Neuron-targeting piezoelectric microneedles disrupt pro-tumorigenic neuron-immune crosstalk and restore anti-tumor immunity in melanoma

Abstract

Neuroimmunology has garnered significant attention due to its role in immune regulation, particularly in cancer, where infiltrating neurons can influence antigen presentation, T-cell activation, and cancer metastasis, ultimately leading to an inadequate immune response. Here, we integrate manganese-doped titanium-based metal-organic framework (MOF) piezoelectric materials (MT), coated with neuron-derived membranes from dorsal root ganglia, into microneedles (MN) to create a piezoelectric microneedle (MT MN) patch designed to disrupt neuron-immune crosstalk in melanoma. A single administration of MT via microneedle patch stably deposits the MT at the melanoma site in female mice to accelerate the nociceptor neurons targeting. Upon moderate ultrasound stimulation, the MT facilitates the internalization of TRPV1 and activates the cGAS-STING pathway, resulting in the reduction of Ca²⁺ influx in nociceptor neurons. This ultimately limits the production of calcitonin gene-related peptide (CGRP) and substance P (SP). Consequently, to rescue the tumor immune microenvironment damaged by infiltrated neurons, MT MN is utilized to inhibit the growth and infiltration of nociceptor neurons, highlighting a promising manner for interfering neuron-immune crosstalk in melanoma to enhance cancer immunotherapy.

Activation of nocireceptors can contribute to immunosuppression in cancer. Here the authors report the design and characterization of nociceptor neuron-targeting piezoelectric microneedles, showing disruption of pro-tumorigenic neuron-immune crosstalk and reinvigoration of anti-tumor immunity in melanoma.

Introduction

Neuroimmunology, a rapidly advancing field in neuroscience, investigates the cross-talk between the nervous system and the immune system^1–4. Numerous recent studies have demonstrated that neuron-immune communication plays a significant role in various diseases, particularly cancer^2,5–8. Balood et al.⁹ discovered that melanoma can activate nociceptor neurons to secrete calcitonin gene-related peptide (CGRP), a neurotransmitter produced by nociceptor neurons, which directly increases the exhaustion of cytotoxic CD8⁺ T cells and impairs cancer immunosurveillance. Furthermore, the released CGRP has been shown to influence the maturation of dendritic cells (DCs), as evidenced by decreased expression levels of the co-stimulatory proteins CD80 and CD86¹⁰. Simultaneously, malignant cells trigger calcium activity in neurons, leading to the release of substance P (SP) and promoting metastasis¹¹. In summary, infiltrating neurons in tumor lesions play a vital role in antigen presentation, T-cell activation, and cancer metastasis, making them a potential target for cancer immunotherapy. Nevertheless, despite the persistent activation of nociceptors in cancer cells contributing to the tumor immunosuppressive microenvironment, a clinically validated and effective neural intervention in malignant lesions for cancer immunotherapy remains elusive. Altogether, it is extremely essential to develop strategies to modulate neuron-immune crosstalk at tumor sites in order to reverse the immunosuppressive microenvironment.

Melanoma, a malignant form of skin cancer, is characterized by its high invasiveness and immunosuppressive properties, which are closely associated with the infiltration of nociceptor neurons^9,12–14. Regulating nerve cells in melanoma presents a promising yet largely unexplored strategy for achieving effective immunotherapy. Transient receptor potential vanilloid-1 (TRPV1) is a nonselective cation channel highly expressed in the peripheral nervous system, primarily responsible for Ca²⁺ influx, and has garnered significant attention in neuroscience^15–18. Upon activation by low pH^19–21, heat^22–24, inflammatory cytokines²⁵, prostaglandins^26–28, and other stimuli²⁹, TRPV1 permits the influx of extracellular Ca²⁺ into cells, resulting in neuronal activation and the release of neuropeptides such as CGRP and SP^30–32. Therefore, TRPV1 serves as a potential target for modulating neuron-immune crosstalk. Additionally, recent studies have shown that activation of the STING pathway induces the release of type I interferons (IFN-I), which suppresses Ca²⁺ influx in nociceptor neurons and downregulates neurotransmitter release^33,34. Consequently, reducing the expression of TRPV1 channel proteins on the surface of nociceptor neurons in cancer, while simultaneously activating the STING pathway to inhibit neuronal activity, represents a promising approach for neuron-immune crosstalk dysregulation in melanoma. However, such approaches remain exceedingly rare.

In this work, to effectively interfere with the neuron-immune interaction in cancer, a piezoelectric microneedle patch is developed by loading manganese-doped titanium-based metal-organic framework (MOF) piezoelectric materials (MT) into the microneedle array, thereby reestablishing immunosurveillance in melanoma (Fig. 1). The MT particles are coated with dorsal root ganglia (DRG)-derived neuron membrane fragments to specifically target nociceptor neurons within the tumor. Unlike traditional intravenous injections, which can easily become trapped in non-target nervous tissues, microneedle (MN) patches, featuring an array of microneedles, provide a minimally invasive method for transdermal drug delivery and demonstrate excellent biocompatibility for melanoma treatment^35–38. Simultaneously, piezoelectric nanomaterials can generate persistent nanoscale local electric fields through continuous ultrasound-triggered pressure, offering significant potential for promoting receptor internalization^39–41 without requiring open wounds, wires, or power supplies, which may cause infections, inflammation, and localized pain^42–44. Therefore, after transdermal delivery of MT piezoelectric nanosheet into tumor tissue and targeted nociceptors, the TRPV1 receptors on the cell membrane are internalized in response to moderate ultrasound stimulation. Subsequently, the piezoelectric materials generate ROS that damage mitochondria, leading to the release of mitochondrial DNA (mtDNA). Concurrently, doped Mn²⁺ ions are released within the nociceptors. Together, these factors activate the STING pathway, resulting in the release of IFN-I^45–48. Following these synergistic effects, the influx of intracellular calcium ions is reduced, leading to decreased nociceptor impulses and diminished secretion of CGRP and SP. This cascade ultimately interferes with nerve activity, thereby enhancing the efficacy of anti-tumor immunotherapy in melanoma.

Fig. 1. Schematic illustration of MT MN patch interference neuron-immune cross-talk for the enhancement of cancer immunotherapy.

a The piezoelectric MOF was synthesized using the hydrothermal method. After coating with a nerve cell membrane, it was encapsulated and formed into a piezoelectric microneedle. b After the piezoelectric microneedle penetrates the tumor area, it releases the encapsulated MT. Under ultrasound stimulation, this promotes the endocytosis of TRPV1, activates STING, inhibits the activity of pain receptors, disrupts neural-immune cross-talk, and enhances the anti-tumor immune response.

