Uncovering the link between synaptic density and mental illness through in vivo imaging

Linking synapses to mental illness: the promise of in vivo imaging

The in vivo study of synaptic density (the indirect estimated sum of presynaptic active zones and postsynaptic densities in brain tissue) through positron emission tomography (PET) is a rapidly growing field of research that has the potential to revolutionize our understanding of mental illness and of neurodevelopmental conditions. The idea that the brain’s synaptic wiring, neural function and behaviour are closely linked is not a new one, but advances in neuroimaging techniques have only recently made it possible to collect direct empirical data linking the status of brain synapses (synaptic density) with neural function, behaviour and disease in living humans.

Forty-five years ago, Irwin Feinberg suggested that excessive synaptic pruning during adolescence could lead to severe mental illness. Unequivocal evidence to support this theory in living humans is still lacking. The overall objective of this editorial is to present our stance on the value of measuring synaptic density in vivo using PET to better understand the development of mental illness. We argue that this novel approach might be more translationally and conceptually useful than related techniques using MRI because it is likely more proximal to function. As such, it could generate an unprecedented level of comprehension on synaptic organization and pruning and its role in the development of mental illness.

We first provide an overview of the keystone neuroscience concepts of synaptic communication and of synaptic pruning; we discuss the popular, though still unproven, theory that abnormal (excessive) synaptic pruning during brain development potentially leads to mental illness. We then provide our perspective on the utility of PET imaging of synaptic density and consider pitfalls and hurdles ahead in translating findings in a clinically meaningful way.

Discovery of the synapse as a potential target for treatment of neuropsychiatric illnesses

The synapse — from the Greek “synapsis,” meaning “to clasp together” — is the fundamental unit of electrochemical communication between 2 neurons. The concepts of the synapse and of synaptic plasticity have been crucial in shaping our understanding of the brain and of modern neuroscience. The quest that led to the emergence of these keystone principles is the same winding quest that anciently sought to identify the seat of the human soul.1

The term “synapse” was coined in 1897 by Sherrington, and the concept of neurotransmission, describing that chemicals travelling through synaptic gaps were the primary means of communication between neurons, was described in 1921 by Otto Loewi and Henry Dale.2 From these breakthrough discoveries emerged a field of medical hypothesis testing around the premise that deficits in synaptic neurotransmission, or malfunction in the communication between neurons at synapses (“synaptopathy”), could explain all neuropsychiatric illnesses. There has always been high expectation that by studying synaptic neurotransmission we would discover science-based treatments in neuropsychiatry. This hope was consolidated following the revolutionary discoveries that led to the treatment of Parkinson disease in the 1960s, when Hornykiewicz established Parkinsonism as a disease of dopamine deficiency and 1 year later showed that symptoms could be treated with L-DOPA.3^,4 At the same time, the concept that dopamine hyperfunction might be involved in the etiology of psychosis received substantial credence with the serendipitous discovery of antipsychotics as drugs that inhibit dopaminergic transmission.5 While this landmark finding established dopamine transmission as a target for mental illness, evidence for dopaminergic dysfunction in psychosis has been harder to come by.6 Currently, although the general theory that abnormal neurotransmission is linked with neuropsychiatric illness is still broadly accepted, there are relatively limited in vivo data directly supporting this in humans. To date, brain imaging with PET has been focused primarily on measuring markers of synaptic function, including key receptors, enzymes and transporters. This work has produced important findings (including robust dopaminergic deficits in Parkinson disease7 as well as monoaminergic abnormalities in schizophrenia6 and major depressive disorder8), but efforts to identify a specific abnormality in neurotransmission as the cause of a given mental illness have not been as successful. Complementary information about synaptic architecture and brain wiring has the potential to advance our understanding of the neural basis of psychiatric disorders.

