Design and characterization of the asymmetric urchin head/hollow tail nanostructures (UHHTNs)
The AuNSs were first added in a isopropyl alcohol (IPA)-H2O mixture (2.5:1 V/V) containing (4-mercaptophenylacetic acid, 4-MPAA) and poly acrylic acid (PAA, M.W.= 5500), followed by stirring at room temperature for 30 min. The two types of ligands competition for surface binding sites enables the Janus like segregation on the surface of AuNSs27, with one side is 4-MPAA region, and the other side is PAA domain. Then, hexadecyltrimethylammonium bromide (CTAB), tetraethyl orthosilicate (TEOS) and ammonium hydroxide were added in sequence. The rich -COOH group of free PAA molecular chain in solution interacted with the NH4+ to form PAA-NH4 complex as a salt28, 29, leading to the mixture phase separation, with one enriches water and salt, the other one is mostly IPA, and thus a large number of small droplets are generated30, 31.
Silica was first deposited on the 4-MPAA side to form half-coated thin SiO2 layer as the carboxylic groups of 4-MPAA could react with the silane monomer (Supplementary Fig. 1), and at PAA side, PAA polymer can be like a ‘sponge’ to absorb and reserve droplets inside its net structure due to their superhydrophilic nature, enabling continuous directional fusion of the droplets extracted from solution upon collision, which has been studied well in our recent work32. The CTAB molecules could adsorb onto the surface of PAA via Coulomb force and electrostatic interaction between CTA+ and PAA-. Due to the charge interaction, the silica oligomers interacted with CTA+. Continuous the small droplet fusion onto PAA along longitude enable silica continuous deposition at the interface between PAA-CTAB to form an open hollow tail connected to the half shell of 4-MPAA-covered surface. The self-nucleated hollow silica nanoparticles formed in solution further confirmed the generation of free droplets (Supplementary Fig. 2). If only 4-MPAA was used, the 4-MPAA covered the entire surface of the AuNSs and no small droplets formed, resulting in a fully encapsulated AuNS@silica core-shell structure (Supplementary Fig. 3).
Asymmetric urchin head/hollow tail nanostructures (UHHTNs) consist of a NIR-absorbed AuNS half-coated with a SiO2 shell in the head region and an open hollow tail connected to the half shell (Fig 2a). The employed AuNSs have nanospikes and an average diameter of ~265 nm (Fig. 2b, Supplementary Fig. 4). Transmission electron microscopy (TEM) images revealed the uniform morphology of the as-synthesized UHHTNs with high yield (almost 100%). High-resolution transmission electron microscopy and selected area electron diffraction results show that the SiO2 shell in the hollow tail is amorphous (Fig. 2c). UHHTNs have a spiky head with a width of ~285 nm and a hollow tail with a dedicated opening of ~100 nm at the end of the hollow tail (Fig. 2d). The average body length of UHHTN is ~351 nm (Fig. 2e). Scanning electron microscopy (SEM) image reveals that there are many nanospikes on the surface of the UHHTNs that resemble the surface spiky morphology of some urchin (Fig. 2f). The thicknesses of silica in the head and tail regions marked with bule rectangular frames are ∼6.78 ± 1.21 nm and 9.06 ± 1.52 nm, respectively (Supplementary Fig. 5). The outlines of the SiO2 and AuNS compartments can be clearly detected by elemental mapping and matched well with the relative positions in the UHHTN (Fig. 2g).
The UV-vis-NIR absorption spectra of as-prepared AuNSs and UHHTNs were performed. The bare AuNSs have a broad absorption range from 800 to 1200 nm and an intense absorption at ~916 nm in the NIR region, resulting from their localized surface plasmon resonance (LSPR). After the SiO2 coating, the peak was redshifted by 23 nm from 916 to 940 nm due to an increase in the refractive index of the medium surrounding of the AuNSs. Note that before and after the SiO2 coating, no appreciable broadening of the absorption spectrum was occurred, indicating that AuNSs did not form aggregates during the growth process (Fig. 2h). The dedicated architecture still remained intact even after long periods of ultrasonication (Supplementary Fig. 6). Moreover, the UHHTN aqueous solution showed obvious light scattering capacity after storage of 2 months and then stirring, while large precipitates were formed for bare AuNS aqueous solution, indicating that the UHHTNs possess good dispersion stability with the protection of thin silica shell (Fig. 2i). This approach easily produced large-scale monodisperse UHHTN colloids (Supplementary Fig. 7).
