SV2A-PET reveals transcriptomic excitatory-inhibitory imbalance in schizophrenia

Background

Contemporary models of schizophrenia implicate a disturbance in the balance between excitatory glutamatergic and inhibitory GABAergic signalling, resulting in altered cortical computation and synaptic function. Convergent evidence from postmortem, cellular, and in vivo studies indicates lower presynaptic terminal density in schizophrenia, particularly in frontal and cingulate cortices. Positron emission tomography (PET) targeting synaptic vesicle protein 2A (SV2A) provides an in vivo proxy measure of presynaptic terminal density [1]. However, whether alterations in SV2A-PET signal reflect selective disruption of excitatory or inhibitory synaptic architecture remains unknown. This in vivo measure can be co-modelled with transcriptomic markers of excitation and inhibition to investigate whether E/I imbalance may be present in schizophrenia and to identify which synaptic subtypes are preferentially affected.

Methods

[11C]UCB-J PET data were acquired from 71 participants (n=42 SCZ, n=29 CTR), including both medicated (n=20) and unmedicated individuals (n=22) with schizophrenia. Regional SV2A binding was quantified as distribution volume ratio relative to centrum semiovale (DVRcs) and parcellated using the Harvard-Oxford atlas (139 bilateral ROIs). Individual-level generalised least squares (GLS) regression modelled regional SV2A binding against transcriptomic predictors from the Allen Human Brain Atlas, focusing on markers of excitatory and inhibitory synapses (VGLUT1, VGLUT2, and VGAT), with Ledoit-Wolf shrinkage applied to stabilise covariance estimation. Spatial spin permutation testing (100,000 rotations) controlled for spatial autocorrelation. Symptom severity was quantified using the PANSS, and exploratory associations with magnetic resonance spectroscopy-derived glutamate measures from anterior cingulate cortex and hippocampus were examined.

Results

Transcriptomic predictors accounted for a substantial proportion of regional SV2A variance across both schizophrenia and control participants (R² = 0.39±0.10). Model fit quality did not differ significantly between groups (m-SCZ: 0.364±0.120; um-SCZ: 0.398±0.088; HC: 0.391±0.08), indicating preserved transcriptomic organisation of synaptic density in schizophrenia. Hierarchical group comparisons revealed a selective reduction of the VGLUT1-related SV2A component in schizophrenia compared with healthy controls (HC: 0.341±0.110; SCZ: 0.279±0.118; t(69) = 2.198, p = 0.031, Cohen’s d = 0.54), driven primarily by medicated patients. VGLUT2- and VGAT-related components showed no group differences, suggesting selective disruption of corticocortical glutamatergic terminals. No robust associations were observed between transcriptomic SV2A components and symptom severity. Exploratory analyses indicated positive associations between VGAT- and VGLUT2-related components and glutamate levels in the anterior cingulate cortex (ρ = 0.45, p = 0.018) and hippocampus (ρ = 0.60, p = 0.001) in healthy participants; these associations were absent in schizophrenia.

Conclusions

Our findings demonstrate that regional SV2A-PET signal can be decomposed into transcriptomic components reflecting excitatory and inhibitory synaptic architecture. The selective reduction of VGLUT1-related coupling in schizophrenia, without corresponding changes in VGAT coupling, points toward a specific vulnerability of corticocortical glutamatergic systems rather than uniform presynaptic loss. The disruption of MRS glutamate–gene expression coupling in schizophrenia further suggests that normative integration between excitatory neurochemistry and inhibitory synaptic organisation is perturbed. Together, these results establish SV2A-PET combined with transcriptomic modelling as a novel in vivo framework for quantifying excitation-inhibition imbalance in schizophrenia, extending this measure from a scalar index of synaptic density to a biologically stratified systems-level readout.

References

Onwordi, E. C., E. F. Halff, T. Whitehurst, A. Mansur, M.-C. Cotel, L. Wells, H. Creeney, D. Bonsall, M. Rogdaki and E. Shatalina (2020). “Synaptic density marker SV2A is reduced in schizophrenia patients and unaffected by antipsychotics in rats.” Nature communications 11(1): 1-11.