Feinberg’s legacy: how the hypothesis of adolescent synaptic pruning and mental illness rose to prominence in psychiatry

Developmental neuroscientists were quick to propose that subtle disturbances in synaptic communication and aberrant neurotransmission, occurring during critical developmental phases, could perhaps explain severe mental illnesses, such as schizophrenia and autism spectrum disorder, and that synaptic pruning could be an important aspect.9^–11 The developing brain undergoes progressive functional maturation through formation, elimination, rearrangement and stabilization of synapses.12 During synaptogenesis, neurons extend axons and dendrites, which make contact with other neurons at synapses. This dynamic and complex process occurs in waves, with the highest levels of synapse formation occurring during the first few years of life. During puberty, through the teen years and into adulthood, synaptic architecture is massively changed, particularly in the prefrontal cortex. Synapses that are not active or that are redundant are lost, engulfed by microglia, a process known as synaptic pruning, whereas the network between other synapses is strengthened and stabilized with myelinisation, a process called synaptic plasticity. Together this reorganization of existing neural connections and the formation of new synapses results in more efficient, streamlined and adaptable brain networks, and in the emergence of higher-order thinking and more complex cognitive function. One of the earliest scientists to study synaptic pruning was the Canadian neuroscientist Donald Hebb, who proposed that neurons that fire together, wire together, and those that don’t, die off.13 Since these discoveries, research has shown many times that small changes in synaptic plasticity can have a substantial impact on behaviour. Studies in animals have shown, for example, that manipulating the strength of synapses through techniques such as optogenetics can alter behaviour.14^–16 Additionally, the scientific literature is abundant in examples of studies showing that genetic and pharmacological manipulations that enhance or reduce synaptic plasticity, either globally or in specific regions and circuits, alter behaviour. Prime examples of synaptic plasticity affecting behaviour come from studies investigating the persistent restructuring of neuronal circuits by drugs of abuse17^–20 as well as Hebbian plasticity involved in memory and learning.21 In this regard, it has been argued that small changes in synaptic spine morphology and changes in long-term potentiation and long-term depression in pre- and postsynaptic elements have a role in the consolidation of memories and could in part be driving addictive behaviour.19 The consensus is that synaptic plasticity affects behaviour and can explain some pathologies, but more research is needed to understand the underlying mechanisms leading to “abnormal” synaptic arrangement.

The theory that schizophrenia and probably other mental health disorders are diseases of synaptic pruning was first proposed by Irwin Feinberg in 1987. Feinberg, a psychiatrist and researcher at the University of California, Los Angeles, argued that schizophrenia might be caused by an overproduction of synapses during adolescence, followed by a period of excessive pruning of these synapses during early adulthood.22 Feinberg based his theory on studies of postmortem brain tissue from individuals with schizophrenia, which showed a reduced number of dendritic spines in prefrontal cortex, temporal lobes and in basal ganglia together with an overabundance of glial cells.23^–25 He did not, however, propose a precise mechanism by which excessive pruning might occur. Several non-mutually-exclusive views have surfaced suggesting that infection, prolonged stress and genetic factors might be at play. The theory that Feinberg conceptualized challenged the prevailing view that schizophrenia was caused by a chemical imbalance in the brain. It was ahead of its time and largely ignored.11^,26

Now, several decades later, studies continue to reveal data that may align with Feinberg’s theory using a range of sophisticated ex-vivo imaging approaches, including confocal microscopy or 2-photon microscopy; immunohistochemistry of antibodies for proteins that are found at synapses, such as SV2A, synaptophysin, PSD-95 and gephryn; Western blotting of synaptotagmin or synapsin protein levels; and in situ hybridization to measure the expression of genes that are involved in synapse formation (e.g., Neurexin or Shank). In this regard, meta-analytic reviews of postmortem brain findings reveal a region-specific decrease in the density of postsynaptic elements in schizophrenia, in line with preclinical research.27^,28 Altogether, this convergent evidence provides strong but indirect support for a synaptic deficit model of schizophrenia without, however, demonstrating that excessive pruning is the root cause of this deficit. Developmental influences on synaptic processes are a very active area of study in model systems, but a major translation gap exists in understanding the trajectory and etiology of psychiatric disease in living humans. Studies performed in postmortem samples, while broadly consistent with detailed models developed in preclinical systems, typically have small sample sizes, limited patient characterization and a time lag between key developmental processes and sample availability, making it difficult to characterize specific influences on neurobiology.29 This presents a particular challenge for studying longitudinally, highly dynamic processes in adolescence and young adulthood, and limits our understanding of neurodevelopmental alterations over time, in early life and early disease.