Tuning of the surface spiky topologies and hollow tail diameters
The surface nanotopologies of nanostructures can be continuously varied by changing only the 4-MPAA concentrations. When the amount of 4-MPAA added is 5 mM, only nanostructures with relatively smooth surfaces (spiky length of ~13 nm) in the head region were obtained (Fig. 3a (left), b, e and supplementary Fig. 8). In the case of 15 mM, the spiky length on the surface was increased to 65 nm (Fig. 3a (middle), c, f and Supplementary Fig. 9) and simultaneously spike number increases. A further increase in the 4-MPAA concentration led to a much more and longer spike with a length of ~94 nm (Fig. 3a (right), d, g and Supplementary Fig. 10). The cavity diameters in b, c, and d are ~240 nm, 115 nm and 100 nm (indicated by yellow line), respectively.
Moreover, it can be noted that the spike lengths increase as the 4-MPAA concentration increases, while hollow tail diameters decrease. The corresponding hollow diameter of the tail decreased from ~240 nm to 115 nm to 100 nm (Fig. 3h). In addition, the resultant hollow tail has varied length, which are mainly attributed different growth rate (Fig. 3i). The smaller the tail diameter was, the faster the growth rate became, and therefore the resulting tails had varied lengths.
Based on the above observations, a convincing mechanism was proposed to explain the formation of spiky surface in head. It has been demonstrated that specific binding locations of surface ligands can produce “active growth sites” and “inactivated growth sites”, leading to specific site-selective growth33, 34, 35. In our system, nanospike length of UNNTH is closely related to silica deposition site in head region, which is determined by 4-MPAA coverage location because the carboxylic groups of 4-MPAA could react with the silane monomer as mentioned above.
A low 4-MPAA concentration cannot ensure that 4-MPAA molecules can fully cover entire site in the 4-MPAA region of AuNS, thus the 4-MPAA molecules preferentially adsorbed on the high-curvature sites of AuNSs such as the vertices due to their high surface energy, and further crosslinked into a thin silica layer along all vertices of 4-MPAA-covered AuNSs (Fig. 3j). Smooth surfaces without nanospikes were obtained in the head area. The intermediates exhibited a large cavity length in the head region (Supplementary Fig. 11) with no silica in the groove, confirming that silica growth did not occur in the groove region.
As the concentration increases, the cavity length in the head was significantly reduced (Supplementary Fig. 12), implying that the 4-MPAA started to appear at the edges and caused the silica to gradually deposit inward (Fig. 3k). When enough amount of 4-MPAA was added, the 4-MPAA molecules can sufficiently cover on the vertices and edges. In this case, silica growth inherited the surface topology from AuNSs, the cavity in the head almost disappeared (Supplementary Fig. 13) and eventually developed into distinctive spiky surface (Fig. 3l). For the variation of hollow tail diameter, it can be explained by ligand competition theory that 4-MPAA concentration not only affect coverage degree, but also affects the two types of ligand occupied area27. Hence, an increased amount of 4-MPAA enable complete coverage on the vertices and edges, and simultaneously results in a smaller PAA region on the AuNS surface. As a result, the surface spiky length gradually become longer, whereas the tail diameter gradually become smaller. These results indicated that the ligand concentration allows fine control over the spiky surface topologies and hollow diameter.
Thermo-responsive drug delivery and motion performance of UHHTN nanorobots
AuNSs with sharp spikes have gained considerable attention as photothermal agents because of their excellent photothermal properties. For this reason, the optical properties of UHHTNs were studied carefully. UHHTNs with more and longer nanospikes (spike length of ~94 nm and body length of ~351 nm shown in Fig. 2c) were chosen as a model structure. The temperature of the UHHTN aqueous solution (the concentration was 100 μg/mL) increased to ~40, 48, 55, and 62°C under 980 nm laser irradiation at varied power densities of 0.5, 1.0, 1.5, 2.0 W·cm−2 for 10 min, respectively (Fig. 4a), while that of the pure water increased only to 35.1°C (Supplementary Fig. 14). The modified thin silica shell body with an opening does not shield the photothermal conversion performance of the AuNS core (Supplementary Fig. 15).
