Understanding the Neurobiology of Mental Illness

Historical Perspectives on Biological Psychiatry

Unlike diseases such as cancer and malaria, which are viewed as intrinsically biological, psychiatric illnesses have long been defined by societal dogma rather than biological basis, said Bill Martin of Janssen. For several centuries, attempts at treating these conditions were dismal. Widespread institutionalization did have a silver lining: it medicalized diseases previously seen as moral failings. But treatments fell short, with “interventions that were at best guided by hypothesis and conjecture that were nearly always wrong,” and, even if occasionally useful, “nonetheless misguided,” Martin said.

The arrival of chlorpromazine—the first effective medicine for schizophrenia—in 1952 led to a sharp drop in institutionalization, ushered in a new era of biological psychiatry, and spurred the development of neuropharmacology as a research discipline. Chlorpromazine and two other drugs, imipramine for depression and lithium for mania in bipolar disorder—all discovered serendipitously—provided early treatment tools. The three classes of medicine they represent remain widely used today.

Martin described three historical elements from the middle of the 20^th century—a drug and two books—that represent forward-looking strands “to guide us to a brighter future” in psychiatric research.   The drug is haloperidol (Haldol), which was discovered by Dr. Paul Janssen, the founder of Janssen Pharmaceuticals, in a Belgian village in 1958. It revolutionized treatment by making it possible for institutionalized patients to be treated at home. It was also the springboard for early success in rational drug design, with what we now call typical and atypical antipsychotics created based on its structural modifications. Martin said that was a crucial step forward, but its potential was held back by the way mental illnesses have been classified. The field needs new frameworks for scientific understanding and treatments that address biological mechanisms rather than poorly defined syndromes, he explained.

Martin traced the ideas the field needs to bring forth those frameworks to two books – The Organization of Behavior, by Donald Hebb, published in 1949, and Perceptrons, by Marvin Minsky and Seymour Papert, published in 1969. Hebb is most famous for his theory of behavior rooted in the physiology of the nervous system. But he also wrote about how the proper synchrony of neuronal activity can be disturbed to cause mental illness, and how genetic and environmental factors are at play. Two decades later, Minsky and Papert’s description of network learning systems introduced the computational basis for understanding how the brain could produce organized behavior.

Network theory of mental disorders maps the connections between symptoms and when they occur.

These ideas are hugely relevant today as researchers begin to understand the dysregulations in circuitry and neural networks that underlie depression and other illnesses. Additionally, current studies using machine learning support a “network theory” of mental illness, in which illnesses described in the Diagnostic Statistical Manual of Mental Disorders (DSM) are not distinct; rather, symptoms are connected across diseases and can act as bridges between them. “It’s not that the DSM is inaccurate, as much as it’s not useful for advancing science,” Martin said. By mapping patterns of connectivity and using functional brain imaging, the researchers can begin to understand how networks of symptoms relate to brain circuit dynamics. “I believe that with a blend of humility and optimism, we will further refine how psychiatric diseases are conceptualized and treated—and maybe even cured, or potentially prevented,” he concluded.

Brain Mechanisms in Psychiatric Disorders

Although researchers have made progress in identifying genes that play a role in psychiatric diseases, the field is still in the early days of understanding the biological effects of those genetic changes, said Consuelo Walss-Bass of the University of Texas Health Science Center at Houston. To identify the cellular disruptions that these disorders cause, Walss-Bass’s lab grows brain cells in a dish using stem cells reprogrammed from skin or other tissue obtained from people with psychiatric disorders. This creates “a virtual brain biopsy” that can be matched with that person’s DNA sequence, she said, allowing the researchers to study the contribution of genetic makeup to cell function.

This reprogramming process is expensive and time-consuming, so most studies using the approach have small sample sizes. To make genetic analysis more powerful in small sample sizes, Walss-Bass’s lab compares the cellular and genetic profile within family members, using blood samples collected from families in Costa Rica in which schizophrenia is highly prevalent. In one recent study, they examined neurons and astrocytes reprogrammed from such samples. They found 454 genes expressed differently in family members who had schizophrenia compared to those who did not. Many of those genes are involved in a signaling pathway linked to GSK3-beta, a gene shown in other studies to be connected to schizophrenia. Chemically blocking this pathway [had opposite effects] in cells grown from healthy siblings’ blood samples compared with those from siblings with schizophrenia.

