GNR was obtained via a common seed-mediated growth strategy according to the previous reports (Shen, et al., 2014). After regulating the concentration of seeds, silver ions for size, and pH value for LSPR, ∼100-nm GNR with a peak at ∼900 nm was synthesized. Transmission electron microscopy (TEM) image results showed the GNR nanocrystals with uniform size and excellent dispersion (Figure 1B). According to a reported method of bio-phase (water/cyclohexane) system, organosilica mesoporous silica was coated on GNR (Yang, et al., 2013). During the silica-coating procedure, CTAB formed as the surfactant around the GNR and served as the single micelle template for the formation of the mesoporous organosilica silica shell. The ratio of conventional silica precursor (tetraethyl orthosilicate) and organosilica precursor [bis-(triethoxysilylpropyl) tetrasulfide)] was set as 4:1. After 48-h reaction at 60°C, oMSN-GNR with the size of ∼300 nm was successfully fabricated, as shown in the TEM image of Figure 1C, the synthesized core–shell nanosphere could be clearly observed, and the amorphous silica shell was evaluated to present a homogeneous thickness with ∼100 nm. Disordered mesopores of ∼5 nm in diameter could be distinctly estimated from TEM in Figure 1D and scanning electronic microscope image in Figure 1E, offering an ideal opportunity for oMSN-GNR to be performed as general drug delivery carriers. Moreover, the dynamic laser scanning result exhibited that the average size of oMSN-GNR was estimated as ∼315 nm (Figure 1F) with pore size evaluated as ∼5.7 nm after calculating by nitrogen adsorption branch of the isotherm via the density functional theory approach (Figure 1G). Importantly, insignificant variation of size distribution was found after different biological buffer treatments (such as PBS, 10% blood, fetal bovine serum, and cell culture medium) for different hours, suggesting the superior stability of GSH-sensitive mesoporous nanoshell nanoparticles (Supplementary Figure S1). Meanwhile, the specific surface area of as-synthesized oMSN-GNR was analyzed as ∼210.6 cm³ g⁻¹ through the conventional Brunauer–Emmett–Teller method (Figure 1H). Both mesoporous channels and appreciable specific surface area facilitated the therapeutic agents' loading and delivery. Subsequently, DOX was then loaded into the mesoporous channels of the organosilica shell at pH 7.4 via electrostatic interactions, and the oMSN-GNR–DOX complex was formed after a 12-h reaction. Owing to –Si-OH on the surface of silica, the zeta potential decreased after organosilica coating. Interestingly, zeta potential reversed into the negative side after drug encapsulation, which is mainly ascribed to the negative charge of DOX (Figure 1I). The results demonstrated that massive chemotherapy agents were successfully loaded, and the negative charge could facilitate cellular endocytosis. Simultaneously, ∼10-nm LSPR red-shift was observed after silica coating, and a further small red-shift (∼5 nm) absorption of the characteristic peak of DOX accompanying slight band damping was also observed in oMSN-GNR–DOX, further proving that DOX was successfully encapsulated in the mesoporous shell (Figure 1J).

A multifunctional drug carry system can directionally deliver drug molecules to tumor sites and should precisely release them under specific conditions of different stimuli in tumor cells (Yang et al., 2018; You et al., 2018). Among the different tumor-specific stimuli, imbalance of redox status in tumor cells is an effective initiator for the cytoplasm burst release of drugs. Due to this, redox responsive drug release systems have been developed extensively. Among the redox-sensitive bonds, the tetra-sulfide bond is the most commonly involved in redox disrupted linkages (Reddy, et al., 2005; Koo, et al., 2008). It can be very stable in an extracellular microenvironment with ∼2-μM GSH, but it is capable of being rapidly broken via exchange reactions of thiol-disulfide in the cytoplasm with ∼10-mM GSH (Cheng, et al., 2011; Bahadur, et al., 2014). Accordingly, organosilica shell reaction with GSH buffer was firstly investigated. As shown in Figures 2A–C, after 2-h GSH incubation and 808-nm laser irritation, silica shell became soft and loose and completely degraded with only residual GNR remaining after incubated with GSH buffer for 6 h, demonstrating a time-dependent biodegradable behavior under tumor redox microenvironment (Figures 2Ai–Ci). In contrast, in the control group of water incubation, the morphology of the oMSN