TRPC4/TRPC5 are critical for neuronal modulation by transcranial focused ultrasound in retrosplenial cortex in male mice

Abstract

Transcranial focused ultrasound (tFUS) enables non-invasive neuromodulation, yet its underlying molecular mechanisms remain largely elusive. Here, we show that transient receptor potential canonical 4 (TRPC4) and transient receptor potential canonical 5 (TRPC5) channels are critical mediators of tFUS-induced neuronal modulation in the mouse brain. Applying tFUS to the retrosplenial cortex (RSC) in male mice desensitizes mechanical and thermal sensitivity while robustly elicits early growth response 1 (Egr1) expression. Inhibiting these tFUS-induced Egr1 ensembles blocks the somatic sensory effects. Transcriptomic analysis identifies Trpc4 enrichment in tFUS-activated Egr1-positive cells. Both pharmacological inhibition and genetic knockdown of TRPC4 abolish tFUS-mediated sensory modulation. Targeted knockdown further demonstrates that the highly homologous TRPC5 plays a comparable role. In situ proximity ligation assay, co-immunoprecipitation, and live-cell calcium imaging confirm that TRPC4 and TRPC5 form a protein complex in the RSC that facilitates the tFUS response. These findings establish TRPC4/TRPC5 as essential molecular components for tFUS neuromodulation.

The authors find that transcranial focused ultrasound non-invasively modulates Egr1+ neuronal activity via TRPC4 and TRPC5 channels in the retrosplenial cortex in male mice, regulating mechanical and thermal sensitivity.

Introduction

Pain is vital for health¹; however, impairment or pathology in the somatic sensory system results in neuropathic pain, affecting approximately 8% of the global population². Effective pain management requires innovative strategies, and transcranial focused ultrasound (tFUS) is emerging as a promising tool. Ultrasound administered to the dorsal root ganglion in nerve-injured rats yields pain desensitization³. Furthermore, tFUS activation of the primary somatosensory cortex in chronic pain mouse models regulates both thermal and mechanical hyperalgesia⁴. In human research, tFUS applied to the right anterior thalamus of healthy individuals remarkably diminished sensitivity to thermal pain⁵, with analogous enhancements noted in patients with chronic therapy-resistant neuropathic pain⁵. Moreover, direct ultrasonic stimulation of the anterior cingulate cortex (ACC) resulted in fast, clinically significant, and lasting reductions in pain severity⁶. These studies highlight the potential of tFUS as a non-invasive and effective method for managing chronic pain by targeting specific pain-related sites. While promising, further research is required to establish its role in pain management fully.

Understanding the molecular mechanisms underlying tFUS stimulation is crucial for investigating its application in targeted brain region stimulation and potential behavioral modification. This research could pave the way for developing non-invasive, targeted treatments for chronic pain and various neurological disorders. tFUS holds significant promise for regulating neuronal activity due to its non-invasive advantages^7,8, hypothesized to act via specific ion channels^9,10. For example, it has been suggested that tFUS regulates neuronal activity by targeting mechanosensitive ion channels⁹, an idea that has gained widespread acceptance. Channels such as Piezo1, TREK-1, TREK-2, TRAAK, and NaV1.5¹¹ are responsive to tFUS, suggesting a direct modulatory mechanism that affects intracellular ion concentrations and potentially gene transcription^12,13. Nonetheless, the ion channels underlying tFUS modulation of pain sensation remain unidentified.

Retrosplenial cortex (RSC), a region known for processing somatosensory information and regulating chronic pain through neuroplastic changes^14–16. Recent studies show that RSC neurons respond to noxious visceral and cutaneous inputs¹⁷; the RSC-ACC glutamatergic pathway facilitates supraspinal nociceptive processing¹⁸, and manipulating RSC activity alters mechanical and thermal pain sensitivity¹⁹. These findings confirm the RSC actively modulates pain signals (not merely correlating with pain), consistent with evidence linking it to neuropathic pain maintenance beyond its traditional spatial and mnemonic roles^14,20. While S1 and ACC are canonical pain hubs, we selected RSC for its role in integrating sensory nociception with cognitive-emotional processing—critical in chronic pain^14,20. Peripheral nerve injury induces RSC transcriptomic and functional plasticity¹⁴, and chronic pain causes long-term structural/functional alterations²⁰, expanding neuromodulation beyond traditional circuits, especially for pain with prominent cognitive-emotional components.

Here, we show that the RSC serves as a focal point for tFUS to modulate the pain threshold. Using a combination of behavioral and electrophysiological assays, we demonstrate that tFUS stimulation of the RSC significantly elevates the pain threshold in mice. This phenomenon is contingent upon a reduction in neuronal activity, primarily facilitated by transient receptor potential canonical (TRPC) channel members 4 and 5 (TRPC4/TRPC5), thereby reinforcing the role of tFUS in modulating neuronal activity via these channels. These findings yield critical insights into the mechanisms underlying tFUS and its potential as a non-invasive intervention for chronic pain.

Results

Transcranial focused ultrasound (tFUS) targeting the retrosplenial cortex (RSC) alleviates pain in mice

To examine the effects of tFUS on pain sensation, we exploited a compact head-mounted ultrasound stimulator as a core device to induce ultrasound neuromodulation^21,22 in the RSC of freely moving mice, of which assessed pain-related behaviors (Fig. 1a–d and Supplementary Fig. 1a). We first used stimulus-evoked methods, including von Frey filaments assay and Hargreaves tests, to detect changes in the pain threshold during tFUS delivered to the RSC at the different inter-stimulus interval (ISI: 1.5, 3, and 10 s) (Fig. 1b, c). Notably, the paw withdrawal threshold (PWT) exhibited a remarkable increase when ISI was 1.5 s, which was further augmented when ISI was 3 s (Fig. 1e). Consistently, the significant thermal withdrawal latency (TWL) increase was comparable between 1.5- and 3-s ISI measured using the Hargreaves test (Fig. 1f). Thus, we selected 3 s of stimulus repetitions for subsequent tests. Comparing PWT and TWL at baseline, during tFUS, and at 1, 2, 6, 12, and 24 h post-stimulation. We found that while the elevated PWT (Fig. 1g) lasted 1 h after tFUS stimulation, the extended TWL continued for at least 6 h after tFUS stimulation (Fig. 1h). These data show that tFUS can suppress both mechanical and thermal pain, with the effects sustainable for at least one hour for mechanical pain, but much longer for thermal pain.

Fig. 1. Pain threshold regulation can be achieved by transcranial focused ultrasound (tFUS) targeting the retrosplenial cortex (RSC).

a In vivo ultrasonic stimulation in free-moving mice. b Head-mounted ultrasound stimulator on the RSC. c Pulse ultrasonic parameter pattern: P₀, the peak acoustic pressure; PD, pulse duration; PRI, pulse repetition interval; PTD, pulse train duration; ISI, inter-stimulus interval; T_off, stop time of ultrasonic stimulation. d Standardized acoustic pressure distribution diagram of ultrasound stimulator. X-axis: horizontal side; Z-axis: stimulation depth. e Paw withdrawal threshold (PWT) detected at different parameters of ultrasonic stimulation acting on RSC. Two-sided multiple unpaired t-tests; baseline (BL): P = 0.8826; ISI = 1.5 s: P = 0.0278; ISI = 3 s: P = 0.0008; ISI = 10 s: P = 0.0635; control (Ctrl): n = 6 mice; tFUS: n = 5 mice. f Thermal withdrawal latency (TWL) measured under different parameters of ultrasonic stimulation applied to RSC. Two-sided multiple unpaired t-tests; BL: P = 0.8899; ISI = 1.5 s: P = 0.0418; ISI = 3 s: P = 0.0241; ISI = 10 s: P = 0.8643; n = 6 mice per group. g PWT detected at different time points before and after ultrasound delivered to RSC in both Ctrl and tFUS groups. Two-way RM ANOVA, interaction, F_{(6, 78)} = 5.08, P = 0.0002; n (Ctrl) = 7 mice, n (tFUS) = 8 mice. h TWL observed in both groups. Two-way RM ANOVA, interaction, F_{(6, 72)} = 1.28, P = 0.2753; n = 7 mice per group. i Common peroneal nerve (CPN) ligation in mouse. Created in BioRender. Wu, C. (2026) https://BioRender.com/2m5cgnz. j PWT detected in sham and CPN ligation mice pre- and post-tFUS targeting RSC. Two-way RM ANOVA, sham vs. CPN: interaction, F_{(2, 16)} = 10.32, P = 0.0013; n = 5 mice per group. k TWL was detected in sham and CPN groups pre- and post-tFUS targeting RSC. Two-way RM ANOVA, sham vs. CPN: interaction, F_{(2, 16)} = 3.937, P = 0.0407; n = 5 mice per group. Source data are provided as a Source Data file. All data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Given that neuropathic pain affects approximately 8% of people, it is of interest to test whether tFUS can alleviate neuropathic pain. To this end, we performed tFUS on the RSC of mice with common peroneal nerve ligation (CPN ligation), a mouse model of neuropathic pain (Fig. 1i). On day 7 after CPN ligation, the mice showed markedly reduced nociceptive threshold compared with sham-operated control animals. tFUS to the RSC significantly increased the PWT and TWL of the injured mice (Fig. 1j, k). These observations indicate that tFUS to the RSC is effective in alleviating neuropathic pain.

To eliminate the possibility that the activation is due to thermal effects of tFUS, we measured the tFUS stimulator’s temperature using a thermal probe. The temperature increase on the stimulator region was less than 1 °C during tFUS-stimulated RSC (Supplementary Fig. 1b–e), which is much lower than the previously reported threshold for thermal-induced neural activation (ΔT > 5 °C)^23,24. To ensure the safety of tFUS, we performed Nissl staining and apoptotic signal detection. We found no difference in Nissl bodies or significant apoptotic signals between the control and tFUS groups (Supplementary Fig. 1f–h). Therefore, our ultrasound system and parameters do not cause obvious tissue heating or damage to the stimulated region.

The RSC’s involvement in regulating anxiety and memory storage²⁵ led us to test for potential side effects of tFUS application. We conducted elevated plus-maze (EPM) and open-field (OFT) tests but observed no significant differences (Supplementary Fig. 2a–h), suggesting that tFUS does not induce anxiety-like changes. Moreover, the novel object recognition test (NOR), reflecting learning and memory, remained unchanged by tFUS stimulation of the RSC (Supplementary Fig. 2i–k). To assess the impact of ultrasound on the motor abilities of mice, we used rotarod and gait analysis and found that neither was influenced by tFUS (Supplementary Fig. 3). We conclude that ultrasound applied to the RSC does not cause anxiety or affect motor ability.

To evaluate the functional specificity of tFUS, a plastic circular pedestal was utilized to precisely center the ultrasound stimulator over the anterior cingulate cortex (ACC) or primary motor cortex (M1). Crucially, the temporal modulation parameters of ultrasound stimulation protocols (PD, duration; PRI, pulse repetition interval; PTD, pulse train duration; ISI) were kept consistent across RSC, ACC, and M1 stimulation to ensure comparable data for analysis across all three regions (Supplementary Fig. 4a, b). We applied tFUS to ACC, a critical component of the pain matrix²⁶. Interestingly, tFUS stimulation to ACC also increased TWL in mice (Supplementary Fig. 4a–e). In contrast, the application of tFUS to the M1 did not induce changes in PWTs (Supplementary Fig. 4f); instead, it significantly decreased the traveling distance in the OFT (Supplementary Fig. 4g, h). These results indicate that the effects of tFUS are related to the functions of the targeted brain regions.

