Cell-Type-Specific Bidirectional Modulation of the Cortico–Thalamo–Cortical Sensory Pathway by Transcranial Focused Ultrasound (tFUS)

Abstract

Introduction

Transcranial focused ultrasound (tFUS) can noninvasively modulate sensory pathways, but the cell-type-specific mechanisms underlying excitatory or inhibitory effects remain unclear. Here, we investigate how tFUS applied to the somatosensory cortex (S1) influences S1 and posterior medial thalamic nucleus (POm) responses to hind paw vibration-tactile stimulation and which neuronal populations mediate these effects.

Methods

Vibration-tactile stimulation evoked potentials (TEPs) and multi-unit activity (MUA) in S1 and POm were recorded from male rats. Optogenetic tagging was used to identify S1 CaMKII-positive, PV-positive, and SST-positive neurons, while waveform features were used to classify putative excitatory (i.e., regular-spiking units - RSUs) and inhibitory neurons (i.e., fast-spiking units - FSUs) in POm.

Results

We found that only S1 CaMKII-positive neurons and POm RSUs responded robustly to tactile stimulation. When tFUS was applied to S1, high pulse repetition frequency (PRF), high duty cycle, and high-pressure stimulation (etFUS) produced excitatory modulation of the sensory pathway, whereas low PRF, low duty cycle, and low-pressure stimulation (itFUS) induced inhibitory effects. Further analyses suggested that excitatory modulation may involve activation of S1 CaMKII-positive neurons, while the inhibitory effect may be associated with their deactivation.

Conclusions

These findings demonstrate that tFUS exerts bidirectional, parameter-dependent modulation of a sensory pathway and highlight the critical role of CaMKII-positive neurons in mediating these effects. This study provides mechanistic insight into cell-type-specific neuromodulation by tFUS, particularly in bidirectional modulation of a sensory pathway, and informs the optimization of stimulation parameters for targeted therapeutic interventions.

Introduction

Transcranial focused ultrasound (tFUS) has emerged as a promising tool for modulating neural circuits [1–7] and developing therapeutic interventions [8–12] by delivering pulsed mechanical energy to targeted brain regions. Compared with other noninvasive neuromodulation techniques, tFUS offers high spatial specificity and the ability to reach deep brain structures.

Previous studies have demonstrated that tFUS can modulate sensory pathways in both animals [13,14] and humans [15,16] depending on stimulation parameters. In human studies, the ability to discriminate tactile vibration frequencies has been shown to improve when excitatory modulation is induced by applying tFUS to the primary somatosensory cortex (S1) [15]. tFUS has also been reported to significantly attenuate the amplitude of somatosensory evoked potentials (SEPs) elicited by median nerve stimulation [16]. In animal studies, targeting specific thalamic nuclei using 1.14 MHz low-intensity focused ultrasound results in suppression of SEPs [14]. In addition, studies in large-animal models have demonstrated that tFUS applied to S1 and its thalamic projections can safely induce transient and reversible suppression of SEPs [13]. These findings suggest that tFUS is capable of modulating sensory processing within the cortico–thalamo–cortical pathway. However, the neuronal mechanisms underlying how different tFUS stimulation parameters produce distinct modulatory effects within sensory pathways remain largely unknown.

Previous studies have suggested that tFUS can produce cell-type-specific effects depending on stimulation parameters. Intracranial recordings in S1 have shown that tFUS can selectively elicit excitatory responses in regular-spiking units (RSUs; putative excitatory neurons) when applied with pulse repetition frequencies (PRFs) above 300 Hz, whereas fast-spiking units (FSUs; putative inhibitory neurons) show no response under these conditions [17]. Activation of RSUs in S1 has also been shown to drive both feedback and feedforward cortico–thalamo–cortical pathways [1]. In addition, studies delivering cell-type-specific doses of tFUS to the cortex have reported activation of several canonical neuronal populations, including Thy1-positive pyramidal excitatory neurons and inhibitory interneurons such as somatostatin-positive (SST) and parvalbumin-positive (PV) neurons [18]. However, whether and how neuronal selectivity is preserved when tFUS modulates sensory processing within the cortico–thalamo–cortical pathway remains unknown.

In this study, vibration-tactile stimulation is applied to the hind paw to activate the sensory pathway. Multi-site intracranial recordings are used to monitor neuronal activity in both S1 and the posterior medial thalamic nucleus (POm) during vibration-tactile stimulation, with or without concurrent tFUS applied to S1 at varying PRFs and duty cycles (DCs) in male rats. Tactile-evoked potentials (TEPs) are analyzed to assess excitatory or inhibitory effects. S1 neurons are classified into CaMKII-positive, PV-positive, and SST-positive populations using optogenetic tagging, while POm neurons are categorized as RSUs or FSUs based on waveform characteristics. Spiking rates from these neuronal populations are measured to evaluate their responses to tactile stimulation and tFUS, providing insight into the mechanisms by which tFUS modulates sensory processing within the cortico–thalamo–cortical (CTC) pathway.

Methods and Materials

Experimental Design

AAV Virus injection

Male rats (Hsd: WI, Envigo, USA), weighing 250–350g, age 3–6 months, were used as subjects. All animal studies were approved by the Institutional Animal Care and Use Committee at Carnegie Mellon University in accordance with US National Institutes of Health guidelines.

