Brain stimulation preferentially influences long-range projections

Abstract

Advances in brain stimulation have made it possible to target smaller and smaller regions for electromagnetic stimulation, in the hopes of producing increasingly focal neural effects. However, the brain is extensively interconnected, and the neurons comprising those connections may themselves be particularly susceptible to neurostimulation. Here, we test this hypothesis by identifying long-range projections in single-unit recordings from nonhuman primates receiving transcranial alternating current stimulation. We find that putative long-range projections are more strongly affected by stimulation than other cells. Specifically, they are both more entrained on average and account for occurrences of extremely strong entrainment. Given that stimulation appears to target the edges, rather than nodes, of neural networks, it may be necessary to rethink how neurostimulation strategies are designed.

Fig. 1.. Unit types can be identified via their extracellular action potentials.

(A) Sketches of action potentials recorded near the soma (green) and along an axon (blue). On the basis of the underlying biophysics, signals recorded at these locations have very different shapes. We extracted features capturing these differences, as indicated on the blue waveform. (B) Distributions of these features and threshold values (dashed red lines) are shown in the histograms; see Materials and Methods. (C) On the basis of these features, we divided our data into five cell classes: putative axonal recordings from long-range projections (blue and black), interneurons (purple), and pyramidal cells (red and green). Individual (thin lines) and average waveforms (thick) are shown for each class, with the threshold values shown beneath. As the three-dimensional scatterplot shows, these classes form distinct clusters in the joint feature space. The clusters differ in average firing rate and distribution across brain areas. *P < 0.05. BW, bandwidth; BG, basal ganglia; HC, hippocampus. Cell in (A) generated from the data of Hallerman et al. (28, 88).

Fig. 2.. Only signals from putative axons persist after local activity was pharmacologically silenced.

(A) Averaged extracellular waveforms (±1 standard deviation) recorded before (left) and 40 min after (right) injecting muscimol, a potent GABA (γ-aminobutyric acid) agonist. Waveforms are color-coded as in Fig. 1 and are shown on the basis of their relative locations on the microelectrode array. Red X’s indicate neurons that were no longer present after injection. (B) Average firing rates for isolated single units before and after muscimol injection. (C) High-pass filtered signal on four example channels, indicated in (A), showing the lack of multiunit activity as well. (D to F) Same as (A) to (C) but in a second animal.

Fig. 3.. Putative long-range projections are more strongly affected by tACS.

(A) Changes in PLVs for each of the 405 classifiable neurons in our dataset, colored by cell class (as in Figs. 1 and 2); shading indicates individually significant changes (P < 0.05) in ΔPLV. Horizonal lines indicate the medians for each class, and arrowheads indicate the 90th percentile. The gray region indicates the 95% CI around the overall mean. n.s., not significant. (B) Raster plots showing the activity of two example neurons: a putative axonal recording from a long-range projection [labeled a in (A)] and a somatic recording from a putative pyramidal cell (labeled b). Spiking activity during sham is shown in gray, and activity during tACS is shown in color. Polar histograms showing the phase of spiking relative to the stimulation (color) and local field potential (gray) demonstrate how phase locking changes; PLV values are shown below each histogram. * indicates significant differences (P < 0.05) in medians; # indicates significant differences in the 90th percentile.

Fig. 4.. Stimulation may cause complex effects on brain-wide activity.

(A) Stimulation is traditionally expected to affect local computations in the targeted region, and the results of that perturbation are passed onto other brain regions (arrowheads) via its outputs (lines). (B) Our data suggest that stimulation not only modulates local processing but strongly affects inputs and outputs to the target area as well. (C) In extreme cases, local and long-range effects may even be qualitatively different, with local activity desynchronized (blue) but long-range connections entrained to the stimulation (red). See text for details.