The researchers are now exploring the function of genes on this pathway whose expression level changed in the schizophrenia cells. They are also investigating how antipsychotic medicines affect the dysregulated GSK3-beta pathway in those cells to determine “how different individuals that have this particular phenotype of GSK3-beta dysregulation respond to particular antipsychotics,” she said.

Walss-Bass’s lab complements their brain cell experiments with studies on postmortem brains. Her lab started a brain collection in 2016, now consisting of about 130 brains–three quarters of which come from people who had a substance use disorder. For each brain specimen, the researchers also collect a skin biopsy to create reprogrammed brain cells, and a blood sample to use in biomarker discovery. Additionally, they also conduct a “psychological autopsy,” interviewing family members to understand the behavior and symptoms of the person who died.

People with opioid use disorder may have an overgrowth of blood vessels in their brains.

In one recent study of postmortem tissue, her team examined the dorsolateral prefrontal cortex—a part of the brain associated with executive function and regulating impulsivity—and identified changes associated with opioid addiction. They also imaged postmortem brains, finding that five out of 11 people with opiate use disorder, but none of the control brains, had an overgrowth of blood vessels in this brain region. Walss-Bass plans to investigate the link between vascular dysregulation and opiate use disorder in follow-up studies in stem cells, she said.

The Genetics of Mental Illness

Benjamin Neale of Massachusetts General Hospital described ongoing large-scale international efforts to identify the genes that contribute to autism, schizophrenia, bipolar disorder, and epilepsy. Researchers use microarrays to test whether variants of common genes are associated with a disorder. Alternatively, they sequence either a genome or just the portions of it encoding genes (called the exome) to identify rare gene mutations involved in a disorder, then use statistical analyses to understand the strength of these mutations’ effects. Some mutations scramble a gene enough to knock out its function completely, while others, called missense mutations, replace a single amino acid in the sequence and have a more variable effect.

The largest exome sequencing study of autism, conducted by the Autism Sequencing Consortium and published in 2020, found 102 genes that raise the risk of the disorder when mutated. Among the 26 most strongly associated genes, some are already under investigation, others are known to cause specific syndromes associated with autism, and still others are known to play a role in balancing excitatory and inhibitory neuron activity, a possible biological pathway in autism. Most genes carry loss of function mutations, but missense mutations are also present.

A meta-analysis identified rare variants in ten genes that confer substantial risk for schizophrenia. As in autism, both loss of function and missense mutations play a role. Among them is a gene previously implicated in schizophrenia and two genes involved in regulating neuron activity. “But it’s kind of complicated to interpret some of the biological and phenotypic consequences of these genes,” Neale said. Microarray studies in schizophrenia also point to changes in how neurons make connections. Several genes flagged in common variant discovery also show rare variants that may contribute to disease risk, the study found, suggesting that many genes cumulatively influence risk for schizophrenia.

A microarray analysis for schizophrenia suggests that many genes involved in the disease play a role in regulating connections between neurons.

Genetic analysis in bipolar disorder is a little behind schizophrenia, but researchers are conducting similar work. A recent analysis identified a gene called AKAP11, also implicated in schizophrenia but not in autism or developmental delay. “I think this is an emerging legitimate adult psychopathological hit,” Neale said. The protein encoded by the gene helps inhibit GSK3-beta, the molecular target of lithium, a drug widely used to treat bipolar disorder.  Epilepsy, although not a neuropsychiatric disorder, occurs at a heightened rate in people with schizophrenia and also frequently co-occurs with autism. A recent study of epilepsy and three subtypes suggests that, as in schizophrenia, many genes are involved.

Overall, said Neale, “the first genes of major impact have been identified. We are super thrilled and excited about that, but we’ve got a lot more work to do.”

Quantifying the Brain-Behavioral Space Across the Psychosis Spectrum

Manifestations of psychiatric illnesses vary tremendously across individuals. This raises a fundamental challenge, said Alan Anticevic of the Yale School of Medicine: individualizing treatment to address a specific patient’s constellation of symptoms. What’s more, as novel therapeutics appear in the pipeline, being able to identify people most likely to benefit from them will be a boon for efficacy in clinical trials.