To rule out the possibility that tFUS affects pain via the auditory system, we used a congenital deafness mouse model featuring a point mutation in the cadherin 23 (Cdh23)²⁷. Wild-type mice (WT) showed a significant increase in mobility time when exposed to a sound in the hearing range (70 dB). In contrast, the Cdh23 mutant mice exhibited no such response, confirming their lack of sound response (Supplementary Fig. 4i). Despite this, the Cdh23 mutant mice still displayed markedly increased PWT and prolonged TWL in response to tFUS exposure (Supplementary Fig. 4j, k). Accordingly, it is unlikely that the increase in PWT and TWL during tFUS stimulation is related to the auditory pathway.

tFUS induced significant changes in neuronal activities in the RSC

To further validate the precise alterations of tFUS, we performed high-temporal-resolution in vivo electrophysiological recordings of neuronal firing in the RSC of freely behaving mice, before, during, and after ultrasound stimulation. Single-neuron spikes were isolated via principal component analysis (PCA) (Fig. 2a, b)²⁸ and classified as pyramidal neurons or interneurons based on spike half-wave width (Supplementary Fig. 5a) and discharge frequency (Fig. 2c). Mice received three sequential tFUS stimulation cycles (S1, S2, S3) followed by post-tFUS intervals (AS1, AS2, AS3) (Fig. 2d). During the 0.3 s tFUS application, the spike rate of pyramidal neurons was significantly reduced (Fig. 2e, f). Post-tFUS, discharge frequency was variable-some increased, some decreased, and some were unchanged-suggesting a heterogeneous, network-coding-like alteration (Fig. 2g, h). Critically, the proportion of neurons with a decreased spike rate was significatly higher than those with an increased spike rate (Fig. 2i). Interneurons showed a consistent fluctuation trend in spike rate (Fig. 2j–m), with a significantly larger proportion exhibiting a decrease than an increase (Fig. 2n). To ensure the reliability of our findings, we conducted a control experiment with baseline recordings (without stimulation) and recordings during ultrasound stimulation of ACC. A multi-channel electrode was unilaterally implanted in the RSC to record neural activity. Concurrently, a head-mounted ultrasound stimulator was positioned and fixed over the ACC to deliver tFUS stimulation. The results showed that the firing rate in the RSC remained unchanged between ultrasound-on and ultrasound-off states (Supplementary Fig. 5b–d). This suggests that the observed reduction in firing rate is not due to ultrasound-related noise, but rather is specific to the stimulation of the target region. Therefore, these findings suggest that tFUS can significantly modulate neuronal activity in the RSC.

Fig. 2. tFUS can temporarily and continuously inhibit RSC neuronal activity.

a Multi-channel electrodes were unilaterally implanted in the RSC with a head-mounted ultrasound stimulator, with correct placement confirmed by Nissl staining (red circle). Scale bar, 1 mm. b Example of two-dimensional principal component analysis showing well-isolated single units. c Putative pyramidal neurons (PNs) and interneurons (INs) were separated by unsupervised cluster analysis (inset: representative waveforms; scale bar: 400 ms). d Experimental timeline. S1: stimulation 1; AS1: after stimulation 1. e Rasters of example 10 PNs and mean rate, 95% confidence intervals. f Spike firing rate for PNs during inter-tFUS interval (ITI) and sustained pulse (US); two-sided paired t-test (n = 10 neurons), P (S1) = 0.000047, P (S2) = 0.02, P (S3) = 0.02. g Rasters of example, PNs during baseline, AS1, AS2, and AS3 periods represent a subset of eight representative neurons. h The PNs’ spike firing rate after stimulation in RSC. i Percentage of decreasing, increasing, and no changing firing rate of PNs (n = 33 neurons, 39% decreasing, 21% increasing, two-sided Fisher’s exact test, P = 0.04; Neurons were classified based on a 20% threshold relative to baseline firing rate: as increasing if the firing rate increased by >20% (P < 0.05, paired t-test); as decreasing if it decreased by >20% (P < 0.05); and as no change if the change was within ±20% or not significant (P ≥ 0.05)). j Rasters of example 8 INs and average spike firing rate, 95% confidence intervals. k Spike firing rate for INs during ITI and tFUS; two-sided paired t-test (n = 8 neurons), P (S1) = 0.02, P (S2) = 0.02, P (S3) = 0.10. l Rasters of example 8 INs. m The INs’ spike firing rate after stimulation in RSC. n Percentage of decreasing, increasing, and no change in the firing rate of INs. (n = 18 neurons, 50% decreasing, 22% increasing, two-sided Fisher’s exact test, P = 0.03). Source data are provided as a Source Data file. All data are presented as mean ± SEM. *P < 0.05, ****P < 0.0001; ns not significant.

Early growth response 1 (Egr1) is a marker of tFUS-sensitive cells in the RSC

To further confirm the direct effects of tFUS on specific brain regions, we examined the expression of immediate-early genes, which reflect instantaneous changes in cellular activity triggered by external stimulation²⁹. We observed significantly increased Egr1 expression in the RSC at 0.5 and 2 h after tFUS stimulation via real-time quantitative fluorescence PCR (Fig. 3a). Under the same conditions, the expression of c-Fos was elevated at 0.5 h but not 2 h after stimulation (Fig. 3b). Applying tFUS to the ACC and M1 also increased the mRNA of Egr1, but it only elevated c-Fos in the ACC but not M1 (Supplementary Fig. 4c, d). These findings collectively demonstrate that the tFUS-induced changes in Egr1 are more sensitive and longer-lasting than those in c-Fos.

Fig. 3. Early growth response 1 (Egr1) as a marker of tFUS-sensitive cells in RSC.

a The expression of Egr1 mRNA at 0.5 h and 2 h after tFUS acting on RSC. One-way ANOVA, n (Ctrl) = 9 mice, n (0.5 h) = 10 mice, n (2 h) = 6 mice, F_{(2, 22)} = 5, P = 0.0092. b The expression of c-Fos mRNA at 0.5 h and 2 h after tFUS targeting RSC. One-way ANOVA, n (Ctrl) = 8 mice, n (0.5 h) = 8 mice, n (2 h) = 6 mice, F_{(2, 19)} = 5.52, P = 0.013. c The number of cells expressing Egr1 in different brain regions. Two-way ANOVA, interaction: F_{(4, 60)} = 3.99, P = 0.0062; Šídák’s multiple comparisons test, RSC: P = 0.0062; ACC (anterior cingulate cortex): P = 0.4294; MCC (middle cingulate cortex): P > 0.9; CA1(field Cornu Ammonis 1 of the hippocampus): P > 0.99; MPtA (medial parietal association cortex): P > 0.5; n = 7 mice per group. d Representative images of Egr1 expression in different brain regions. Scale bar, 100 μm. e Representative images of glutamatergic (anti-CaMKIIα, red) and GABAergic (anti-GAD65&67, purple) neurons in RSC background (Ctrl)-labeled and tFUS-labeled Egr1 cells (cyan in glutamatergic groups, green in GABAergic groups) via immunostaining. Nuclei in blue (DAPI). White arrowheads indicate double-positive neurons: overall scale bar, 100 μm. The zoom-in images correspond to the areas demarcated by the white dashed lines. Scale bar of the enlarged area, 10 μm. f Percentage of CaMKIIα- and GAD65&67-positive cells among Ctrl- or tFUS-labeled RSC Egr1 cells. Two-way ANOVA, interaction, F_{(2, 19)} = 5.52, P = 0.013; Šídák’s multiple comparisons test, Ctrl vs. tFUS (CaMKIIα/Egr1): P = 0.0025, n (Ctrl) = 8 mice, n (tFUS) = 7 mice; Ctrl vs. tFUS (GAD65 + GAD67/Egr1): P = 0.9786, n (Ctrl) = 8 mice, n (tFUS) = 9 mice. Source data are provided as a Source Data file. All data are mean ± SEM. *P < 0.05, **P < 0.01; ns not significant.

To confirm protein synthesis after 2 min of tFUS stimulation in the RSC, we performed immunofluorescence staining (IF) for Egr1 and c-Fos at 50 min post-tFUS. Mice receiving tFUS exhibited a significantly elevated ratio of Egr1-positive (Egr1⁺) cells in the RSC compared to controls (Fig. 3c, d). In contrast, c-Fos expression within the RSC showed no substantial difference between the control and tFUS groups (Supplementary Fig. 4n, o). Furthermore, the number of tFUS-activated Egr1⁺ cells was significantly enhanced in the RSC of Cdh23 mutant mice (Supplementary Fig. 4l and m). Additionally, when comparing potential tFUS-responsive brain regions based on the stimulator’s acoustic pressure range (Fig. 1d and Supplementary Fig. 1a), the RSC exhibited a significant increase in the number of Egr1⁺ cells within the RSC, surpassing the changes observed in other brain regions (Fig. 3c, d). Given the more sensitive and durable changes in mRNA and protein expression, these results collectively indicate that Egr1 is a potential marker for tFUS-sensitive cells, responding more robustly to tFUS than c-Fos.

To further investigate the neuronal mechanism of tFUS, we employed neuronal markers to validate the cell types of tFUS-activated cells. Approximately 70% of Egr1⁺ cells were co-labeled with glutamatergic neurons (CaMKIIα), a significantly higher proportion in the tFUS group compared to the control group (Fig. 3e, f). Conversely, approximately 10% of Egr1⁺ cells were co-labeled with GABAergic neurons (GAD65&67), with no significant change after tFUS (Fig. 3e, f). These data indicate that tFUS primarily activates CaMKIIα neurons.

tFUS-induced Egr1⁺ cells are involved in the regulation of pain

To determine whether tFUS-activated Egr1⁺ cells are involved in pain regulation, we used the activity-dependent neural tagging strategy, known as targeted recombination in the active population (TRAP)³⁰, which employs 4-hydroxytamoxifen (4-OHT) to initiate recombinase activity, thereby providing permanent genetic access to neurons activated by specific external or behavioral stimuli. The TRAP system integrated the ER^T2-Cre-ER^T2 sequence downstream of the Egr1 promoter, ensuring that the cyclization recombination enzyme (Cre) was activated upon Egr1 induction. 4-OHT induced the Cre enzyme to translocate from the cytoplasm to the nucleus, where it excised the sequence flanked by LoxP sites. Three weeks after virus injection in the RSC, mice were intraperitoneally injected with 4-OHT 30 min after tFUS. The mice were euthanized 72 h after the final procedure. Subsequent analysis confirmed the expression of tFUS-TRAPed cells. These cells were identified as those expressing tFUS-driven Egr1, which drove the subsequent Cre-LoxP recombination (Fig. 4a, b).

Fig. 4. tFUS-induced Egr1-positive (Egr1⁺) cells in RSC are necessary for pain behavior.

a Example diagram of the targeted recombination in the active population (TRAP) system design: bilateral microinjection of AAV9-Egr1-ER^T2CreER^T2-mCherry and a Cre-dependent AAV2/9-Syn-DIO-hM4Di-eGFP hybrid virus (Egr1TRAP) into RSC; tFUS activates Egr1 expression, resulting in subsequent expression of ER^T2CreER^T2-mCherry. Intraperitoneal injection of 4-hydroxytamoxifen (4-OHT) allows Cre to enter the nucleus and drive the expression of recombinant Cre-dependent hM4Di-eGFP. Created in BioRender. Wu, C. (2026) https://BioRender.com/vpmqvag. b Schematic diagram of the timeline for verifying the TRAP system’s efficiency. Three weeks after injecting a viral mixture of AAV9-Egr1-ERT2-Cre-ERT2-mCherry and AAV9-Syn-DIO-GFP, we administered an intraperitoneal injection of 4-OHT 30 min following tFUS. We then euthanized the mice three days later to verify the expression of cells captured by tFUS TRAP. Created in BioRender. Wu, C. (2026) https://BioRender.com/5ineha7. c Quantification of tFUS TRAPed cells in Ctrl and tFUS groups. Two-sided unpaired t-test, t = 2.595, P = 0.0169, Ctrl: 12 slices from 8 mice, tFUS: 11 slices from 8 mice. d Confocal images of tFUS TRAPed cells in RSC. White arrowheads indicate tFUS TRAPed cells (cyan): overall scale bar, 200 μm. The magnified images correspond to the regions enclosed by the white dashed lines. Scale bar of the enlarged area, 20 μm. e Experimental design for the TRAP system with the behavioral study. Created in BioRender. Wu, C. (2026) https://BioRender.com/ea0ua8l. f The effect of the TRAP system inhibits tFUS-activated Egr1⁺ cells on PWT. Two-way RM ANOVA, interaction: F_(2,42) = 7.01, P = 0.0024; n (TRAP-Ctrl) = 11 mice, n (TRAP) = 12 mice. g The effect of the TRAP system inhibits tFUS-activated Egr1⁺ cells on TWL. Two-way RM ANOVA, interaction: F_{(2, 42)} = 3.16, P = 0.053; n (TRAP-Ctrl) = 11 mice, n (TRAP) = 12 mice. Source data are provided as a Source Data file. All data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001; ns not significant.