Rats were anesthetized with isoflurane before being mounted on a standard stereotaxic apparatus for viral infusion. Their body temperature was maintained with a heating pad, and heart rate (270–400 bpm) and respiratory rate (50–80 bpm) were monitored during the infusion. The isoflurane was kept at 2.5% with 0.4–0.6 L/min flow rate.

Based on the rat atlas [19], a burr hole was made in the skull over left S1 (AP:−2, ML: 2.5) for each rat. The dura covering the tissue in the hole was carefully removed. PV-ChR2/mCherry (Addgene, No.213945-PHPeB, N = 10), SST-ChR2/mCherry (Addgene, No. 213941-PHPeB, N = 12) or CaMKII-ChrimsonR/mScarlet (Addgene, No. 124651-AAV9, N = 12) were selected to label PV-positive, SST-positive or CaMKII-positive neurons, respectively. 1 μL of the selected virus was injected into the S1 from the left side (DV:1.5mm, angle: 40°) with a speed of 0.2μL/min using a 10 μL microsyringe (Hamilton, No. 7635–01) and a 33-gauge needle (Hamilton, No. 7762–06). After virus infusion, the burr hole was covered by tissue adhesive (3M Vetbond) and the skin was sutured.

Surgical Procedures for Anesthetized Recordings

Four weeks following viral infusion, rats underwent surgery to implant electrodes. Fig. 1(A) illustrates the setup for the experiment. Rats were anesthetized with isoflurane before being mounted on the stereotaxic apparatus. Body temperature was maintained with a heating pad and heart rate (270–400 bpm) and respiratory rate (50–70 bpm) were monitored throughout the surgery. Isoflurane was kept at 2% with 0.4–0.6 L/min flow rate during recordings.

Fig. 1.

Experimental setup and ultrasound parameter. (A) Experimental setup of intracranial electrodes for recording multi-region signals, optical fiber for activating specific type of neurons and 128-element random array ultrasound transducer. The 64-channel optrode was inserted into left somatosensory cortex (S1) and 32-channel electrode was inserted into left posterior medial thalamic (POm). Only bubble-free ultrasound gel was used as the coupling medium between the transducer and the exposed skull. Vibration-tactile stimulation was applied to the right hind paw. (B) Ultrasound parameters used in this study. The inter-sonication interval (ISI) was set at 2.5s with 10% jitter to avoid neuronal adaptation to fixed ISI. Ultrasound duration (UD) was set at 100ms with vibration stimulation or 67ms for tFUS only. Pulse repetitive frequency (PRF) and duty cycle (DC) tested in this study were 30Hz with 0.6%DC and 3000Hz with 60% DC. Pulse duration (PD) were calculated by PRF*DC. (C-D) Ex-vivo hydrophone peak-to-peak pressure amplitude z-axis (upper) and y-axis (bottom) scan at the estimated targeted brain region with low pressure (C, ~98kPa) and high pressure (D, ~163kPa).

To explore the neuronal responses to tFUS, neuronal signals from the S1 and POm were recorded. The POm is a higher-order thalamic area that projects to the cortex and receives driver input from the cortex [21]. The S1 burr hole was remade at the virus injection site by enlarging the original virus injection hole (AP: −2, ML: 2.5), and the POm hole was made at AP: −4.5 and ML: 2.6 in the skull for each rat. The dura covering the holes was carefully removed. A 64-channel Optrode (A1×64-Edge-6mm-22.5–177-OXA64LP, Neuronexus, Ann Arbor, MI, USA) was inserted into S1 from the left side (angle: 40°) to make sure the signals were recorded from the virus injection site. The 32-channel NeuroNexus microelectrode (A1×32-Poly3-10mm-50-177, NeuroNexus) was inserted into POm from the back side (angle: 28°, DV:5.8) to leave space for the transducer. The transducer was set vertically over the skull after the insertion of the electrodes.

Vibration-Tactile Stimulation

To test the tFUS effect on the sensory CTC pathway, vibration-tactile stimulation was used as it has been widely applied to study the functional network integrity of ascending sensory pathways [20–23]. The vibration motor was placed under the right hind paw. A waveform generator was used to drive the motor producing vibration-tactile stimulation. The output was set at 5V with frequency of 100Hz or 500Hz. The duration of stimulation was 100ms.

Transcranial focused ultrasound stimulation (tFUS)

The customized 128-element random array ultrasound transducer H276 (f₀: 1.5 MHz, −3dB axial specificity: 1.36 mm and lateral specificity: 0.46 mm, manufactured by Sonic Concepts, Inc., Bothell, WA, USA) was used. The transducer was driven by a Vantage 128 research ultrasound platform (Verasonics, Kirkland, WA, USA) using a DL-260 connector to steer the ultrasound beam. Ex-vivo scanning was performed using the customized hydrophone-based 3D ultrasound pressure mapping system (HNR500, Onda Corporation, Sunnyvale, CA, USA). To mimic the in-vivo experimental setup, a rat skull sample was placed in degassed water, with the transducer placed vertically over the skull and the hydrophone measuring the ultrasound pressure beneath the skull. The peak-to-peak in-situ ultrasound pressure magnitude with beam steering was 98 kPa for the low-pressure level (Fig. 1C) and 163kPa for the high-pressure level (Fig. 1D) and spatial-peak temporal average (I_SPTA) was 32.76 mW/cm² and 54.49mW/cm² respectively with 1500 kHz PRF, 30% DC and 67ms UD. All tFUS parameters used in this study are summarized in Table 1.