In vivo imaging is needed to address the questions of “synaptopathy” in the development of mental illnesses: innovation in in vivo measurement of synaptic density

Various MRI techniques, such as cortical thickness, grey matter volume, cortical surface area and white matter measurements using diffusion tensor imaging, can be applied to measure patterns of gross brain structure in living humans and circumvent many of the limitations of postmortem brain studies. MRI-based studies of adolescent neurodevelopment describe a canonical trajectory of increasing then decreasing grey matter volume, sometimes interpreted to reflect synaptogenesis and pruning processes and therefore to lend support to these models, as well as alterations in people expressing or at risk of schizophrenia and other developmental disorders.29^–32 However, these measures can have an indirect and often ambiguous association with brain function and microstructure, so specific hypotheses about synaptic development cannot be probed. More precise measures are necessary to critically evaluate theories of synaptic development and pruning in humans and understand the influence of real-world factors on these processes.

The recent development of PET imaging methods to measure the synaptic vesicle glycoprotein 2A (SV2A) provides an opportunity to bridge these scales of analysis and close the knowledge gap around our understanding of human neurodevelopment. SV2A is expressed at near-constant levels on synaptic vesicles in all neuron types throughout the central nervous system and is part of a neuron’s vesicle release machinery directly linked to vesicle trafficking and, therefore, to neuronal communication.33 The most widely used of these radiotracers, [¹¹C]UCB-J and [¹⁸F]SynVesT-1, are highly selective for SV2A, as shown in blocking and displacement studies with the antiepileptic drug levetiracetam, but they were developed to have the high brain penetrance, target affinity and fast kinetics necessary for PET radiotracers.34^–37 Initially, [¹¹C]UCB-J and [¹⁸F]SynVesT-1 were applied to quantify synapse loss in neurologic disorders, with detailed validation studies showing the exceptional quantitative precision of these tools in vivo and their close correlation with in vitro measures of synaptic markers.34^,38 Research is ongoing to refine our understanding of biological influences on these signals. In a mouse model of Alzheimer disease, SV2A PET was sensitive to both baseline synaptic deficits and drug-induced increases in synaptic density in the hippocampus after treatment with the synaptogenic kinase inhibitor saracatinib.39 A pilot study in humans found that while a single dose of ketamine did not induce a measurable change in the SV2A PET signal in healthy people, it may have reversed deficits in those with severe depression.40 A recently announced multi-centre initiative aims to characterize the synaptic biology of SV2A radiotracer binding further, including detailed pre- and postmortem measurements in patients with Alzheimer disease (https://fnih.org/our-programs/pre-competitive-analytical-validation-sv2a-pet-biomarker-synaptic-density-sv2a).

In the meantime, SV2A PET has been applied to study synaptic changes in a wide range of brain disorders. Studies in neurodegenerative disorders suggest that SV2A PET shows higher sensitivity and unique information about early brain changes relative to other neurophysiological measures.41^–43 The body of literature in psychiatric disorders is more preliminary, with initial studies identifying synaptic deficits across mood, substance use and psychotic disorders.44^–47 Lower synaptic density in the hippocampus and prefrontal cortex is a common theme, though the overall pattern and magnitude varies across studies, and correlations with cognitive measures have been reported in patient samples including those with obesity and cannabis use disorder.45^,48 Direct comparison of SV2A PET to structural and functional brain measures show both overlapping and shared features, suggesting that synaptic density is partially correlated to, but distinct from, gross anatomic changes.46^,49