Results

The synthesis and characterization of MT

The MT nanoparticles were synthesized by coating the Mn-doped Ti-MOF with the neuron membrane (Fig. 1a). The Mn-doped Ti-MOF nanoparticles were synthesized by dynamic homogenous reaction, as previously reported. Scanning electron microscopy (SEM) and Transmission electron microscopy (TEM) revealed that the MOF were square in shaped (Fig. 2a, b and Supplementary Fig. 1). The element distribution of the MOF was also evaluated by elemental mapping via scanning transmission electron microscopy (STEM). The results confirmed the present and uniform distribution of Ti, Mn, and N in the MOF (Fig. 2c). The surface element composition and the metal valence states were further evaluated by the X-ray photoelectron spectroscopy (XPS). From the spectrum of Ti 2p (Fig. 2d), the two peaks at the binding energy of 464.45 eV and 458.65 eV correspond to Ti⁴⁺ from the titanium-oxo cluster. In the spectrum of Mn 2p (Fig. 2e), the peak of Mn²⁺ can be divided into three peaks at the binding energy of 653.78 eV, 647.83 eV, and 641.67 eV. Furthermore, the crystal structure of the Ti-MOF doped with and without Mn was evaluated trough X-ray diffraction (XRD). And the Mn-doped Ti-MOF showed the same tetragonal structure as Ti-MOF, demonstrating that the dope of Mn not change the structure of Ti-MOF (Fig. 2f). Additionally, the Raman pattern also proved the same structure of Mn/Ti-MOF and Ti-MOF (Fig. 2g). Simultaneously, Mn/Ti-MOF and Ti-MOF were characterized by Fourier Transform Infrared spectroscopy (FT-IR). The peak at 771, 638, and 425 cm⁻¹ were due to the stretching vibration of the O-Ti-O (Supplementary Fig. 2). The vibrational peak of the C-N in aromatic amines at 1257 and 1377 cm⁻¹ (Supplementary Fig. 2). Meanwhile, the peak at 1642 cm⁻¹ proved the exist of the C=O in carboxyl group (Supplementary Fig. 2). And The peak at 3440 cm⁻¹ were due to the stretching vibration of the-NH₂(Supplementary Fig. 2). The same FT-IR pattern and the exist of different groups proved the successful synthesis of Mn/Ti-MOF and the same structure of Mn/Ti-MOF and Ti-MOF. After the successfully synthesis of the Mn/Ti-MOF, the next step is coat the neuron cell membrane fragments onto the Mn/Ti-MOF. The neuron cells were collected from the DRG tissue and the cell membrane were extracted. SEM and TEM were used to observe whether the cell membrane is encapsulated on the surface of the MOF. And the results proved that the cell membrane successfully adhered to the surface of the material (Fig. 2h and Supplementary Fig. 3). After the coating of membrane, the ζ-potential was from −17.43 mV (Mn/Ti-MOF) to −30.8 mV (MT), which approach to the value of mere cell membrane (−37.3 mV) (Supplementary Fig. 4a). Concurrently, the results of the dynamic light scattering (DLS) indicated that the hydrodynamic diameter of MT slightly increased after the membrane coating (Supplementary Fig. 4b). Subsequently, the protein composition of MT was analyzed using sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), revealing a profile consistent with that of the corresponding cancer cell membranes (Fig. 2i). The results demonstrated that the characteristic proteins of neuron cell were reserved on the cell membrane. Afterward, the stability of the MT was measured by dispersing the MT nanoparticles into 10%FBS solution and DMEM at different time intervals and evaluating their change of diameter (Supplementary Fig. 5). The outcomes confirmed that MT have good stability. Subsequently, the MT nanoparticles were dispersed in solutions of different pH values. The morphology of MT was characterized by SEM (Supplementary Fig. 6a), and the release kinetics of Mn ions from MT was evaluated (Supplementary Fig. 6b). The results demonstrated that MT remained stable in the pH range of 6-7, while it began to degrade and release Mn ions at pH below 4. Furthermore, the piezoelectric effect occurs when external stress is applied to a material, causing the separation of electric charges within the material and generating a local electric field. In this situation, the piezoelectric force microscopy (PFM), a scan-probe technique, was performed to confirm the piezoelectricity of MT. Under the continuous voltage stimulation, the amplitude variation of MT showed a ferroelectric butterfly hysteresis loop, indicating that the strain is constantly changing (Fig. 2j). The presence of a phase voltage hysteresis loop demonstrated the excellent piezoelectricity of the MT (Fig. 2k).

Fig. 2. Characterization of piezoelectric nanoparticle MT.

Typical transmission electron microscopy (a) and scanning electron microscopy (b) image of Mn/Ti-MOF. Scale bars, 200 nm. The representative images of a and b are shown from n = 3 biologically independent experiments. c EDS element mapping of Mn/Ti-MOF. Scale bars, 200 nm. The representative images of c are shown from n = 3 biologically independent experiments. Zn 2p (d) and Mn 2p (e) XPS spectra as well as peak fitting curves. The representative images of d and e are shown from n = 3 biologically independent experiments. f XRD pattern of Mn/Ti-MOF and Ti-MOF. The representative images of f are shown from n = 3 biologically independent experiments. g Raman spectra of Mn/Ti-MOF and Ti-MOF. The representative images of g are shown from n = 3 biologically independent experiments. h The scanning electron microscopy image of Mn/Ti-MOF covered by cell membrane (MT). Scale bars, 200 nm. The representative images of h are shown from n = 3 biologically independent experiments. i SDS-PAGE protein assay of Mn/Ti-MOF, MT nanosheet, and DRG cell membranes (Mem). The representative images of i are shown from n = 3 biologically independent experiments. The amplitude image (j) and phase image (k) of MT nanosheet observed by PFM. The representative images of j and k are shown from n = 3 biologically independent experiments. Source data are provided as a Source Data file.

The design and fabrication of MT MN

As previously mentioned, the MT piezoelectric nanoparticles were successfully synthesized. To adapt them for melanoma treatment and enhance their biological safety, we encapsulated the MT nanoparticles within soluble microneedle patches. In this study, the microneedles were fabricated using the traditional mold method. Hyaluronic acid served as the base material for preparing the microneedles, with nanoparticles incorporated into the structure. First, a 10 mg/mL MT solution was mixed with hyaluronic acid and dispersed into the PNMS mold. The mold was then placed under vacuum for 3 min, followed by centrifugation to remove air bubbles; this process was repeated two to three times. After filling the microneedle cavities, excess hyaluronic acid solution containing MT was removed, and the microneedles were dried. Then, add the hyaluronic acid solution without MT as the backing and let it dry overnight (Supplementary Fig. 7a). The microneedle patch array consists of 10 × 10 needles, measuring 8.7 × 8.7 mm (Supplementary Fig. 7b). Furthermore, the morphology of the microneedles was observed using a microscope (Fig. 3a, b). Each microneedle is conical in shape, with a distance of 600 µm between adjacent needles. The base of each microneedle is square, measuring 260 × 260 µm, and the length of each microneedle is 500 µm. Additionally, the morphology was also examined using SEM (Fig. 3c). During the synthesis of the MT microneedle patch, MT nanoparticles were loaded into the microneedles. Elemental analysis of Ti and Mn distribution in the MT microneedles by EDS revealed that MT is primarily localized within the microneedles (Fig. 3d). Previous research has demonstrated that the hardness of microneedles needs to be in the range of 0.1 N/needle to 0.2 N/needle, which is essential for effective penetration of the stratum corneum. The hardness value of the MT MN patch was 0.15 N/needle, indicating that the mechanical strength of the MT MN patch is sufficient for skin penetration. Simultaneously, we used agarose gel to simulate skin tissue in order to observe the penetration ability of the microneedles. The results confirmed that the microneedles were indeed capable of effectively penetrating the agarose gel (Fig. 3f). Furthermore, we used Balb/c mice as a model to investigate the penetration ability of microneedles in living skin tissue, as well as the skin’s self-repair ability after penetration. Figure 3g demonstrates that microneedles can leave small holes on the surface of the mouse’s skin to release the drugs contained within the microneedles. However, after 10–20 min, the microneedle holes on the skin surface disappear, indicating that the microneedles maintain good mechanical strength in animal skin tissue and do not impair the skin’s self-repair ability. Hematoxylin and eosin staining confirmed that the MT MN patch can penetrate the stratum corneum (Fig. 3g). Moreover, no obvious micro-indents can be observed 20 min after the MT MN patch has been removed. Additionally, we assessed the solubility of the microneedles by observing their changes after insertion into the skin of mice (Supplementary Fig. 8). The microneedles dissolved upon penetrating the skin, further demonstrating that MT can be effectively released.

Fig. 3. Characterization of piezoelectric microneedle patches.

Representative bright-field microscopy images showing the front (a) and side (b) views of the MT MN patch. Black arrows indicate the relevant parameters of the microneedles. The representative images of a and b are shown from n = 3 biologically independent experiments. c Morphology of the MT MN patch observed under a scanning electron microscope (SEM). Scale bars, 500 µm. The representative images of c are shown from n = 3 biologically independent experiments. d The corresponding element mapping of Ti and Mn in the MT MN patch characterized by EDS. Scale bars, 500 µm. The representative images of d are shown from n = 3 biologically independent experiments. e Force-displacement curves of the MT MN patch. The representative images of e are shown from n = 3 biologically independent experiments. f 3D reconstruction of images of RhB fluorescence on agarose after application of MT-RhB-loaded microneedles patch. Scale bars, 200 µm. The representative images of f are shown from n = 3 biologically independent experiments. g Representative images of the MT MN patch on the surface of live Balb/c mouse skin. After removal, the needle hole recovered within 10 min. Scale bars, 2 mm. The representative images of g are shown from n = 3 independent mice. h H&E staining images of Balb/c mouse skin with MT MN patch applied (left) and post-removal (right). The representative images of h are shown from n = 3 independent mice. Black arrows indicate the indents on the mouse skin. Scale bars, 100 µm. The representative images of h are shown from n = 3 biologically independent experiments. Source data are provided as a Source Data file.