We then encapsulated fatty acids into the UHHTNs through the opening of the hollow tails. Briefly, UHHTNs were dispersed in a eutectic mixture of lauric acid (melting point (m.p.): 44°C) and stearic acid (m.p.: 69°C) at a weight ratio of 4:1). The loaded UHHTNs (denoted as PCM-UHHTN) were retrieved by centrifugation, followed by gentle washing with DMSO to remove the surface adsorption and free fatty acids. Then water was added to solidify the DOX-trapped fatty acids inside the cavity. Owing to their excellent biocompatibility and biodegradability, these two fatty acids can be safely used in biomedicine. The differential scanning calorimetry (DSC) curve exhibited a sharp melting point at ~40℃, which is slightly higher than the normal temperature of the human body (37℃) (Supplementary Fig. 16). The thermal gravimetric analysis (TGA) of the PCM-UHHTNs showed a loading amount of 11.3 wt% for PCM, indicating the PCM can be loaded into the cavity through the opening (Fig. 4b). Furthermore, the PCM and DOX were co-encapsulated into the cavity for NIR-triggered drug release (Supplementary Fig. 17). The UV-vis-NIR spectra and fluorescence images confirmed the successful encapsulation of the DOX (Fig. 4c and Supplementary Fig. 18). TEM results further showed that PCM/DOX was loaded into the cavity of UHHTNs and no drug-loaded fatty acid particles are observed, implying that the final product is only PCM/DOX-trapped UHHTNs (Supplementary Fig. 19). The loading amount was calculated to be 110 mg per gram of the UHHTNs (Supplementary Fig. 20).
The release of encapsulated DOX could be readily tuned by varying the power density and/or the duration of laser irradiation. The amount of released DOX reached 61% at a 1 W·cm−2 power density within 8 h, corresponding to an equilibrium temperature of 45℃ (Fig. 4d). In contrast, less than 3% and 5% of DOX were released
at 0.3 W·cm−2 and 0 W·cm−2, respectively, indicating that the drug was retained inside the interior of the UHHTNs due to the limited diffusion of DOX molecules by the solid PCM.
We further explored the motion performance of PCM-UHHTNs. The trajectories of randomly selected nanorobots (n = 10) were tracked from the recorded videos by Image J under the condition of NIR irradiation with varied power densities and corresponding mean square displacement (MSD) were calculated (Fig. 4e and Supplementary Fig. 21). The average MSD of nanorobot trajectories showed an increase with time (Supplementary Fig. 22). The effective diffusion coefficient was obtained by applying the equation 1 and 2 shown in Supplementary Information. The diffusion coefficient of the UHHTN nanorobots increases from the Brownian diffusion value (~0.82 μm2/s) in the absence of NIR laser to 3.08 μm2/s at 0.5 W·cm−2 laser power, 4.28 μm2/s at 1.0 W·cm−2 laser power, 5.12 μm2/s at 1.5 W·cm−2 laser power, and 5.90 μm2/s at 2.0 W·cm−2 laser power (Fig. 4f). These results confirmed that the movement of nanorobots can be facilely tuned by the incident NIR laser power.
To prove that the propulsion is attributed to asymmetry of UHHTNs, we use symmetric AuNS@SiO2 core-shell nanoparticles with similar size and NIR-responsive photothermal effect for comparison (Supplementary Fig. 23). The trajectory shows random motion within short distance, suggesting that the motion is indeed morphology dependent (Supplementary Fig. 24). Based on these results, the motion of the nanorobot can be attributed to the generation of the local thermal gradients across the PCM/nanorobot, where the AuNSs in the head region convert the adsorbed photons to heat. The simulation results show that the temperature near the head region was higher than those near the tail opening (Fig. 4g and Supplementary Fig. 25). When a temperature gradient is formed along the interface between the solvent and the nanorobot, an osmotic pressure gradient parallel to the interface and antiparallel to the temperature gradient is created36. This is mainly due to the difference in ion concentration caused by the thermal gradient. The concentration is slightly higher at the cold side and thus the pressure is slightly higher at the cold side, and its gradient is opposite to ∇T. As a consequence, there is a thermoosmotic fluid flow along the surface toward higher temperature37. The fluid velocity field profile shows that thermo-osmotic flow on the outer surface of the UHHTNs drives the fluid to the hot end in the head (Figure 4h). As the fluid is stationary within the laboratory frame of reference, this implies that the particle has to move along the temperature gradient opposite to the interfacial fluid flow and thus drives UHHTNs via a mechanism of thermophoresis (Figure 4i).
