-synaptic” (axonal) side and the “post-synaptic” (dendritic) side of the connection. Some of these proteins contribute to the synapse’s structure. Some on the pre-synaptic side coordinate the release of chemicals called neurotransmitters from blobs called vesicles, while some on the postsynaptic side form or manage the receptors that receive those messages. Neurotransmitters may compel the receiving neuron to take in more ions (hence building up electric charge), but synapses aren’t just passive relay stations of current. They adjust in innumerable ways according to changing conditions, such as the amount of communication activity the host cells are experiencing. Across many synapses the pace and amount of neurotransmitter signaling can be frequently changed by either the presynaptic or postsynaptic side. And sometimes, especially early in life, synapses will appear or disappear altogether.

Moreover, plasticity doesn’t just occur at the level of the single synapse. Combinations of synapses along a section of dendrite can all change in coordination so that the way a neuron works within a circuit is altered. These numerous dimensions of plasticity help to explain how the brain can quickly and efficiently accomplish the physical implementation of something as complex as learning and memory, Nedivi said.

A 2018 study in Sur’s lab illustrated learning occurring at a neural circuit level. His lab trained mice on a task where they had to take a physical action based on a visual cue (e.g. drivers know that “green means go”). As mice played the game, the scientists monitored neural circuits in a region called the posterior parietal cortex where the brain converts vision into action. There, ensembles of neurons increased activity specifically in response the “go” cue. When the researchers then changed the game’s rules (i.e. “red means go”) the circuits switched to only respond to the new go cue. Plasticity had occurred en masse to implement learning.

Many mechanisms 

To carry out that rewiring, synapses can change in many ways. Littleton’s studies of synaptic protein components have revealed many examples of how they make plasticity happen. Working in the instructive model of the fruit fly, his lab is constantly making new findings that illustrate how changes in protein composition can modulate synaptic strength.

For instance, in a 2020 study his lab showed that synaptotagmin 7 limits neurotransmitter release by regulating the speed with which the supply of neurotransmitter-carrying vesicles becomes replenished. By manipulating expression of the protein’s gene, his lab was able to crank neurotransmitter release, and therefore synaptic strength, up or down like a radio volume dial. 

Other recent studies revealed how proteins influence the diversity of neural plasticity. At the synapses flies use to control muscles, “phasic” neurons release quick, big bursts of the neurotransmitter glutamate, while tonic ones steadily release a low amount. In 2020 Littleton’s lab showed that when phasic neurons are disrupted, tonic neurons will plasticly step up glutamate release, but phasic ones don’t return the favor when tonic ones are hindered. Then last year, his team showed that a major difference between the two neurons was their levels of a protein called tomosyn, which turns out to restrict glutamate release. Tonic ones have a lot but phasic ones have very little. Tonic neurons therefore can vary their glutamate release by reducing tomosyn expression, while phasic neurons lack that flexibility. 

Nedivi, too, looks at how neurons use their genes and the proteins they encode to implement plasticity. She tracks “structural plasticity” in the living mouse brain, where synapses don’t just strengthen or weaken, but come and go completely. She’s found that even in adult animal brains, inhibitory synapses will transiently appear or disappear to regulate the influence of more permanent excitatory synapses.

Nedivi has revealed how experience can make excitatory synapses permanent. After discovering that mice lacking a synaptic protein called CPG15 were slow learners, Nedivi hypothesized that it was because the protein helped cement circuit connections that implement learning. To test that, her lab exposed normal mice and others lacking CPG15 to stretches of time in the light, when they could gain visual experience, and the dark, where there was no visual experience. Using special microscopes to literally watch fledgling synapses come and go in response, they could compare protein levels in those synapses in normal mice and the ones without CPG15. They found that CPG15 helped experience make synapses stick around because upon exposure to increased activity, CPG15 recruited a structural protein called PSD95 to solidify the synapses. That explained why CPG15-lacking mice don’t learn as well: they lack that mechanism for experience and activity to stabilize their circuit connections. 