We first assessed the efficacy of the tFUS-TRAP system. Three weeks after the injection of an AAV9-Egr1-ER^T2-Cre-ER^T2-mCherry and AAV9-Syn-DIO-GFP viral mixture, 4-OHT was administered 30 min post-tFUS. A significantly higher number of TRAPed cells was observed in the tFUS group compared to the control group, confirming the system’s reliable and effective labeling of tFUS-activated Egr1⁺ cells (Fig. 4b–d). Then, the Egr1-TRAP system was combined with Syn-DIO-hM4Di-eGFP to inhibit ultrasound-activated Egr1⁺ cells (Fig. 4a and e). Following CNO injection, the control group showed a significant increase in both PWT and TWL compared with baseline, whereas the TRAP group showed no such increase (Fig. 4f, g). Thus, the inhibition of tFUS-activated Egr1⁺ cells further attenuated tFUS’s impact on pain perception. These results reflected that tFUS modulates pain perception through Egr1⁺ cells in the RSC.

Transcriptome characteristics of the tFUS-sensitive cells

Based on our findings, we identified Egr1⁺ cells as an essential marker of tFUS-sensitive cells in the RSC. To screen for critical factors, we employed Egr1-GFP mice combined with fluorescence-activated cell sorting (FACS) to isolate cells without Egr1-GFP (Egr1-negative, Egr1⁻) and cells with Egr1-GFP (Egr1-positive, Egr1⁺) from both control and tFUS-treated samples (n = 6 mice per group) for transcriptomic sequencing (Fig. 5a and Supplementary Fig. 6). Using a stringent selection criterion |log₂FC| > 2 and P_adj < 0.001, differentially expressed genes (DEGs) were identified through comparative analysis across control and tFUS groups (Fig. 5b, Supplementary Figs. 7 and 8, and Supplementary Data 1). Of the DEGs found in the Egr1⁺ (tFUS vs. control), 2 downregulated DEGs and 118 upregulated DEGs exhibited overlap with the other comparisons (Fig. 5c, d).

Fig. 5. Trpc4 as a key gene in tFUS-activated cells.

a Flow chart displaying the transcriptome sequencing strategy for screening tFUS-activated cells in the RSC. Created in BioRender. Wu, C. (2026) https://BioRender.com/9dik64d. b The number of differentially expressed genes (DEGs) across several groupings. c Venn diagram of downregulated DEGs in different groups. d Venn diagram of upregulated DEGs in different groups. e UpSet plot describing the overlap of DEGs from Gene Ontology (GO) enrichment terms relating to ion channel and neuronal activity. Each row represents a GO enrichment term, while each column denotes a potential intersection of these terms. The filled dots indicate which sets are included in an intersection, and the bar charts at the top show the size of each set. f Gene frequency in GO enrichment terms associated with ion channel and neuronal activity. g Gene-GO enrichment term network. Round rectangles depict GO enrichment terms, while ellipses designate genes. The line colors match the rectangles representing the GO enrichment terms. h Heatmap showing the expression levels of each DEG identified from the intersection of the GO enrichment terms connected with ion channels and neuronal activity. Source data are provided as a Source Data file.

Then, we conducted Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses on the DEGs overlapping between the tFUS (Egr1⁺ vs. Egr1⁻) and Egr1⁺ (tFUS vs. control) groups, excluding DEGs from the control (Egr1⁺ vs. Egr1⁻) group. The GO analysis revealed that biological process terms primarily encompassed the regulation of membrane potential and calcium ion transport. Molecular function included active transmembrane transporter activity and ion channel activity. Cellular component terms primarily comprised the neuron-to-neuron synapse and postsynaptic specialization. The GO enrichment findings indicate that tFUS might modulate the activity of ion channels to regulate neuronal activity, offering potential applications in pain management (Supplementary Figs. 9 and 10 and Supplementary Data 2 and 3).

We refined the gene set by focusing on GO terms related to ion channels and neuronal activity (e.g., regulation of membrane potential, various transmembrane transporter activities) based on GO enrichment analysis and in vivo electrophysiological data. Intersecting the genes from these GO terms led us to identify transient receptor potential canonical 4 (Trpc4) (Fig. 5e–h). We prioritized Trpc4 because it is a member of the transient receptor potential (TRP) family, which is linked to ultrasonic neuromodulation^31–33 and plays a critical role in neuronal calcium signaling³⁴. Additionally, Trpc4 showed changes in expression in our RNA-seq results after tFUS stimulation. While other TRP channels (e.g., TRPM2³¹, TRPC1³², TRPC6³³) are involved in ultrasonic neuromodulation, their mechanisms are less relevant to the specific neural circuit model of this study.

TRPC4 facilitates ultrasonic activation of neurons in the RSC

To confirm the ultrasonic sensitivity of the TRPC4 ion channel, in situ hybridization (ISH) combined with an IF study was performed in the RSC. Our findings showed that Trpc4-expressing cells colocalized with Egr1, and tFUS significantly enhanced Trpc4 expression in Egr1⁺ cells (Fig. 6a, b and Supplementary Fig. 11a, b). Moreover, the Trpc4-expressing cells in the RSC were primarily glutamatergic neurons (Supplementary Fig. 11c–e). Based on our RNA-seq results, other ion channel genes (such as K2P, HCN, and VGSC family members) were also upregulated, suggesting potential synergistic contributions from multiple channels.

Fig. 6. TRPC4 facilitates the ultrasonic activation of RSC.

a Representative images showing co-localization of Trpc4 mRNA (red) and Egr1 (cyan) in the RSC of Ctrl and tFUS groups. Nuclei are stained with DAPI (blue). White arrowheads indicate double-positive cells: overall scale bar, 100 μm. The boxes delineated by the white dashed lines are shown in the zoom-in views (scale bar, 10 μm). b mRNA expression of Trpc4 in RSC’s Egr1-GFP⁺ cells between Ctrl and tFUS groups. Two-sided unpaired t-test, t = 2.935, P = 0.0324, n (Ctrl) = 3, n (tFUS) = 4. c Schematic of inside-out patch-clamp recording with tFUS. d, e Representative traces (d) and quantification of currents (e) at −60 mV in Ctrl, TRPC4, and TRPC4 + ML204 (10 μM) groups (one-way ANOVA, P = 6.33 × 10⁻⁸), n (Ctrl) = 5 patches from 3 cell cultures, n (TRPC4) = 12 patches from 4 cell cultures, n (TRPC4 + ML204) = 5 patches from 3 cell cultures. f Chronological representation of the experimental protocol for the vehicle control and ML204 cohorts. Created in BioRender. Wu, C. (2026) https://BioRender.com/s0x1erp. g PWT was detected in the vehicle control and ML204 cohorts before and upon tFUS targeting of the RSC. Two-way RM ANOVA, vehicle control vs. ML204: interaction, F_{(1, 8)} = 8.940, P = 0.0173; n = 5 mice per group. h TWL was identified in the vehicle control and ML204 cohorts before and during tFUS targeting of the RSC. Two-way RM ANOVA, vehicle control vs. ML204: interaction, F_{(1, 8)} = 39.92, P = 0.0002; n = 5 mice per group. i, j Representative action potential traces (i) and spike numbers (j) before and after Englerin A (EA, 200 nM) application in RSC neurons, recorded from 11 neurons across 10 mice. Two-way RM ANOVA, BL vs. EA: interaction, F_{(6, 70)} = 5.37, P = 0.0001. All data are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

To investigate the impact of tFUS on TRPC4, we focused on its electrophysiological properties, as this channel exhibits inherent ion permeability. Patch-clamp recording in the inside-out configuration (Fig. 6c) showed a substantial increase in current in TRPC4 cells following ultrasound stimulation compared to the control group. Furthermore, treatment with ML204 (10 μM), a selective TRPC4 inhibitor with an IC₅₀ value of 0.96 μM, markedly suppressed this tFUS-evoked current enhancement (Fig. 6d, e). These results, which suggest tFUS may activate the TRPC4, provide evidence that TRPC4 is regulated by ultrasound stimulation. This highlights the potential of TRPC4 as a primary factor in ultrasonic neuromodulation, advancing understanding of the mechanisms underlying ultrasonic activation of RSC.

TRPC4 facilitates ultrasonic effects on pain

To investigate the potential regulatory mechanism that TRPC4-mediated modulation of pain threshold in RSC upon action of tFUS, we inhibited TRPC4 activity with ML204 (2 mg/kg, twice daily via intraperitoneal injection) (Fig. 6f). tFUS successfully increased PWT and prolonged TWL in the vehicle control group; however, it failed to enhance PWT or TWL in the ML204-treated cohort (Fig. 6g, h), suggesting the involvement of TRPC4. Furthermore, IF results showed that tFUS-induced Egr1 expression was significantly reduced in the RSC of ML204-treated mice compared to the vehicle control group (Supplementary Fig. 11f, g), indicating that TRPC4 controls the cellular response to tFUS. To determine if TRPC4 mediates the analgesic effect across multiple brain regions, we pre-administered ML204 and then ultrasonically stimulated the ACC. As seen in the RSC, tFUS failed to extend TWL in the ACC in the ML204 cohort, with the significant prolongation observed in the vehicle control group (Supplementary Fig. 4e). In vitro whole-cell patch-clamp recordings in the RSC showed that the TRPC4 agonist Englerin A (EA, 200 nM) substantially reduced the frequency of action potential firing in pyramidal neurons, indicating that activation of TRPC4 leads to decreased neuronal excitability (Fig. 6i, j). Collectively, these behavioral, cellular, and electrophysiological results demonstrate that tFUS-mediated regulation of the pain threshold depends on TRPC4 activity across different brain regions.

To further investigate the role of TRPC4 in modulating neuronal activity and pain threshold during the stimulation of tFUS, we employed short-hairpin RNA (shRNA) to reduce the expression levels of TRPC4 in glutamatergic neurons of the RSC (Fig. 7a–c and Supplementary Fig. 11h). Consistent with the findings from ML204 treatment, TRPC4 knockout prevented tFUS from altering the PWT or the TWL (Fig. 7d, e). While a scrambled shRNA control group showed the expected tFUS-induced continuous decline in RSC neuronal activity, knocking down TRPC4 significantly reduced this sustained decrease (Fig. 7f–h). To further confirm these findings, we employed Trpc4-T2A-iCre mice with chemical genetics suppression to inhibit the activity of TRPC4-expressing neurons (Fig. 7i), which also blocked tFUS from elevating PWT or TWL (Fig. 7j, k). Additionally, comparing the CNO + tFUS group with the baseline in the PWT results, there appears to be a trend toward increased PWT; however, the effect did not reach statistical significance (Fig. 7j). In summary, these findings demonstrate that the expression of TRPC4 is necessary for tFUS to regulate the continuous decline in RSC neuronal activity, thereby influencing the control of pain sensation by tFUS.