Cellular cytotoxicity, internalization and in vitro penetration
To evaluate the effect of surface spikes and self-motile capability on cellular internalization, nanoparticles with smooth surfaces were synthesized for comparative purposes (Supplementary Figs. 26 and 27). The CTAB of UHHTNs was removed by extraction before usage to ensure biocompatibility. MDA-MB-231 cells were treated with 100 μg/mL DOX-loaded nanoparticles without nanospikes, spiky nanoparticles and spiky nanoparticles with laser irradiation for 0-30 min. Red fluorescence could already be found in the cytoplasm in the Spike group, however, the fluorescence was still scarce at 30 min in the No Spike group. When nanoparticles with spiky heads were explored under NIR irradiation (1 W·cm−2) for 0-30 min, the confocal results showed that more cells were stained with red fluorescence
than those without laser irradiation (Fig. 5a and Supplementary Fig. 28). The cellular uptake efficiency was quantitatively evaluated by calculating the red fluorescence intensity (Fig. 5b). Similar results were demonstrated in three other cell lines, A549, PANC-1 and B16F10 (Fig. 5c, d and e; Supplementary Figs. 29-31). To confirm whether the UHHTN/DOX were internalized inside cellular interiors, the cell membranes of four cell lines were stained and the results show that more UHHTN/DOX nanorobots were located in the cytoplasm and nucleus (Supplementary Fig. 32). In addition, we also explored the internalization efficiency of UHHTN/DOX with different degrees of spikes shown in Fig 3. The results show that the UHHTN/DOX with a much more and longer spike with a length of ~94 nm have higher internalization efficacy, which possibly attributed to more spikes with certain length of spikes in nanoparticle surface enhance the nano-bio interfacial interaction (Supplementary Fig. 33).
Considering that the UHHTN could have a higher spike-mediated mechanical stress due to smaller contact area between nanospikes and cell membrane, which could disrupt membrane to enter into interior of cells. We first confirm whether UHHTN/DOX entering inside cell is dependent on endocytosis, cells were incubated with the UHHTN/DOX at 4°C, a temperature at which endocytosis is generally inhibited38. Our results show certain decrease, suggesting that internalization is endocytosis-dependent. We further observed changes of cellular uptake behavior with specific endocytosis inhibitors. Pretreatment with mycoplasmin (fractal protein endocytosis inhibitor), chlorpromazine (lattice protein endocytosis inhibitor), and monensin (lattice protein and fractal protein non-dependent endocytosis inhibitor) had no effect on UHHTN/DOX internalization, while very weak red fluorescent signals were visible around cells treated with cytochalasin D (macropinocytosis inhibitor), so we can infer that UHHTN/DOX internalization is mainly dependent on the macropinocytosis-mediated endocytosis pathway (Supplementary Fig. 34).
We further explore the distribution of UHHTN/DOX inside the cell. We labeled lysosomes and Golgi apparatus separately and observed the co-localization of nanorobots with both. Under NIR irradiation, more UHHTN/DOX entered the cell and were distributed in the Golgi apparatus, lysosomes and nucleus (Supplementary Fig. 35). These results suggest that UHHTN/DOX can efficiently enter the intracellular compartment and promote nuclear delivery of DOX to exert antitumor effects.
To exclude the possibility that NIR irradiation (1 W·cm−2) may cause damage to the cells that could have an effect on cell internalization, CCK-8 analysis was performed and showed that the cell viability among these three groups after 0-30 min of incubation was similar (Supplementary Fig. 36). Furthermore, the cell viability remained greater than 90% under treatment with UHHTN/DOX for 24 and 48 h (Supplementary Fig. 37), demonstrating the cytocompatibility of UHHTN/DOX. Similar results were also observed in A549, PANC-1 and B16F10 cells (Supplementary Figs. 38).