has negligible transformation even for 6-h coculture (Figures 2Aii–Cii). In addition, the biodegradable rate of oMSN-GNR is faster than traditional mesoporous silica with total decomposition for nearly 5 days, possibly due to heat from GNR that facilitates the redox-sensitive reaction. Meanwhile, considering the GNR exhibits absorption coefficients, the photothermal effect for GNR-based nanocomposites was subsequently estimated in an aqueous solution after 808-nm diode NIR laser illumination. As we can clearly see in Figures 2D,E, after NIR laser irradiation for 2 min, the temperature of oMSN-GNR nanoparticles with concentration at 200 μg/ml drastically increased to ∼70°C and reached the maximum temperature at ∼75°C for 5-min exposure. In contrast, the pure water group maintained the room temperature, even after 5-min irradiation. Taking together, all these hyperthermia evidence proved that the core–shell nanoparticles had a surprisingly optimal photothermal effect for tumor ablation. DOX release behavior was then investigated by ultraviolet–visible spectrum under different buffers at 37°C. We found that in the absence of NIR laser irradiation and GSH treatment, the cumulative DOX release is ignorable even after 12-h incubation, suggesting that electrostatic interaction and π−π stacking force could induce little drug leakage in our system. Interestingly, in comparison with the GSH treatment group with ∼60.12% drug release in 4 h, a faster ∼89.1% DOX release was obtained after irradiation with a 300-mM NIR laser (with 808-nm wavelength) (Figure 2F). Except for GSH triggered silica framework decomposition, under the NIR laser irradiation, the DOX release behavior became the fastest for pure GSH and water treatment. This is mainly attributed to two aspects, laser-concerted hyperthermia could dissociate the strong connection between organosilica and DOX, ^[28] and furthermore, the corresponding heat was capable of boosting the reaction between GSH and the co-doped tetra-sulfide bond. Finally, the biocompatibility of GNR-based nanocarriers was studied against liver tumor cell lines. The viability of tumor cells maintained up to 90% even cocultured nanocarrier's concentration reached as high as 200 μg/ml for 24 h, proving the superior biocompatibility and its further application for in vivo tumor eradication (Figure 2G).

Inspired by the ideal biocompatibility, we explored the cellular internalization of our GNR-based nanocomposite toward the huh7 cell line. Firstly, DOX was absorbed on the GNR surface through charge adsorption (GNR-DOX), and common mesoporous silica-coated GNR was prepared for DOX loading (MSN-GNR–DOX). During the endocytosis experiment, the GNR concentration of GNR-DOX, oMSN-GNR–DOX, and MSN-GNR–DOX was equivalent. The cytoplasm fluorescence of DOX in GNR-DOX was remarkably lower than MSN-GNR–DOX and oMSN-GNR–DOX, resulting from unsatisfactory drug loading efficiency of free GNR (Figures 3A,B). Moreover, analogous DOX fluorescence can be observed in oMSN-GNR–DOX and common MSN-coated GNR-based DOX carriers, suggesting that tetra-sulfide bond doping in the silica scaffold had a negligible effect on drug loading capacity. Latterly, the cytotoxicity of different concentrations of oMSN-GNR–DOX toward huh7 cells with/without 808-nm laser irradiation was carefully studied. As shown in Figure 3C, concentration-dependent cell viability was detected in both groups. However, after 5 min of 808-nm NIR laser exposure, fewer tumor cells were alive, especially with high concentration treatment in comparison with no laser irradiation. This result could be ascribed to the chemotherapy enhancement via hyperthermia, demonstrating the advantage of the synergistic effect of PTT and chemotherapy in our drug delivery platform. Simultaneously, after being treated with PBS, free DOX, oMSN-GNR–DOX, and oMSN-GNR–DOX + laser, cells were then stained by apoptosis assay, annexin V–fluorescein isothiocyanate/propidium iodide before flow cytometry analyzing analysis. We found that in comparison with other groups, the oMSN-GNR–DOX + laser treatment group exhibited the highest percentage of apoptotic/necrotic tumor cells. In contrast, the percentage of apoptotic/necrotic tumor cells in the oMSN-GNR–DOX group was only one-third of that in the oMSN-GNR–DOX + laser treatment group (Figure 3D, Supplementary Figure S2). This flow cytometry results were consistent with cell killing assays, further demonstrating the optimal tumor cell ablation capability of a combination of PTT and chemotherapy.