The problem is challenging because it requires integrating gene alterations with changes in circuitry, cellular functioning, systems-level alterations, and symptom profiles into an individualized, quantitative, neuro-behavioral map, Anticevic explained. Although much of the molecular data is not yet available, brain imaging can provide a powerful enough picture of neural systems to map complex brain health symptoms.

Anticevic and his colleagues developed a computational framework called N-Bridge (neuro-behavioral relationships in dimensional geometric embedding) using a publicly available data set describing 436 people with schizoaffective disorder, schizophrenia, or bipolar disorder with psychosis. The data set had extensive symptom data as well as functional magnetic resonance imaging for each person.

The researchers first quantified the behavioral variation by creating a three-dimensional representation for each person, revealing that peoples’ symptoms overlap extensively across syndromes, with no clear separation. They then calculated each person’s brain connectivity using their fMRI data and developed a quantitative method for mapping their symptoms onto the brain map. The result is “a brain map that tells us how much symptoms co-vary with the dysconnectivity in that region,” Anticevic said. This data-driven mapping yielded a much clearer picture than the map achieved using traditional Diagnostic and Statistical Manual of Psychiatric Disorders (DSM) diagnoses.

Brain maps derived using patients’ symptom profiles provides clearer information about affected circuitry than those derived using traditional diagnoses.

The researchers also added gene expression data for the different brain regions—taken from the Allen Human Brain Atlas—onto these maps. These data come from post-mortem brain studies, “but they are powerful in that they can give us spatial location about gene expression,” Anticevic said. They can then look at how specific genes of interest—for example, those associated with specific neurotransmitters or receptors—affect a person’s symptom profile maps. He and his colleagues are now examining how brain circuitry in healthy people who receive a pharmacological intervention, such as ketamine, psilocybin, or LSD, can inform appropriate treatment for people with psychosis. By leveraging the link between brain circuitry and symptoms in a given individual in these ways, he said, it will be possible to target treatments more accurately.

Mechanisms of Motivational and Cognitive Deficits

Impairments in motivation and goal-directed action is a poorly understood, and understudied aspect of schizophrenia, depression and other psychiatric disorders. Deanna Barch of Washington University in St. Louis studies how different neurobiological mechanisms might contribute to such impairments.

Researchers can break down motivation and goal-directed action into multiple components, such as the feeling of liking something or experiencing sensory enjoyment and pleasure, being able to anticipate that an activity will bring on that feeling, and the process of implicitly or explicitly learning the stimuli or actions that lead to it. Drumming up motivation and completing a goal-directed action also requires calculating the effort required, conducting a cost-benefit analysis on whether to allocate it, then generating and enacting a plan to achieve it.

The process of accomplishing goal-directed tasks can be broken down into multiple cognitive components. People with schizophrenia struggle with a subset of them.

These components are driven by different brain networks and might be affected differently depending on the disorder, Barch explained. “While behavior is hugely important, we want to be able to parse different mechanisms, pathways or processes that might contribute to what in theory looks like the same behavior across different forms of mental illness.” Research in her lab and others suggests that people with schizophrenia have no difficulties experiencing pleasure or implicitly understanding how to attain it. However, they struggle to predict that a future experience will be pleasurable and learn which activities or tasks they find pleasurable. This stymies their ability to allocate effort to the task.

Barch’s team has showed that people with schizophrenia generally underestimate the pleasure they expect to experience from a future reward compared to people without a mental illness, although how the two experience the reward does not differ. The findings hold true in the lab, when people are asked to play a gambling game in which they can receive monetary rewards, and in everyday life, when people report their enjoyment and expected enjoyment of events using a cell phone app. The researchers also used brain imaging to show that people with schizophrenia had reduced activation in key brain areas to a predicted reward.

In pursuing rewards, people often must create models of the world that calculate the likelihood of expected outcomes based on past experience. Using a video game in which participants had to choose a planet to which they sent space travelers to retrieve treasure, Barch and her colleagues showed that people with schizophrenia tended not to create such a model, but to choose based on results of just their previous choice. Additionally, in a different choice, they found that people with schizophrenia tend to exert less effort because they underestimated the reward’s value. People who underestimated more tended to be ranked by clinicians as having difficulties with motivation and pleasure, and vice versa. These differences were reflected in brain activity patterns. Dips in these cognitive modalities are present to varying degrees in people with various mental illnesses, Barch said.