Activating STING pathway in nerve cell by MT

As previously reported, the autocrine or paracrine STING-mediated IFN-I signaling suppresses nociceptors, leading to a reduction in Ca²⁺ influx. Therefore, we first evaluated the ability of MT to activate the STING pathway in nerve cells. In this situation, we use HT-22 nerve cells as an example to investigate the mechanism by which MT activates STING. Firstly, the in vitro biosafety of MT was evaluated using the methyl thiazolyl tetrazolium (MTT) assay. The survival rate of 3T3 cells remained above 90% when the concentration of MT was within the range of 0-80 µg/ml (Supplementary Fig. 9a). Additionally, ultrasound did not have a significant effect on normal cells (Supplementary Fig. 9b). As a piezoelectric nanomaterial, MT can generate reactive oxygen species (ROS) under ultrasound irradiation. Figure 4a illustrates that intracellular ROS levels were assessed using the DCFH-DA probe, and the results confirmed that MT is capable of producing ROS when exposed to ultrasound. Furthermore, the killing ability of MT nanoparticles against B16 melanoma cells under ultrasound irradiation was investigated (Supplementary Fig. 9c). The results showed that MT nanoparticles exhibited moderate cytotoxicity toward tumor cells at a concentration of 80 μg/mL. Meanwhile, the generated ROS have the potential to disrupt the redox balance within the cell, leading to mitochondrial damage. Therefore, we utilized JC-1, a fluorescent probe for detecting mitochondrial membrane potential, to evaluate the level of mitochondrial damage. The confocal fluorescence images indicate that the control, ultrasound (US), and mitochondrial-targeted (MT) groups exhibit red fluorescence (JC-1 aggregates); in contrast, the MT + US group displays green fluorescence (JC-1 monomers) (Supplementary Fig. 10), demonstrating mitochondrial damage in the MT + US group. During the activation of the STING pathway, damaged mitochondria and/or metal ions, particularly Mn²⁺, can initiate the phosphorylation of the STING protein. Subsequently, the phosphorylated STING (p-STING) induces the generation of phosphorylated IRF3 (p-IRF3) and phosphorylated TBK1 (p-TBK1), which promote the secretion of interferon type I (IFN-I). As demonstrated by Western blot results, the expression levels of p-STING and p-IRF3 were elevated in the MT group (Fig. 4b), confirming the activation of STING. This activation is attributed to the presence of Mn²⁺ doped in the MT. Additionally, compared to the MT group, the expression levels of p-STING and p-IRF3 were further increased in the MT + US group (Fig. 4b), proving the produced ROS damage mitochondria could enhance the activation of STING. Furthermore, immunofluorescence experiments were conducted to characterize the activation of STING. Fluorescence images demonstrate an increased fluorescence intensity of p-TBK1 and p-IRF3 in the MT and MT + US groups (Fig. 4c, d), further supporting the notion that MT activates STING by inducing oxidative stress through ROS generated in the mitochondria, along with the presence of Mn²⁺ doped within it. Subsequently, DRG cells were isolated from mouse dorsal root ganglia for subsequent experiments, among which the proportion of TRPV1⁺ cells reached 92% (Supplementary Fig. 11). The expression level of IFN-I in DRG cells was evaluated using qRT-PCR. Compared to the control and US groups, an increased expression of IFN-I mRNA was observed in the MT + US group (Fig. 4e). Enzyme-Linked Immunosorbent Assay (ELISA) further confirmed the elevated expression of IFN-I in the MT + US group (Fig. 4f), indicating that, compared to the other groups, the activation of the STING pathway was enhanced in the MT + US group. The increased secretion of IFN-I in the MT + US group demonstrates the significant potential of MT in inhibiting nociceptors. IFN-I, produced as a result of STING activation, can inhibit the influx of Ca²⁺.

Fig. 4. Inhibition of HT22 and DRG cell function in vitro by MT piezoelectric nanoparticles.

a Representative fluorescence images of intracellular ROS staining by DCFH-DA. Scale bars, 50 µm. The representative images of a are shown from n = 3 biologically independent experiments. b Western blotting analysis of p-IRF3 and p-STING with different treatments. The representative images of b are shown from n = 3 biologically independent experiments. Typical immunofluorescence images of p-TBK1 (c) and p-IRF3 (d) in HT-22 neurons. Scale bars, 50 µm for (c, d). The representative images of c and d are shown from n = 3 biologically independent experiments. The expression of IFN-β in DRG cells was evaluated by qRT-PCR (e) and ELISA (f). Data are presented as mean ± s.d. for (e, f) (n = 3 biologically independent samples). Representative neurite outgrowth pictures (g) and average neurite length (h) of DRG cells after treatments. Scale bars, 50 µm for (g). Data are presented as mean ± s.d. for (h) (n = 4 biologically independent samples). The representative images of g are shown from n = 3 biologically independent experiments. Western blotting analysis (i) and immunofluorescence images (j) of TRPV1 in DRG cells. Scale bars, 50 µm for (j). The representative images of i and j are shown from n = 3 biologically independent experiments. The Ca²⁺ images (k) and the mean fluorescence intensity (l) of DRG cells after treatments. Scale bars, 50 µm for (k). The representative images of k are shown from n = 3 biologically independent experiments. Data are presented as mean ± s.d. for (l) (n = 3 biologically independent samples). The expression of neurotransmitters CGRP (m) and SP (n) after different treatments. Data are presented as mean ± s.d. for (m, n) (n = 3 biologically independent samples). o Mechanisms by which MT piezoelectric nanoparticles mediated neuron inhibition. Statistical comparisons were performed using one-way analysis of variance (ANOVA), followed by Tukey’s multiple-comparison test. A value of p < 0.05 was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant. Source data are provided as a Source Data file.

Restraining DRG cell activities in vitro

Next, we will assess the neuroinhibitory effects of the piezoelectric material MT. First, we experimentally tested the targeting ability of the neurons in MT (Supplementary Fig. 12). And the MT demonstrated a strong ability to target nerve cells due to the properties of the DRG cell membrane. Melanoma has been shown to induce the progression of nociceptors. Therefore, take DRG cells as an example, we evaluated the effect of B16 melanoma cells on DRG cells. The results indicate that nerve length significantly increased in the B16 co-culture group (Supplementary Fig. 13a, b), demonstrating a promoting effect on DRG cells. Furthermore, to determine the influence of MT on DRG cells, we treated the DRG cells with control, US, MT, and MT + US, and evaluated neurite synapse (Fig. 4g, h). The MT + US group significantly inhibited the growth of neural synapses. Additionally, melanoma can activate nociceptors (Supplementary Fig. 14), resulting in the influx of Ca²⁺ and the secretion of CGRP and SP. TRPV1 is a non-selective cation channel that is highly expressed in nociceptors and can respond to various stimuli, inducing Ca²⁺ influx and leading to the secretion of CGRP and SP. The downregulation of TRPV1 presents a promising approach to inhibit calcium influx, which subsequently decreases the secretion of CGRP and SP. The DRG cells were treated with control, US, MT, and MT + US. Western blot analysis was employed to evaluate the expression of TRPV1. As demonstrated in Fig. 4i, the protein expression levels in the MT + US group were significantly downregulated, indicating that the nanoscale electric field generated on the cell membrane by MT under moderate ultrasound irradiation can trigger the internalization of TRPV1. The immunofluorescence assay of TRPV1 further supports this conclusion (Fig. 4j). Meanwhile, the cell surface biotinylation assay also proved that the MT have the potential to induce the endocytosis of TRPV1 (Supplementary Fig. 15a, b). Pharmacological rescue experiments confirmed that MT induces the downregulation of TRPV1 expression via the endocytosis pathway (Supplementary Fig. 16). In Supplementary Fig. 17, the level of TRPV1 decreased with increasing concentrations of MT, whereas the LDH level showed no significant change. This indicates that the reduction in TRPV1 was not caused by endocytosis of the nanoparticles, but may instead be attributed to the piezoelectric properties of the material. Capsaicin, a small molecule, has been utilized in the field of analgesia. It can activate nociception and promote calcium influx. Therefore, we employed capsaicin to activate DRG cells and further assessed the inhibitory effects of various materials on calcium influx following different treatments. The concentration of calcium ions in the cells was measured using Fluo-4 AM, a fluorescent probe that emits green fluorescence in response to calcium ions. After applying different treatments in the presence of capsaicin, strong green fluorescence was observed in capsaicin and capsaicin + US group (Fig. 4k, l), indicating the influx of Ca²⁺ induced by capsaicin. In contrast, the capsaicin + MT group exhibited relatively weak green fluorescence intensity (Fig. 4k, l). This reduction is attributed to the manganese ions in MT, which activate the STING pathway to promote the secretion of IFN-I and inhibit calcium ion influx (Supplementary Fig. 18). The green fluorescence intensity in the capsaicin + MT + US group was the lowest and comparable to that of the control group, demonstrating that the combined effect of MT + US activating the STING pathway and reducing the expression of TRPV1 significantly inhibited capsaicin-induced calcium ion influx (Fig. 4k, l). Furthermore, after inhibiting the activity of TRPV1 channels with capsazepine, the effect of MT on DRG cells disappeared, and no changes in intracellular Ca ions were observed, demonstrating the specificity of TRPV1 (Supplementary Fig. 19). Consistently, ELISA was employed to assess the secretion of CGRP and SP (Fig. 4m, n). The results indicated that capsaicin enhances the production of CGRP and SP. Furthermore, the capsaicin + MT group exhibited decreased secretion of both CGRP and SP, while the capsaicin + MT + US group demonstrated secretion levels comparable to those of the control group. In conclusion, the activation of the STING pathway in DRG cells by MT, along with its downregulation of TRPV1, highlights its significant neuroinhibitory capacity. This results in a reduction of calcium ion influx and the subsequent generation of CGRP and SP (Fig. 4o). Additionally, the STING inhibitor SN-011 was administered to suppress the STING pathway (Supplementary Fig. 20). The results demonstrate that inhibition of the STING pathway has the potential to reduce the effect of MT.

