Imaging the Cytoskeleton to Understand Schizophrenia

Upper left: Prof. Takeo Yoshikawa, RIKEN Center for Brain Science; Upper right: Prof. Nobutaka Hirokawa, University of Tokyo; Lower, members of Dr. Hirokawa’s lab – from left to right – Dr. Momo Morikawa, Shogo Yoshihara and Dr. Yosuke Tanaka.

Schizophrenia is a mental disorder in which patients often suffer hallucinations, delusions and disordered thinking that can severely impair their daily lives. Approximately 20 million people worldwide are affected with schizophrenia. Despite its prevalence, very little is understood about the molecular mechanisms causing this disease.

A collaborative group of scientists have combined their expertise and demonstrated in a recent article in Cell Reports (right) that schizophrenia is a disease of the cytoskeleton.

The work was directed by Prof. Nobutaka Hirokawa of the University of Tokyo; he has been researching the cytoskeleton and characterizing the kinesin superfamily molecular motors (KIFs) for 40 years. His team worked in collaboration with Prof. Takeo Yoshikawa, a molecular psychiatrist from the RIKEN Center for Brain Science who provided insights from the side of experimental psychiatry.

We interviewed these scientists along with key members from Prof. Hirokawa’s lab: graduate student and lead author Shogo Yoshihara, lab lecturer Dr. Yosuke Tanaka who supported Yoshihara’s cell biology and mouse research and Dr. Momo Morikawa, a postdoctoral research scientist who established the schizophrenia model mouse and assisted with Yoshihara’s mouse behavioral tests. We learned more about their discovery and how they used confocal microscopy and scanning electron microscopy to generate supporting data.

What new discoveries did you present in this publication?

We have demonstrated that schizophrenia is a disease of cytoskeleton. Cytoskeletal structures are very dynamic and small, which makes them difficult to image because they require high resolution and high speed. We mainly use a ZEISS confocal microscope with an Airyscan detector as it beautifully depicts the actin bundle alignment and the gaps between the actin bundles. In order to observe individual actin fibers, we utilized a ZEISS scanning electron microscope, which is able to very clearly image not only the actin bundles but also individual actin fibers. These microscopes have helped us understand actin dynamics in neurons, which turned out to be one of the most important pathogenic mechanisms of schizophrenia.

Live cell confocal movie of Lifeact-mRuby- and EB1-YFP-expressing primary hippocampal neurons of the indicated genotypes at DIV1 treated with or without 500 µM betaine.

We identified that both elevated carbonyl stress in the body and the KIF3 kinesin motor dysfunction in cells are synergistically leading to neuronal cytoskeletal disorders and dendritic hyperbranching. These can be reversed by the carbonyl stress scavenger, betaine. Based on these data, we propose betaine as a new remedy for schizophrenia.

How did you use confocal and scanning electron microscopy to study the cytoskeleton?

Microscopy was used to elucidate the molecular mechanism of the neurite hyperbranching.

By observing immature Kif3b+/- neurons, we noticed marked abnormalities in lamellipodia, which are the peripheral membrane-like structure of developing neurons. Normal lamellipodia continuously showed erratic movements due to the folding of actin bundles, but these movements were significantly impaired in Kif3b+/- neurons. Furthermore, we found that this hyperstabilization of the lamellipodial movement lead to abnormal invasion of microtubules into the peripheral actin-rich domain of lamellipodia, by simultaneously imaging F-actin and microtubules.

Double-color time-lapse recording of microtubule invasion from the C- into the P-domains of lamellipodia in EB1-YFP-expressing (corresponding to microtubule plus ends, Magenta) and Lifeact-mRuby-transgenic (corresponding to F-actin, Cyan) hippocampal neurons of the indicated conditions at DIV1 with or without administration of 500 μM betaine, which are represented by the time lapse sequence (Left three columns), time stacks during 10 min (Right column). Scale bar, 2 μm. Arrows, movements of typical puncta of microtubule plus ends. Note that the microtubule plus ends tended to be excluded from the P domain but the Kif3b+/- neurons, where they specifically invaded into the P domain of the lamellipodia. Imaged using confocal microscopy.

Double-color time-lapse recording of microtubule invasion from the C- into the P-domains of lamellipodia. Arrows show movements of typical puncta of microtubule plus ends. Imaged using confocal microscopy.