Table 1.

Summary of tFUS parameters used in this study^a

tFUS conditionsb	Pressure	ISPTA (mW/cm2)	ISPPA (mW/cm2)	MI	Temperature risec (°C)
PRF 30Hz DC 0.6%	98 kPa	0.66	109.2	0.08	0.08±0.03
PRF 30Hz DC 0.6%	163 kPa	1.09	181.63	0.16	0.04±0.02
PRF 30Hz DC 60%	98 kPa	65.52	109.2	0.08	0.29±0.06
PRF 1500Hz DC 30%	98 kPa	32.76	109.2	0.08	0.13±0.10
PRF 1500Hz DC 30%	163 kPa	54.49	181.63	0.16	0.17±0.10
PRF 3000Hz DC 60%	98 kPa	65.52	109.2	0.08	0.09±0.03
PRF 3000Hz DC 60%	163 kPa	108.98	181.63	0.16	0.26±0.04
PRF 3000Hz DC 0.6%	98 kPa	0.66	109.2	0.08	−0.02±0.06

^aDetails of the acoustic metric calculations and temperature measurement procedures are provided in the Supplementary Materials. Abbreviations: I_SPTA, spatial-peak temporal-average intensity; I_SPPA, spatial-peak pulse-average intensity; MI, Mechanical index.

^bExcept for ultrasound pulse repetition frequency (PRF) and duty cycle (DC), all other ultrasound parameters were kept constant, including the fundamental frequency (f₀ = 1.5 MHz), ultrasound duration (UD = 67 ms), and inter-stimulation interval (ISI = 2.5 s with 10% jitter).

^cTemperature increases are reported as the maximum temperature change observed during 500 consecutive tFUS trials (mean ± SEM, n = 3 rats).

In our study, the inter-sonication interval (ISI) was set at 2.5s for each trial with 10% jitter to avoid entraining any response, and the UD was kept at 100 ms for conditions with tactile stimulation and 67ms for tFUS only condition (Fig. 1B). 3000Hz PRF with 60% DC and 30Hz PRF with 0.6% DC were used with either low pressure or high pressure. Each recording contained 250 trials for tFUS with tactile stimulation or 500 trials for tFUS only. The order of conditions was randomized. In addition, 200 trials of optical stimulation were interleaved between every two tFUS recording sessions to account for recording instability and to ensure accurate neuronal classification for each tFUS condition.

Data acquisition and preprocessing

All data were recorded using a commercial multi-channel neural signal acquisition system (Tucker-Davis Technologies, Alachua, FL, USA). For evoked potential analysis, recorded local field potentials (LFP) were bandpass filtered between 10Hz and 300Hz with a zero-phase infinite impulse response filter then notch filtered with 60Hz for better tFUS-induced vibration artifacts removal (Supplementary Materials and Fig. S1). For spike analysis, recorded MUAs were bandpass filtered between 300 Hz and 6 kHz. Spike sorting was performed using Offline Sorter (Plexon). Spike waveforms were detected beyond the threshold of −3.5 standard deviations from the Mean of Peak Heights Histogram, and clustering was performed using k-means. The spike waveforms and timestamps were stored for further analysis in MATLAB (R2022a, MathWorks, USA).

Signal Processing

Tactile Evoked Potentials (TEPs)

Tactile-evoked potentials (TEPs) were analyzed to evaluate neural responses to vibration-tactile stimulation with the epoch from 100 ms before the trial onset time to 500 ms after. The filtered LFPs were then normalized using a z-score transformation relative to a baseline period of 100 ms prior to the onset of tFUS stimulation for each trial. The z-scored signals were subsequently averaged across recording channels and across subjects to obtain the mean TEP waveform for each experimental condition.

Neuronal type classification and response quantification

Only neurons that fired more than 800 spikes per recording session were included for further clustering analysis. To assess neuronal responses to vibration-tactile stimulation and tFUS, neuronal types were first identified. For neurons recorded in S1, optical stimulation was applied in opto-tagged rats for cell-type identification. Opto-tagged neurons were identified by a significant increase in firing rate during optical stimulation compared with the 100 ms pre-stimulation baseline (p < 0.05) and a maximum spike latency of < 8 ms following stimulation onset [24]. Optical stimulation was delivered using red light (630 nm) for CaMKII-positive neurons or blue light (465 nm) for SST-positive or PV-positive neurons (50 ms duration; PlexBright LED Module, Plexon, Dallas, TX, USA). Optical stimulation was conducted immediately prior to the tFUS session. Sorted neurons were putatively identified as the same neuron types identified by opto-tagging when their waveforms during the tFUS session showed high similarity to those of previously identified opto-tagged neurons (correlation coefficient > 0.9) [25].

