Despite extensive research, psychiatric disorders remain poorly understood, mainly because their pathophysiology is far too complex to be elucidated using previous technology. Recent works have identified contributory factors for psychiatric disorders to be susceptible gene variants (molecular layer), synaptopathy (subcellular and cell layer), and alteration in neuronal circuits (circuit layer), which probably results in behavioral manifestations (individual layer). However, the relationship between each layer remains unknown, which hinders the integrative and causal mechanistic understanding of behavior. It seems likely that each layer can affect one another, from the macroscale to the mesoscale and then to the microscale layer, or vice versa (Figure 1). Thus, my team aims to gain a constructive understanding of the multiscale hierarchical nature of psychiatric disorders. First, we will establish a new imaging technique (Aim 1), which will be combined with already available state-of-the-art techniques to develop a multiscale analysis that can identify causal relationships between the molecular/cellular events and the pathophysiology of these disorders (Aim 2). Furthermore, physical conditions are crucial components that can determine brain (dys)function, and exacerbation of psychiatric disorders by metabolic diseases are an emerging global burden. We will therefore investigate how brain functions are modulated by the physical environment (Aim 3). Finally, elucidation of the genuine causal factors for psychiatric disorders would have a long way to go, and to bypass these processes we are now developing a systems-oriented drug discovery approach, with which we will be able to take quantitative measurements of synaptic deterioration and stress-related metabolites (Aim 4). Taken together, the goal of this proposal is to elucidate how specific neuronal circuits are altered and by what cellular mechanisms in the disease state. Findings based on our strategy would causally identify the contributory factors for psychiatric diseases, which could provide the knowledge necessary to establish circuit-centric therapeutics as well as molecular- (and chemistry-) based drug designs
We focus on the synapse because various lines of evidence, including human genetics, brain imaging, and postmortem brain studies, have repeatedly suggested that disturbances in neuronal connectivity (i.e., synaptopathy) underlie a variety of psychiatric disorders (Hayashi-Takagi, 2017, Neurosci Res). However, it is not yet known whether synaptopathy is an underlying mechanism of disease or a secondary consequence. It is more likely that these are not mutually exclusive, and that the vicious cycle of pathology starts with genetic predisposing vulnerabilities and environmental factors, and results in circuit dysregulation as a final common pathophysiology. However, because of the lack of a technique that can manipulate individual synapses, the links between synapses and psychiatric disorders have been correlational. Furthermore, the involvement of synapses in learning and memory, which is one of the most advanced research topics in neuroscience, have also remained correlational. Memory research is elegant because scientists can directly and quantitatively assess the relationship between the stimulus input and the learning/memory output, as measured by behavioral responses, together with molecular and cellular events. Thus, while our final goal is to elucidate the pathophysiology of psychiatric disorders, we first examined the causal relationship between synapses and learning/memory. For this purpose, we have developed a novel synaptic optoprobe, Activated Synapse targeting PhotoActivatable Rac1 (AS-PaRac1). AS-PaRac1 is unique; not only can it specifically label recently potentiated dendritic spines, but it can also selectively induce shrinkage in spines containing AS-PaRac1 (Hayashi-Takagi et al., 2015, Nature (Article)). However, the biggest drawback of AS-PaRac1 is that it also labels learning-independent spine potentiation, which is often activity-independent. Spines that are generated in the absence of learning (spontaneous generation) are unstable and more likely to be eliminated (Xu et al, 2009, Nature). This indicates that AS-PaRac1 also labels transient spine potentiation that is unrelated to the learning-induced rearrangement of neural circuitry. Of the many forms of synaptic plasticity, Hebbian plasticity is an activity-dependent reorganization of neuronal circuits that results from various aspects of learning, and its induction requires synchronized firing of presynaptic and postsynaptic neurons. This led me to hypothesize that simultaneous visualization of both presynaptic and postsynaptic activations in an activity-dependent manner could be useful imaging method, with which double-labeled presynaptic and postsynaptic connections represents Hebbian plasticity (Figure 2A). For this purpose, the presynapses in the activated neuron can be labeled with vesicle-associated membrane protein 2 (VAMP2) that is fused with mTurquoise2. All probes, including that of the presynapse (VAMP2-mTurquoise), potentiated spine (AS-PaRac1-APEX-mClover2), and postsynaptic neuron (tdTomato), are expressed in an activity-dependent manner, and can be designed for rapid decay by the destabilizing sequence. We developed various fluorescence probes with distinct time constants of protein expression, transportation, and degradation, and optimized this activity-dependent triple labeling method (Figure 2B). One configuration is a conditional expression of a presynaptic marker with use of a double-floxed inverted open reading frame (DIO) and Cre recombinase. This projection pathway-specific optogenetics will help to elucidate the dynamic nature of synaptic potentiation in a specific neuronal circuit chosen from the vastly complex brain (functional connectomics). In addition, a combination of this method with FAST (block-face serial microscopy tomography) imaging (in collaboration with Dr Hitoshi Hashimoto, Osaka University) will allow us to perform a comprehensive analysis of how such potentiation is distributed throughout the brain in control and disease mouse models at a single-synapse resolution.