Fig. 7. Specific inhibition of TRPC4 blocks tFUS-mediated regulation of pain.

a Schedule for behavioral test. Created in BioRender. Wu, C. (2026) https://BioRender.com/eq7leg6. b Illustrations of CaMKIIα-TRPC4-short-hairpin RNA (shRNA)-RFP virus (red) expression in RSC. Scale bar, 200 µm. The rectangular region is the enlarged partial RSC, scaled to 50 µm. c TRPC4 shRNA knocks down Trpc4 expression in RSC via ISH. Two-sided unpaired t-test, P = 0.0116, n = 4 mice per group. d Effects of tFUS on PWT. Two-way RM ANOVA; scrambled vs. TRPC4 shRNA, BL: P = 0.9891, virus: P = 0.9741, tFUS: P = 7.66 × 10⁻⁷; n (scrambled) = 8 mice, n (TRPC4 shRNA) = 13 mice. e Effects of tFUS on TWL. Two-way RM ANOVA; scrambled vs. TRPC4 shRNA, BL: P = 0.9474, virus: P = 0.9987, tFUS: P = 0.0068; n (scrambled) = 8 mice, n (TRPC4 shRNA) = 13 mice. f Representative raster plots of pyramidal neurons (PNs) and interneurons (INs) after scrambled or TRPC4 shRNA expression. g TRPC4 knockdown attenuates tFUS-induced changes in spike firing rates in PNs and INs. h Percentage of neurons showing decreased firing rates after tFUS (PN: scrambled, n = 19 neurons, 37% decrease, TRPC4 shRNA, n = 17 neurons, 24% decrease, two-sided Fisher’s exact test, P = 0.04; IN: scrambled, n = 25 neurons, 52% decrease, TRPC4 shRNA, n = 12 neurons, 17% decrease, Fisher’s exact test, P < 0.001). i Timetable for behavioral assessment. Created in BioRender. Wu, C. (2026) https://BioRender.com/jnl5c1l. j Impact of tFUS on PWT to hM4Di-mediated inhibition of TRPC4 expression. One-way RM ANOVA; BL vs. tFUS: P = 0.0011; tFUS vs. tFUS + CNO: P = 0.037; BL vs. tFUS + CNO: P = 0.13; n = 7 mice. k Effect of tFUS on TWL to hM4Di-induced suppression of TRPC4 expression. One-way RM ANOVA; BL vs. tFUS: P = 0.0122; tFUS vs. tFUS + CNO: P = 0.0024; BL vs. tFUS + CNO: P = 0.9995; n = 7 mice. Source data are provided as a Source Data file. All data are presented as mean ± SEM; *P < 0.05, **P < 0.01.

Transient receptor potential canonical 5 (TRPC5) mediates ultrasound-induced analgesia

In this study, we found that TRPC4 facilitated the effects of ultrasound on pain. Given the significant functional overlap between TRPC4 and TRPC5 channels, both of which are potentiated by G protein-coupled receptors (GPCRs), and the fact that these two channels exhibit similar current-voltage relationships36, we investigated whether TRPC5 in the RSC exhibits a comparable function in response to ultrasound stimulation. To explore this, we used shRNA to knock down TRPC5 expression, and the knockdown efficacy in targeted cells was confirmed by ISH (Fig. 8a, b) and qPCR (Supplementary Fig. 12a). The analgesic effects of tFUS were significantly reduced in the TRPC5 shRNA group compared to the control group (Fig. 8c, d). To further validate the role of TRPC5 in tFUS neuromodulation, we performed in vivo multichannel electrophysiological recordings; the scrambled shRNA control group showed the anticipated gradual decrease in RSC neuronal activity induced by tFUS, whereas reducing TRPC5 notably mitigated this persistent decline (Fig. 8e–g). The above results indicate that, similar to TRPC4, TRPC5 plays a key role in mediating the analgesic response to ultrasound.

Fig. 8. TRPC5 in RSC is involved in ultrasonic stimulation.

a Representative images show Trpc5 mRNA expression in the RSC scrambled and CaMKIIα-TRPC5-shRNA samples, detected by ISH (red). White arrowheads indicate positive cells (scale bar, 100 μm). The zoom-in images correspond to the areas demarcated by the white dashed lines (scale bar, 20 μm). b TRPC5 shRNA knockdown of Trpc5 expression in RSC, as determined by ISH analysis. Two-sided unpaired t-test, P = 0.0001, n = 8 mice per group. c Effects of tFUS on PWT following shRNA knocking down the expression of TRPC5. Two-way RM ANOVA, interaction: F_{(1, 18)} = 12.96, P = 0.002; n (scrambled) = 9 mice, n (TRPC5 shRNA) = 11 mice. d Effects of tFUS on TWL after shRNA knocking down TRPC5 expression. Two-way RM ANOVA, interaction: F_{(1, 16)} = 17.86, P = 0.0006; n (scrambled) = 8 mice, n (TRPC5 shRNA) = 10 mice. e Raster plots for PNs (left) and INs (right) during different periods after expression of TRPC5 knockdown (bottom) and scrambled (top) shRNA. This plot displays representative examples. f TRPC5 knockdown reduced the tFUS-induced decrease in the spike firing rate after stimulation in both PNs (left) and INs (right). g TRPC5 knockdown reduced the percentage of tFUS-induced decreases in spike firing rate after stimulation in both PNs and INs (PN: scrambled, n = 44 neurons, 46% decrease, TRPC5 shRNA, n = 28 neurons, 3% decrease, two-sided Fisher’s exact test, P < 0.0001; IN: scrambled, n = 13 neurons, 77% decrease, TRPC5 shRNA, n = 20 neurons, 15% decrease, two-sided Fisher’s exact test, P < 0.001). Source data are provided as a Source Data file. All data are presented as mean ± SEM; *P < 0.05, **P < 0.01, ****P < 0.0001; ns not significant.

TRPC4 and TRPC5 interact endogenously, and co-expression synergistically potentiates ultrasound-evoked calcium responses

Given that both TRPC4 and TRPC5 play an important role in mediating ultrasound-induced analgesia, we next investigated whether these two channels form a functional heteromeric complex to exert their analgesic effects. Using ISH and IF, we confirmed that both TRPC4 and TRPC5 are co-expressed in the same neuronal populations within the RSC (Supplementary Fig. 12b–e), suggesting potential synergistic interactions in pain modulation. To further examine whether TRPC4 and TRPC5 physically interact, we performed an in situ proximity ligation assay (PLA). A notable PLA signal was detected in the TRPC5 scrambled shRNA control, with significantly stronger signal intensity than in the TRPC5 shRNA knockdown group (Fig. 9a, b), indicating a close physical proximity between the two channels. To further confirm their direct interaction, we performed co-immunoprecipitation (Co-IP) assays using RSC tissue lysates. The results revealed that TRPC4 and TRPC5 physically associate and form a stable protein complex under physiological conditions (Fig. 9c, d). Collectively, these biochemical and histological findings provide strong evidence for the existence of a TRPC4/TRPC5 heteromeric complex in vivo.

Fig. 9. Endogenous interaction of TRPC4 and TRPC5 and synergistic enhancement of ultrasound-evoked calcium responses upon co-expression.

a Representative images show the in situ proximity ligation assay (PLA, red) signal of TRPC4 and TRPC5 (scale bar, 200 μm). The white dashed lines indicate the locations of the zoomed-in images (scale bar, 20 μm). b PLA signal of TRPC4 and TRPC5 detected in the RSC scrambled and CaMKIIα-TRPC5-shRNA groups. Two-sided unpaired t-test, P = 0.0036, n = 4 mice per group. c Co-immunoprecipitation (Co-IP) analysis was performed to detect the association between TRPC5 and TRPC4 in RSC lysates. Protein complexes were immunoprecipitated (IP) with an anti-TRPC5 antibody or normal rabbit IgG (negative control). The resulting precipitates were analyzed by Western blotting (IB) using antibodies against TRPC5 and TRPC4. Input (5% of total lysate) represents the initial protein expression. The data show that TRPC4 specifically co-precipitated with TRPC5, indicating an endogenous physical interaction between the two channels. Uncropped blots in Source Data. d The relative expression of TRPC4 in IgG and TRPC5 IP groups (two-sided paired t-test, P = 0.0119, n = 10 mice). e Representative calcium imaging heatmaps of Cal-630 fluorescence in HEK-293T cells (scale bar, 20 μm). f The line chart displays fluorescent intensity (ΔF/F%) in the calcium imaging experiment. Two-way RM ANOVA, interaction, F_{(2495, 328342)} = 78.66, P < 1 × 10⁻¹⁰⁰; US, ultrasound stimulation. g The violin plot exhibits ΔF/F% at 5 s before US. Two-way ANOVA, interaction, F_{(2, 658)} = 1.026, P = 0.3590. h The violin plot shows ΔF/F% at 10.5 s at the beginning of US. Two-way ANOVA, interaction, F_{(2, 658)} = 4.561, P = 0.0108. i The violin plot displays ΔF/F% at 100 s during the US. Two-way ANOVA, interaction, F_{(2, 658)} = 78.66, P = 1.1655 × 10⁻⁶. Calcium imaging sample sizes: n = 99–128 cells per group from 7 to 8 dishes. Source data are provided as a Source Data file. All data are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns not significant.

Finally, to determine whether TRPC4 and TRPC5 can assemble into a functional heteromeric complex in response to ultrasound stimulation, we performed live-cell calcium imaging using the fluorescent calcium indicator Cal-630. HEK-293T cells were transfected with TRPC4-mCherry (5 µg), TRPC5-EGFP (5 µg), or co-transfected with both constructs (2.5 µg each). Both TRPC4- and TRPC5-expressing cells exhibited clear increases in intracellular calcium upon ultrasound stimulation (Fig. 9e–i). Notably, co-expression of TRPC4 and TRPC5 exhibited the most pronounced calcium elevation during ultrasound stimulation, indicating the formation of a functional heteromeric complex (Fig. 9f, h, and i). Application of the selective TRPC4/C5 inhibitor ML204 significantly suppressed the ultrasound-induced calcium signals across all transfection groups, including the co-transfected cells.

Taken together, these findings demonstrate that TRPC4 and TRPC5 form a functional heteromeric complex in the RSC, which is integral to ultrasound-activated pain modulation, highlighting their potential as therapeutic targets for pain treatment.

In summary, our study suggests that TRPC4 and TRPC5 are critical for tFUS-induced analgesia in the RSC. Knockdown of TRPC4 in glutamatergic neurons of the RSC and ML204 inhibition both had similar analgesic effects. The knockdown significantly reduced the sustained decline in pyramidal and interneuron activity caused by tFUS. Meanwhile, inhibiting TRPC4 significantly reduced tFUS effects on pain thresholds, supporting its role in mediating the tFUS response and controlling tFUS-induced pain sensation. Similar to TRPC4, TRPC5 is also essential for ultrasound analgesia, as evidenced by changes in pain sensitivity in TRPC5 knockdown mice, and in vivo electrophysiological studies confirmed that it mediates tFUS-induced reduction in RSC activity. In RSC neurons, TRPC4 and TRPC5 physically interact to form a stable heteromeric complex, with co-expression driving the strongest ultrasound-induced calcium elevation, confirming functional synergy. Together, these channels underpin tFUS pain modulation and represent promising therapeutic targets.

Discussion

In this study, we demonstrate that tFUS application in the RSC alters neuronal activity and increases the pain threshold in mice. The effects of tFUS are mediated by TRPC4 and TRPC5 in the RSC, where these channels colocalize and form a stable heteromeric complex to exert their synergistic effects. Meanwhile, we confirm that TRPC4 plays a critical role in ultrasound-induced pain relief, with TRPC5 additionally validated as essential, as evidenced by in vivo electrophysiological recordings showing mitigated tFUS-induced reduction in RSC activity upon TRPC5 knockdown. Overall, our study provides further evidence supporting the use of tFUS as a non-invasive, promising method for pain management, while elucidating the core role of TRPC4/TRPC5 heteromeric complexes in mediating tFUS-induced analgesia (Fig. 10).

Fig. 10. tFUS modulates neuronal activity via TRPC4/TRPC5 in RSC, altering spike rates and desensitizing mechanical and thermal sensations in mice.

