Direct activation of sparse, distributed populations of cortical neurons by electrical microstimulation

Abstract

For over a century, electrical microstimulation has been the most direct method for causally linking brain function with behavior. Despite this long history, it is still unclear how the activity of neural populations is affected by stimulation. For example, there is still no consensus on where activated cells lie or on the extent to which neural processes such as passing axons near the electrode are also activated. Past studies of this question have proven difficult because microstimulation interferes with electrophysiological recordings, which in any case provide only coarse information about the location of activated cells. We used two-photon calcium imaging, an optical method, to circumvent these hurdles. We found that microstimulation sparsely activates neurons around the electrode, sometimes as far as millimeters away, even at low currents. Our results indicate that the pattern of activated neurons likely arises from the direct activation of axons in a volume tens of microns in diameter.

Figure 1. Using two-photon imaging to measure the effects of cortical microstimulation

(A) Schematic location of imaging sites in cortex, primary visual cortex of mouse, rat (not shown), and cat (area 18). (B) Two-photon bulk-loaded calcium imaging in vivo. Femtosecond-pulsed laser light is used to measure calcium-induced fluorescence changes in neurons. A single plane is imaged at one time. Lower panel: example image. All cells are loaded with OGB-1 AM (green), and astrocytes are labeled with SR101 (red/yellow). (C) Relationship between calcium concentration and spiking activity. Top: a simulated train of 5 spikes. Middle: spike rate, computed by smoothing the spike train with a Gaussian kernel. Bottom: expected somatic calcium concentration, computed by convolving an exponential describing calcium influx with the spike train.

Figure 2. Measured time courses in neurons in response to microstimulation

(A) Anatomical view of neurons and astrocytes in mouse visual cortex. Electrode is positioned 25 μm to right of image. Arrows point to somas of 5 cells: four neurons (1,2,3,5) and one astrocyte (4). Image frames were collected at 2.5 Hz. (B) Time courses (average of 15 repetitions) of the differential fluorescence signal (ΔF/F₀) from the 5 cells labeled in (A). Only cell 1 responded. Stimulation with glass pipette: 100 ms train, 16 μA. (C) Time courses of single trial responses from cell 1 in (A). Each of 15 repetitions is plotted in a single color and the mean is plotted in black. (D–E) As in (A–B), for a second experiment in mouse visual cortex. Here image frames were collected at 31 Hz. Stimulation with glass pipette: 100 ms train at 10 μA. (F) Expanded view of cell 1's average trace in (E). (G) Time courses of individual trials from cell 1 in (E,F). Conventions as in (C).

Figure 3. The pattern of activated cells is sparse

(A) Anatomical images with overlay showing activated neurons for two currents; metal electrode, cat area 18. Pink indicates cells with greater than 20% ΔF/F₀ average response. While no cells are activated at 7 μA, several are activated at 9 μA. (B) Average ΔF/F₀ responses of all cells for experiment in (A). Pink bars are cells that showed greater than 20% ΔF/F₀ average responses. (C) As in (A); glass pipette, mouse visual cortex. Number of activated cells increases with current. (D) Summary of 8 experiments in which current was systematically changed. X-axis, current. Y-axis, number of cells activated in imaging plane. Dotted lines: stimulation applied with pipette; solid lines, metal electrode. Bottom: vertical lines show inferred threshold for each experiment. (E) As in (A, C); metal electrode, cat area 18. Electrode was positioned 4 mm away, to bottom right of image.

Figure 4. Moving the tip slightly yields a different set of activated cells

(A) Image showing the position of cells relative to the electrode. Only the region well-loaded by the calcium indicator is shown. Glass pipette, mouse visual cortex. (B) Schematic diagram indicating which cells were activated by stimulation before and after withdrawing the electrode by 15 μm. While some cells respond both before and after moving the tip (purple), many respond exclusively before (magenta) or after (cyan). Note: cell outlines enlarged for clarity. Stimulation: 100 ms train at 10 μA (interleaved with 25 μA trains, results shown in D). (C) Responses of all cells, before and after moving the tip. X-axis: average ΔF/F₀ response before moving the electrode. Y-axis: response to stimulation after the tip was moved. Gray data points did not reach activation threshold. Others colored as in (B). (D) Distribution of responses for low current (A–C) and high current (25 μA) conditions. Higher currents activate more cells, with more overlap between before and after populations. (E) Time courses of responses for another experiment in which electrode was moved 15 μm away and then repositioned to its original location. Individual trials shown as different colors. Glass pipette, mouse visual cortex. Stimulation: 100 ms train at 12 μA. Three example cells are shown here, out of 136 imaged cells. A total of 7 cells were activated at position 0 (left), 14 at the deeper position (middle) and 8 when tip was returned to position 0 (right). Of these, 1 cell active at position 0 was no longer activated on return to position 0, and two additional cells were activated, presumably because the electrode was not restored to the exact same (micron-level precision) position in the tissue. (F) Fraction of cells activated at both electrode positions as a function of displacement, for four experiments (expt. 1 – 4). Expt.. 1 is data in (B–C). Expts. 2 – 4 are control experiments in which displacement was increased from 5 to 30 μm. All experiments were done at near-threshold currents (10, 12 and 10 μA). Fraction at position 0 is defined at 100%.