For the identification of neurons in POm, spike waveform features were analyzed. Specifically, the initial phase (IP, from spike onset to the first re-crossing of the baseline) and the afterhyperpolarization period (AHP, the interval from the end of the IP to its subsequent re-crossing of the baseline) were measured. A k-means clustering algorithm was then applied to classify neuronal types based on spike waveform characteristics, as inhibitory neurons typically exhibit shorter spike durations and narrower waveforms. Based on this classification, neurons were categorized as regular-spiking units (RSUs) or fast-spiking units (FSUs).

Neuronal firing rates were calculated, and peri-stimulus time histograms (PSTHs; bin size: 25 ms) were generated to quantify neuronal responses to tactile stimulation and tFUS.

Statistical analysis

All the results shown in figures with statistics are mean ± S.E.M. The TEPs were compared using the Kruskal-Wallis test and post hoc pairwise comparisons were performed using the Wilcoxon signed-rank test. The spiking rates were normalized to the pre-stim period (100 ms pre-stim) for comparison across different conditions using the Kruskal-Wallis H test. Post hoc two-tailed Wilcoxon rank-sum tests with Bonferroni correction were employed. Additionally, normalized spike rates within defined response windows were compared using one-way ANOVA with Bonferroni post hoc correction. The statistical significance level was set at 0.05.

Results

Tactile stimulation evoked potentials (TEPs) and neuronal responses

Vibration-tactile stimulation reliably evoked responses in both S1 and POm, as evidenced by clear tactile-evoked potentials (TEPs; Fig. 2A, N = 34 rats). In both S1 and POm, the first negative peak occurred at approximately 22 ms (N22) after stimulation onset in both the 100 Hz (v100) and 500 Hz (v500) vibration groups, and both were significantly different from baseline (p < 0.05), with no significant difference observed between the two frequencies. Given 100 Hz stimulation produced a slightly larger N22 response, subsequent experiments with tFUS focused primarily on this frequency.

Fig. 2.

Vibration-tactile stimulation evoked potentials (TEPs) and neuronal responses. (A) TEPs in S1 (top) and POm (bottom) to baseline (grey, no tactile stimulation), tactile stimulation with 100Hz (blue) and 500Hz (orange) frequencies. The stimulation duration is 100ms. The blue, orange and pink bars indicate significant difference between baseline vs v100Hz, baseline vs v500Hz and v100Hz vs v500Hz, respectively (p<0.05). The absolute N22 amplitude increased by 330% and 300% in S1, and by 415% and 390% in POm relative to baseline. (B-D) S1 (upper) and POm (bottom) neuronal responses to baseline (B), tactile stimulation with 100Hz (C) and 500Hz (D) frequencies. The * indicates a significant difference in the normalized firing rate of the bin/bins compared to the 100 ms pre-stimulation baseline firing rate (p < 0.05). Bin size is 25 ms. Red bars indicate bins during the stimulation period. The peak normalized firing rates following stimulation onset were 162% and 195% for v100Hz and v500Hz stimulation, respectively, in S1 neurons, and 110% and 105%, respectively, in POm neurons.

Analysis of multi-unit activity (MUA; Fig. 2B–D) revealed that neurons in both S1 (v100: n = 56; v500: n = 54 from 34 rats) and POm (v100: n = 124; v500: n = 107 from 34 rats) responded significantly to 100 Hz and 500 Hz vibration stimulation compared with their respective baseline groups (S1: n = 101; POm: n = 375 from 34 rats). Median onset latency was 8 ms in S1 neurons and 6 ms in POm neurons (1ms bins), with median peak latencies of 32 ms and 21 ms, respectively, indicating earlier responses in POm consistent with ascending thalamocortical transmission (Fig. S2).

Following neuronal type identification, the responses of specific neuron types were analyzed (Fig. 3). CaMKII-positive (n = 24 from 12 rats), SST-positive (n = 15 from 12 rats), and PV-positive (n = 10 from 10 rats) neurons in S1 were identified based on their responses to optical stimulation (Fig. 3A–C, Fig. S3). Histological verification further confirmed viral expression (Supplementary Materials and Fig. S4). Tactile-evoked responses showed distinct patterns across these cell types (Fig. 3D–F), with only CaMKII-positive neurons exhibiting significant responses to vibration-tactile stimulation, while SST-positive and PV-positive neurons showed no detectable responses. In POm, RSUs and FSUs were classified based on their IP and AHP waveform features (Fig. 3G). Only RSUs (n = 63 from 34 rats) displayed robust tactile-evoked responses, while FSUs (n = 61 from 34 rats) remained unresponsive.

Fig. 3.