The diagram depicts the mechanism by which tFUS, when applied to the RSC of mice, modulates neuronal activity. This modulation is achieved through its action on both pre-existing TRPC4/TRPC5 channels, which are critical for ion movement and neuronal modulation, and through the de novo synthesis of TRPC4 in RSC neurons at later stages. The process desensitizes mechanical and thermal sensations, as demonstrated through von Frey and Hargreaves tests. Multi-channel in vivo recordings highlight significant changes in the spike rates of pyramidal neurons and interneurons under tFUS stimulation, with sustained (US) and inter-tFUS interval (ITI) pulses driving these effects. The model underscores that inhibiting TRPC4/TRPC5 or its expression reduces tFUS’s impact, providing evidence that TRPC4/TRPC5 is essential for ultrasonic neuromodulation. Created in BioRender. Wu, C. (2026) https://BioRender.com/18b0tt4.

It is widely accepted that tFUS sensitivity is driven by the diverse, heterogeneous expression of mechanosensitive channels across cell types. Intrinsic calcium-permeable ion channels in cortical neurons, such as TRPC1, TRPC6, and TRPM2, are key ultrasound-activated ion channels that facilitate pathways in ultrasonic neuromodulation^31–33. Transcriptome sequencing of Egr1-GFP⁺ cells (GO enrichment analysis) and in vivo multi-channel electrophysiological recordings consistently show that tFUS regulates neuronal functions by modulating ion channel activity. Specifically, TRPC4, a TRP family member, is strongly linked to ultrasonic neuromodulation^31–33. tFUS significantly up-regulates Trpc4 expression, as confirmed by RNA-seq and ISH. However, this is a use-dependent increase in expression, meaning TRPC4 is clearly detected under basal conditions in Egr1-GFP⁺ cells of the RSC neurons. This basal expression forms pre-existing TRPC4 channels that the tFUS treatment likely utilizes, necessitating the upregulation to prepare additional reserves, a common interpretation for most upregulated DEGs. This view differs from the conventional understanding of inducible genes, which require de novo synthesis for function. Crucially, inhibiting TRPC4 with ML204 blocks the tFUS effect on the RSC, suggesting that tFUS activates both pre-existing TRPC4 channels and de novo-synthesized channels at later stages in RSC neurons. This mechanism is not limited to a single brain region, as inhibiting TRPC4 also prevented tFUS from extending the TWL in the ACC.

TRP channels act as crucial sensors for diverse intracellular and extracellular stimuli³⁵. Within this superfamily, TRPC channels regulate various physiological functions and are primarily triggered by phospholipase C. While most TRPC channels are activated by G_q/11-coupled receptor stimulation, TRPC4 necessitates concurrent activation of G_i/o-coupled receptors³⁶. TRPC4 activation in neurons has been shown to induce a strong and sustained depolarization block, which subsequently inhibits neuronal firing, likely through the inactivation of voltage-gated sodium channels^37,38. This depolarization block potentially explains how TRPC4 serves as a molecular nexus for tFUS action in neurons of the RSC, simultaneously mediating both action potential inhibition and Egr1 expression. To experimentally validate this suppressive effect, patch-clamp recordings in brain slices demonstrate that activating TRPC4 with the agonist EA significantly reduced action potential firing in RSC pyramidal neurons (Fig. 6i, j). Despite causing depolarization block, TRPC4 is also well-established to facilitate calcium influx into neurons^39,40, which increases intracellular calcium concentrations, thereby promoting Egr1 expression. Thus, by boosting calcium entry independently of neuronal firing, TRPC4 activation highlights the intricate relationship between increased expression of an indicator of neural activation (Egr1) and the inhibition of electrical activity. Future research is crucial to investigate the specific pathways and molecular mechanisms that link TRPC4 activation to Egr1 expression. This will be key to fully understanding the nuanced roles of calcium signaling in both general neuronal function and the specific response to tFUS stimulation.

Our in vivo multi-channel electrophysiological recordings indicated a substantial decline in neuronal spike rate within the RSC during tFUS stimulation. Following tFUS application, the neuronal discharge frequency showed heterogeneous post-tFUS treatment effects. While TRPC4 inhibition did not alter neuronal activity during tFUS stimulation of the RSC, it notably reduced the percentage of neurons displaying a sustained drop in activity post-tFUS. As discussed before, TRPC4 activation typically decreases, rather than increases, firing rates. The lack of an immediate, clear effect of TRPC4 inhibition on firing rates during stimulation means that this treatment did not overtly silence action potential generation. While further analysis is required to quantify and account for subtler alterations in firing rates rigorously, it is plausible that any direct changes induced by TRPC4 inhibition might be masked or compensated for by parallel mechanisms (such as mechanosensitive ion channels or alternative signaling pathways), thereby obscuring a clear assessment of their direct impact.

TRP channels are critical regulators of pain signaling. Traditional targets (TRPV1, TRPA1, TRPM8) are found in nociceptors^41,42, whereas newer TRP targets (TRPM3, TRPC4, TRPC5, TRPV3, TRPV4, TRPM2) exhibit a wider distribution, such as sensory neurons, brain nuclei, epithelia, and immune cells⁴³. A study using TRPC4 knock-out rats demonstrated tolerance to visceral pain, while somatic pain responses were unaffected⁴⁴. This involvement was corroborated by the fact that the TRPC4 inhibitor ML204 effectively suppressed visceral pain responses in wild-type rats⁴⁵ and, when applied locally, ML204 also raised the neuropathic pain threshold in rats following spared nerve injury⁴⁵. This evidence suggests that tFUS simulation, which may activate TRPC4, could impact pain responses. Future study is required to understand the mechanism of tFUS-mediated TRPC4 activation.

Given the significant functional homology and potential overlap between TRPC4 and TRPC5⁴⁶, we hypothesized that TRPC5 is integral to ultrasound-induced analgesia. Our study demonstrates that TRPC5 knockdown significantly attenuated the analgesic effects of ultrasound, indicating a crucial role of this channel in analgesia. Since TRPC4 is also implicated in this process, we explored its mechanism of action. Co-expression of TRPC4 and TRPC5 in the same neuronal populations within the RSC provides a structural basis for their synergistic interaction, while in situ PLA confirmed specific endogenous TRPC4/TRPC5 binding. Co-IP assays with RSC tissue lysates corroborated these findings, confirming that TRPC4 and TRPC5 form a stable endogenous heteromeric complex in vivo. Consistent with previous research, TRPC4 and TRPC5 can assemble into either homotetrameric or heterotetrameric⁴⁷ non-selective cation channels. This function partnership likely serves to fine-tune neuronal excitability and pain perception. Live-cell calcium imaging further characterized the heteromer’s functionality: co-expression of TRPC4 and TRPC5 elicited a more robust calcium influx in response to ultrasound stimulation than either channel alone, confirming that the heteromeric complex is functionally superior at transducing ultrasound signals. Future studies would focus on elucidating the precise mechanisms underlying the interaction between TRPC4 and TRPC5 and their combined contribution to the overall analgesic effects of ultrasound. Meanwhile, exploring the potential for pharmacological modulation of these channels could yield innovative strategies to enhance pain relief in clinical settings.

Diverse ultrasonic parameters, encompassing acoustic pressure amplitude, pulse train duration (PTD), duty cycle (DC), and pulse repetition frequency (PRF, 1/PRI)^21,22, closely affect the ultrasonic intensity and yield distinct outcomes of ultrasonic neuromodulation. Ultrasonic absorption can elevate tissue temperature, while optimized parameters can avoid excessive warming. During our measurements with tFUS stimulation of the RSC, the temperature increase at the target tissue was less than 1 °C. The minimal elevation is significantly below the reported threshold for thermal-induced neural activation (ΔT > 5 °C)^23,24. Thus, our results suggest that the observed effects of tFUS on RSC neuronal activity and pain threshold are attributable to a non-thermal mechanism.

The spatial-peak, temporal-averaged intensity (I_spta), which denotes the intensity averaged throughout the experimental duration, is the optimal metric for assessing ultrasound-mediated heat transfer to tissue⁴⁸. I_spta values varied across the different ISIs tested. The lowest intensity, observed with the 10-s ISI (22.3 mW/cm²), was deemed insufficient to elicit behavioral changes, suggesting inadequate intensity for significant neural modulation⁴⁹. Conversely, the I_spta for the 3-s ISI (74.3 mW/cm²) markedly elevated the pain threshold, whereas that for the 1.5-s ISI (148.7 mW/cm²) produced a lesser elevation in pain threshold than the 3-s ISI. Collectively, these findings demonstrate that the change in pain threshold induced by tFUS does not display a linear relationship with the escalation in ultrasonic intensity.

PRF and DC also play essential roles in tFUS neuromodulation. The PRF governs the rhythm of tFUS, shaping the temporal dynamics of the neuronal response⁵⁰, while the DC defines the active-to-rest time ratio⁵¹. Adjustments to these parameters alter neuronal excitability and modulation, highlighting the intricate balance required for effective neuromodulation. Although a systematic mapping of excitatory and inhibitory effects induced by different tFUS parameters (PRF and duty cycle) across brain regions has not yet been fully established, emerging literature indicates that these parameters can bidirectionally modulate cortical excitability. Higher PRF (typically 500–3000 Hz) combined with moderate-to-higher duty cycles (e.g., 20–60%) appears to favor net excitatory effects, often by preferentially activating regular-spiking putative excitatory neurons, whereas lower PRF (e.g., <100 Hz) with low duty cycles (e.g., 1–10%) tend to produce more inhibitory effects^52,53. In the RSC, a region implicated in integrating sensory and affective aspects of pain, excitatory stimulation (e.g., optogenetic or chemogenetic activation of RSC glutamatergic neurons or their projections to the ACC) has been reported to facilitate nociceptive behaviors, such as decreased paw withdrawal thresholds and shortened thermal latencies¹⁸. Conversely, inhibitory modulation of RSC neuronal activity has been associated with antinociceptive effects¹⁹. These observations suggest that the analgesic effects observed in our study may result, at least in part, from tFUS-induced net inhibition of RSC activity via TRPC4/TRPC5-mediated pathways. As shown by our patch-clamp recordings, TRPC4 activation induces a strong depolarization block that suppresses neuronal firing while still permitting calcium influx to promote Egr1 expression. Future experiments will systematically vary tFUS parameters (e.g., PRF: 100, 500, 1000, 2000 Hz; duty cycle: 5, 10, 20, 40%) using TRPC4/TRPC5 genetic models, combined with qPCR, immunofluorescence, and electrophysiological recordings, to further clarify the net effects of our tFUS protocol in the RSC and its interaction with TRPC4/TRPC5 in pain signaling.

Low-intensity focused ultrasound has demonstrated a reversible capacity to both excite and inhibit neural activity⁵⁴ in both in vitro and in vivo studies, affecting neuronal firing rates and conduction speeds in a bidirectional manner^32,55,56. Our findings indicate that the application of tFUS significantly decreased the spike rates of both pyramidal neurons and interneurons during tFUS application. Crucially, the post-treatment alterations in discharge frequency of these two neuron types differed. This complexity suggests that the impact of tFUS on neural activity is multifaceted, likely modulated by factors including the specific parameters used and the characteristics of the targeted tissue. Additional research is required to elucidate the mechanisms underlying tFUS-induced neuromodulation and to determine how different parameters can be optimized to achieve specific therapeutic outcomes.

Targeting pain-associated regions such as the dorsal root ganglion (DRG), primary somatosensory cortex (S1), insula, and anterior thalamus with tFUS has yielded encouraging outcomes in alleviating pain sensitivity in both animal models and human participants. Specific applications have demonstrated efficacy: ultrasound application to the DRG successfully desensitized pain in nerve-injured rats³. tFUS activation of the S1 reduced heat pain sensitivity in WT and regulated heat and mechanical hyperalgesia in chronic pain mouse models⁴. In human studies, tFUS targeting the right anterior thalamus significantly reduced thermal pain sensitivity in healthy individuals⁵⁷ and provided comparable improvements in patients with chronic, therapy-resistant neuropathic pain⁵⁸. Unlike previous research, our work demonstrates a significant alteration in the pain threshold following tFUS targeting the RSC, a pain-associated cortical subregion. Implementing tFUS on the RSC resulted in substantial alterations in the PWT and TWL in both WT and chronic pain mouse models, illustrating the promising potential of tFUS as a non-invasive approach for chronic pain management.