Microglia in Mental Illness

As the brain matures, synapses—connections between neurons—form abundantly, then get pruned and remodeled to sculpt precise connections and build brain circuitry. The process may also make the brain vulnerable to stressors that can spark mental illnesses such as schizophrenia, which often first manifests in adolescence, said Beth Stevens, of Harvard Medical School and Boston Children’s Hospital.

Genetic research linking common and rare variants of synapse-related genes to schizophrenia and multiple studies suggest that differences in the number of synapses or their function might characterize the disease. Scientists have also linked immune-related pathways to schizophrenia, but which genes underlie that link has been unclear. However, a 2016 study Stevens co-authored found that increased expression of one version of the gene C4, which encodes an important immunity protein called complement component 4, explains most of the immune system link. C4 is part of the complement system, which consists of several proteins that help the immune system destroy microbial invaders. The finding dovetailed with her lab’s work showing that C4 promotes synaptic pruning during visual system development in mice.

To investigate the connection, collaborating researchers created mice carrying multiple copies of the schizophrenia-linked C4 variant. They showed that these mice had more synapse pruning—and fewer synapses in the prefrontal cortex—than controls. Immune support cells called microglia do the pruning, engulfing less active synapses. Complement proteins also preferentially interact with these inactive synapses, and Stevens’s lab is exploring whether C4 may act as an “eat me” signal and what other signals may be present.

Her lab is also looking at the role of another schizophrenia risk gene called CSMD1. Mice that lack CSMD1 have excessive synaptic pruning, they found, suggesting (along with other clues) that the protein tags synapses to act as a brake on pruning. Hypothetically, defects in molecules like CSMD1 or C4 could lead to excessive binding of complement to synapses, resulting in over-pruning, she said.

Molecules like CSMD1 and C4 may act as cues instructing microglia on whether to maintain or prune a synapse.

Stevens and her colleagues are investigating how these signals may affect neuron activity during adolescence in the prefrontal cortex, a higher-order brain area affected in schizophrenia. They measured electrical activity in neurons of mouse prefrontal cortexes at different developmental ages and found that the input these neurons receive shifts from excitatory to inhibitory when synapses get pruned. They now plan to find out if this pattern changes in the absence of C4 or CSMD1. Additionally, they are examining how social isolation affects these patterns in mice to model how environmental factors might spark mental illness. Early studies suggest the normal shift in activity disappears in mice housed alone. Stevens will conduct similar studies in marmosets, whose brains have a level of complexity more similar to humans. She also plans to study how these biological mechanisms affect various cognitive tasks.

Clinical Staging in Psychiatry

The Holy Grail in psychiatry is to develop a system of diagnosis that works well for patients, researchers, and clinicians, said Patrick D. McGorry from the University of Melbourne. The current system, based on the Diagnostic and Statistical Manual of Psychiatric Disorders (DSM), is flawed on all three counts and has been criticized extensively over the past few years.

McGorry studies how mental disorders develop and how to predict their trajectory in an individual. He is working to create a more effective system of diagnosis that maps the relationship between symptoms as they emerge at different stages of illness. As a starting point for this “staging” approach, he and others in the 1990s defined a prepsychotic or prodromal stage in illnesses like schizophrenia, acknowledging that most people experiencing a psychotic break had shown signs of illness for months or years. This idea “led to a very significant paradigm shift” in both research and clinical services by defining the first stage of mental illness requiring treatment.

At this stage, symptoms are very undifferentiated, McGorry said. Although a third of people with prepsychotic symptoms develop psychosis, many of the rest develop symptoms of other mental disorders. That means researchers and clinicians must take a transdiagnostic and inclusive approach to identify young people with a range of problems like distress, anxiety, and depression. “Out of that psychopathological soup you then start to precipitate out some more stable syndromes,” he said. For most people, early symptoms will be transient, but a subset will progress to sustained mental illness.

Early symptoms of mental illness are undifferentiated and often transient, but for some, they progress to into more stable syndromes.

Researchers can use network-based approaches to map day-to-day symptoms over time and identify the relationship between symptoms, which is quite unstable initially but in some people becomes specific syndromes like depression or psychosis. McGorry and others are also working to identify biomarkers for distinct stages of mental illness transdiagnostically, rather than for specific syndromes. Drugs or psychosocial treatments that are effective across syndromes can potentially target the biological mechanisms at play, he says.