Identification of neuronal subtypes and their responses to tactile stimulation. (A-C) S1 neuronal subtype identification using optical stimulation in opto-tagged rats. The PV-positive (A), CaMKII-positive (B), and SST-positive (C) neurons exhibited more than twofold increase in firing rate relative to the pre-stimulation period. Insets display the representative spike waveforms of the recorded neurons. Solid gray lines indicate the initial phase (IP), and dotted lines indicate the afterhyperpolarization (AHP) duration. The average IP and AHP were 0.39 ± 0.02 ms and 0.97 ± 0.03 ms for PV-positive neurons, 0.68±0.03 ms and 1.23±0.03 ms for CaMKII-positive neurons, and 0.41±0.02 ms and 1.20±0.04 ms for SST-positive neurons, respectively. (D-F) Responses of identified S1 PV-positive (D), CaMKII-positive (E), and SST-positive (F) neurons to 100Hz vibration-tactile stimulation. S1 CaMKII-positive neurons showed peak firing rate change of 196%. (G) POm neuronal subtype identification based on waveform characteristics. RSU and FSU clusters based on IP and AHP calculation of waveforms using k-means. The average IP and AHP for classified neurons were 0.65±0.01 ms and 0.95±0.01 ms for RSUs, and 0.21±0.01 ms and 0.22 ±0.01 ms for FSUs, respectively. (H-I) Responses of classified POm RSUs (H) and FSUs (I) to 100Hz vibration-tactile stimulation. POm RSU neurons exhibited a peak normalized firing rate of 135%. The * indicates a significant difference in the normalized firing rate of the bin/bins compared to the 100 ms pre-stimulation baseline firing rate (p < 0.05). Bin size is 25 ms. Red bars indicate bins during the stimulation period.

TEPs and neuronal responses to tFUS with High PRF and DC

The TEPs and neuron type-specific responses to tFUS with a 3 kHz PRF and 60% duty cycle (DC) were analyzed (Fig. 4) to examine the modulatory effects of tFUS. Both high- and low-pressure levels were tested in combination with tactile stimulation to evaluate how tFUS intensity influences cortical and thalamic responses. Based on the TEPs, both S1 and POm exhibited a significant increase in the responsive peak (N22) in the high-pressure tFUS plus tactile stimulation group (v100+H3kHz) compared with tactile stimulation plus low-pressure tFUS group (v100+L3kHz) as well as tactile stimulation alone (Fig. 4A), indicating a pressure-dependent modulatory effect of tFUS.

Fig. 4.

TEPs and neuronal responses to 100 Hz vibration-tactile stimulation combined with tFUS (3 kHz PRF, 60% duty cycle). (A) TEPs recorded in S1 and POm during 100 Hz vibration-tactile stimulation alone (v100Hz, gray), 100 Hz stimulation combined with low-pressure tFUS (v100Hz+L3kHz, blue), and 100 Hz stimulation combined with high-pressure tFUS (v100Hz+H3kHz, orange). Stimulation duration was 100 ms. Blue, orange, and pink bars indicate significant differences between v100Hz vs. v100Hz+L3kHz, v100Hz vs. v100Hz+H3kHz, and v100Hz+L3kHz vs. v100Hz+H3kHz, respectively (p < 0.05). The absolute N22 amplitude increased by 206% during v100+H3kHz stimulation in S1, and by 129% in POm, relative to tactile stimulation alone. (B-D) POm RSU (upper) and FSU (bottom) neuronal responses to v100Hz only (B), v100Hz+L3kHz(C) and v100Hz+H3kHz (D). The * indicates a significant difference in normalized firing rate of the bin(s) compared to the 100 ms pre-stimulation baseline (p < 0.05). Bin size was 25 ms. Red bars denote bins during the stimulation period. The peak response levels within the first 25 ms following stimulation onset were 135%, 130%, and 145% for v100Hz alone, v100Hz+L3kHz, and v100+H3kHz respectively. (E) Comparison of normalized RSU firing rate in the first 25 ms window among v100Hz, v100Hz+L3kHz, and v100Hz+H3kHz conditions. * indicates significant difference between v100 and v100+H3kHz groups (p<0.05). POm RSU instant response increases by 8% during v100Hz+H3kHz stimulation relative to v100Hz alone.

As ultrasound pressures above 130 kPa can induce displacement of intracranial electrodes, which can be detected in the recorded signals [1,26], we focused our analysis on POm neuronal responses to tactile stimulation combined with tFUS. Consistent with previous observations, only POm RSUs showed significant responses to tactile stimulation with or without tFUS, whereas FSUs exhibited no detectable responses (Fig. 4B–D). To further quantify the effect, normalized firing rates within the 0–25 ms window following tactile stimulation and tFUS onset were compared across groups. High-pressure tFUS elicited significantly higher responses than the other groups (Fig. 4E), indicating that tFUS with high PRF, high DC, and high pressure produces an excitatory modulatory effect on the sensory pathway.

TEPs and neuronal responses to tFUS with Low PRF and DC

The TEPs and neuron type-specific responses to tFUS with a 30 Hz PRF and 0.6% duty cycle (DC) were analyzed (Fig. 5) to examine modulatory effects at both high- and low-pressure levels. Based on the TEPs, both S1 and POm exhibited a significant decrease in the responsive peak (N22) in the low-pressure tFUS plus tactile stimulation group (v100+L30Hz) compared with tactile stimulation alone and tactile stimulation combined with high-pressure tFUS (v100+H30Hz) (Fig. 5A).

Fig. 5.