Considering the transport of nanorobots in the body from blood vessels to specific tumor sites, we analyzed the transvascular and transcellular capacity. Two different models, including the penetration of vascular endothelial cells (endothelial cells in the upper chamber and tumor cells in the lower chamber) and tumor cells (tumor cells in both chambers) were used in a Transwell test (Fig. 5f). The result exhibited that large amount of UHHTN/DOX nanoparticles with laser irradiation can be transferred to MDA-MB-231 cells in the lower chamber than that without irradiation (Fig. 5g, left panel and Supplementary Fig. 39a). Similar results were shown in transcellular penetration (Fig. 5g, right panel and Supplementary Fig. 39b). Such penetration performance of the UHHTN/DOX nanorobots was also exhibited in the three constructed Transwell models of A549, PANC-1 and B16F10 cell lines, indicating that our nanorobot system function in a variety of tumors with stiff stroma (Supplementary Figs. 40-42).
In order to further investigate the penetration of UHHTN nanorobots, three-dimensional multicellular tumor spheroid (3D MTS) similar to solid tumors were established. It was found that more and higher-density red fluorescence throughout the 3D MTS was visible in Spike+NIR group compared with either Spike or No Spike group (Fig. 5h). The mean optical density (MOD) values were quantitative data on the red fluorescence signal generated by the three groups after penetration into the 3D MTS (Fig. 5i and (Supplementary Fig. 43). The fluorescence intensity of the nanorobots in the equatorial plane was measured to be about 1.6 and 2.9 times higher than the other two groups, which suggests that nanorobots with the combination of active motility and nano-bio interaction can effectively promote their tumor penetration.
Therapeutic efficacy and safety of UHHTN/DOX nanorobots
The therapeutic effect of UHHTN/DOX nanorobots in vitro was investigated. Considering the drug carrying capacity and photothermia effect of UHHTN/DOX nanorobots, the cytotoxicity of UHHTN/DOX nanorobots with 1.5 W·cm−2 laser irradiation for 30 min was confirmed by the live/dead cell analysis (Supplementary Fig. 44) and colony formation assay (Supplementary Fig. 45). Furthermore, the migration and invasion test of MDA-MB-231 cells demonstrated that UHHTN/DOX nanorobots exerted a powerful synergistic anti-TNBC cell efficacy, which might be ascribed to the enhanced cellular internalization of UHHTN nanorobots (Supplementary Figs. 46-48).
Spinal vertebral body is the main axial bone. It has been reported that spinal metastasis accounts for approximately 60% of TNBC distant bone metastasis with stiff stroma39, therefore we selected a spinal metastasis model to analyze the biological functions of UHHTN nanorobots. To evaluate the in vivo therapeutic efficiency of UHHTN/DOX nanorobots for spinal metastasis, the mouse TNBC spinal metastasis models were established by intraspinal vertebral inoculation of MDA-MB-231 cells into mice, followed by intravenous nanoparticles injection and 1.5 W·cm−2 laser irradiation for 30 min. Fourteen days after cell injection, these mice were randomly divided into 7 groups, which were intravenously given the following treatments: saline, DOX, Laser (980 nm laser with 1.5 W·cm−2 for 30 min), UHHTN, UHHTN+Laser, UHHTN/DOX, and UHHTN/DOX+Laser, followed by the bioluminescence imaging, survival status analysis, tumor measurement, histology, computed tomography (CT) and magnetic resonance imaging (MRI) examination at different stages (Fig. 6a). The blood retention rate of UHHTN/DOX was measured using DOX and DOXIL as controls. The half-live (t1/2) of UHHTN/DOX is slightly lower compared to DOXIL. While the UHHTN/DOX significantly prolonged the circulation of DOX compared to free DOX after intravenous administration, and is twice as high as free DOX, indicating the ability of UHHTN/DOX to maintain a long circulation time (Supplementary Fig. 49).