Several factors guided our selection of the RSC. Crucially, existing literature^17,18,20, alongside our own published study^14,19, has increasingly implicated the RSC in the processing and maintenance of neuropathic pain. The RSC, although not a primary sensory region, exhibits notable structural and functional changes in chronic pain conditions, with its activity linked to pain-related behaviors in neuropathic pain models. This strongly suggests a significant function for the RSC in the cognitive and emotional dimensions of the pain experience. Notably, the presence of TRPC4 in the RSC, as observed in S1 and ACC, provides a plausible molecular target for tFUS-mediated neuromodulation. Given the increasing evidence for TRPC4’s involvement in various pain modalities^44,59,60, including neuropathic pain, investigating its specific role in a region where it is abundantly expressed and functionally implicated in pain provides valuable mechanistic insight.

The RSC is a highly interconnected cortical region that plays a crucial role in spatial navigation, memory, and contextual processing^61,62. However, emerging evidence also highlights its involvement in pain processing, particularly in the integration of sensory-discriminative and affective-motivational aspects of pain^14,18,19. The RSC integrates sensory and limbic inputs to contextualize pain, modulate pain perception, and form pain-related memories. The current study employed tFUS parameters that selectively influenced RSC pain circuits without significantly affecting memory or anxiety networks. These specific tFUS settings likely enhanced modulation of pain-related neuronal populations. At the same time, behavioral tests for pain proved more sensitive to these changes than those for memory or anxiety, which may have been too subtle to detect.

The RSC plays a multifaceted role in nociception, involving the processing of aversive information, reactivity to pain in puts within neuronal circuits, and the activation of pain-mitigating inhibitory processes. Afferent and efferent RSC connections were investigated using both retrograde and anterograde tracing to elucidate their potential circuitry in controlling spinal reflex withdrawal behaviors^15,63. These studies indicate a descending connection between the RSC and the periaqueductal grey (PAG), a major pain regulatory center. The PAG exerts its influence primarily via the rostral ventromedial medulla (RVM). At the same time, the PAG has limited direct projections to the spinal cord; it extensively projects to the RVM, which subsequently descends to the spinal dorsal horn via the dorsolateral funiculus. Non-selective stimulation of this PAG-RVM induces antinociception. The RSC is therefore hypothesized to modulate spinal reflexes by projecting to the PAG and activating the descending PAG-RVM pathway. Consequently, manipulating RSC activity, such as via tFUS, is expected to affect pain metrics such as PWT and TWL.

To determine whether tFUS directly modulates neurons in the RSC or acts indirectly via a sensory pathway, we treated Cdh23 mutant mice with tFUS targeting the RSC. Both PWT and TWL were significantly increased, accompanied by a substantial rise in Egr1⁺ cell count in the RSC. Furthermore, multi-channel recordings showed concurrent alterations in the activity of pyramidal neurons and interneurons during tFUS. These data indicate that tFUS directly modulates neuronal activity rather than acting indirectly.

Based on our multi-channel electrophysiological recordings, the observed variation in neural activity suggests that tFUS differentially affects distinct neuronal groups, including excitatory and inhibitory neurons (Fig. 2). These acute, single-neuron changes appear to trigger long-lasting synaptic plasticity, potentially leading to broader network reorganization. Prior research indicates that tFUS can modulate neural circuit activity by fine-tuning parameters like intensity and frequency. This modulation can manifest as either long-term potentiation or long-term depression of synaptic connections, thereby reshaping brain function^64–66 and impacting pain neural circuits to produce lasting behavioral effects^67,68 (Fig. 1g, h).

Our study shows that tFUS applied to the RSC modulates neuronal activity and alters pain sensitivity in mice, laying a foundation for optimizing chronic pain treatment. Neuroimaging, like fMRI, could guide tFUS target selection by identifying aberrant pain-related RSC activity, enabling personalized therapy. In addition, tFUS may complement existing treatments, such as pharmaceuticals and cognitive behavioral therapy, with future trials needed to explore synergistic effects. These findings also set the stage for investigating tFUS effects in other pain-related brain regions to advance non-invasive neuromodulation therapies.

In this study, we demonstrate that stimulating RSC with tFUS significantly modulates neuronal activity and elevates the pain threshold in mice. This analgesic effect is primarily mediated by Egr1⁺ cells and associated with altered Egr1 expression. RNA-sequencing analysis of sorted Egr1-GFP⁺ cells identified TRPC4 as a key regulator of the tFUS response. Further investigation revealed that both TRPC4 and TRPC5 are indispensable for ultrasound-mediated analgesia, as they co-express in RSC neurons, form a stable endogenous heteromeric complex in vivo, and function together to elicit robust calcium influx upon ultrasound stimulation. Collectively, our findings establish tFUS as a powerful, non-invasive, non-viral tool for in vivo brain targeting, with successful application to the RSC, highlighting its potential to advance neuroscience research and clinical neuromodulation.

Methods

Animals

Adult male mice (6–8 weeks of age, body weight 20–35 × g) were group-housed (4–5 per cage) under controlled environmental conditions (room temperature 21 ± 1 °C, relative humidity 60 ± 5%, 12-h light/dark cycle with lights on at 07:00). Food and water were provided without restriction. Mice were allowed to acclimate to the behavioral testing environment for 15 min daily over 7 consecutive days prior to experiments. All procedures were performed in compliance with the guidelines of the Animal Care and Use Committee of Zhejiang University (approval number ZJU20250555) and followed Chinese regulations on medical waste management. Male C57BL/6J mice were purchased from Zhejiang Vital River Laboratory Animal Technology Co., Ltd (Zhejiang, China). Egr1-GFP male mice were kindly provided by Dr. Wei Yang (Zhejiang University), B6(V)-Cdh23v-2J/J male mice (Cdh23) by Dr. Wei Xiong (Chinese Institute for Brain Research), and Trpc4-T2A-iCre male mice (Strain NO. T055395) from GemPharmatech (Nanjing, China).

Ultrasound stimulation system

tFUS was delivered using a custom head-mounted piezoelectric stimulator (5 mm diameter, 0.74 MHz center frequency)²¹ driven by a programmable brain stimulation controller (UPG-100, Hyus Meditec, Shenzhen, China)²². Mice were anesthetized with pentobarbital sodium (50 mg/kg, intraperitoneally (i.p.) injected, Sigma, Cat# P3761). A pedestal was secured to the skull over the RSC (anterior/posterior (AP): −1.58 mm; medial/lateral (ML): ± 0.2 mm) along with three anchoring screws, and the assembly was fixed with dental cement. Both control and tFUS groups underwent identical surgical implantation. After 7 days of post-surgical recovery, the stimulator was coupled to the pedestal using ultrasound gel. Pulsed ultrasound (sound pressure 214 kPa at 100% amplitude, pulse duration (PD) 0.5 ms, pulse repetition interval (PRI) 1 ms, pulse train duration (PTD) 300 ms) was applied for 2 min at inter-stimulus intervals (ISIs) of 1.5, 3, or 10 s. Its acoustic pressure distribution was mapped using a 3D acoustic scanning system to evaluate the head-mounted ultrasound stimulator.

The calculation of the spatial-peak temporal-average intensity (I_spta) is a two-step process that accounts for both the intensity of the ultrasound pulse and the fraction of time the pulse is active⁶⁹. First, the spatial-peak pulse-average intensity (I_sppa) is calculated from the peak acoustic pressure (P0) and the medium’s acoustic properties (density ρ and speed of sound c). This value represents the intensity when the ultrasound is actively firing a pulse. Next, this value is multiplied by the stimulus’s total duty cycle (DC_total). The DC_total is a product of two nested cycles: the pulse DC, which is the ratio of the PD to the PRI, and the burst DC, which is the ratio of the PTD to the ISI. Combining these components, the final I_spta value represents the intensity averaged over the entire on-and-off cycle. The formula is below:

$$I_{\mathrm{spta}}=I_{\mathrm{sppa}}*{DC}_{\mathrm{total}}=\frac{P_0^2}{2\mathrm{\rho }\mathrm{c}}*\left(\frac{PD}{PRI}\right)*\left(\frac{PTD}{ISI}\right)$$ 1

Measurement of brain temperature

Mice were anesthetized with 5% isoflurane in an induction chamber and maintained at 2% isoflurane via a facemask. They were then immobilized in a stereotaxic frame (RWD Life Science, Shenzhen, China). The hair on the mouse’s head was shaved, and was disinfected with iodine and alcohol. A small incision was made, the skull was cleaned with hydrogen peroxide, the anterior fontanelle was located, the height of the fontanelle was measured, and the RSC was found (AP: −1.58 mm; ML: 0 mm; dorsal/ventral (DV): −2 mm) according to specific coordinates. After localizing the brain and performing stereoscopic surgery, the cranial drill was used to drill into the RSC carefully. A temperature probe was then inserted into the center of the RSC to measure the brain temperature in mice using a flexible probe (Physitemp, Cat# IT-24P) and machine (Physitemp, Cat# BAT-12R). The skin was then carefully sutured, and all mice were kept in a thermostatic device until they fully recovered before being returned to their home cages.

Electrophysiological recordings

Patch-clamp recordings in HEK-293T cells

Human embryonic kidney cells (HEK-293T, Oricell®, Cat# H4-1401) were maintained in DMEM (Corning Cellgro, Cat# 10-017-CVA) supplemented with 5% fetal bovine serum (Excel. Bio, Cat# FCS500), 5% F12 (Gibco^TM, Cat# 11765-054) and 1% penicillin/streptomycin (Gibco^TM, Cat#15140122) at 37 °C and 5.0% CO₂. Cells were passaged using 0.25% trypsin (Gibco^TM, Cat#25200056) every 2–3 days. pcDNA 3.1-TRPC4-mCherry plasmid (GuanNan. Co, Ltd, Hangzhou, China) was transfected using Lipofectamine^TM 3000 (Invitrogen^TM, Cat# L3000075) according to the manufacturer’s instructions. After 12–14 h, fluorescent cells were selected for recording. Inside-out patch-clamp recordings were performed with a HEKA EPC-10 amplifier and PatchMaster software, using fire-polished borosilicate pipettes (~6 MΩ resistance). Bath and pipette solutions were standard for TRPC recordings. Voltage protocol: 0 mV holding, stepped to −60 mV for 200 ms, returned to 0 mV for 50 ms. Currents were sampled at 10 kHz and filtered at 2.9 kHz at room temperature (22 ± 1 °C). ML204 (10 μM final, MCE, Cat# HY-12949) was applied via the bath. tFUS (0.74 MHz, 214 kPa at 25% amplitude) was delivered every 3 s for 5 min directly to the recording chamber. We used 25% of the maximum amplitude because this was the optimal intensity for achieving and maintaining a stable seal with the cell membrane. At other amplitudes, we found the cells were prone to either lysing easily or detaching from the pipette. This 25% amplitude allowed us to clamp the cells and obtain reliable recordings successfully.