Another Sur Lab study in 2018 helped to show how multiple synapses sometimes change in concert to implement plasticity. Focusing on a visual cortex neuron whose job was to respond to locations within a mouse’s field of view, his team purposely changed which location it preferred by manipulating “spike-timing dependent plasticity.” Essentially right after they put a visual stimulus in a new location (rather than the neuron’s preferred one), they artificially excited the neuron. The reinforcement of this specifically timed excitement strengthened the synapse that received input about the new location. After about 100 repetitions, the neuron changed its preference to the new location. Not only did the corresponding synapse strengthen, but also the researchers saw a compensatory weakening among neighboring synapses (orchestrated by a protein called Arc). In this way, the neuron learned a new role and shifted the strength of several synapses along a dendrite to ensure that new focus.

Lest one think that plasticity is all about synapses or even dendrites, Nedivi has helped to show that it isn’t. For instance, her research has shown that amid monocular deprivation, inhibitory neurons go so far as to pare down their axons to enable circuit rewiring to occur. In 2020 her lab collaborated with Harvard scientists to show that to respond to changes in visual experience, some neurons will even adjust how well they insulate their axons with a fatty sheathing called myelin that promotes electrical conductance. The study added strong evidence that myelination also contributes to the brain’s adaptation to changing experience.

It’s not clear why the brain has evolved so many different ways to effect change (these examples are but a small sampling) but Nedivi points out a couple of advantages: robustness and versatility.

“Whenever you see what seems to you like redundancy it usually means it’s a really important process. You can’t afford to have just one way of doing it,” she said. “Also having multiple ways of doing things gives you more precision and flexibility and the ability to work over multiple time scales, too.”

Insights into illness

Another way to appreciate the importance of plasticity is to recognize its central role in neurodevelopmental diseases and conditions. Through their fundamental research into plasticity mechanisms, Bear, Littleton, Nedivi and Sur have all discovered how pivotal they are to breakdowns in brain health.

Beginning in the early 1990s, Bear led pioneering experiments showing that by multiple means, post-synaptic sensitivity could decline when receptors received only weak input, a plasticity called long-term depression (LTD). LTD explained how monocular deprivation weakens an occluded eye’s connections to the brain. Unfortunately, this occurs naturally in millions of children with visual impairment, resulting in a developmental vision disorder called amblyopia. But Bear’s research on plasticity, including mechanisms of LTD, has also revealed that plasticity itself is plastic (he calls that “metaplasticity”). That insight has allowed his lab to develop a potential new treatment in which by completely but temporarily suspending all input to the affected eye by anesthetizing the retina, the threshold for strengthening vs. weakening can be lowered such that when input resumes, it triggers a newly restorative connection.

Bear’s investigations of a specific form of LTD have also led to key discoveries about Fragile X syndrome, a genetic cause of autism and intellectual disability. He found that LTD can occur when stimulation of metabotropic glutamate receptor 5 (mGluR5) causes proteins to be synthesized at the dendrite, reducing post-synaptic sensitivity. A protein called FMRP is supposed to be a brake on this synthesis but mutation of the FMR1 gene in Fragile X causes loss of FMRP. That can exaggerate LTD in the hippocampus, a brain region crucial for memory and cognition. The insight has allowed Bear to advance drugs to clinical trials that inhibit MGlur5 activity to compensate for FMRP loss.

Littleton, too, has produced insight into autism by studying the consequences of mutation in the gene Shank3, which encodes a protein that helps to build developing synapses on the post-synaptic side. In a 2016 paper his team reported multiple problems in synapses when Shank was knocked out in fruit flies. Receptors for a key form of molecular signaling from the presynaptic side called Wnt failed to be internalized by the postsynaptic cell, meaning they could not influence the transcription of genes that promote maturation of the synapse as they normally would. A consequence of disrupted synaptic maturation is that a developing brain would struggle to complete the connections needed to efficiently encode experience and that may explain some of the cognitive and behavioral outcomes in Shank-associated autism. To set the stage for potential drug development, Littleton’s lab was able to demonstrate ways to bypass Wnt signaling that rescued synaptic development.

By studying plasticity proteins Sur’s lab, too, has discovered a potential way to help people with Rett syndrome, a severe autism-like disorder. The disease is caused by mutations in the gene MECP2. Sur’s lab showed that MECP2’s contribution to synaptic maturation comes via a protein called IGF1 that is reduced among people with Rett. That insight allowed them to show that treating Rett-model mice with extra IGF1 peptide or IGF1 corrected many defects of MECP2 mutation. Both treatment forms have advanced to clinical trials. Late last year IGF1 peptide was shown to be effective in a comprehensive phase 3 trial for Rett syndrome and is progressing toward FDA approval as the first-ever mechanism-based treatment for a neurodevelopmental disorder, Sur said. 