McGorry then described how staging can help rethink mental health care to provide treatment to young people who need it. Investment in children by families, communities and society overall peaks at about age 22, an especially vulnerable time for mental illness. Yet health systems tend to focus on older adults because most health problems correlate with aging. That means clinicians often miss the opportunity to intervene early, when treatment is safer, simpler, and cheaper. “We need major system reform,” he said. “We’ve already got effective treatments, but we are not delivering them very well, and certainly not in a timely manner.”

In 2006, McGorry launched a youth mental health service called Headspace in Australia that could provide a model for other countries, he said. The network consists of about 130 centers across the country and has provided mental health care to about 150,000 young people who would not otherwise have received treatment. In addition, it serves as a research platform for developing this new, flexible diagnostic model. “It’s very aspirational, what we are doing, but also very practical because we are building a system that will make research, reform, and better outcomes possible,” said McGorry.

Panel Discussion: The Future of Neuroscience R&D: Deepening Insights into Patients through Conceptual and Scientific Advances

The panel discussion began with the issue of early intervention. Data on psychosis offers strong support for early intervention, McGorry noted, and treatments that shorten the duration of a person’s first psychotic episode can significantly improve outcomes. However, early symptoms can be hard to ascribe to specific mental illnesses, so it’s important to meet patients where they are, he said. As the field’s understanding of biological mechanisms improves, it should be possible to individualize early treatment.

Gaining such understanding in the face of the complexity of mental illness requires collaboration and humility from researchers, the panelists said. The link Stevens described between immune-related genes C4 and CSMD1 in schizophrenia risk is just one of many proposed mechanisms being explored, none of which are proven. But as genetic studies yield more targets, researchers are finally in the exciting position of having many mechanistic clues to pursue using cross-disciplinary tools, said Stevens.

The genetic landscape is extremely complex, and so is its relationship with environmental risk factors, said Neale. We all carry some risk-increasing and some risk-decreasing variants. He suggested envisioning each person’s risk as a vessel containing a mix of genetic and environmental risk factors. When the vessel overflows, a severe mental health episode occurs. Some people are born with the vessel mostly empty, and others with the vessel mostly full. But the stressors that can cause levels to rise are varied and dynamic.

To wrap up, the speakers discussed their thoughts on how the pandemic has affected mental illness–particularly in young people. Stevens noted that it’s ironic that her lab is studying the effects of social isolation on the prefrontal cortex in animals. In a way, the pandemic has provided a kind of natural experiment on how social isolation affects people with mental illness. “It has the potential to have a serious impact,” said Stevens. Another important question the pandemic raises, in light of reports of the neurological sequelae of COVID, is whether psychiatric symptoms might turn out to be part of that picture.

McGorry noted that his team conducted some modeling that predicted a heightened suicide risk in the case of economic recession. That hasn’t happened in Australia, although they did see a sharp rise in self-harm among young people. “That’s probably more down to the distress and disruption of education, which will probably settle,” he said.

One thing the field of psychiatry has lacked so far, said McGorry, is a “moonshot” mentality–a concerted commitment to funding large-scale research that could help elucidate the biology of mental illness and identify new treatments. One big gap such a moonshot could tackle is to map biomarkers of normal human development during adolescence, when vulnerability to mental illness is high, Stevens explained. This could be done through some already ongoing studies, such as the Adolescent Brain Cognitive Development study co-led by Deanna Barch. Researchers could then rely on biomarkers to help stratify patients for treatment.

McGorry agreed that identifying biomarkers in adolescence, both in health and disease, is paramount. Headspace, the national youth mental health service he and his colleagues launched in Australia, sees 150,000 youth age 10-24. Studying that group of people could yield important data. Additionally, rational drug design has not worked well in psychiatry, where the best drugs are not visible. Instead, the best drugs, such as lithium and clozapine, were discovered serendipitously, and researchers still don’t know how they work. Pinning down those drugs’ mechanisms of action could help open the door to more effective compounds, he said.

Yet perhaps the biggest moonshot would be working out the biological consequences of mutations that raise the risk of mental illness. “We’ve got phenotypes and traits,” said Neale, “and we have very little capacity to figure out what their biological consequences really are.”