Challenges of SV2A PET imaging

As always, major limitations remain. PET imaging provides a spatial resolution on the order of millimetres; finer scales of analysis, including cell-type-specific changes, remain unreachable. On a practical level, PET is technologically demanding and, even with longer-lived radiotracers like [¹⁸F]SynVesT-1, is limited to a handful of centres in Canada. This may restrict its uptake as a clinical technology and, in research, limits the kind of big data initiatives available with other neuroimaging techniques. Simplified quantification methods using white matter reference regions, which would facilitate more widespread use, have been investigated with [¹¹C]UCB-J50^,51 but have not yet been evaluated for [¹⁸F]SynVesT-1. Multicentre studies and data sharing are becoming more common with PET, but careful attention must be paid to acquisition, data processing and analysis procedures. Image normalization, as is standard practice for many MRI measures, can help address these issues and facilitate data pooling, but with PET this comes at a cost of losing some of the quantitative information that helps to situate molecular measurements within biological models.

On a theoretical level, important caveats and open questions persist regarding biological interpretation and application of synaptic density PET radiotracers. SV2A PET measures only presynaptic, not postsynaptic, alterations. Synaptic vesicle number varies with brain region and synapse size, meaning that the SV2A PET signal may not strictly represent synapse number alone.52 On the other hand, it may be argued that synaptic vesicle density is the measure more relevant to understanding the brain’s functional capacity. SV2A expression may be higher in inhibitory than in excitatory neurons,33 which could result in systematic differences in sensitivity to different developmental stages or influences. Most broadly, it remains to be determined to what extent synaptic density variation will identify features specific to a given disease or causal factor. Region-specific differences in synaptic density, if observed, might suggest specific circuits altered in a given psychiatric disorder, while broad or whole-brain changes might reflect general impairments in synaptic development or plasticity processes. In vivo PET imaging carries the general strength of being an inherently whole-brain measure, mitigating against the so-called “streetlight” problem of tending to focus study on regions where measurements are easier. For example, many human cortical regions have less direct analogues in rodent models compared with subcortical regions, while in postmortem studies the quality of tissue measurements can vary based on regional influences.29 Synaptic density PET may allow detection, for future targeted study, of effects in brain areas that are not hypothesized a priori, perhaps because they are more challenging to evaluate. However, in the literature to date, regionally specific effects in any neuropsychiatric sample have been the exception, with widespread or global effects commonly reported. Larger studies in thoroughly characterized samples may help resolve this by starting to untangle subtler patterns of change or overlapping influences on brain development. Either a marker of broadly altered synaptic function or a tool for investigating disease-specific pathways may have utility for research in psychiatry, but it will be crucial to appreciate this distinction in order to design meaningful clinical studies.

Conclusion

We have argued in this editorial that in vivo measurements of synaptic density might be more translationally and conceptually useful than MRI because they are likely more proximal to function. We have to acknowledge that this assertion is, for now, speculative — in vivo measurement of synapse density might not provide a meaningful practical advance over existing approaches. At the very least, we expect that interest in this new tool will stimulate new paths of discoveries. In the best case, this work can help us translate, refine and apply models of neurodevelopmental factors in psychiatry directly in people and understand crucial periods of brain growth at an unprecedented level of detail. Because synapses influence every aspect of brain function, it is incumbent upon researchers to be rigorous in designing studies and interpreting results based on a nuanced understanding of biology. Characterizing the range of potential influences on synaptic density and its plasticity throughout the lifespan can help guard against the risk of interpreting any neurobiological measure as a determinative indicator of well-being or ability by adding, rather than removing, complexity to our understanding of human developmental neurobiology. Hornykiewicz’s discoveries linking dopaminergic neuron loss to Parkinson disease and subsequently demonstrating efficacy of dopamine substitution to treat its symptoms were completed within a remarkable 1-year span. More than half a century of neuropsychiatry research since then has shown that such a timeframe from bench to clinic is unlikely to be repeated. A new source of information about the living human brain can nevertheless help strengthen the links between neurobiology, mental illness and, ultimately, treatment.