Regulation of immune cells and cell migration in vitro

As previously described, MT nanoparticles have the potential to inhibit neuronal activity, resulting in decreased secretion of SP and CGRP. CGRP and SP have been shown to suppress immune cell activity by restraining the maturation of DCs and exhausting CD8 + T cells, thereby promoting cancer cell metastasis. Therefore, it is essential to evaluate the impact of MT on nerve cell behavior, as well as their interactions with immune and cancer cells. In this situation, the BMDCs and CD8⁺ T cells were collected from bone marrow and spleen, respectively, and co-cultured with treated DRG cells were then analyzed using flow cytometry (Fig. 5a). Mature DCs play a crucial role in the immune response, including antigen presentation and T cell activation. However, the secretion of CGRP by neurons suppresses DC maturation. Mature DCs are characterized by the expression of CD80 and CD86 on their surface. To evaluate DC maturation, DCs were treated with lipopolysaccharide (LPS) to induce maturation and then co-cultured with differently treated dorsal root ganglion (DRG) cells. The proportions of CD80- and CD86-expressing cells decreased from 39.6% and 39.7% to 26.7% and 28.4%, respectively, in mature DCs co-cultured with untreated and capsaicin-treated DRG cells (Fig. 5b, c), indicating that CGRP secretion inhibits DC activation. Additionally, the percentages of CD80 and CD86 in the capsaicin+MT group increased (Fig. 5b, c), demonstrating that activation of the STING pathway in neurons downregulates CGRP secretion. Furthermore, the capsaicin + MT + US group exhibits the highest expression levels of CD80 and CD86 (Fig. 5b, c), demonstrating that STING activation and TRPV1 downregulation via the piezoelectric effect significantly suppress neuronal activity and promote DCs maturation. As described in the literature, the production of CGRP can impair immunosurveillance in cancer by inducing exhaustion in CD8⁺ T cells. PD-1 and Tim-3 are two membrane proteins highly expressed on the surface of exhausted T cells; therefore, their co-expression was used to evaluate CD8 + T cell exhaustion. The DRG treated with capsaicin alone showed CD8⁺ T cell exhaustion, likely due to activated DRG cells producing CGRP, which inhibits CD8⁺ T cell function (Fig. 5d). In contrast, DRG cells treated with capsaicin + MT exhibited a reduced ability to clear CD8⁺ T cells, as indicated by lower expression levels of PD-1 and Tim-3 (Fig. 5d). In the capsaicin + MT + US group, PD-1 and Tim-3 levels were the lowest and returned to normal levels (Fig. 5d), demonstrating that MT combined with US can effectively inhibit DRG cells and maintain CD8⁺ T cell activity. Additionally, IFN-γ and IL-2 serve as two key effector cytokines for evaluating the functional exhaustion of CD8 + T cells. Therefore, we analyzed the subsets of activated CD8 + T cells expressing IFN-γ and IL-2 (Fig. 5e, f). Upon capsaicin treatment, the percentages of cells positive for IFN-γ and IL-2 decreased from 46.7% to 21.9% and from 81.6% to 20.8%, respectively, suggesting that elevated CGRP expression can promote CD8 + T cell exhaustion. In contrast, the restored expression of IFN-γ and IL-2 in the capsaicin + MT and capsaicin +  MT + US treatment groups demonstrated that activation of the STING pathway and downregulation of TRPV1 can rescue the efficacy of immunotherapy. Furthermore, quantitative analyses of CD80, CD86, PD-1/Tim-3, IFN-γ, and IL-2 are also presented (Fig. 5g–k). These results further suggest that MT can potentially suppress the immunosuppressive function of neurons in vitro. Additionally, the data presented in the heatmap illustrate the ability of MT to reestablishment immunosurveillance by inhibiting the CGRP secretion capacity of nociceptors. (Fig. 5l). Notably, neurons in the tumor microenvironment not only have the ability to impair immunosurveillance but also possess the potential to promote cancer cell metastasis. As described in the literature, the secretion of SP from neurons within cancer lesions can enhance metastasis through the RNA-TLR7 axis. Meanwhile, nociceptors can also facilitate cancer progression via the CGRP-RAMP1 axis. Therefore, the nervous system plays a crucial role in cancer metastasis. To evaluate the ability of MT to suppress cancer metastasis by inhibiting neuronal activity, a cell scratch assay was performed (Fig. 5m). The results demonstrated that the capsaicin + MT + US treatment effectively eliminated the ability of nerve cells to promote tumor metastasis. In general, MT nanoparticles have significant neuron inhibitory abilities, downregulating the secretion of CGRP and SP, which leads to the restoration of immunosurveillance and a decrease in cancer metastasis.

Fig. 5. The in vitro neuro-immune regulatory ability of MT.

a Schematic of the in vitro immune regulation detection process. Flow cytometry analysis of CD80 (b) and CD86 (c) expression on mature BMDCs. Flow cytometry analysis of PD-1/ Tim-3 (d), IFN-γ (e) and IL-2 (f) expression on activated CD8⁺T cells. g, h Quantitative analysis of CD80 (f) and CD86 (g) expression on mature BMDCs. Data are presented as mean ± s.d. for (g, h) (n = 5 biologically independent samples). Quantitative analysis of PD-1/Tim-3 (i), IFN-γ (j) and IL-2 (k) expression on activated CD8⁺T cells. Data are presented as mean±s.d. for (i–k) (n = 5 biologically independent samples). l Heatmap of the expression levels of CD80, CD86, PD-1/Tim-3, IFN-γ and IL-2with different treatments. The representative images of l are shown from n = 3 biologically independent experiments. m Cell scratch experiments conducted on different treatment groups. Scale bars, 50 µm for (m). The representative images of m are shown from n = 3 biologically independent experiments. Statistical comparisons were performed using one-way analysis of variance (ANOVA), followed by Tukey’s multiple-comparison test. A value of p < 0.05 was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant. Source data are provided as a Source Data file.

In vivo biosafety of MT

It has been proved that MT nanoparticle have the potential to inhibit the activity of the nociceptor in vitro. Therefore, the next were evaluated the nerve cell regulate ability and tumor suppression efficiency of MT nanoparticle in vivo. However, the biosafety of MT should be evaluated firstly. We assessment the hemolysis rate of MT in vitro firstly. As shown in Supplementary Fig. 21, the hemolysis rate of MT nanoparticles at concentrations ranging from 0 to 80 µg/mL is less than 1%, demonstrating that the MT nanoparticles are safe for subsequent treatments. Subsequently, the biosafety of MT was evaluated in vivo. Kunming mouse were divided into two groups (Control and MT). The bodyweight of the Kunming was measured and recorded every 4 days (Supplementary Fig. 22). The results demonstrated that the MT nanoparticles have no significant influence on the body weight of mice, illustrating the good biosafety of MT. Immediately, after 30 days, the mouse were sacrificed and the blood and organs were collected for the next biochemical and hematological and Hematoxylin and Eosin (H&E) staining assay. First, a complete blood count (CBC) analysis was conducted on the mice (Supplementary Fig. 23). The results indicated that all blood parameters of the mice treated with MT were within the normal range and showed no significant differences compared to the control group. These findings demonstrate that MT has no significant effect on the blood of mice. Subsequently, the liver and kidney functions of the mice were evaluated by analyzing their serum (Supplementary Fig. 24). The results showed that all relevant numerical indicators were within the normal range, with no significant differences compared to the control group. These findings indicate that MT does not exhibit obvious toxicity to the liver and kidney functions of mice and demonstrate that MT has good biocompatibility. In addition, the five internal organs of the mice were collected for histopathological analysis using H&E-stained sections (Supplementary Fig. 25). The histological data show that the organs of mice treated with MT exhibit no obvious damage, indicating that the material does not cause significant toxicity to the structure of various organs and demonstrating its good biocompatibility. Meanwhile, the in vivo biodistribution of MT tracked after transdermal delivery via microneedles were evaluated (Supplementary Fig. 26). These results demonstrated that MT could accumulate at the tumor site with no obvious off-target effects. In conclusion, MT demonstrates good biological safety, allowing for subsequent tumor treatment experiments to be conducted.