TEPs and neuronal responses to 100 Hz vibration-tactile stimulation combined with tFUS (30 Hz PRF, 0.6% duty cycle). (A) TEPs recorded in S1 and POm during 100 Hz vibration-tactile stimulation alone (v100Hz, gray), 100 Hz stimulation combined with low-pressure tFUS (v100Hz+L30Hz, blue), and 100 Hz stimulation combined with high-pressure tFUS (v100Hz+H30Hz, orange). Stimulation duration was 100 ms. Blue, orange, and pink bars indicate significant differences between v100Hz vs. v100Hz+L30Hz, v100Hz vs. v100Hz+H30Hz, and v100Hz+L30Hz vs. v100Hz+H30Hz, respectively (p < 0.05). The absolute N22 amplitude decreased by 80% during v100+L30Hz stimulation in S1, and by 73% in POm, relative to tactile stimulation alone. (B-D) POm RSU (upper) and FSU (bottom) neuronal responses to v100Hz only (B), v100Hz+L30Hz(C) and v100Hz+H30Hz (D). The * indicates a significant difference in normalized firing rate of the bin(s) compared to the 100 ms pre-stimulation baseline (p < 0.05). Bin size was 25 ms. Red bars denote bins during the stimulation period. The peak response levels within the first 25 ms following stimulation onset were 135%, 130%, and 135% for v100Hz alone, v100Hz+L30Hz, and v100+H30Hz respectively. (E) Comparison of normalized RSU firing rate in the first 25 ms window among v100Hz, v100Hz+L30Hz, and v100Hz+H30Hz conditions. No significant difference was observed among groups (p>0.05).

Consistently, only POm RSUs responded to tactile stimulation with or without tFUS (Fig. 5B–D), and there was no significant difference in normalized firing rates within the 0–25 ms window following tactile stimulation and tFUS onset across groups (Fig. 5E).

Overall, these results indicated that tFUS with low PRF, low duty cycle, and low pressure exerted an inhibitory modulatory effect on the sensory pathway.

Cell-type-specific mechanisms underlying bidirectional modulation of the sensory pathway by tFUS

We found that tFUS modulates the sensory pathway in a parameter-dependent manner, producing excitatory effects with high PRF, high DC, and high pressure, and inhibitory effects with low PRF, low DC, and low pressure, raising the question of which neurons are responsible for these differential effects.

The excitatory effects could not be directly assessed in S1 neurons due to stimulation-induced vibration [1,26]. Therefore, we analyzed and compared POm neuronal responses while selectively activating CaMKII-positive, PV-positive, or SST-positive neurons in S1 to identify the neuronal populations most closely associated with the effects of tFUS (Fig. 6). POm RSUs responded to tactile stimulation both with and without optical activation of CaMKII-positive (n = 24 from 12 rats), PV-positive (n = 20 from 10 rats), or SST-positive (n = 17 from 12 rats) neurons in S1, whereas POm FSUs showed no responses (CaMKII-positive, n = 23 from 12 rats; PV-positive, n = 20 from 10 rats; SST-positive, n = 25 from 12 rats) (Fig. 6A–D). Additionally, optical activation of S1 CaMKII-positive neurons induced a significant increase in firing of POm RSUs within the 0–25 ms window following tactile stimulation onset (Fig. 6E). These results provide indirect evidence that the excitatory modulation of the sensory pathway by tFUS is mediated through activation of CaMKII-positive neurons in S1.

Fig. 6.

POm RSU and FSU neuronal responses to 100 Hz vibration-tactile stimulation combined with selective activation of S1 neurons via optical stimulation in opto-tagged rats. (A-D) POm RSU (upper) and FSU (bottom) neuronal responses to v100Hz only (A), v100Hz with S1 CaMKII-positive neuronal activation (B), v100Hz with S1 PV-positive neuronal activation (C) and v100Hz with S1 SST-positive neuronal activation (D). The * indicates a significant difference in normalized firing rate of the bin(s) compared to the 100 ms pre-stimulation baseline (p < 0.05). Bin size was 25 ms. Red bars denote bins during the stimulation period. The peak response levels within the first 25 ms following stimulation onset were 135%, 148%, 130%, and 117% for v100Hz alone, v100Hz with S1 CaMKII-positive activation, v100Hz with S1 PV-positive activation, and v100Hz with S1 SST-positive activation, respectively. (E) Comparison of normalized RSU firing rate in the first 25 ms window among v100Hz, v100Hz+S1 CaMKII-positive activation, v100Hz+ S1 SST-positive activation and v100Hz+ S1 PV-positive activation conditions. * indicates significant difference between v100 and v100Hz+S1 CaMKII-positive activation groups (p<0.05). Relative to v100Hz alone, S1 CaMKII-positive neuronal activation increased the initial POm RSU response by approximately 10%.

POm neuronal responses did not reveal the mechanism underlying the inhibitory effect. Therefore, we directly analyzed S1 neuron-type-specific responses to low-pressure tFUS to investigate this effect (Fig. 7). S1 CaMKII-positive neurons (n = 20 from 12 rats) exhibited a decrease in firing rate approximately 25 ms after tFUS onset, whereas SST-positive (n = 15 from 12 rats) and PV-positive neurons (n = 14 from 10 rats) showed no responses. To further determine which tFUS parameters were critical for this inhibitory effect, CaMKII-positive responses were examined under high PRF with high DC (n = 25 from 12 rats), high PRF with low DC (n = 22 from 12 rats) and low PRF with high DC (n = 19 from 12 rats) at low pressure (Fig. S5). No decrease in firing rate was observed under these conditions, indicating that the inhibitory effect is specifically induced by the combination of low PRF, low DC, and low pressure.