Whole-cell patch-clamp recordings in mouse brain slices

Mice were deeply anesthetized with isoflurane (5% induction in oxygen). Depth of anesthesia was confirmed by the absence of the pedal withdrawal reflex, after which mice were sacrificed by decapitation. After decapitation, brains were rapidly dissected from wild-type C57BL/6J mice and cut into 300-µm-thick coronal slices on ice. The slices were then transferred to and continuously submerged in oxygenated (95% O₂/5% CO₂) artificial cerebrospinal fluid (aCSF) (92 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 2 mM MgSO₄, 30 mM NaHCO₃, 1.25 mM NaH₂PO₄, and 25 mM glucose), and allowed to recover at room temperature for 1 h. Whole-cell patch-clamp recordings were performed exclusively on the somata of pyramidal neurons located in the RSC. Under differential-interference-contrast microscopy, baseline action potentials were first recorded; subsequently, the TRPC4/5 agonist Englerin A (200 nM, MCE, Cat# HY-133168) was added, and the slices were incubated for 20 min before a second round of action-potential recording. Recording pipettes (4–6 MΩ) were filled with an internal solution (124 mM K-gluconate, 5 mM NaCl, 1 mM MgCl₂, 0.2 mM EGTA, 10 mM HEPES, 2 mM Mg-ATP, 0.1 mM Na₃-GTP, and 10 mM phosphocreatine disodium (pH adjusted to 7.2 with KOH)). After membrane rupture, the initial access resistance ranged from 15 to 30 MΩ and was monitored throughout the experiment. Data were excluded if the access resistance changed by >15%. All action-potential signals were filtered at 1 kHz and digitized at 10 kHz.

Calcium imaging of living cells

The plasmids pcDNA3.1-TRPC4(NM_016984)-mCherry and pcDNA3.1-TRPC5(NM_009428)-EGFP were custom-constructed by GuanNan Biotechnology Co., Ltd (Hangzhou, China). The full sequences and maps of these plasmids are available from the corresponding author upon reasonable request. pcDNA3.1-TRPC4-mCherry and pcDNA3.1-TRPC5-EGFP plasmids were transfected into HEK-293T using Lipofectamine^TM 3000 transfection reagent (Invitrogen^TM, Cat# L3000075). Cells were plated on 3.5 cm glass-bottom dishes. After washing with calcium imaging buffer (CIB), cells were loaded with 2 μM Cal-630 (Shanghai Maokang Biotechnology, Cat# MX4536-500UG) for 30 min at 37 °C. Imaging was performed on an Olympus FV3000 confocal microscope, and data were analyzed in Fiji/ImageJ.

Electrode implantation and surgery

Mice were anesthetized with isoflurane (5% induction, 2% maintenance) and placed in a stereotaxic frame (RWD Life Science, Shenzhen, China). Jewerler’s screws were implanted to anchor a combined electrode-ultrasound pedestal. Due to spatial constraints of the ultrasound stimulator, 8-channel electrodes were implanted at a 30° angle targeting the RSC (AP: −1.58 mm; ML: ±0.2 mm; DV: −2.0 mm). After 4–7 days of recovery, neuronal activity was recorded at 30 kHz (bandpass 1–5 kHz) using the NeuroStudio system (BrainTech Inc). Single units were sorted offline with OfflineSorter (Plexon, USA) and analyzed in MATLAB R2020b (MathWorks). Neurons were classified based on a 20% threshold relative to baseline firing rate: as increasing if the firing rate increased by >20% (P < 0.05, paired t-test), as decreasing if it decreased by >20% (P < 0.05), and as no change if the change was within ±20% or not significant (P ≥ 0.05). The neuronal classification was based on an unsupervised clustering approach using Euclidean distances between all cell pairs in a two-dimensional space defined by half-wave width and baseline firing rate.

Neuropathic pain model

The common peroneal nerve (CPN) ligation model was established under pentobarbital anesthesia (50 mg/kg, i.p.). A small incision was made lateral to the left knee, the peroneal muscles were separated, and the CPN was gently isolated and ligated with 5-0 silk suture. Sham animals underwent identical procedures without ligation. Muscle and skin were sutured, and the animal recovered postoperatively.

von Frey behavioral assay

Mechanical sensitivity was assessed using the up-down method with calibrated von Frey filaments applied to the plantar surface of the left hindpaw⁷⁰. Positive responses included brisk withdrawal, flinching, or licking. The 50% paw withdrawal threshold (PWT) was calculated according to the standard formula:

$$50\%\mathrm{g\; threshold}={10}^{\left(X_f+k\delta \right)}/10000$$ 2

where X_f is the value (in log units) of the final von Frey hair used; k is the tabular value for a pattern of positive/negative responses, and δ is the mean difference (in log units) between stimuli. tFUS effects were evaluated by applying three 2-min stimulation trains separated by 5 min, with PWT measured immediately after each train.

Hargreaves test

Thermal paw withdrawal latency (TWL) was measured using the Hargreaves apparatus (Ugo Basile, Gemonio, Italy, Cat# 37370-001). Mice were habituated to the glass platform, and an infrared heat source was focused on the left hind paw. Cut-off time was 20 s. Measurements were repeated at least three times, and tFUS effects were tested using the same 2-min train protocol as above.

Open-field test

Locomotor activity and anxiety-like behavior were evaluated in an open-field test (OFT). Animals were allowed to explore freely for 5 min. In the tFUS group, ultrasound stimulation was delivered for 2 min, starting 1 min into the test session. Total distance traveled and time spent in the center zone were analyzed offline using ANY-maze software (Stoelting, Illinois, USA).

Elevated plus-maze test

Anxiety-related behavior was assessed using an elevated plus-maze (EPM) apparatus consisting of two open arms (30 × 5 cm, no wall) and two enclosed arms (30 × 5 cm, with 15 cm-high walls), elevated 60 cm above the floor. Each mouse was placed individually in the center of the maze and allowed to explore for 5 min. In the tFUS group, ultrasound was applied for 2 min beginning at the 1-min mark. Movement trajectories were recorded and quantified using ANY-maze software (Stoelting, Illinois, USA).

Novel object recognition

Recognition memory was tested in a 45 × 45 cm plastic arena. Mice were first habituated to the empty arena for 10 min. Two identical objects were then placed symmetrically in the arena (10 cm from the walls), and the mouse was allowed to explore for 5 min, during which tFUS was delivered in the second minute for the stimulated group. After a 24-h retention interval, the mouse was returned to the arena where one familiar object had been replaced with a novel object of a different shape and color. Exploration time for each object was recorded over another 5-min session using ANY-maze software. The recognition index was calculated as the percentage of time spent exploring the novel object relative to the familiar object, divided by total exploration time.

Accelerating rotarod

Habituation was repeated every day for 30 min just before the session. Training: At 4 rpm/min, the mouse was placed on a rotating rod and accelerated to the preset endpoint. The initial speed was 4 rpm/min, and the acceleration was 20 rpm/min² (acceleration time was 120 s) until the maximum speed of 40 rpm/min was reached. Mice were trained to continuously move on a spinning stick for 5 min, with the speed gradually accelerated from 0 rpm/min to 40 rpm/min over the first 2 min and maintained at 40 rpm/min for the last 3 min. Mice were trained for three days, 3–5 times a day, with a minimum interval of 10 min between sessions.

Formal testing was performed 24 h after the training was completed. Testing was set as: the back one-third of the mouse’s tail and placed on a stationary rotating rod, facing away from the direction of the rod’s rotation. One tested mouse was placed in each channel. Time was counted simultaneously after the spinning stick was started. The initial speed was 4 rpm/min, and the acceleration rate was 20 rpm/min² (acceleration time was 120 s) until the maximum speed of 40 rpm/min was reached. The speed was gradually accelerated to 40 rpm/min over the first 2 min and was maintained at this rate for the last 3 min. Each mouse was tested 3–5 times, with a minimum interval of 10 min between each test. The latency period of mice falling off the rotating rod (the time the mice stay on the rotating rod) was recorded, and the maximum score of 300 s was recorded if mice completed the entire testing cycle.

Gait analysis

Mice were habituated to the experimental environment for one week to reduce the interference of stress response on gait experiment results. Training was set as follows: mice were placed in the experimental setup (SansBio, Cat# SA-114) with red and green lights turned on for adaptation. Mice were trained three times a day for 3 min each, for 3 consecutive days, until all animals were able to walk on the track without stopping under unstimulated conditions. Formal testing was conducted at least 24 h after the training session was completed. Testing was performed as follows: the mouse was placed at the entrance of the treadmill and allowed to pass through continuously without interruption. The gait video was recorded using a high-frequency camera. Each mouse was tested once, and the calculation was based on the number of successful movements, which included three to five complete movements. The quantified parameters were averaged.

Stereotaxic injection

For virus delivery, adult mice (2–3 months, 22–30 × g) were injected with pentobarbital sodium (50 mg/kg, i.p.) and fixed in a stereotaxic frame equipped with a heating pad. The scalp was shaved, disinfected with iodine and 75% ethanol, and a small craniotomy was performed after cleaning the skull with 30% hydrogen peroxide. Bregma was identified, and coordinates for the RSC were determined (AP: −1.58 mm; ML: ±0.2 mm; DV: -1.2 mm). Bilateral injection (≦ 1 μL per side) was performed at 0.05 μL/min using a microinjection pump. The needle remained in place for 10 min post-injection. The incision was sutured, and mice recovered in a heated chamber before returning to their home cages.

Targeted recombination in the active population test

To selectively target and inhibit Egr1⁺ cells activated by tFUS, we employed a targeted recombination in the active populations (TRAP)³⁰ strategy. Egr1 promoter-driven ER^T2CreER^T2 was used to link Cre activity to ultrasound-induced Egr1 expression⁷⁰. Chemogenetic inhibition was achieved by co-expression of DIO-hM4Di. Virus mixtures were injected into the RSC (AP: −1.58 mm; ML: ±0.2 mm; DV: −1.2 mm). Three weeks after injection, mice received 2 min of tFUS stimulation. Thirty minutes later, 4-hydroxytamoxifen (4-OHT, 25 mg/kg, i.p., Med Chem Express, Cat# HY-16950) was administered to induce nuclear translocation of Cre. Three days after TRAP labeling, clozapine nitrogen oxide (CNO, 2 mg/kg, i.p. Med Chem Express, Cat# HY-17366) was injected 30 min before behavioral testing to inhibit the targeted neurons.

Real-time quantitative fluorescence PCR experiment

Primer sequences for β-actin, Egr1, c-Fos, and Trpc5 were designed based on GenBank entries (Supplementary Table 1) and synthesized by Hangzhou Youkang Biological Co., Ltd. Following tFUS, mice were anesthetized with pentobarbital sodium (50 mg/kg, i.p.) and decapitated on ice. The RSC was rapidly microdissected under a stereomicroscope. For shRNA validation, WT mice received bilateral RSC injections of CaMKIIα-EGFP-TRPC5-shRNA (250 nL per hemisphere). After 3 weeks of viral expression, brains were dissected and cut into 300-µm-thick slices on ice. Slices were maintained in oxygenated (95% O₂/5% CO₂) aCSF (92 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 2 mM MgSO₄, 30 mM NaHCO₃, 1.25 mM NaH₂PO₄, and 25 mM glucose). Virus-infected regions were delineated and collected for fluorescence microscopy, followed by quantification of Trpc5 mRNA levels. Total RNA was extracted using the TransZol Up Plus RNA kit (TransGen Biotech, Cat# er501-01). A total RNA of 1.0 μg was reverse-transcribed using HiScript®Ⅱ Q RT SuperMix (Vazyme^TM, Cat# R223). Real-time PCR was performed with the cycling conditions listed in Supplementary Table 2. Relative expression levels were quantified by the 2^−ΔΔCt method, with β-actin as the internal reference. ΔCt was determined according to the formula:

$$\mathrm{\Delta }{C}_t=C_{t,target}-C_{t,\beta -actin}$$ 3

Immunofluorescent staining

Mice were deeply anesthetized with pentobarbital sodium (50 mg/kg, i.p.) and transcardially perfused with ice-cold phosphate-buffered saline (PBS) followed 4% paraformaldehyde (PFA). Brains were post-fixed in 4% PFA overnight at 4 °C, cryoprotected in graded sucrose solutions (15% then 30%), and sectioned coronally at 30 μm on a cryostat. Free-floating sections containing the RSC were blocked with 0.3% Triton ×-100 in donkey serum for 1 h at room temperature, and then incubated overnight at 4 °C with primary antibodies (Egr1, 1:1000, Cell Signaling Technology^TM, Cat# 4154, Clone name 44D5; Lot 6; c-Fos, 1:1000, Synaptic Systems, Cat# 226308, Clone name Gp108B5; Lot 1–14; CaMKIIα, 1:200, Cell Signaling Technology^TM, Cat# 50049, Clone name 6G9; Lot 4; Anti-GAD65 + GAD67, 1:100, Abcam, Cat# ab183999, Clone name EPR19366; Lot 1071027-6; Anti-TRPC5, 1:500, NeuroMab, Cat# 75–104, Lot 455-9JD-55; Anti-TRPC4, 1:200, Alomone labs, Cat# ACC-119, Lot ACC119A0102). After washing, sections were incubated with fluorescent secondary antibodies for 1 h at room temperature in the dark, counterstained with DAPI, and mounted (Abcam, Cat# ab104139). Images were acquired on an Olympus FV3000 confocal microscope and analyzed using Fiji/ImageJ. Regions of interest were selected according to stereotaxic coordinates corresponding to the tFUS target.