The tumor suppression efficiency of MT MN

Subsequently, the antitumor therapeutic efficacy of MT MN was further evaluated. The B16-bearing C57BL/6 mouse model was established, and different treatments were implemented after 7 days (Fig. 6a). Basic tumor indicators were monitored during the treatment, and tumor tissues were collected on day 14 for correlation analysis. Throughout the therapeutic process, the body weights of mice in the various treatment groups were recorded, and the results indicated that MT MN demonstrated good biosafety and minimized the influence of confounding factors on the therapeutic effect (Fig. 6b). Tumor volume and weight were further measured (Fig. 6c–e and Supplementary Fig. 27). The MT MN + US group exhibited significant tumor inhibition compared to the control group. Importantly, we combined anti-PD-1 therapy with MT MN treatment. By comparing the anti-PD-1 group with the anti-PD-1 + MT MN + US group, it was observed that the latter demonstrated superior tumor inhibition compared to the anti-PD-1 group, highlighting the potential of MT MN to enhance immune checkpoint blockade (ICB) therapy in clinical practice. Additionally, the survival duration of mice in the MT MN and MT MN + US groups showed slight and significant increases, respectively (Fig. 6f), indicating that the inhibition of nociceptors in cancer enhances antitumor therapeutic efficacy. The principle of Hematoxylin and Eosin (H&E) staining involves using hematoxylin and eosin staining solutions to color tissues. Hematoxylin stains chromatin in the nucleus and nucleic acids in the cytoplasm blue or purple, while eosin stains proteins in the cytoplasm and extracellular matrix red or pink. This staining method allows for clear visualization of the cellular structure and composition in tissue samples. Therefore, the collected tumor tissues were examined using H&E stains to assess the extent of tumor damage. As illustrated in Fig. 5g, there was marked injury in the MT MN + US treated group, demonstrating the damage to the tumor caused by MT MN + US. TUNEL staining is a common technique used to detect DNA fragmentation during apoptosis. The principle of this technique is to link labeled dUTP to the 3’-OH ends of fragmented DNA in apoptotic cells using terminal deoxynucleotidyl transferase (TdT), which allows for the visualization of apoptotic cells under a microscope. In apoptotic cells, DNA breaks generate a significant number of 3’-OH ends that can be specifically detected through TUNEL staining, while normal or proliferating cells exhibit few DNA breaks and, therefore, are not stained. This technique is widely utilized in biomedical research to investigate the mechanisms of cell death and related diseases. Additionally, tumor tissues were analyzed using TUNEL staining (Fig. 6h). The MT MN group exhibited slight fluorescence, while the fluorescence in the MT MN + US group was enhanced, demonstrating the potential of inhibiting nociceptors in cancer for antitumor therapy. Furthermore, when combined with anti-PD-1, this approach showed significant antitumor efficacy, highlighting its potential application in practical settings.

Fig. 6. Evaluation of the anti-tumor and neurosuppressive effects of MT MN patches in vivo.

a Schematic illustration of anti-cancer therapy. b The body weights of C57BL/6 mice treated with different treatments. Data are presented as mean ± s.d. for (b) (n = 8 mice per group). Digital picture of tumors (c), tumor weight (d) and tumor volume (e) of B16 tumor-bearing C57BL/6 mice at 14 days after different treatments. Data are presented as mean ± s.d. for (d, e) (n = 8 mice per group). f Kaplan–Meier survival plot of B16 tumor-bearing C57BL/6 mice after different treatments (n = 8 mice per group). g H&E staining images of tumor from the B16 tumor-bearing C57BL/6 mice with different treatments. Scale bars, 100 µm for (g). The representative images of g are shown from n = 3 independent mice. Representative immunofluorescence images (h) and quantification analysis (i) of Tunel. Scale bars, 100 µm for (h). Data are presented as mean ± s.d. for (i) (n = 3 fields in total from 3 mice). Representative immunofluorescence images of GAP43 (j), TRPV1 (k), CGRP (l), and SP (m). Scale bars, 100 µm for (j–m). The representative images of j–m are shown from n = 3 independent mice. Quantification analysis of GAP43 (n), TRPV1 (o), CGRP (p), and SP (q). Data are presented as mean ± s.d. (n = 3 fields in total from 3 mice). r Heatmaps of GAP43, TRPV1, CGRP, and SP. The representative images of r are shown from n = 3 independent mice. The expression of neurotransmitters CGRP (s) and SP (t) after different treatments in melanoma. Data are presented as mean±s.d. for (s, t) (n = 4 biologically independent samples). Statistical comparisons were performed using one-way analysis of variance (ANOVA), followed by Tukey’s multiple-comparison test. A value of p < 0.05 was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant. Source data are provided as a Source Data file.

Regulation of the neural environment within the tumor

Melanoma, a malignant form of skin cancer, is characterized by its high invasiveness and immunosuppressive properties, which are infiltrated by nociceptor neurons. Growth-associated protein 43 (GAP43) is a kind of classical neuron mark and the expression of GAP43 in tumor lesion were evaluated by immunofluorescence staining (Supplementary Fig. 28). The results proved that the tumor were infiltrated with neurons. Next, the neurosuppressive ability of the MT MN patch were tested in vivo. Firstly, the nerve density was evaluated in tumor tissues after different treatments through testing the expression of GAP43 by immunofluorescence staining. MT MN significantly decreased the expression of GAP43 in tumor tissues (Fig. 6j, n), which might be due to the release of Mn²⁺ and the activation of STING pathway. MT MN + US and MT MN + US + anti-PD-1 further reduced the nerve density in tumor tissues and showed the strongest inhibition (Fig. 6j, n), which perhaps due to the generation of ROS and the activation of STING pathway. Furthermore, it has been proved that the MT can induce the internalization of TRPV1 in vitro. Therefore, we further exam the TRPV1 expression in tumor tissues treated with different treatments (Fig. 6k, o). Control, US showed no obvious fluorescence intensity change, demonstrating that US and single MT MN have no significant influence on TRPV1 expression. However, under the stimulation of US, MT MN exhibited inhibition on TRPV1 further proved that the generation of local electric field on the cell membrane could induce the internalization of receptor (Fig. 6k, o). TRPV1 as a nonselective cation channel primarily responsible for Ca²⁺ influx and promote the secretion of neurotransmitters, including SP and CGRP. As immunosuppressive neuropeptide, SP and CGRP have been proved to damage the activity of T-cell and elicit the metastasis of cancer cells. In this situation, we have proved that the MT MN patch can induce the internalization of TRPV1 under the stimulation of US in vivo. Therefore, the secretion of SP and CGRP were further evaluation to prove the influence of MT MN. MT MN exhibited decreased fluorescence intensity of SP and CGRP (Fig. 6l–q), which might be due to the activation of STING induce the produce of IFN-I, suppressing the Ca²⁺ influx, by the released Mn²⁺ from the MT nanoparticles. Furthermore, MT MN + US exhibited the lowest fluorescence intensity (Fig. 6l–q), demonstrating that the piezoelectric effect induced TRPV1 internalization and activated STING in nerve cells through the released Mn²⁺ and the generated ROS. This synergistically inhibited the influx of Ca²⁺, thereby reducing the secretion of SP and CGRP. Additionally, the relevant data shown in the heatmap illustrate that MT MN have the potential to inhibit the neuron environment of tumor, suppressing the growth of neurons and the secretion of CGRP and SP (Fig. 6r). The expression of CGRP and SP in melanoma was also evaluated using an ELISA kit (Fig. 6s, t). The results demonstrated that the MT MN patch could inhibit nerve cell activity and downregulate the expression of CGRP and SP. From the above experimental results, it is evident that MT can effectively inhibit the activity of nerve cells in tumor tissues and suppress the secretion of CGRP and SP. This conclusion is supported by the detection of IFN-β levels in the tumor tissue (Supplementary Fig. 29), demonstrating that MT activates the STING pathway to induce IFN-β secretion, which, to some extent, helps inhibit nerve cell activity. Meanwhile, Ifnar1-KO mice were inoculated with tumors for in vivo experiments assessing type I IFN signaling-dependent effects. The Ifnar1-KO/MT + US group exhibited tumor inhibition compared to the control group. However, the therapeutic effect in the Ifnar1-KO/MT + US group was not as strong as that observed in the WT/MT + US group (Supplementary Fig. 30a–c). These results indicate that the absence of type I IFN signaling weakens the therapeutic effect, demonstrating that the therapeutic process is partially dependent on type I IFN. Furthermore, B16 tumor‑bearing mice were treated with resiniferatoxin (RTX) to ablate TRPV1 expressing nerves (Supplementary Fig. 31a–c). Notably, both the MT + US + RTX-treated and RTX-treated groups exhibited significantly slower tumor growth compared with the vehicle control group. These findings confirm the critical role of the TRPV1 nerve-dependent pathway in tumor progression. Additionally, the paw withdrawal threshold was evaluated (Supplementary Fig. 32). The increased paw withdrawal threshold verified that the intervention modulates nociceptive behaviors in parallel with alterations in tumor immunity, supporting a nerve-driven effect. Simultaneously, we evaluated the direct effects of the treatment on immune cells and melanoma cells using Rag1-KO mice. As shown in Supplementary Fig. 33, the WT/MT + US group exhibited the greatest therapeutic efficacy compared to the control group. In contrast, there was no significant difference in therapeutic efficacy between the Rag1-KO/MT + US group and the control group. These results confirm the immune-mediated mechanism underlying the observed antitumor effect and indicate that, in the absence of an intact immune system in vivo, MT does not exert a prominent therapeutic effect on melanoma. In conclusion, all of the results demonstrate that the piezoelectric nanomaterial MT suppresses neural activity through TRPV1- and type I interferon-dependent signaling pathways, thereby influencing the immune system in vivo.