Nedivi’s plasticity studies, meanwhile, have yielded new insights into bipolar disorder. During years of fundamental studies, Nedivi discovered CPG2, a protein expressed in response to neural activity that helps regulate the number of glutamate receptors at excitatory synapses. The gene encoding CPG2 was recently identified as a risk gene for bipolar disorder. In a 2019 study her lab found that people with bipolar disorder indeed had reduced levels of CPG2 because of variations in the SYNE1 gene. When they cloned these variants into rats, they found they reduced the ability of CPG2 to locate in the dendritic “spines” that house excitatory synapses or decreased the proper cycling of glutamate receptors within synapses.

The brain’s ever-changing nature makes it both wonderful and perhaps vulnerable. Both to understand it and heal it, neuroscientists will eagerly continue studying its plasticity for a long time to come.

Mating is instinctual for a mouse but sometimes, for instance when his potential partner smells sick, a male mouse will keep away. When Mark Hyman Jr. Career Development Associate Professor Gloria Choi and colleagues published a study in Nature in April revealing how this primal form of social distancing occurs, they provided an exquisite (and timely) example of how brain circuits factor context into behaviors, making them adaptive and appropriate even when they are innate, or “hardwired.” 

When the odor of illness enters the mouse’s nose, that stimulates neurons in its vomeronasal organ to send an electrical signal through a nerve to the brain’s olfactory bulb. Cells there, Choi’s team discovered, relay the signal on to neurons in a region called the cortical amygdala that govern the mating instinct. Finally, completing the health-preserving circuit that will inhibit the mating instinct, those neurons pass on the message to brethren in the neighboring medial amygdalar nucleus. In so doing, this sequence feeds a sensory context, the female’s ill odor, into a circuit to override the default context of an internal state, the instinct to mate. The researchers even showed that by artificially stimulating cortical amygdala neurons they could prevent a mouse from mating with a healthy partner and by artificially silencing those same cells they could make a mouse mate with an ill-smelling one.

As you can learn below, the brain has much greater flexibility in how it operates than the electrical circuits that power your house or even the chips that drive your cell phone. But fundamentally it is the routing of electrical signals from neuron to neuron that forms the basis not only for how we behave, but also how we match behavior appropriately to the circumstances we encounter, Choi said.

“The closest component to behaviors and internal states, and changes in those, are still believed to be neurons and circuits,” she said.

Understanding how brain circuits produce behavior is an exciting area of neuroscience research, including in many Picower Institute labs. Their studies are helping to elucidate how the brain’s anatomy is arranged to process information, and how the many dimensions of flexibility that the central nervous system overlays upon that infrastructure can integrate context to guide appropriate behavior. Context, after all, comes from many sources in many forms—from the senses, like scents and sounds and sights; from internal states, like mating drive or hunger or sleepiness; and even from time and place and from what we’ve learned and remember.

So what were you thinking when you did “this” instead of “that”? You were thinking about the context and relying on your brain’s ability to account for it.

Chemical control

The popular “circuit” metaphor makes it easy to think of neurons as merely switches and wires that pass electrical transmissions from one point to another. And indeed they do that, although instead of being screwed and soldered to metal contacts, they use molecules called neurotransmitters to send signals across tiny junctions called synapses. But if that were all that was going on, the brain would be pretty static and it is anything but. Many members of the Institute’s faculty study how learning occurs and memories are formed when the brain changes its synapses to create or edit circuit connections, but none of that is strictly necessary for existing circuits to flexibly control behaviors that we’ve already learned or that are innate. The brain has other ways to flexibly change how it operates. Choi’s team, for instance, found that the behavioral change of inhibiting mating could not occur without the cortical amygdala neurons also sending a chemical, thyrotrophin releasing hormone (TRH), to the medial amygdalar nucleus neurons. 

