Reticular formation responses to magnetic brain stimulation of primary motor cortex

Abstract

Transcranial magnetic stimulation (TMS) of cerebral cortex is a popular technique for the non-invasive investigation of motor function. TMS is often assumed to influence spinal circuits solely via the corticospinal tract. We were interested in possible trans-synaptic effects of cortical TMS on the ponto-medullary reticular formation in the brainstem, which is the source of the reticulospinal tract and could also generate spinal motor output. We recorded from 210 single units in the reticular formation of three anaesthetized macaque monkeys whilst TMS was performed over primary motor cortex. Short latency responses were observed consistent with activation of a cortico-reticular pathway. However, we also demonstrated surprisingly powerful responses at longer latency, which often appeared at lower threshold than the earlier effects. These late responses seemed to be generated partly as a consequence of the sound click made by coil discharge, and changed little with coil location. This novel finding has implications for the design of future studies using TMS, as well as suggesting a means of non-invasively probing an otherwise inaccessible important motor centre.

Figure 1. Antidromic identification and artefact removal

A, example identification of a reticulospinal neurone. Successful collision of a spontaneous spike (a) from a neurone within PMRF with an antidromic spike (b) elicited by spinal cord stimulation at a spike-stimulus interval of 1.3 ms (bottom) but not at 1.4 ms (top). B, example application of the artefact removal algorithm. The unprocessed recording in response to a single stimulus is shown as a thick line. The artefact may be divided into a period where the amplifier has saturated (a), and a subsequent phase where neural activity can be detected on a shifted baseline (b, shaded grey). This baseline was fitted to a double exponential curve (dotted line) and subtracted, yielding the trace shown underneath (thin line) in which a spike may be discerned responding to the stimulus (*). Dashed vertical lines in all panels represent stimulus time.

Figure 2. Individual PMRF cell responses to TMS

Example raster plots of single cells responding with spikes in the early latency window (A), early and late windows (B) and with suppression of firing following TMS (C). Dotted line denotes stimulus onset. Corresponding peri-stimulus time histograms are shown for each cell during TMS at an intensity of 100% MSO; dashed horizontal lines mark the mean, and mean + 2 SD, of the baseline (pre-stimulus) firing. Overlain waveforms for the discriminated spikes are shown as insets.

Figure 3. Population PMRF responses to TMS

A, plot of response probability for all cells across the population which respond to TMS (n = 159). Each horizontal bar represents a single cell response to a single stimulus intensity; response is indicated using the false colour scale on the right, which gives the probability of firing in a 1 ms bin. The dotted red line denotes stimulus onset. Rows have been sorted by response onset latency. Cells which responded at multiple intensities contribute multiple rows to this figure. B, distribution of onset latency (first elevated PSTH bin post-stimulus), assessed from A. Bins have been colour coded according to the different response latency windows (green, early; blue, middle; red, late). C, scatter plot of response jitter against latency (standard deviation, and mean, of single-trial first-spike latency respectively). Each point in C describes the response from one cell at an intensity 100% MSO; n = 124 cells were used to compile this plot.

Figure 4. Response characteristics

A, spinal epidural recordings following ascending intensities of TMS from monkey P, recorded from a ball electrode on the spinal dura between the C4 and C5 vertebrae. Stimulus onset is marked with a dashed black line and arrow; direct (D) and indirect (I) responses are shown within the dashed grey lines. Grey circles denote an early D response, probably due to white matter activation. B, histograms of response thresholds for responses of 1–3 ms (column 1), 3–7 ms (column 2) and 7–25 ms (column 3) after the stimulus. Black and white shading indicates responses to contralateral and ipsilateral M1 stimulation, respectively (see inset schematic diagram of head, in which recording side is indicated by an arrow). In each plot, n indicates the number of cells responding. Arrows indicate, for comparison, the threshold for a spinal volley in each animal (30% for animal P, 30% for V, and 35% for U). C, cumulative probability distribution of response thresholds for each response window; data from ipsilateral and contralateral responses has been combined for this figure. D, spikes elicited per stimulus for cells which responded to TMS at a given intensity. Error bars designate SEM. E, the product of the plots in C and D. This can be interpreted as the spikes elicited per stimulus averaged across all cells, including (as zeros) those cells which did not respond. Error bars represent SEM, determined by the Monte Carlo resampling procedure described in Methods.

Figure 5. Anatomical reconstruction of recording sites

Representative parasagittal sections through the brainstem showing the location of recording sites and cells responding in early (1–3 ms, A), middle (3–7 ms, B) and late (7–26 ms, C) latency windows. Red triangles indicate cells with significant responses; black circles cells with no responses. Symbols outlined in green mark identified RST cells. The location of some symbols have been shifted to allow all recording sites to be seen more clearly without overlap; such shifts for display clarity were always less than 0.5 mm. Gi, gigantocellular reticular nucleus; Cu, cuneate; IO, inferior olive; PnC, caudal pontine reticular nucleus; 6N, abducens nucleus.

Figure 6. Evidence on which pathways contribute to PMRF response to TMS

A, a cell which responded strongly to contralateral TMS in the late latency window, but which did not respond to electrical stimulation of the scalp through percutaneous needles inserted beneath the coil (10 mA, 0.2 ms pulse width). B, responses from a single cell during contralateral TMS. Note that the responses persisted even at the lowest non-zero intensity tested of 5% MSO. C, responses from a different cell following contralateral TMS, with the coil positioned 1 cm above the scalp. D, responses following contralateral TMS at 60% MSO, as the coil was raised off the head by the distances indicated. In the final two periods, a plastic disc (6.35 mm thickness) was inserted to fill the gap between the coil and the head, restoring the response in the late window. In all panels, responses are shown as raster displays, with stimulus conditions fixed within the regions denoted by horizontal lines. Dashed vertical line marks time of stimulus.

Figure 7. Single cell response to TMS and click stimuli

A, raw recordings of responses to TMS stimuli over contralateral motor cortex at 60% MSO. B, raster plots showing how responses varied with TMS intensity. C, single sweep responses from the same cell, following click stimuli delivered through a bone vibrator placed at the same location as the coil in A and B. D, raster plot showing changes in response to clicks at different intensities. Click intensity is expressed as eTMS%, the intensity of TMS which produced comparable sound intensities in separate calibration experiments.