A 20-day survival recording was performed to observe the survival rate (Fig. 6b). The survival rates of nude mice treated with UHHTN/DOX+Laser were improved to 80%, compared with 60% and 40% in the UHHTN+Laser and UHHTN/DOX groups, respectively, demonstrating the synergistic effects of chemotherapy and photothermal therapy on the survival rate. In contrast, the survival rate of the DOX group was 0% (mice died on 2, 4, 7, and 13 days, respectively) due to the systemic toxic effects and adverse reactions. In an in vivo biodistribution study, mice carrying TNBC spinal tumors were collected for in vitro fluorescence imaging of spinal tumors and major organs after 24 h, 48 h, 72 h, and 96 h of intravenous injection of UHHTN/DOX nanorobots (Fig. 6c). The fluorescence intensity of spinal tumor tissue was significantly higher than that of other tissues, indicating that the UHHTN/DOX nanorobot can passively target tumor sites. We used high performance liquid chromatography to determine the concentration of DOX in different organs to further validate the distribution of UHHTN/DOX. After 24 h of injection, the accumulation of UHHTN/DOX in the tumor increased significantly, and higher DOX was detected compared to other organs (Supplementary Fig. 50). Bioluminescence imaging and gross observation showed the progressive growth of spinal metastasis in the saline control group, indicating successful model establishment (Fig. 6d and Supplementary Fig. 51). Compared with the saline group, the UHHTN/DOX+Laser treatment significantly inhibited tumor growth, as shown by the bioluminescence imaging, tumor weight and volume (Fig. 6e, f and g), suggesting the high therapeutic efficiency of UHHTN/DOX nanorobots with photothermal effects. Compared with the UHHTN+Laser and UHHTN/DOX groups, the UHHTN/DOX+Laser group had smaller tumor in the spine (Fig. 6d-g and Supplementary Figs. 51 and 52). Therefore, the combination of nanorobot-mediated chemotherapy and photothermal effects was more suitable for the treatment of spinal metastasis.
To further verify whether the nanorobots can promote bone regeneration after tumor elimination. Following twenty days after injection, the MRI imaging showed that tumors with high levels of T2 signaling in the spine were obviously reduced in the UHHTN/DOX+Laser group, resulting in a nearly normal morphology in mice (Fig. 6h). Next, the bone destruction in the tumor was analyzed by 3D reconstruction of CT analysis (Fig. 6i). Osteolytic lesions in the spine could be observed in the saline and laser groups, but the internal osteolysis was significantly relieved after treatment with UHHTN/DOX nanorobots combined with NIR, indirectly suggesting the inhibitory effect of the nanorobots on the bone destruction of TNBC cells. Quantitative morphometric analysis showed that the bone mineral density (BMD) and mean bone volume/tissue volume ratio (BV/TV%) (Fig. 6j, k) in the UHHTN/DOX+Laser treatment group were 2.4 and 1.3 times higher than those in the control group, respectively. Taken together, these results demonstrated the superior efficacy of UHHTN/DOX nanorobots in the reduction of TNBC spinal metastasis.
To confirm that the synergistic chemo- and phototherapeutic effect of UHHTN/DOX nanorobots can function in a variety of tumors with stiff stroma, we carried out in vivo anti-tumor experiments in lung cancer, pancreatic cancer and melanoma using tail vein injection of UHHTN/DOX nanorobots. In the subcutaneous tumor models of lung cancer, pancreatic cancer and melanoma (PANC-1, A549 and B16F10) in mice, UHHTN/DOX+laser treatment significantly inhibited tumor growth and led to almost complete elimination of subcutaneous tumors, which was much better than the UHHTN+Laser and saline groups (Supplementary Figs. 53-55), suggesting that UHHTN/DOX nanorobots can be applied to the treatment of other rigid tumors.
We next explored whether UHHTN/DOX could be eliminated from the body. During one treatment cycle after UHHTN/DOX injection, urine and feces samples were collected. Si was detected in the urine and both Au and Si were detected in the feces. the possible reason is that Si in UHHTN/DOX can be excreted through the kidneys, while a small amount of Si and Au can be excreted through liver processing and bile after ingestion of the liver, and finally through the feces (Supplementary Figs. 56). To detect the long-term toxicity of Au and Si, we took different organs of mice at the end
of the treatment to detect the content of Au and Si elements. It was found that a small amount of Au elements was present in the liver, spleen and lung (Supplementary Fig. 57). Finally, hematoxylin and eosin staining (H&E) of the major organs and hematological examination indicated that our nanorobots is biocompatible and no significant side effects were shown (Supplementary Figs. 58-63).