Anti-tumor immunotherapy in vivo

Next, were further evaluated the immunotherapy efficacy in vivo. As previously described, the activated nociceptor secretes CGRP and SP to influence antigen presentation, T-cell activation, and cancer metastasis. The results indicate that piezoelectric microneedles MT MN have demonstrated potential neuroinhibitory and anti-tumor effects in vitro. Consequently, we further evaluated the antitumor immunotherapeutic efficacy of MT MN in vivo. During immunotherapy, the activation of mature dendritic cells (DCs) enables antigen presentation to T cells, thereby initiating an antitumor effect. Therefore, the proportion of DC cells plays a crucial role in this process. It was discovered through flow cytometry that the MT MN group exhibited a significantly increased proportion of mature DCs at 21.4%. An even greater increase in the proportion of mature DCs was observed in the MT MN + US group, which reached 24.5% (Fig. 7a, b). This increase is attributed to the neuroinhibitory capacity of MT MN, which reduces the secretion of CGRP, thereby enhancing the maturation of DCs. Furthermore, the proportion of DCs in the anti-PD-1 + MT MN + US group (30.4%) was higher than that in the anti-PD-1 group (26.8%) (Fig. 7a, b). This finding demonstrates that MT MN decreases the activity of nerve cells in the cancer environment and enhances antigen presentation by restoring the function of DCs. Subsequently, we evaluated the proportion of CD3⁺CD8⁺ T cells to further assess the ability of MT MN to restore immune surveillance within the tumor microenvironment. The proportion of CD3⁺CD8⁺ T cells in the MT MN + US group (14.1%) was significantly improved compared to the control group (Fig. 7c, d), demonstrating that the suppression of nociceptors in cancer tissue partially reestablishment the immunosurveillance effect of tumor tissues. Additionally, we found that the anti-tumor efficacy of anti-PD-1 was significantly enhanced when combined with MT MN + US, as evidenced by the increase in the proportion of CD3⁺CD8⁺ T cells from 15.5% to 22.4% (Fig. 7c, d). Next, we further evaluated the proportion of CD3⁺CD4⁺ T cells and observed a similar phenomenon (Fig. 7e, f). Heat map and immunofluorescence analyses indicated that MT MN + US can alleviate the immunosuppressive effects at the tumor site and enhance the therapeutic efficacy of anti-PD-1 (Fig. 7g–i). Taken together, these results further demonstrate that suppression of neural activity by the MT nanomaterial leads to complete reestablishment of immunosurveillance in the tumor microenvironment. Furthermore, we established a model for distant tumors to simulate cancer relapse and metastasis (Fig. 7j). After various treatments, tumor tissue was collected for further evaluation. Initially, we monitored several basic indicators of the tumor, including size, volume, and H&E staining (Fig. 7k–n and Supplementary Fig. 34). Furthermore, heterogeneity can be observed in the individual tumor growth curves, which is attributed to inherent biological variation among animals, including differences in tumor seeding efficiency, immune status, and individual responsiveness to the neuroimmunomodulatory treatment (Supplementary Fig. 34). Meanwhile, it was found that the MT MN + US group demonstrated the ability to inhibit distal tumors, highlighting the considerable potential of MT MN in preventing tumor metastasis. Moreover, in contrast to the anti-PD-1 group, the anti-PD-1 + MT MN + US group exhibited an enhanced ability to inhibit distal tumors, further demonstrating that the regulation of the tumor neural environment can suppress cancer metastasis and promote anticancer immunotherapy. Subsequently, the anti-tumor immunotherapeutic effect of MT MN was further validated by establishing a lung metastasis model (Fig. 7o, p). The results revealed that the number of metastatic nodules in the MT MN + US group was significantly decreased, demonstrating the anti-metastatic capability of piezoelectric microneedles by inhibiting neural activity in the tumor environment. By comparing the anti-PD-1 group with the anti-PD-1 + MT MN + US group, it is evident that the anti-PD-1 + MT MN + US group possesses a stronger ability to resist lung metastasis, reflecting the potential clinical application of MT MN in enhancing immune checkpoint therapy. In summary, the therapeutic efficacy observed in the distal tumor and lung metastasis models further demonstrated the successful establishment of long-term immune memory and the reestablishment of immunosurveillance in vivo. In addition, we collected organs from mice in different experimental groups for H&E staining. The results showed that, throughout the treatment process, the organs of mice in all groups remained in a normal state, further demonstrating the biosafety of the material (Supplementary Fig. 35).

Fig. 7. The in vivo anti-tumor immune therapeutic effect of MT MN patch.

Flow cytometry analysis of mature DCs (a), CD3 + CD8+ T cells (c), CD3 + CD4+ T cells (e) in vivo. Quantitative analysis of mature DCs (b), CD3 + CD8+ T cells (d), CD3 + CD4+ T cells (f) in vivo. Data are presented as mean±s.d. (n = 5 biologically independent samples). g Heatmaps of mature DCs, CD3 + CD8+ T cells, CD3 + CD4+ T cells in vivo. The representative images of g are shown from n = 3 independent mice. h, i Representative immunofluorescence images of CD3 + CD8+ T cells (h) and CD3 + CD4+ T cells (i) in vivo. Scale bars, 100 µm for (h, i). The representative images of h and i are shown from n = 3 independent mice. j Schematic illustration of anti-cancer immunotherapy. Digital photographs (k), distal tumor weight (l), and distal tumor volume (m) of the dissected tumor from the B16 tumor-bearing C57BL/6 mouse. Data are presented as mean ± s.d. for (l, m) (n = 8 mice per group). n H&E staining images of distal tumor from the B16 tumor-bearing C57BL/6 mice with different treatments. Scale bars, 100 µm for (n). The representative images of n are shown from n = 3 independent mice. Digital photographs (o) and H&E staining (p) images of lung tissue. Scale bars, 100 µm for (p). The representative images of o and p are shown from n = 3 independent mice. Statistical comparisons were performed using one-way analysis of variance (ANOVA), followed by Tukey’s multiple-comparison test. A value of p < 0.05 was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant. Source data are provided as a Source Data file.

