Fig 1.
Representation of the experimental paradigm studied.
An electrode on the cortical surface (squared, size 150 μm) delivers current to cells embedded in the cortical tissue (shown is one example reconstruction of pyramidal cells in layer II/III). The current pulse is monopolar, with amplitude ranging from 0 to 150 μA, and lasts 200 μs.
Fig 2.
Anatomical reconstructions of the main types of cortical neurons.
(a) Typical anatomical profiles for the main types of cortical neurons. Green denotes axon, purple–apical dendrite, blue–basal dendrite, red dot shows soma position. Top row exhibits excitatory cells (PY—pyramidal neurons, SC—spiny stellate cell). Bottom row contains inhibitory interneurons (BC—basket cell). (b) Averaged axonal densities formed by the neurons of each specific type. Color denotes logarithm of averaged axonal density (AD), computed over a set of available reconstructions of cortical cells. Logarithmic scale was used for better visualization of axonal arborization. This provides a general intuition on the generic shape of the axonal arborization for distinct types of cortical cells, which is crucial for the analysis.
Fig 3.
Estimation of the activation probability induced by surface stimulation.
An example of typical layer IV pyramidal cell is shown. For each cell, we assigned R, and Z (depth) parameters. Activating function identifies its trigger area (red markers), where the effective current is above threshold. Action potentials can be initiated in these segments and propagate along the axonal arborization. To populate a statistical set (to find the average probability of spiking), each cell reconstruction was shuffled by rotating and shifting along the vertical axis (indicated by bold arrows), and multiple reconstructions were considered for each cell type (up to a total of 561 cells, see S1 Table and Methods: Selecting cell reconstructions within available databases).
Fig 4.
Probability of stimulation-induced activation is different across layers and cell types.
The top row shows direct activation probability for 3 distinct types of layer I interneurons. Rows 2–4 (top to bottom) correspond to layers II—V. The left column contains probability for excitatory cells (pyramidal and spiny stellate), middle column contains data on soma/proximal dendrite-targeting interneurons (basket cells) and the right column contains probability for tuft/proximal dendrite-targeting interneurons (Martinotti and bitufted cells). The data represent anodal stimulation (I = 275 μA).
Fig 5.
Different cell types have distinct preferences for stimulation type (anodal or cathodal).
(a-c) Dependence of the activation probability on the net electrode current I for excitatory cells (a), basket cells (b), Martinotti and bi-tufted cells (c). (d1,2) Anodal stimulation (d1) activates pyramidal cells LII/III more effectively than cathodal stimulation (d2). (e1,2) Pyramidal LIV/V and spiny stellate cells have no preference for any type of stimulation. Because of the rich axonal arborization in supragranular layers, both types of stimulation provide large activation area. (f1,2) Cathodal current is more effective in activation of non-myelinated horizontal axons of Martinotti cells in supragranular layers.
Fig 6.
Numerical simulations predict that feedback inhibition controls response properties.
(a): Schematic representation of the network model structure, which consists of 3 types of cells, located in 3 different layers (canonical circuit). PY stands for pyramidal neuron, SpS–spiny stellate cell, BC–basket cells, MC–Martinotti cells. Lines with circles denote excitatory AMPA connections (solid–strong, dashed—weak), whereas bars denote inhibitory GABA connections. (b): Two electrophysiological classes of neurons were used in our simulations: top voltage trace (green) corresponds to regular spiking neurons (used for pyramidal, spiny stellate cells and Martinotti cells) and bottom voltage trace demonstrates activity of fast spiking interneurons (used for basket cells). (c): Spike raster plots exhibit network activity for cathodal (left panel) and anodal (two right panels) stimulations. The cells were activated during first 1 ms of simulation according to activation probability (see text for details). Green dots–PY and SpS cell spikes, red dots–interneuron spikes. Left panel shows weak response to cathodal stimulation (-100 μA). Middle panel shows response to moderate anodal stimulation (75 μA), which induced a large population response. The right panel shows response to a strong positive current (300 μA), which activated a large number of basket cells in layer II/III and Martinotti cells in all layers, which prevented the activation of excitatory cells in layers II-IV. (d, e): Population responses as a function of net electrode current for layer II-IV excitatory cells.