The microtubule plus ends tended to move horizontally in Kif3b+/+ neuron lamellipodia, but rather perpendicularly in Kif3b+/- ones toward the periphery of the lamellipodia. These excessive peripheral microtubules hyperstabilized the processes on the Kif3b+/- lamellipodia, which was considered to be the basis of dendritic hyperbranching. Very intriguingly, a significant loss of actin bundling was found there. According to both confocal microscopy with Airyscan and scanning electron microscopy, the density of actin bundles in Kif3b+/- lamellipodia was significantly lower than that of Kif3b+/+ lamellipodia. These data suggested that the continuous actin bundle dynamics may be essential for excluding microtubules from the peripheral region of the lamellipodia and for suppressing the dendrite hyperbranching. These cytoskeletal phenotypes could be reversed by betaine administration to the culture medium.

Imaging of actin cytoskeleton in the lamellipodia of primary hippocampal neurons of the indicated conditions at DIV1 either by Airyscan superresolution microscopy labelled with fluorescent phalloidin (top), or by scanning electron microscopy (bottom). Scale bars, 2 μm (top) and 1 μm (bottom).

Imaging of actin cytoskeleton in the lamellipodia of primary hippocampal neurons of the indicated conditions at DIV1 either by Airyscan superresolution microscopy labelled with fluorescent phalloidin (top), or by scanning electron microscopy (bottom). Scale bars, 2 μm (top) and 1 μm (bottom).

As a candidate effector cargo of KIF3 for actin bundling, we identified the neuronal regulator protein, collapsin response mediator protein 2 (CRMP2), which had previously been identified as a major target of carbonylation in the brain (Toyoshima et al. 2019). KIF3 was bound to CRMP2, and KIF3-CRMP2 complex was nicely localized to microtubule and actin cytoskeleton using 4-color Airyscan microscopy imaging. This CRMP2 distribution was significantly reduced in the periphery of Kif3b+/- neurons.

Immunofluorescence micrographs of a Kif3b+/+ primary hippocampal neuron at DIV1 labeled with an antibody against CRMP2 (Magenta), an antibody against KIF3A (Yellow), fluorescent phalloidin (Cyan), and an antibody against α-tubulin (Green), at low (Upper) and high magnifications (Lower), imaged using confocal microscopy.

Immunofluorescence micrographs of a Kif3b+/+ primary hippocampal neuron at DIV1 labeled with an antibody against CRMP2 (Magenta), an antibody against KIF3A (Yellow), fluorescent phalloidin (Cyan), and an antibody against α-tubulin (Green), at low (Upper) and high magnifications (Lower), imaged using confocal microscopy.

To find the mechanistic link between betaine and CRMP2 activity, we finally conducted biochemical assays. We synthesized recombinant CRMP2 protein using E. coli, modified them with carbonylation, and mixed them with fluorescent actin bundles to observe with TIRF microscopy by the help of our previous colleague biochemist Dr. Tadayuki Ogawa. Very intriguingly, the unmodified CRMP2, but not the carbonylated CRMP2 showed strong actin bundling activity. This study suggested that decarbonylation of CRMP2 by betaine could functionally compensate the CRMP2 deficiency in Kif3b+/- lamellipodia by improving its actin bundling activity.

Actin bundling assay for CRMP2 or AGE-CRMP2 using TIRF microscopy. Note that CRMP2 (Upper) but not hypercarbonylated AGE-CRMP2 (Lower) could facilitate actin bundling. Scale bar, 5 μm.

Actin bundling assay for CRMP2 or AGE-CRMP2 using TIRF microscopy. Note that CRMP2 (Upper) but not hypercarbonylated AGE-CRMP2 (Lower) could facilitate actin bundling. Scale bar, 5 μm.

How does this research fit into your overall goals for studying schizophrenia?

Prof. Nobutaka Hirokawa: Since our discovery of KIFs, we have studied the molecular mechanism of intracellular transports. Among our approaches we especially like to visualize the molecular mechanisms using various kinds of microscopy. Really “Seeing is believing”. Recently we have uncovered that defects in KIFs are deeply related with pathogenesis of human diseases and we will pursue this direction. I’m very honored and glad to see that we finally got a nice approach to the fundamental remedy to schizophrenia in this study as one of the most fruitful outcomes of our KIF research. I hope this remedy will be further appreciated by the clinicians, because there have been still few good medications for schizophrenia.

Prof. Takeo Yoshikawa: I started my career as a psychiatrist. Among psychiatric illnesses, schizophrenia is the severest disease and the effects of available therapeutics are limited. I have dedicated my research to schizophrenia. Although this disease is very common with 1% prevalence, the molecular etiology has been largely elusive. My research goal is the ultimate identification of the disease etiology and development of fundamental remedy, based on cellular and biochemical understanding of molecular pathogenesis. I really appreciate and enjoyed the collaboration with these cell biologists from the University of Tokyo in these five years that provided an ultimate progress in this field.

Read Next

Visits on this Page:4615