Discussion

In this study, we have constructed MT MN patch for interfering the neuro-immune interaction in cancer to reestablish the immunosurveillance in melanoma. The targeted delivery of MT to nociceptors in melanoma tissue is achieved through transdermal administration using microneedles and the targeting capability of the cell membrane. Under the irradiation of ultrasound, the generation of local electric field on nerve cell membrane through the piezoelectric material induce charge imbalance and the internalization of TRPV1 protein. Furthermore, the generation of ROS by the US activated MT damage mitochondrial and combined with the released of doped Mn²⁺ ions activated the STING pathway to elicit the secretion of IFN-I. The downregulation of TRPV1 and the secretion of IFN-I suppress the influx of Ca²⁺, leading to the decrease of CGRP and SP to rescue immunosuppress environment in melanoma. Inhibition of neural cell activity through the piezoelectric effect and activation of the STING pathway effectively disrupt neuro-immune crosstalk to restore tumor immunosurveillance. This approach addresses a critical gap of neuroimmunotherapy in cancer and provides a valuable reference method. Furthermore, compared with traditional STING agonists and TRPV1 inhibitors, which carry a risk of off-target effects and may induce severe side effects, this strategy exhibits favorable biosafety. Furthermore, conventional exogenous electrical stimulation typically requires open wounds, wires, and power supplies, potentially leading to infection, inflammation, and local pain. The present strategy can effectively overcome these limitations. Although MT piezoelectric nanomaterials exhibit excellent neural activity inhibition ability by downregulating TRPV1 through nanoscale electric fields, direct measurement of the nanoscale electric fields at the membrane would better support this conclusion. Nevertheless, these approaches have limitations, and such refinements will be pursued in our future studies. Meanwhile, we did not include an empty microneedle patch group or an MT-without-Mn group, because the microneedle patch alone does not reach the tumor microenvironment, and Mn ions are involved in regulating the piezoelectric properties of MT. Although this does not affect the final conclusion, it limits our ability to fully dissect the individual contributions of mechanical, electrical, and Mn-mediated effects. Therefore, this could become a direction of considerable research value in the future. In summary, these therapies may also prove effective in treating other cancers in the future, such as glioma, pancreatic tumors, and lung cancer, highlighting their significant potential for clinical application.

Methods

Materials

Titanium tetraisopropanolate (TTIP, catalog number A34013, >98%) was purchased from innochem. 2-Aminoterephthalic Acid (H₂BDC-NH₂, catalog number A151463, >98%) and Arac (catalog number C111218) were purchased from Aladdin. N,N-Dimethylformamide (DMF, catalog number PHR1553) was purchased from Sigma-Aldrich. MnCl₂·4H₂O (catalog number 221279, >98%) was purchased from Sigma-Aldrich. Phosphate buffer solution (PBS, catalog number G4202) and DMEM/F12 (catalog number G4610) were purchased from Servicebio. Penicillin-streptomycin (catalog number C0222) was purchased from Beyotime. trypsin-EDTA (catalog number 15400054) and Neurolbasal medium (catalog number 21103049) were purchased from Gibco. Fetal bovine serum (FBS, catalog number 35-016-CV) was purchased from Corning. B-27 supplement (catalog number C0350) and l-glutamine (catalog number C0212) were purchased from Beyotime. PE Anti-mouse CD8 (catalog number 100707, dilution factor 1:200), APC anti-mouse CD279 (PD-1) (catalog number 109112, dilution factor 1:200), PE/CY7anti-mouse CD366 (Tim-3) (catalog number 134009, dilution factor 1:200), FITC anti-mouse CD11c (catalog number 117305, dilution factor 1:200), PE anti-mouse CD80 (catalog number 600055, dilution factor 1:200), APC anti-mouse CD86 (catalog number 159215, dilution factor 1:200), FITC anti-mouse CD3 (catalog number 100339, dilution factor 1:200), PE/CY7 anti-mouse IL-2 (catalog number 503831, dilution factor 1:200), APC anti-mouse IFN-γ (catalog number 505809, dilution factor 1:200),IL-2 (catalog number 575402), anti-mouse CD28 (catalog number 102116, dilution factor 1:200) were purchase from Biolegend.

Ethical statement

All studies were conducted using healthy female C57 mice (6–8 weeks old, 14–16 g), female Kunming mice (6–8 weeks old, 20 g) obtained from the Laboratory Animal Center of Jilin University, China. Rag1-KO mice (6–8 weeks old, 14–16 g) and Ifnar1-KO mice (6–8 weeks old, 14–16 g) were purchased from Cyagen Biosciences Inc (Suzhou, China). All animal procedures were approved by the Changchun Institute of Applied Chemistry Animal Care and Use Committee (Approval No. 20220035). According to the approved protocol, the maximum allowable tumor burden was 2000 mm³, and all tumor sizes in our study were kept below this limit. The study design did not take sex into consideration. Mice were housed under specific pathogen‑free conditions at 26 ± 1 °C with 50 ± 5% humidity and a 12 h light–dark cycle. And all animal care and handling procedures were in accordance with the guidelines approved by the ethics committee of Changchun Institute of Applied Chemistry.

Measurement and characterization

The morphology of MT was observed by SEM (S-4800 FS-SEM, Hitachi) and TEM (FEI TECNAI G2 20). The elements analysis was acquired by energy dispersive X-ray spectroscopy (EDS) (S-4800 FS-SEM, Hitachi) at the working voltage of 20 kV and working current of 15 μA. The hydrodynamic diameter and zeta potential of the MT was measured by DLS (Zetasizer 3000HS). The in vitro stability of MT was also evaluated by monitoring their hydrodynamic diameter in PBS and DMEM with 10% FBS for different times by DLS analysis. Piezoelectric properties were verified using an atomic force microscope (Dimension Edge, Bruker). Analyze the crystal structure of MT through XRD (D8 FOCUS, Bruker). The elements in MT were analyzed using an X-ray photoelectron spectrometer (ESCALAB 250). The immunofluorescence analysis, fluorescence quantification and observation was acquired by confocal laser scanning microscopy (Nikon Eclipse Ni-E, Japan). Perform immunophenotype analysis using flow cytometry (BD LSRFortessa Cell Analyzer).

Isolation of DRG cells

C57BL/6 mice were euthanized, and collecting spin on ice to dissect DRG out into DMEM/F12 (Servicebio G4610), completed with 10% FBS (Corning 35-016-CV) and 1% Penicillin-Streptomycin Solution (Beyotime, C0222). Cells were then dissociated in Hanks’ balanced salt solution (Servicebio G4203) completed with 1 mg/ml collagenase I (Beyotime, ST2294) and incubated for 30 min under 37 °C. And the collagenase solution was removed and DRG washed with Hanks’ balanced salt solution three times. The 2 ml 0.05% trypsin-EDTA (Gibco, 15400054) were used to further dissociate the DRG tissue for 30 min under 37 °C. The DRG neurons were collected and cultured with Neurolbasal medium (Gibco, 21103049) completed with 10% FBS, 0.05 ng/ml NGF (MCE, HY-P70530), B-27 supplement (Beyotime, C0350), 200 mM l-glutamine (Beyotime, C0212) and 0.01 Mm Arac (Aladdin, C111218).

Preparation of neuron membrane fragments

The DRG cells were initially cultured in Neurolbasal medium supplemented with 10% FBS, 0.05 ng/mL NGF, B-27 supplement, 200 mM L-glutamine, and 0.01 mM AraC. To obtain cell membrane fragments, the cells were detached using a cell scraper and isolated by centrifugation at 700 × g for 5 min. The collected cells were resuspended in precooled PBS buffer (pH 7.4) and centrifuged again at 600 × g for 5 min. The resulting cell pellets were suspended in a hypotonic lysing buffer containing membrane protein extraction reagent and phenylmethanesulfonyl fluoride (PMSF) (Beyotime, P0033) and incubated on ice for 10–15 min. Subsequently, the cells in this solution were disrupted repeatedly using a freeze-thaw method, followed by centrifugation at 700 × g for 10 min at 4 °C. The supernatant was then subjected to further centrifugation at 14,000 × g for 30 min to collect the cell membrane fragments. The membrane products were lyophilized overnight, weighed, and stored at –80 °C. Prior to use, the lyophilized membrane materials were rehydrated in ultrapure water or PBS buffer (pH 7.4).

Preparation of MT

To obtain MT, the Ti-based piezoelectric MOF was synthesized following a previously reported method. First, a mixture of 9 mL DMF and 1 mL methanol was prepared. Then, 0.543 g H₂BDC-NH₂, 0.23 mL TTIP, and 0.0074 g MnCl₂·4H₂O were added to the solution. The mixture was stirred for 20 min at room temperature before being transferred into a Teflon-lined stainless-steel autoclave. The autoclave was heated to 150 °C for 36 h. After cooling to room temperature, the resulting powder was collected by centrifugation, washed three times with DMF and methanol, and dried under vacuum at 80 °C. MT was fabricated by coating the Ti-based piezoelectric MOF with neuron membrane fragments using an extrusion method. Specifically, the Ti-based piezoelectric MOF was mixed with the cell membrane fragments and subjected to an ultrasonic water bath for 60 s. The mixture was then sequentially extruded through 400 nm and 200 nm polycarbonate porous membranes using a mini-extruder. As a result, MT was obtained.