Nissl staining

After rinsing in distilled water for 2–5 min, the frozen brain slices were incubated in Nissl staining solution (Sangon Biotech, Cat# E607316) for 10–30 min at room temperature. The sections were then washed twice in distilled water and visualized using a light microscope.

TUNEL assay

Following the instructions of the TUNEL apoptosis detection kit (Elabscience, Cat# E-CK-A324) to assess cell apoptosis, slices were washed with PBS and then fixed with 4% paraformaldehyde (diluted in PBS) for 20 min at room temperature. After washing twice with PBS, slices were treated with 0.2% Triton ×-100 (prepared in PBS) for 20 min at room temperature to make the cells transparent, then washed twice with PBS. Slices were then processed with the TUNEL reaction according to the manufacturer’s instructions. Finally, the number of TUNEL-positive cells was recorded using a fluorescence microscope.

PinpoRNA^TM multiplex fluorescent RNA in situ hybridization

The mRNA in situ hybridization (ISH) was performed manually using a PinpoRNA multiplex Fluorescent RNA ISH kit (PIT0002, GD Pinpoease Biotech Co., Ltd., Guangdong, China). Short probes complementary to the target RNA sequence were designed using patented algorithms (China patent number ZL202110581853.9). The tissue was fixed with 4% PFA, and endogenous peroxidase was inhibited with Pre-A solution at room temperature. Protease treatment was applied to expose the target RNA molecules, followed by probe hybridization at 40 °C for 2 h. Signal amplification was carried out through reactions 1, 2, and 3, with HRP molecules added in reaction 3 (China patent number ZL202110575231.5). A tyramide fluorescent substrate was used in the final step to label the target RNA fluorescently using the tyramide signal amplification assay.

Duolink® proximity ligation assay

One drop of Duolink® Blocking Solution (Sigma-Aldrich, Cat# DUO2007) was added to each sample, and the slides were incubated in a humidity chamber at 37 °C for 60 min. Primary antibody Anti-TRPC4 (Alomone Labs, Cat# ACC-119, Lot ACC119A0102) was diluted to 1:50, and Anti-TRPC5 (NeuroMab, Cat# 75–104, Lot 455-9JD-55) was diluted to 1:100 in the Duolink® Antibody Diluent (Sigma-Aldrich, Cat# DUO82008). For Duolink® PLA Probe incubation, the Duolink® In Situ PLA Probe Anti-Rabbit PLUS (Sigma-Aldrich, Cat# DUO92002) and Anti-Mouse MINUS (Sigma-Aldrich, Cat# DUO92004) probes were diluted 1:5 in the antibody diluent. The primary antibody solution was tapped off from the slides, and then the slides were washed twice for 5 min each in 1× Wash Buffer A (Sigma-Aldrich, Cat# DUO82046) at room temperature. The wash buffer was tapped off, and the PLA probe solution was applied to the slides in a humidity chamber for 1 h at 37 °C. The probe was tapped off, and the ligase was added to the samples in a humidity chamber for 1 h at 37 °C. The ligation solution was tapped off from the brain slides, and the slides were washed twice for 5 min each in 1× Wash Buffer A at room temperature. After that, Polymerase was added to the 1× Amplification buffer, and the slides were incubated in a humidity chamber at 37 °C for 1 h. For the final washes, the amplification solution was tapped off from the slides. Then, the slides were washed twice for 10 min each in 1× Wash Buffer B (Sigma-Aldrich, Cat# DUO82048) at room temperature. Finally, the slides were rinsed in 0.01× Wash Buffer B for 1 min. Excess wash buffer was tapped off from the slides, and the samples were mounted with DAPI-Aqueous (Abcam, Cat# ab104139). Images were obtained by using a confocal microscope (OLYMPUS IX83-FV3000).

Co-Immunoprecipitation (Co-IP)

Mice were deeply anesthetized with isoflurane (5% induction in oxygen). Depth of anesthesia was confirmed by the absence of the pedal withdrawal reflex, after which mice were sacrificed by decapitation. Tissues from the mouse retrosplenial cortex were lysed in ice-cold RIPA buffer supplemented with protease and phosphatase inhibitor cocktails. After centrifugation at 12,000 × g for 5 min, the supernatant was collected. For IP, 5 µg of anti-TRPC5 antibody (NeuroMab, Cat# 75–104, Lot 455-9JD-55) (or Mouse (E7Q5L) Monoclonal Antibody IgG2b Isotype Control, Cell Signaling Technology^TM, Cat# 53484, Clone name E7Q5L, Lot 5) was incubated with the lysates overnight at 4 °C, followed by the addition of Protein A/G magnetic beads overnight. The beads were washed three times with chilled wash buffer, and the bound proteins were eluted by boiling in SDS loading buffer. The samples were then subjected to SDS-PAGE and transferred to PVDF membranes for Western blot analysis using an anti-TRPC4 antibody (1:200, Alomone Labs, Cat# ACC-018, Lot ACC018AN0402), an anti-TRPC5 antibody (1:200, Alomone Labs, Cat# ACC-020, Lot ACC020AN1350), and β-actin (1:1000, Diagbio, Cat# db10001 Lot W0407P).

RNA-seq based on fluorescence-activated cell sorting (FACS)

Mice were anesthetized with isoflurane. Depth of anesthesia was confirmed by the absence of the pedal withdrawal reflex, after which mice were sacrificed by decapitation. RSC tissue was rapidly dissected on ice, minced, and enzymatically in medium containing 3 mg/ml pronase (Sigma, Cat# p6911-1g, 37 °C, 180 rpm, 30 min)⁷⁰. After mechanical dissociation and centrifugation steps, cells were resuspended in PBS and viability assessed by trypan blue exclusion. Egr1-green fluorescent protein (GFP)-positive and -negative populations were sorted using a BD FACS ORP ARIA Ⅱ. Sorted cells were processed for cDNA synthesis and amplification. Libraries were prepared with TruePrepTM Index Kit V3 for Illumina® (Vazyme^TM, Cat# TD203), quality-checked on Agilent Bioanalyzer and Qubit, and sequenced on Illumina HiSeq X-TEN (LC Bio, China, 150 bp paired-end).

Bioinformatics analysis included adapter trimming (Cutadap), quality control (FastQC), alignment to the mouse genome (GRCm38) with HISAT2, transcript assembly with StringTie, and differential expression analysis using the DEGseq method (∣log₂ (fold change)∣> 2 and P_adj < 0.001). The GO (http://www.geneontology.org) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were then performed and graphically visualized in R (version 4.2.1). Cytoscape software version 3.9.1 (https://cytoscape.org/) was used to construct the Gene-GO enrichment term network.

Statistical analysis and reproducibility

In all experiments, data acquisition and analyses were performed blindly. Data were presented as mean ± SEM, and were analyzed using GraphPad Prism 8. Paired, unpaired Student’s t-test, and Tukey’s multiple comparisons test were used to measure differences between the two groups, and comparison between groups was performed by chi-square test (χ² test), Fisher’s exact test; one-way ANOVA, two-way ANOVA, two-way repeated measures ANOVA, and Šídák’s multiple comparisons test were used for comparison between various groups. P < 0.05 was considered statistically significant.

Representative images presented in Figs. 2a, 3d, e, 4d, 6a, 7b, 8a, and 9a and e are from independent experiments with similar results (2a, n = 11 mice; 3d, n = 7 mice per group; 3e, Ctrl (CaMKIIα/Egr1) n = 8 mice, tFUS (CaMKIIα/Egr1) n = 7 mice; Ctrl (GAD65 + GAD67/Egr1) n = 8 mice, tFUS (GAD65 + GAD67/Egr1) n = 9 mice; 4d, n = 8 mice per group; 6a, Ctrl n = 3 mice, tFUS n = 4 mice; 7b, n = 13 mice; 8a, n = 8 mice per group; 9a, n = 4 mice per group; 9e, TRPC4 (Vehicle) n = 7 dishes, TRPC5 (Vehicle) n = 8 dishes, TRPC4 + 5 (Vehicle) n = 7 dishes, TRPC4 (ML204) n = 7 dishes, TRPC5 (ML204) n = 7 dishes, TRPC4 + 5 (ML204) n = 7 dishes). All replicates were biological, derived from different experimental subjects.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

C.W., J.Y., and T.S. contributed equally to this work. C.W. undertook the behavioral tests, qPCR, ISH, and data analysis. J.Y. conducted patch-clamp tests and RNA-seq analysis. T.S. accomplished multi-channel recording experiments and data analysis. W.B.Q. and G.F.L. produced and optimized the ultrasonic neuromodulation system. C.Z. conducted the behavioral tests, qPCR, and immunofluorescence staining. L.L. carried out the fluorescence-activated cell sorting test. L.Z.X. conducted patch-clamp tests. W.X., F.Y., and W.Y. designed experiments, performed data analysis, and approved the draft. W.B.Q., H.R.Z., and X.Y.L. designed experiments, analyzed data, and wrote the paper.

Peer review

Peer review information

Nature Communications thanks Kai Yu and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This study was supported by the National Natural Science Foundation of China (32271042 to X.Y.L., 82101250 to T.S., and 82327805 to W.B.Q.), the National Key Research and Development Program of China (2023YFC2506203 to X.Y.L.), the STI 2030—Major Projects (2021ZD0200401 to H.R.Z. and W.B.Q.), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB0930000 to H.R.Z. and W.B.Q.), the Guangdong Provincial Key Laboratory of Multimodality Non-Invasive Brain-Computer Interfaces (2024B1212010010 to W.B.Q.), and the Shenzhen Science and Technology Program (JSGGZD20220822095602005 to H.R.Z., KCXFZ20230731093959009 to W.B.Q., SGDX20230821092359003 to W.B.Q., and ZDSYS201802061806314 to H.R.Z. and W.B.Q.).

Data availability

The data underpinning the findings of this investigation are accessible in the article and its supplementary materials. Source Data are provided with this paper. All sequencing data generated in this study have been deposited in GSA under accession CRA024354. Schematic figures and illustrations were created by the authors using Microsoft PowerPoint (Fig. 1a, b and Supplementary Fig. 2a, e, i), Adobe Illustrator 2026 (Fig. 2a and Supplementary Fig. 5b), BioRender (Figs. 1i, 3a, b, d, 4a, b, e, 5a, 6f, 7a, i, 7i and 10 and Supplementary Figs. 1b, d, 3a, b, d, and 4a, b) and EdrawMax (Fig. 6c). All schematics are the authors’ original work and remain the intellectual property of the authors. For the illustration produced with EdrawMax, the official support team has expressly confirmed that self-designed figures used for non-commercial academic manuscript submission are fully owned by the authors and do not require any third-party copyright authorization. Source data are provided with this paper.

Code availability

The analysis code is publicly available on GitHub (https://github.com/awnsjjj/TRPC4-TRPC5-are-critical-for-neuronal-modulation-by-tFUS-in-retrosplenial-cortex-in-male-mice) and archived on Zenodo (10.5281/zenodo.15128201). Executable compute capsule on CodeOcean, 10.24433/CO.1809197.v2.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74779-2.