Fig. 7.

S1 neuronal responses to low-pressure tFUS (30 Hz PRF, 0.6% duty cycle). (A–C) Responses of S1 CaMKII-positive (A), SST-positive (B), and PV-positive (C) neurons to tFUS. Yellow *(pointed by yellow arrow) indicates a significant decrease in normalized firing rate compared to the 100 ms pre-stimulation baseline, whereas black * indicates significant increases relative to baseline (p < 0.05). Bin size was 25 ms. Red bars denote bin windows during the stimulation period. L30 Hz tFUS induced an approximately 22% decrease in firing rate of S1 CaMKII-positive neurons during the first 25 ms of stimulation.

Discussion

In this study, we recorded neuronal activity in both S1 and POm during vibration-tactile stimulation, with and without tFUS. We observed that only S1 CaMKII-positive neurons and POm RSUs responded robustly to tactile stimulation. We then examined neuronal responses to high- and low-pressure tFUS applied to S1 in combination with vibration-tactile stimulation, using both high PRF, high DC, and low PRF, low DC protocols. High PRF, high DC, and high-pressure tFUS produced excitatory modulation of the sensory pathway, whereas low PRF, low DC, and low-pressure tFUS induced inhibitory effects. Mechanistically, these bidirectional effects were consistent with the involvement of S1 CaMKII-positive neurons, with excitatory modulation resulting from their activation and inhibitory modulation from their deactivation (Fig. 8). Overall, these results demonstrate that tFUS can bidirectionally modulate the cortico–thalamo–cortical sensory processing pathway by selectively engaging excitatory neurons, providing mechanistic insight into parameter-dependent neuromodulation and guiding the application of tFUS as a therapeutic tool for sensory disorders.

Fig. 8.

Summary of this work. Gray indicates resting-state activity. Peripheral input (vibration–tactile stimulation) excites CaMKII-positive neurons (Py) in S1 and excitatory neurons (Ex) in POm. Excitatory tFUS enhances sensory-related CTC pathway by activating CaMKII-positive neurons in S1 (indicated by darker colors and thicker arrows). Inhibitory tFUS suppresses this pathway by deactivating CaMKII-positive neurons in S1 (indicated by lighter colors and thinner arrows). Excitatory tFUS (etFUS) was delivered at a pulse repetition frequency (PRF) of 3 kHz with a 60% duty cycle at high pressure, whereas inhibitory tFUS (itFUS) was delivered at a PRF of 30 Hz with a 0.6% duty cycle at low pressure.

Bidirectional tFUS parameter-dependent modulation

In this study, we demonstrated that tFUS exerts bidirectional, parameter-dependent modulatory effects on sensory processing. High PRF, high DC, and high-pressure stimulation produced excitatory effects, whereas low PRF, low DC, and low-pressure stimulation induced inhibitory effects (Fig. 4–5). Although auditory confounds related to PRF have been reported [27–29], our previous work using 30Hz and 3kHz PRF suggested low PRF was more likely to induce such effects post-stimulation [1], making it unlikely to be the main contributor to the effects observed in this study. The changes in sensory evoked potentials support tFUS-induced neural modulation [16,30–32]. Legon et al. reported that tFUS applied to the S1 could attenuate amplitudes of somatosensory evoked potentials and enhance behavioral performance in sensory discrimination tasks [16]. Studies reported that the bidirectional neuromodulatory effects of tFUS depend critically on stimulation parameters, including PRF, DC, and acoustic pressure levels. For example, Yoon et al. demonstrated that high DC levels (30%, 50%, and 70%) tFUS targeting the motor cortex (M1) and thalamus induced excitatory effects, whereas low DC levels (3–5%) applied to S1 and the thalamus produced inhibitory responses [13]. In another study using different PRFs to stimulate the motor cortex of rodents, high PRF (>500 Hz) resulted in a marked increase in motor evoked potentials (MEP) [33]. In humans, Zadeh et al. reported that low PRF stimulation at 10 Hz produced inhibitory effects on MEP amplitudes[34]. In our study, we observed that modifying either PRF or DC alone could differentially affect responses of CaMKII-positive neurons (Fig 7, Fig. S5), suggesting that combined parameter regimes may play an important role in shaping tFUS effects. Additionally, Kim et al. investigated the effects of tFUS on the mouse S1 using two different Ispta and observed increased hemodynamic response amplitudes with higher acoustic pressure [35]. However, increasing pressure levels may also raise safety concerns, such as potential neuronal damage or localized tissue heating [36]. Therefore, careful selection and optimization of tFUS parameters are essential to achieve the desired neuromodulatory effects while maintaining safety.