Encouraged by the in vitro results discussed earlier, before in vivo tumor ablation investigation, we firstly estimated tumor passive targeting capability of our GNR-based nanocarriers via enhanced permeability and retention effect in huh7 tumor-bearing mice. To realize high penetration depth and low autofluorescence imaging modality, clinically approved ICG was encapsulated into the organosilica shells to act as NIR-II (the second NIR, 1,000–1,700 nm) fluorescent contrast agents (oMSN-GNR–ICG). After tail vein injection of oMSN-GNR–ICG, NIR-II signals in tumor site could be observed after 6-h post-injection. Surprisingly, tumor tissue had the strongest fluorescence intensity and distinct profile within 24-h post-injection. Meanwhile, tumor signals continuously weakened from this time point and still existed after intravenous injection at 72 h, suggesting that laser illumination for PTT should be carried out at this massive tumor accumulation time (Figure 4A). Moreover, the resected major organs and tumors also displayed that maximum tumor accumulation could be detected at 24-h post-injection (Supplementary Figure S3). In contrast, the free ICG group presented no NIR-II signals in the tumor site, demonstrating that oMSN-GNR–ICG was capable of predominantly improving the tumor-targeting ability and possessed longer tumor retention behavior (Figure 4A). Based on this result, we then collected the major organs and tumors after 36-h treatment of free ICG and oMSN-GNR–ICG to perform ex vivo NIR-II fluorescence imaging. Undoubtedly, tumor signals in GNR-based nanoparticle groups exhibited stronger signals in comparison with the free ICG group, which was consistent with in vivo NIR-II imaging results (Figure 4B). Finally, owing to the facile accessibility and simple manipulation of the photothermal camera and NIR laser device, photothermal imaging was performed on the mice after being intravenously injected with oMSN-GNR–ICG. The liver tumor exhibited a slight temperature increase in the PBS group (6.2°C) during the whole imaging durations. Contrarily, in the oMSN-GNR–ICG group, the tumor tissue exhibited significantly higher temperature at each time point in comparison with the PBS group. The highest value at 24-h post-injection was detected as 75.2°C, further demonstrating superior passive targeting behavior of this GNR core–shell nanoparticle (Figure 4C).

After determining the time point of the highest tumor targeting ability, the antitumor efficacy of oMSN-GNR–DOX was tested. Liver tumor-bearing nude mice were randomized into four groups (n = 4) which were undergone with the following administrations: PBS, DOX, MSN-GNR-DOX, and MSN-GNR-DOX + laser. The important data to estimate tumor therapeutic effects, including tumor weight, tumor volume, and body weight, were recorded every 3 days. The digital images presented that tumor size dramatically increased in the PBS group over the whole inhibition time. Compared with the PBS group, some inhibition effects had displayed in the free DOX and oMSN-GNR–DOX groups. Excitingly, the tumor almost disappeared in the oMSN-GNR–DOX + laser group, which was much more effective than the single chemotherapy group (Figure 5A, Supplementary Figure S4). Tumor weight and tumor volume curve showed that oMSN-GNR–DOX + laser treatment exhibited significantly higher tumor elimination capability (Figures 5B,C), showcasing excellent therapeutic efficiency when combining GSH-sensitive chemotherapy with PTT. Furthermore, tumor slices were prepared for H&E staining. As displayed in Figure 5E, the oMSN-GNR–DOX + laser group obviously exhibited massive tumor cell interspace and lots of abnormal cellular morphology, which was consistent with the TUNEL image, indicating that tumor cells suffered from synergistic cytotoxicity of hyperthermia and DNA damage in the oMSN-GNR–DOX + laser group (Figure 5F). Furthermore, in comparison with the oMSN-GNR–DOX + laser groups, partial antitumor effects could be observed in oMSN-GNR + laser, whereas no significant tumor inhibition performance was detected in the pure laser group, hinting at the ideal antitumor effect of synergistic therapy in oMSN-GNR–DOX + laser treatment (Supplementary Figure S5). Simultaneously, the 808-nm laser has no adverse effect on the skin after 5-min irradiation under 1 W/cm², hinting at the biosafety of this NIR laser for PTT (Supplementary Figure S6). No surrounding tissue damage was observed after PTT, indicating the little adverse effect of PTT against normal tissues (Supplementary Figure S7). More importantly, the bodyweight of all mice in four groups had little fluctuation during 15 days of treatment (Figure 5D); meanwhile, the main organs of each group were finally excised for pathological analysis. Compared with the PBS group, unnoticeable histomorphology changes were found in the other three groups, proving that our GNR-based nanocarriers had ignorable adverse effects, thus acquiring potential clinical application value in the future (Supplementary Figure S8). Finally, all corresponding blood biochemistry (Supplementary Figures S9A–E) and blood routine factors (Supplementary Figures S9F–J) in oMSN-GNR–DOX + laser after 15-day treatment presented negligible fluctuations in comparison with the PBS group, demonstrating the low systematic biotoxicity of oMSN-GNR–DOX for synergistic therapy.