Preparation of MT MN patch

The MT microneedle patch was prepared using a two-step micromolding process. First, MT was dispersed in 1 ml of distilled water to form a 10 mg/ml MT suspension. Then, sodium hyaluronate and super active hyaluronic acid were mixed with the MT suspension to create an HA solution (180 mg/ml; sodium hyaluronate to super active hyaluronic acid ratio = 1:5, w/w). A 100 µl aliquot of this solution was applied to the surface of a PDMS mold and subjected to vacuum five times (3 min each) to remove bubbles. The PDMS mold was then dried at 37 °C to form the MT-loaded needles. In the second step, an MT-free HA solution was prepared and added to the surface of the PDMS mold to form the base of the microneedles.

Isolation of CD8⁺ T cells

Six-to-eight-week-old C57BL/6 mice were euthanized, and their spleens were collected in ice-cold PBS supplemented with 10% FBS. The collected spleens were mechanically dissociated. The cells were filtered through a 70 µm cell strainer, centrifuged at 300 × g for 5 min, and resuspended in extraction buffer. CD8⁺ T cells were magnetically sorted (BioLegend, 480007).

To generate cytotoxic CD8⁺ T cells, CD8⁺ T cells were seeded in plates coated with anti-mouse CD3 (BioLegend, 100339, 5 µg/ml) and stimulated for 48 h with 20 ng/ml IL-2 (BioLegend, 575402), 2 µg/ml anti-mouse CD28 (BioLegend, 102116), 0.05 mM β-mercaptoethanol (Aladdin, M301574), 2 mM L-glutamine, and 20 mM HEPES (Aladdin, H657463). Subsequently, the cells were cultured in medium without anti-mouse CD28 to promote expansion.

Isolation of bone marrow-derived dendritic cells (BMDCs)

BMDCs were collected from C57BL/6 mice. The mice were euthanized, and their bones were harvested for further processing. Bone marrow was flushed repeatedly from the bones and filtered through a 70 µm cell strainer. BMDCs were then isolated after depleting red blood cells using Red Blood Cell Lysis Buffer (Beyotime, C3702). Following centrifugation and washing, BMDCs were seeded into 6-well plates and cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin solution, 20 ng/mL GM-CSF (Beyotime, P6006), and 10 ng/mL IL-4 (Beyotime, P5916). BMDCs were subsequently stimulated with lipopolysaccharide (LPS) to induce maturation.

Co-culture CD8⁺ T cells with DRG cells

Naïve DRG neurons (2 × 10⁵) were seeded into a 24-well plate containing T cell medium. One day later, the DRG neurons were treated with control, capsaicin, capsaicin + US, capsaicin + MT, and capsaicin + MT + US, respectively (US irradiation 1 MHz, 50% duty cycle, 0.4 W cm⁻², 30 min). Activated CD8 + T cells (2 × 10⁵) were then seeded into the upper compartment of a transwell system. After 96 h, the CD8 + T cells were collected by centrifugation (300 × g, 5 min), stained with anti-mouse CD8 (BioLegend, 100707), anti-mouse CD279 (PD-1) (BioLegend, 109112), and anti-mouse CD366 (Tim-3) (Biolegend, 134009), and immunophenotyped by flow cytometry.

Co-culture BMDCs with DRG cells

Naïve DRG neurons (2 × 10⁵) were seeded into a 24-well plate containing T cell medium. One day later, the DRG neurons were treated with control, capsaicin, capsaicin + US, capsaicin + MT, and capsaicin + MT + US, respectively (US irradiation 1 MHz, 50% duty cycle, 0.4 W cm⁻², 30 min). Activated BMDCs (2 × 10⁵) were then seeded into the upper compartment of a transwell system. After 96 h, the BMDCs were collected by centrifugation (300 × g, 5 min), stained with anti-mouse CD11c (BioLegend, 117305), anti-mouse CD80 (PD-1) (BioLegend, 600055), and anti-mouse CD86 (BioLegend, 159215), and immunophenotyped by flow cytometry.

Migration assay

For the wound healing assay, the Cell Scratch Plugin was first installed on the bottom of a 24-well plate, and B16F10 cells were seeded into the compartment of the Cell Scratch Plugin. After 24 h, the Cell Scratch Plugin was removed to create the scratch. Simultaneously, DRG cells were seeded into the upper chamber of a transwell with medium for 24 h. The DRG cells were then treated with control, capsaicin, capsaicin + US, capsaicin + MT, and capsaicin + MT + US. After treatment, the DRG cells were co-cultured with the scratched B16F10 cells for 12 h (US irradiation 1 MHz, 50% duty cycle, 0.4 W cm⁻², 30 min). The scratch area was then observed and photographed using a microscope.

In vivo biosafety evaluation

To evaluate the physiological effects caused by MT NPs, C57BL/6 mice were randomly divided into two groups (n = 5). Blood samples were collected 28 days after injection with either PBS (100 μL) or MT NPs (10 mg/kg, 100 μL) for blood biochemistry and routine blood analysis. Additionally, major organs (heart, liver, spleen, lungs, and kidneys) were harvested for histological examination.

Hemolysis test

One milliliter of whole blood was collected in tubes containing lithium heparin from C57BL/6 mice. The blood was then diluted with an appropriate volume of 1× PBS, centrifuged at 2500 rpm for 5 min, and the supernatant was discarded. This washing step was repeated 3–4 times until the supernatant became colorless. The precipitated erythrocytes were resuspended in 1× PBS to prepare an erythrocyte suspension. MT nanoparticles (NPs) were added to the erythrocyte suspension to achieve final concentrations ranging from 0 to 80 μg/mL. The negative control consisted of erythrocyte suspension diluted with 1× PBS, while the positive control was erythrocyte suspension diluted with ultrapure water. The tubes were incubated for 8 h at 37 °C to observe and record hemolysis. Meanwhile, the absorbance of hemoglobin at 540 nm was measured spectrophotometrically. The hemolysis rate (HR%) was calculated using the following equation:

$$HR\%=\left(A_{MT}-A_N\right)\times 100\%/\left(A_P-A_N\right)$$ 1

Where A_MT, A_P, and A_N represent the absorbance of the sample, the positive control, and the negative control, respectively.

In vivo antitumor effect

One week after orthotopic implantation of B16 cells, C57BL/6 mice were treated every 3 days for a total of six treatments with one of the following: control, US, MT MN patch, MT MN patch + US, anti-PD-1, or anti-PD-1 + MT MN patch + US. Body weight, tumor volume, and survival were recorded every 2 days (US irradiation 1 MHz, 50% duty cycle, 0.4 W cm⁻², 30 min). After the treatments, some mice were sacrificed, and tumors and major organs were collected. The tumors were weighed, fixed in 4% paraformaldehyde (PFA), and analyzed by hematoxylin and eosin (H&E) staining as well as immunofluorescence staining for TUNEL, GAP43, TRPV1, CGRP, and SP. Major organs were also fixed in 4% PFA, sectioned, and stained with H&E.

Statistics and reproducibility

In this article, the results of all experiments were presented as mean ± s.d. All figures illustrated were obtained from several independent experiments with similar results. Data are expressed as mean ± standard deviation (SD). Normality was tested using the Shapiro–Wilk test, and homogeneity of variance was verified using Levene’s test. Statistical comparisons were performed using one-way analysis of variance (ANOVA) using the Graph Pad Prism 9.0 software (Graph Pad Software), followed by Tukey’s multiple-comparison test. A value of p < 0.05 was considered statistically significant. *p < 0.05, ***p < 0.001, ****p < 0.0001; ns, not significant.

Reporting summary

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

Author contributions

A.S. conceived and designed the experiments, performed the experiments, analyzed the data, and wrote the original manuscript; Y.Z. performed the experiments; Y.J. and H.Y. participated in the analysis of data; J.R. and X.Q. conceived and designed the experiments, supervised the study, and revised the manuscript.

Peer review

Peer review information

Nature Communications thanks Moran Amit, Sebastien Talbot, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the Natural Science Foundation of China (22237006, 22437006, and T2495262).

Data availability

The data supporting the findings of this study are available within the Article, Supplementary Information, or Source Data file and/or from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74550-7.