Cortico-thalamo-cortical sensory processing pathway

Cortico-thalamo-cortical (CTC) pathways are highly involved in sensory processing, including visual, somatosensory, and auditory information [37]. These pathways involve precise interactions between specific cortical regions and corresponding thalamic nuclei, allowing for both feedforward transmission of ascending sensory signals and feedback modulation from cortex to thalamus [38]. In the somatosensory system, first-order thalamic nuclei, such as the ventral posterior medial nucleus (VPM), primarily relay peripheral sensory inputs to S1, whereas higher-order nuclei, including the posterior medial nucleus (POm), integrate inputs from multiple cortical and subcortical sources to mediate context-dependent processing, attention, and sensorimotor coordination [39–41]. Within these pathways, distinct neuron types contribute differentially: excitatory pyramidal neurons, such as CaMKII-positive neurons in S1, provide the principal output driving feedforward thalamocortical activation while in the thalamus, RSUs transmit excitatory signals back to cortex [41]. Consistent with this functional organization, our results showed that during vibration-tactile stimulation of the sensory pathway, only S1 CaMKII-positive neurons and POm RSUs responded robustly, whereas inhibitory interneurons (PV-positive and SST-positive) in S1 and FSUs in POm exhibited no responses (Figs. 3–6, Fig. S3). These findings highlight the critical role of excitatory neurons in mediating sensory information along the CTC pathway.

Cell-type selectivity of tFUS

tFUS has been shown to exhibit cell-type selectivity in a parameter-dependent manner. Our results demonstrate that S1 CaMKII-positive neurons display distinct responses to tFUS under different stimulation parameters (Fig. 7, Fig. S5), consistent with previous studies showing that high PRF selectively excites excitatory neurons [17]. This selective responsiveness may arise from differences in membrane properties and ion channel composition across neuronal subtypes. Several studies have suggested that ultrasound neuromodulation interacts with mechanosensitive ion channels and voltage-gated channels embedded in neuronal membranes. For instance, low PRF tFUS has been reported to primarily affect potassium (K⁺) channels [42–45] whereas high PRF stimulation can additionally modulate sodium (Na⁺) and calcium (Ca²⁺) channels [46]. Because different neuronal populations express distinct combinations of these ion channels, they may exhibit differential sensitivity to specific tFUS parameters. Pyramidal neurons, including CaMKII-positive excitatory neurons in S1, express multiple tFUS-sensitive ion channels such as K⁺ [43] and Ca²⁺ [47] channels, which may account for their distinct responses to tFUS with different parameters.

Future Applications

Bidirectional modulation of the sensory pathway could have therapeutic potential for various sensory-related disorders. For conditions characterized by overactive sensory processing, low-intensity tFUS could be used to suppress activity, such as sensory hypersensitivity in autism spectrum disorder; whereas high-intensity tFUS could enhance activity in cases of sensory deficits, such as stroke-induced sensory loss.

Limitations and future work

In this study, only neuronal activation was assessed using optogenetics, which limits the precision of our conclusions regarding inhibitory effects. Future studies incorporating pharmacological approaches, optogenetic identification of additional neuronal populations and brain regions, or optogenetic silencing of specific neuronal populations could provide stronger evidence for the mechanisms underlying tFUS-induced inhibition.

Due to the sensitivity of silicon electrodes to tFUS-induced vibration artifacts, S1 spiking activity could not be directly recorded during high-pressure tFUS stimulation. This technical limitation restricted our ability to directly characterize neuronal responses underlying the bidirectional effects of tFUS. Future advances in recording technologies, such as flexible electrodes, may help overcome these challenges.

In addition, only male subjects were used in this study. Although previous study has reported no sex differences in tFUS-induced pain modulation [11], sex-dependent differences in neuronal number and neuronal processes may still influence neuronal responses to tFUS and therefore need further investigation [48].

Furthermore, although a previous study has suggested that anesthesia does not substantially alter neuronal responses to tFUS [49], its effects on sensory processing and brain network dynamics during tFUS require further investigation. In particular, isoflurane has been shown to disrupt awareness by suppressing cortico-thalamo-cortical connectivity [50], which may influence the magnitude of the responses observed under anesthetized conditions.

Finally, several studies have reported plasticity-related effects following prolonged tFUS stimulation [51–54]. Future studies systematically investigating tFUS parameters that may induce long-term plasticity are warranted to determine whether tFUS can produce more robust and long-lasting plastic changes within the cortico-thalamo-cortical pathway, potentially enhancing its therapeutic and clinical applications.

Conclusions

By combining optogenetic tagging, multi-site electrophysiology, and waveform-based neuronal classification, we found that S1 CaMKII-positive neurons and POm RSUs responded robustly to vibration-tactile stimulation. We further demonstrated that tFUS applied to S1 can bidirectionally modulate the cortico–thalamo–cortical sensory pathway in a parameter-dependent manner: high PRF, high DC, and high-pressure stimulation elicited excitatory effects, whereas low PRF, low DC, and low-pressure stimulation produced inhibitory effects. Mechanistically, these bidirectional effects were consistent with modulation of S1 CaMKII-positive neurons, with excitatory modulation driven by their activation and inhibitory modulation arising from their deactivation. This study provides mechanistic insight into the cell-type-specific effects of tFUS on sensory processing and highlights the critical role of CaMKII-positive neurons in mediating these effects. Importantly, these findings have translational implications, suggesting that tFUS parameters can be optimized to either enhance or suppress sensory processing, providing a foundation for the safe and effective development of tFUS-based therapies for sensory disorders.

Data availability

Data supporting the scientific findings are included in the paper and supplementary materials. Additional experimental data are available at: https://dandiarchive.org/dandiset/001877.