Reduced inhibition in depression impairs stimulus processing in human cortical microcircuits

Abstract

Cortical processing depends on finely-tuned excitatory and inhibitory connections in neuronal microcircuits. In major depressive disorder (depression), a disrupted balance due to weaker inhibition by somatostatin-expressing interneurons is implicated in cognitive deficits and rumination symptoms. Here, we tested the impact of reduced somatostatin interneuron inhibition on cortical processing in human microcircuits in depression using a data-driven computational approach. We integrated human cellular, circuit and gene-expression data to generate detailed models of human cortical microcircuits in health and depression. We simulated microcircuit baseline and response activity and found reduced signal-to-noise ratio of cortical processing, and increased false/failed detection of stimuli, due to a higher baseline activity (noise) in depression. Our results thus demonstrate mechanistically how reduced inhibition in human neuronal microcircuits impairs cortical processing in depression, thus establishing a target mechanism for novel treatments and providing quantitative links between inhibition and cognitive deficits which could improve the diagnosis of depression.

EEG biomarkers of reduced inhibition in human cortical microcircuits in depression

“…Resting-state activity simulations. We simulated eyes-closed resting-state by injecting the microcircuit models with background excitatory input representing cortical and thalamic drive, as used previously [24]. The background input was generated by random Orstein-Uhlenbeck (OU) point processes [90] placed on all neurons at the halfway point of each dendritic arbor, and 5 additional OU processes placed along the apical trunk of Pyr neurons in equal intervals, from 10% to 90% apical dendritic length.…”

“…However, experimental methods are limited in their ability to characterize the effects of the cellular changes in depression on human brain signals in vivo, thus meriting the use of computational models. The increased availability of human neuronal electrical and synaptic connectivity data [22,36,37], provides important constraints on detailed models of human cortical microcircuitry [24] which can be used to simulate microcircuit activity in health and disease, including local circuit-generated EEG signals [38][39][40].…”

Age-dependent increased sag amplitude in human pyramidal neurons dampens baseline cortical activity

“…We simulated L5 microcircuits comprised of 1000 neurons distributed along a layer 5 volume (500x500x700 μm 3 , 1600 to 2300 μm below pia; Mohan et al, 2015) using NEURON (Carnevale & Hines, 2006) and LFPy (Hagen et al, 2018). In addition to Pyr and PV neurons, the microcircuit models included somatostatin (SST) and vasoactive intestinal peptide (VIP) interneurons, using previous human L2/3 neuron models (Yao et al, 2021), because there was no human data for these interneurons from L5. Neurons of a given type had the same model (morphology and biophysical properties) but differed in the randomization of their synaptic connectivity and background input.…”

Functional Contribution of the Medial Prefrontal Circuitry in Major Depressive Disorder and Stress-Induced Depressive-Like Behaviors

“…Furthermore, SST-KO mice exhibit depressive- and anxiety-like behaviors along with reductions of BDNF and GAD67 expression in GABA interneurons and elevated corticosterone blood levels ( Lin and Sibille, 2015 ). By compiling human cellular, synaptic, and gene expression data from recent studies, Yao et al (2021) generated computational models of human layer II/III local circuit. Among these models, they created a depression microcircuit model by reducing SST inhibition according to previous gene expression data from MDD patients ( Seney et al, 2015 ).…”

“…Among these models, they created a depression microcircuit model by reducing SST inhibition according to previous gene expression data from MDD patients ( Seney et al, 2015 ). In this model, a reduction in SST inhibition leads to an increase in the pyramidal firing rate, increasing baseline activity, disrupting signal-to-noise ratio, and the increasing probabilities of both false signal detection and failed signal detection over background noise ( Yao et al, 2021 ).…”