Neurogenesis-independent mechanisms of MRI-detectable hippocampal volume increase following electroconvulsive stimulation

Abstract

Electroconvulsive therapy (ECT) is one of the most effective psychiatric treatments but the underlying mechanisms are still unclear. In vivo human magnetic resonance imaging (MRI) studies have consistently reported ECT-induced transient hippocampal volume increases, and an animal model of ECT (electroconvulsive stimulation: ECS) was shown to increase neurogenesis. However, a causal relationship between neurogenesis and MRI-detectable hippocampal volume increases following ECT has not been verified. In this study, mice were randomly allocated into four groups, each undergoing a different number of ECS sessions (e.g., 0, 3, 6, 9). T2-weighted images were acquired using 11.7-tesla MRI. A whole brain voxel-based morphometry analysis was conducted to identify any ECS-induced brain volume changes. Additionally, a histological examination with super-resolution microscopy was conducted to investigate microstructural changes in the brain regions that showed volume changes following ECS. Furthermore, parallel experiments were performed on X-ray-irradiated mice to investigate the causal relationship between neurogenesis and ECS-related volume changes. As a result, we revealed for the first time that ECS induced MRI-detectable, dose-dependent hippocampal volume increase in mice. Furthermore, increased hippocampal volumes following ECS were seen even in mice lacking neurogenesis, suggesting that neurogenesis is not required for the increase. The comprehensive histological analyses identified an increase in excitatory synaptic density in the ventral CA1 as the major contributor to the observed hippocampal volume increase following ECS. Our findings demonstrate that modification of synaptic structures rather than neurogenesis may be the underlying biological mechanism of ECT/ECS-induced hippocampal volume increase.

Fig. 1. ECS induces hippocampal volume increases as measured by MRI.

A The time course of ECS. B The EEG was recorded bilaterally during the session of ECS. C A whole brain voxel-wise group comparison between CTL (0×ECS; n = 12) and 9×ECS (n = 12). The red color represents the volume increases while the blue color represents the volume decreases. D The significant clusters (green: 9×ECS > CTL) were overlayed on the identified dorsal (cyan) and ventral (yellow) hippocampal regions. E The significant clusters (green: 9×ECS > CTL) were overlayed on the identified DG (purple), CA3 (yellow), CA2 (blue), and CA1 (cyan). F Results of the whole brain voxel-wise regression analysis, including the gray matter volume as a dependent variable, the number of ECS sessions as an independent variable, and the TBV as a nuisance covariate. The red color represents a significant positive correlation while the blue color represents a significant negative correlation. G Scatter plots of normalized voxel values (i.e., divided by the TBV and then multiplied by 1000) in the left and right clusters identified by the whole brain regression analysis. There were significant correlations of the number of ECS sessions with the left (r = 0.71, df=46, p < 0.001) and right (r = 0.63, df=46, p < 0.001) hippocampal volumes.

Fig. 2. Neurogenesis is not required for ECS-induced hippocampal volume increase.

A The time course of ECS. The number of the newborn neurons was counted on the hippocampal coronal slices for dDG and horizontal slices for vDG. B Representative images of DCX staining in the dDG and vDG. C The numbers of DCX⁺ newborn neurons in the dDG and vDG were compared between CTL (n = 4) and ECS (n = 4). **p < 0.05, ***p < 0.01 (Student’s t-test, vs. CTL). D The time course of ECS and X-ray irradiation. A lead plate with a 4-mm slit was placed above the mouse head. The slit in the lead plate was placed just above the whole hippocampus. The number of the newborn neurons was counted on the hippocampal horizontal slices of the dDG and vDG. E Representative images of DCX staining in the dDG and vDG. F The number of DCX⁺ newborn neurons in the dDG and vDG were compared among CTL (n = 4), X-ray (n = 4) and X-ray+ECS (n = 4). A whole brain voxel-wise group comparison between X-ray (n = 12) and X-ray+ECS (n = 12) (G), and between CTL (n = 12) and 9×ECS (n = 12) (H). The red color represents the volume increases while the blue color represents the volume decreases. I Scatter plots of normalized voxel values in the left and right overlapping regions between Figures G and H. A significant main effect of group was observed on both sides (Left: F_3,38 = 22.6, p < 0.001; Right: F_3,38 = 17.0, p < 0.001). The graphs display the results of post-hoc Tukey’s tests (*p < 0.05, **p < 0.01).

Fig. 3. ECS increases the expression level of neuronal microstructure-related genes.

A The time course of ECS and RNAseq. B A volcano plot of RNAseq data. The red color represents significant increases in gene expression levels, while the blue color represents decreases (p < 0.05, fold change>1.5). C, D Significant GO terms for biological processes and cellular components of the upregulated genes. There were no significant GO terms for down regulated genes.

Fig. 4. ECS increases the length of the stratum radiatum layer in the vCA1.

A Representative VGluT1 staining in the dHip and vHip. SO: stratum oriens; SP: stratum pyramidale; SR: stratum radiatum, SLM: stratum lacunosum moleculare. B, C Each layer thickness and total layer thickness (SO + SP + SR + SLM) in the dCA1 and vCA1 were compared between CTL (n = 8) and ECS (n = 8). A two-way repeated ANOVA was performed (dCA1: ECS p = 0.95, interaction p = 0.091, vCA1: ECS p = 0.24, interaction p = 0.033, Total length: ECS p = 0.078, interaction p = 0.094). D Representative vCA1 neurons with Golgi staining. The red circles represent the somas of the vCA1 neurons. E The length of the apical and basal dendrites of the vCA1 neurons were compared between CTL and ECS (17 neurons were measured from three mice of each group). Two-way repeated ANOVA was performed (ECS p = 0.0025, interaction p = 0.064). F Interaction numbers were plotted in a Sholl analysis of the apical dendrites in the vCA1 neurons. A two-way repeated ANOVA was performed (ECS p = 6.5 × 10⁻¹⁰, interaction p = 0.19). *p < 0.05, **p < 0.01 (Student’s t-test, p-values were Bonferroni corrected, vs. CTL).

Fig. 5. ECS increases excitatory synaptic density.

A Representative fluorescence microscope image of VGluT1 staining (left) and SRM images (right) in each layer of the vCA1. B, C The density and size of the VGluT1⁺ excitatory terminals were compared in each layer between CTL (n = 8) and ECS (n = 8). A two-way repeated ANOVA was performed (Density: ECS p <0.001, interaction p = 0.031, Size: ECS p = 0.77, interaction p = 0.97). D Representative SRM images of VGluT1 and PSD95 in each layer of the vCA1. E The excitatory synaptic density (pairs of VGluT1⁺ puncta and PSD95⁺ puncta) was compared in each layer of the vCA1 between CTL (n = 6) and ECS (n = 6). A two-way repeated ANOVA was performed (ECS p <0.001, interaction p = 0.015). F The sizes of the pairs of the VGluT1⁺ and PSD95⁺ puncta were compared in each layer of the vCA1 between CTL (n = 6) and ECS (n = 6). G Scatter plots of the density or size of the VGluT1⁺ excitatory terminals in the SR and normalized voxel values in the vCA1. There was a significant correlation of the voxel values in the vCA1 with the VGluT1⁺terminal density (r = 0.67, df=34, p < 0.001, R² = 0.44), but not with VGluT1⁺teminal size (r = –0.06, df=34, p = 0.73, R² = 0.004). The voxel values were normalized using the TBV of each animal, and then the values of the left and right vCA1 were averaged. *p < 0.05 (Student’s t-test, p values were Bonferroni corrected, vs. CTL).