Figure 5. Activation is similar after blocking excitatory transmission

(A) Cellular responses before (left), during (center), and after (right) blockade of excitatory glutamatergic synapses with CNQX and APV. Each row shows the responses of a single cell. Visual stimulus was a drifting square-wave grating in a direction (0 deg) chosen to most strongly excite this region. Electrical stimulus was a 100 ms train at 250 Hz. (B) Average time courses of all cells shown in B. Responses to visual stimuli were abolished by drug application but responses to electrical stimulation were left intact. Note that electrical responses are larger than visual responses; the amplitude of the visual response is consistent with earlier work (Ohki et al., 2005; Sohya et al., 2007). (C) Time course of responses during drug application. Here we averaged ΔF/F₀ responses over neuropil and all cells in the imaged region to be resistant to small pipette movements. (D) Results from 3 experiments (visual cortex; 1 mouse, 2 cats). Y-axis: percent change between pre-drug baseline and drug application of ΔF/F₀ responses to microstimulation, averaged over entire imaged region. Error bars: 3 standard deviations. Dashed line indicates no change. Stars (*) indicate results from experiment in (A–C); diamond, results from experiment in (E). (E) Control experiment in which the same cells were imaged throughout the experiment. Stimulation: 100 ms trains at 25 μA.

Figure 6. Many cells show large responses or no response, while neuropil activation is homogeneous

(A) Maps of activation as current is increased; ΔF/F₀ was calculated on a per-pixel basis and shown for all pixels (color scale shown at right). Same experiment as shown in Figs 3B and 5. Stimulation train: 100 ms. (B) Enlarged view of area indicated with dotted line in A, 10 μA (rightmost panel of A). Shown here is the average anatomy image, with neurons green and astrocytes red/yellow. Electrode tip is visible at top right. (C) ΔF/F₀ map, same region shown in B and indicated with dotted line in A. White arrows: non-activated cells; pink arrows: cells that respond strongly to stimulation. Color scale as in A. Note the activated neuropil region immediately around the tip, which is masked by two less-active cells, one on each side. (D) Panel showing our method for computing cell responses: cells (white lines) were identified from anatomy image. Average ΔF/F₀ value was computed within each white region and plotted here in color. Electrode tip position and pink and white arrows, same conventions as B, C. Same color scale as A and C.

Figure 7. Neuropil activation shows a slow spatial falloff

(A) Anatomical image, with high signal-to-noise area indicated by black outline. Mouse visual cortex, glass pipette, same experiment as in Fig. 4. White lines indicate contours of constant distance from the tip. (B) Neuropil response, plotted as a function of distance from the tip. X-axis, distance from tip; Y-axis, ΔF/F₀, averaged over all pixels in the neuropil at that distance; cell regions are masked out. Note the large neuropil peak near the tip, and the slow falloff at larger distances. (C) Response as current is increased. Color scale: average ΔF/F₀ response to 30 repetitions of stimulation.

Figure 8. Model of cell recruitment by local axonal activation

(A) Model of effects at small scales. A small region of directly-activated neural processes near the tip yields sparse activated cell bodies at a distance. (B) Model of effects at large scales. Activating processes near the tip gives a ball of activated cells, but even near threshold this ball is sparse. Increasing current causes the ball to fill in as more cells are activated throughout. (C) Schematic showing the large number of potential axons near the electrode tip. Left, two-photon anatomy image showing neural cell bodies, electrode (white) and neuropil regions between cell somata. Middle: electron micrograph of a 20 μm square region of mouse cortex, from a different animal. Pipette tip is shown schematically and drawn to approximate scale; two cell bodies are labeled. Right: an enlarged view of a 4 micron square region. N, nuclei; d, dendrites; a, probable axons; m, mitochondria.