To further determine the causal and secondary consequence of each event, we have established a multiscale analysis (Figure 3). Our previous work demonstrated that a Disc1 knockdown mouse model of schizophrenia exhibited a decrease in dendritic spine density (Hayashi-Takagi et al., 2010, Nat Neurosci; Hayashi-Takagi et al., 2014, Proc Natl Acad Sci USA). We also found a significantly greater number of large dendritic spines compared to wild-type mice. The presence of the large spines in our model mirrors findings from another schizophrenia mouse model, calcineurin knockout mice (Okazaki et al, 2018, Neurosci Lett). It is well-known that there is a strong correlation between spine head size and its synaptic efficacy, whereby larger spines generate larger synaptic currents. This led me to hypothesize that large spines can affect the dendritic computation, causally resulting in subsequent behavioral alterations. To test this hypothesis, we set up a multiscale analysis that consisted of an electrophysiological method and Ca2+ imaging to visualize the synaptic input (synaptic level), dendritic event (dendritic level), action potential (cell level), and behavioral manifestations (individual level) (Figure 3A-C). Combination of this method with the functional connectomics (developed in Aim 1) should accelerate the multiscale understanding of brain functions because this imaging method can label potentiated synapses (synapse level) as well as connections between different brain regions (macro-circuit level). In addition, using in vivo optical manipulation (Figure 3D, developed in Aim 1) and in silico abduction (in collaboration with Dr Shoji Tanaka, Sophia University), as well as in vitro validation of these findings in patient-derived iPS cells, we will examine what kind of synaptic pathology underlies the pathology of psychiatric disorders.
Over the past two decades, the prevalence of comorbid psychiatric and physical diseases has increased dramatically (Figure 4). For example, most people over the age of 60 years suffer from the simultaneous presence of two or more diseases. Comorbidity does not mean the mere addition of two diseases that independently follow their own trajectory; comorbidity worsens the prognosis of each condition, and makes the treatment of each more difficult and less effective. For example, depression frequently occurs comorbidly with diabetes mellitus, and up to 80% of patients with comorbidity will experience a relapse of depressive symptoms over a 5-year period, resulting in worse overall clinical outcomes. However, molecular and cellular mechanisms of how physical perturbation would enhance the prevalence of depression and exacerbate its disease trajectory are largely unknown, and this represents a serious unmet need in medicine. To elucidate this issue, we have established a comorbid mouse model (Figure 5A), and longitudinal and multi-axis analyses are currently being performed (Figure 5B, C). We will focus on the relationship between synapse pathology (spine density, size, and turnover rate), behavioral manifestations, and physical parameters such as corticosterone and cytokines (Data not shown). Once we identify candidate exacerbating or protective factors for the condition, the validation of these factors will be examined by manipulative experiments. Ultimately, we aim to identify exacerbating factors and prognosis biomarkers of the comorbidity.
Pharmacological treatment for psychiatric disorders started with chlorpromazine, whose efficacy as an anti-psychotic was serendipitously discovered. Because the pathophysiology of psychiatric disorders remains largely unknown, drug discovery has been limited to chemical modifications of chlorpromazine, all of which have focused on blocking the dopamine and serotonin transmission. It is now widely accepted that drugs that specifically target a single molecule are likely to be less effective or to cause adverse side effects, which highlights the disappointing limitations of drug discovery by a single molecule target.Based on epidemiology and human genetics, psychiatric disorders such as schizophrenia and depression are heterogeneous and polygenic disorders with shared characteristics at the cell and circuit level. In other words, psychiatric disorders are systemic diseases because malfunctions of systemic control result in the symptoms of these disorders. This therefore warrants a systems-oriented approach to more effectively control the robustness of living systems. These include cell-based phenotypic biology assays, in which in vitro neuron culture models can provide substantive information on various cellular responses following exposure to a library of small molecular compounds. To mimic the excitotoxicity, which is supposed to trigger the onset of schizophrenia, we use the NMDA receptor antagonist phencyclidine (PCP) to create animal models of schizophrenia in vitro. Many drugs can cause hallucinations and delusions, but the ability of PCP to mirror almost all aspects of the symptomatology of schizophrenia is unparalleled. The administration of PCP preferentially suppresses the activation of inhibitory neurons, which results in a dramatic disinhibition of pyramidal neuron activity and consequential neurotoxicity. In our primary cortical neuron cultures, PCP has consistently been found to induce a drastic decrease in spine density and the accumulation of the schizophrenia-related metabolite, pentosidine. We are currently using a validated compound library, which consists of 1280 off-patent drugs, to screen for a neuroprotective compound that ameliorates spine deterioration and the accumulation of disease-related metabolites. For this, we have established a semi-automated and quantitative cell-based phenotypic assay (Figure 6). This screening method has now been extensively improved by the combination of deep learning-based automated assessment (in collaboration with Dr. Komatsu Yu, NINS AstroBiology Center). The top 81 compounds in the first screening was subjected to a second screening. We have now obtained 18 compounds as the hit compounds. While some are already known to be neuroprotective, others are quite new in the neuroscience literature. After a series of validation of these hit compounds, we will aim at drug repositioning as a novel therapeutic application for psychiatric disorders. Furthermore, after attaining a complete automation of deep learning-based image acquisition/quantification, we will expand the screening to focus on a cellular event that would be abducted as a causal mechanism in Aims 1 and 3. Through this complementary strategy (Aims 1-4), we aim to identify the underlying pathophysiology of psychiatric disorders and develop the therapeutic implications of these findings.