The effects of motion and TSM remodeling of nanorobot on extravasation and penetration
Although many studies have shown that the motility ability of nanorobots can effectively enhance tumor penetration and accumulation, when applied in vivo to access tumors with stiff stroma, the dense TSM around the tumor could impede the motion of nanorobots40, 41. It has been reported that photothermia could reduce the dense TSM, therefore we further verified whether UHHTN nanorobots can effectively remodel TSM that further contribute to the inhibitory effect of nanorobots on tumor growth. UHHTN-mediated photothermia was obviously triggered by laser irradiation for 10 min, leading to a temperature of approximately 55.5°C in the backs of mice (Supplementary Fig. 64).
The spinal tumors were then analyzed by histology and immunofluorescence to detect the changes in the TSM. Generally, HE staining showed that the ECM became loose with more extracellular space in the UHHTN+Laser and UHHTN/DOX+Laser groups compared to saline and the corresponding groups without laser irradiation (Fig. 7a), and meanwhile exhibited similar degradation efficacy, indicating the role of laser irradiation in regulating ECM composition. In addition, more necrotic tumor and stromal cells with karyorrhexis were observed in the nanorobots groups with laser irradiation. TUNEL staining and quantitative analysis showed that the apoptotic cells, including stromal and tumor cells with red fluorescence, were significantly enhanced in the UHHTN+Laser and UHHTN/DOX+Laser groups compared with the UHHTN and UHHTN/DOX groups (Fig. 7a and d).
Cancer-associated fibroblasts (CAFs) with α-SMA and CD31 markers and tumor-associated macrophages (TAMs) with F4/80 marker are the main stromal cells of the TSM barrier42, 43, CAFs and TAMs were abundantly disturbed in the spinal metastasis of MDA-MB-231 cells in the saline group (Fig. 7b). These cells were scarce in the UHHTN+Laser and UHHTN/DOX+Laser groups than in the UHHTN and UHHTN/DOX groups, indicating that stromal cells were significantly injured by the nanorobot-mediated hyperthermia. Importantly, the quantitative analysis showed that the MOD between the UHHTN+Laser and UHHTN/DOX+Laser groups was similar (Fig. 7e), excluding the effect of DOX on stromal cell death. The ECM surrounding stromal and tumor cells is a complex and dense network in TNBC, with collagen I and fibronectin as the dominant components42, 44. The green fluorescence intensity representing the density of collagen I and fibronectin was significantly lower in the UHHTN+Laser and UHHTN/DOX+Laser groups than in the group without laser irradiation, demonstrating the nanorobot mediated superior denaturation efficacy of ECM (Fig. 7c). The quantitative MOD values of collagen I and fibronectin in the UHHTN+Laser group were 52.67 and 137.2, respectively, in the UHHTN+Laser group and 44.48 and 136.6 in the UHHTN/DOX+Laser group, indicating that differences between these two groups were low (Fig. 7e), therefore the hyperthemia mediated by the nanorobots rather than DOX was the main factor for ECM degradation. The photothermal remodeling capability of nanorobots also functions in other three cellular subcutaneous tumor models in mice (A549, PANC-1 and B16F10) (Supplementary Figs. 65-67).
To prove whether TSM remodeling could open intratumoral pathways that sustain nanorobot motion in vivo and enhance accessibility to distal cancer cells for deeper tumor penetration. We next compared vascular extravasation and tumor penetration in vivo of nanorobots with TSM remodeling capability and passive core-shell structured AuNS@SiO2 nanoparticles with the same TSM remodeling capability but no motion performance. The real-time extravasation and tumour penetration in subcutaneous inoculation of the MDA-MB-231 tumor model was monitored. The result shows the nanorobots were rapidly diffusion into deep tumor tissue from the vascular system and distributed throughout the observation area. The entire tumour tissue showed a nearly equal fluorescence intensity during the 120 min. In contrast, AuNS@ SiO2 nanoparticles was still restricted in the blood vessels at 120 min post injection, and did not diffuse well into the tumor tissue (Fig. 7f). These results reveal that nanorobots with both motion and TSM remodeling have a superior tumour-penetration ability and can extravasate efficiently into distal tumour tissues, significantly contributing to the therapy efficacy.