“Since 1986, that wiring diagram has been staring at researchers,” Flavell quipped. “Many of the small circuits embedded in the wiring diagram have been closely studied, while others haven’t. But a key question that we are trying to answer is how does the whole system work. How are these circuits coupled together to give rise to so-called ‘brain states’?”

In several studies Flavell has shown how a small number of neurons encode contexts and then signal that those circumstances are afoot by releasing chemicals called “neuromodulators” to many other neurons, giving rise to a brain state. Just as TRH may be doing in the circuit Choi uncovered, neuromodulators such as serotonin and dopamine, which are also ubiquitous in humans, add an extra dimension of tuning that can change, or “modulate,” how hardwired circuits process information and output behaviors, Flavell said. Neuromodulators can make neurons more or less electrically excitable given the same degree of input, Flavell explained. They can also make transmission at individual synapses more or less effective.

“The physical connections are like a roadmap, but the way that traffic is actually flowing on the road, the way that neurons are coupled to each other, is dynamic and changes with the animal’s context,” Flavell said. Neuromodulators are one way to make that happen.

For instance, in a 2019 paper in Cell, Flavell’s lab showed how a hungry worm knows to slow down and savor a patch of yummy bacteria when it finds one. A single neuron called NSM extends a little tendril called a neurite into the worm’s pharynx. Equipped with bacterial sensors (that turn out to also be present in the human intestine), the neurite detects when the worm has started to ingest and mash up its food. NSM releases serotonin, which finds its way to many of the neurons in worm’s brain that control locomotion. Upon sensing the serotonin, they hit the brakes.

In a more recent study in bioRxiv, the lab takes their investigation of neuromodulators even further. The study characterizes exactly how serotonin release from NSM modulates that activity of specific neurons in the C. elegans brain. In addition, Flavell’s group found that a neuron called AIA integrates information from sensory neurons about the smell of food. NSM can help determine what it does with that information, depending on whether it detects that the worm is eating or not. If it is, the smell of food (detected by AIA) reinforces that it should stick around to continue dining, a state maintained with serotonin. If the worm isn’t eating, the food smells signal that the animal should go exploring to find the source of that enticing odor. AIA, in that case, can instead trigger neurons that produce a different neuromodulator, called PDF, that cause the worm to start roaming (toward the food odor). Even in the simple circuitry of C. elegans, context changes how neurons interact, giving the animal flexibility to process sensory information.

That neurons capable of emitting neuromodulators can exert far-flung influence over behavior is illustrated by research in Newton Professor Mriganka Sur’s lab, too. There Sur’s team has a focus on a deeply situated, tiny brain region called the locus coeruleus (LC) that happens to supply most of the brain’s norepinephrine. Classically, neuroscientists have regarded norepinephrine from the LC as increasing the brain’s internal state of general arousal, but recent research in the Sur lab suggests it has profound, context-dependent effects on learning and behavior.

For instance, members of the lab have trained mice to expect a reward if they push a lever after hearing a high-pitched tone; the mice also receive an unexpected and irritating puff of air if they mistakenly press the lever after a low-pitched tone. By varying the loudness of the tones, the researchers can also vary the certainty the mice have about what tone they heard. Sur’s lab has found that the louder a high-pitched tone, the more norepinephrine a mouse will send to the motor cortex, which plans movement, before pushing the lever – as if greater certainty prompts it more strongly to push the lever. 

Once the lever has been pushed and the mouse gets its feedback of reward or air puff, LC neurons producing norepinephrine then act to fine-tune learning by calling attention to any surprising feedback, Sur’s team has seen. For instance, if the tone was high pitched and faint, but the mouse took the risk to push the lever, the neurons will send a burst of norepinephrine to the prefrontal cortex to note that pleasant surprise. The biggest post-push surge of the neuromodulator, however, occurs when the mouse guesses wrong: that norepinephrine release to the prefrontal cortex appears to signal that the adverse result must be noted. Sure enough, Sur said, the team has seen that the mouse’s performance typically improves after making an error. The LC’s neuromodulatory actions may contribute to that behavioral improvement, though more research is needed to prove it.

Sur’s is not the only research in The Picower Institute showing that the LC communicates with the prefrontal cortex to improve task performance, though. Last November in the Proceedings of the National Academy of Sciences, Picower Professor Susumu Tonegawa’s lab showed that LC norepinephrine neurons connect via distinct circuits to two different parts of the prefrontal cortex to endow mice with both the ability to curb impulses (i.e. to not “jump the gun” when waiting to perform tasks) and to ignore distractions, such as false cues. 

Rhythms among regions

Much as the Sur and Tonegawa labs have been investigating the LC, Fairchild Professor Matt Wilson’s lab studies how a different region appears to be a key hub for integrating contexts such as location, motion and memories of reward into behaviors such as navigation: the lateral septum (LS). As rats learn to find and return to the location of a reward in a maze, the lab’s extensive measurements of electrical activity among neurons in the LS shows that those cells are taking in and processing crucial contextual input from many other regions. The LS then appears to package that context to help direct the rat’s navigational plans and actions.

Over the past two years, Wilson and former graduate student Hannah Wirtshafter have published papers in Current Biology and in eLife showing that populations of LS neurons distinctively encode place information coming from the hippocampus, reward information coming from the ventral tegmental area and speed and acceleration information coming from the brainstem. The encoding is apparent in changes in the timing and rate at which the neurons “fire,” or electrically activate, in these different contexts. Some LS neurons, for example, become especially active specifically when the rat nears the reward location. In a new article published in Neuroscience and Biobehavioral Reviews in July, Wilson and Wirtshafter combined their observations with those of other labs to propose that the lateral septum packages all this contextual information into an “integrated movement value signal.”

“The lateral septum has a ton of different inputs,” Wirtshafter said. “What could the animal be doing with place-related firing that’s reward modulated and then velocity and acceleration? The answer, we think, based on where the LS outputs to, is that it is sending a signal about the context and whatever reward is part of that context. It includes what movement needs to be done and whether that movement is worth it in that context.”

While there are ample signs in the research that neuromodulators such as dopamine help the LS communicate about contexts like the feeling of reward, the studies also highlight the key role of another mechanism of flexibility: brain rhythms. Also known as brain waves or oscillations, these rhythms arise from the coordinated fluctuation of electrical activity among neurons that are working in concert. They allow neurons in brain regions to broadcast information and neurons in other regions to tune into those broadcasts, so that they can work together to perform a function, Wilson said.  

“These brain dynamics ensure that whoever is sending the information and whoever is receiving the information are doing it at the same time,” Wilson said.

In fact, Picower Professor Earl Miller, who has published numerous studies on how brain rhythms guide the flow of information across the many regions of the brain’s cortex, uses much the same kind of traffic analogy in talking about the function of rhythms that Flavell uses when talking about neuromodulators. Much as those chemicals can, oscillations also flexibly direct the flow of information on the network of “roads” that physical circuit connections create. The traffic metaphor perhaps combines well with the broadcasting one: Just like drivers who tune into a radio traffic report can decide to take an alternate route when they hear about an accident ahead, neurons in a brain region may act differently when they tune into new contextual information coming in from another brain region.

Wilson and Wirtshafter’s research, for example, demonstrates that lateral septum neurons tune into the hippocampus’s broadcast of location information via a specific “theta” frequency of brain waves. In particular, movement through a place is represented by the phase (peak or trough) of the theta waves with which neurons spike. 

“In the hippocampus, the phase at which a cell fires during theta can communicate information about the current, prospective, or retrospective spatial location,” Wilson and Wirtshafter wrote in their article. “For instance, …firing of individual hippocampus place cells begins on a particular phase of theta rhythm and progressively shifts forward as the animal moves through the place field.”

So maybe you are not a mouse deciding whether to mate or a rat rooting through a maze for a treat, but you are a person who has stayed out late at a friend’s house. Your internal state is that you are tired. You could head out on long drive home to the reward of your clean, warm bed, or you could sleep on your friend’s notably mustier couch and explain it your spouse the next morning. Then you remember from the drive to your friend’s place earlier, that there was an all-night rest stop along the highway where you could get coffee. Whether you decide to take the wheel or your friend’s offer of the couch will come from how a combination of neuromodulators and rhythms route information along circuits through key brain regions to integrate all this context—your internal state of tiredness, the memory of where that rest stop was, and the reward of your bed (or the punishment of an angry spouse who might ask “What were you thinking?”). Your brain gives you all the flexibility you need.