Pyroptosis induction with nanosonosensitizer-augmented sonodynamic therapy combined with PD-L1 blockade boosts efficacy against liver cancer

Abstract

Induction of pyroptosis could promote anti-PD-L1 therapeutic efficacy due to the release of pro-inflammatory cytokines, but current approaches can cause off target toxicity. Herein, we design a phthalocyanine-conjugated mesoporous silicate nanoparticle (PMSN) for amplifying sonodynamic therapy (SDT) to augment oxidative stress and induce robust pyroptosis in tumors. The sub-10 nm diameter structure and c(RGDyC)-PEGylated modification enhance tumor targeting and renal clearance. The unique porous architecture of PMSN doubles ROS yield and enhances pyroptotic cell populations in tumors (25.0%) via a cavitation effect. PMSN-mediated SDT treatment efficiently reduces tumor mass and suppressed residual tumors in treated and distant sites by synergizing with PD-L1 blockade (85.93% and 77.09%, respectively). Furthermore, loading the chemotherapeutic, doxorubicin, into PMSN intensifies SDT-pyroptotic effects and increased efficacy. To our knowledge, this is the first report of the use of SDT regimens to induce pyroptosis in liver cancer. This noninvasive and effective strategy has potential for clinical translation.

Graphical abstract

We design a nanosonosensitizer, PMSN, for amplifying sonodynamic therapy (SDT). The sub-10 nm structure and ligand modification enhance tumor targeting and renal clearance. The unique porous architecture of PMSN doubles ROS yield via a cavitation effect and enhances pyroptotic cell populations in tumors. PMSN-mediated SDT efficiently reduces tumor mass in treated and distant sites by synergizing with PD-L1 blockade.

1. Introduction

Despite the enormous therapeutic potential of immune checkpoint inhibitors (ICIs), such as anti-PD-L1 agents, many patients do not benefit from these treatments due to the side effects and limited efficacy in some applications. Converting ‘immune-cold’ tumors into ‘immune-hot’ tumors can synergize with ICIs by promoting tumor antigen release, initiating antigen-processing cells, and increasing T-cell infiltration[1, 2]. Yet, safe and universal therapeutic platforms to promote immunity remain limited[3]. Thus, novel immunogenic treatment strategies are required to generate antitumor immunity and boost anti-PD-1 efficiency.

Recently, the induction of pyroptosis has drawn much attention due to its potency to trigger robust and durable anti-tumor immune responses for efficient inhibition of cancer development and occurrence by releasing antigens, lactate dehydrogenase (LDH) and pro-inflammatory mediators, such as interleukin (IL)-1β and IL-18[4]. However, pyroptosis-based immunotherapy has certain challenges. Pyroptosis is an inflammatory form of gasdermin-executed lytic programmed cell death. Gasdermin is expressed ubiquitously, and its expression level in most cancer cells is usually lower than in normal cells. Several anticancer therapeutic modalities, such as radiation therapy, chemotherapy (i.e. doxorubicin and cisplatin) [5–7] or granzymes (GzmA or GzmB) [8, 9], can induce pyroptosis. Many groups have also demonstrated the correlation between reactive oxygen species (ROS) and pyroptosis induction via Caspase 3/GSDME [10, 11]. PDT[12–15] and some ROS-related treatments[16] have been reported to produce pyroptosis in cancer to enhance inflammation and increase immune cell activity. However, the therapeutic agents without tumor targeting are inevitably distributed into normal tissues, triggering pyroptosis and causing systemic side effects and toxicity. Although acute pyroptosis in the tumor microenvironment would enhance antitumor immunity and suppress tumor progression, chronic tumor inflammation resulting from chemotherapy can cause central hypoxia and accelerate tumor development[17]. Therefore, additional efforts are required to induce acute and tumor-specific pyroptosis to minimize undesirable side effects during treatment.

As a promising oncologic intervention option, sonodynamic therapy (SDT) produces ROS, especially singlet oxygen that is recognized as the major cytotoxic agent[18], by activating sonosensitizers upon application of low-intensity focused ultrasound (US) for irreversible ablation of cancer cells[19]. The PDT and chemotherapy-induced ROS is a driving force of pyroptosis, and some studies have also showed the potential of SDT to trigger an immune response [20, 21]. Acoustic cavitation is the growth and collapse of bubbles in liquids under an ultrasonic field. This process releases a large amount of energy, intensifying and speeding up sonochemical reactions of sonosensitizers, and increasing the formation of sensitizer-derived free radicals[22, 23]. Therefore, with a greater penetration depth of ultrasound waves as compared with PDT, SDT could have potential to induce pyroptosis by robust ROS generation in deeply target lesions [24], particularly in liver cancer. This noninvasive and targeted treatment to induce pyroptosis would have a greater advantage over other approaches. At the present time, SDT-induced pyroptosis in liver cancer has not been reported, possibly due to the lack of targeting guidance for SDT application in deeply located tumors in preclinical models and efficacious sonosensitizers, which can be delivered at a high concentration to the tumor.

Conventional sonosensitizers, such as phthalocyanines, have limited tumor targeting ability, hepatotoxicity due to first pass effect and insufficient ROS generation in tumors. To address these limitations, we design and construct a tumor-targeted nanosonosensitizer built from phthalocyanine-conjugated mesoporous silicate nanoparticles (PMSN) with a sub-10 nm diameter and a c(RGDyC)-PEGylated modification to amplify sonosensitization for possible therapeutic effects. These structural modifications improve the pharmacokinetics over phthalocyanine (Pc) alone, attenuates systemic toxicities, enhances targeting, penetration and accumulation in the tumor, and hence induces ROS accumulation (Figure 1a). Furthermore, we intensify the SDT process and ROS production by loading doxorubicin (DOX) into PMSN (DPMSN), and present a DPMSN-triggered chemo-SDT combinational strategy for treating tumors with increased pyroptotic cell deaths and boosting anti-PD-1 efficiency. Ultrasound stimulation is also optimized to augment ROS production for precise and acute pyroptosis using murine liver tumor models.

Figure 1: Characterization of PMSN and its ROS generation efficiency.

a Schematic illustration of PMSN structure, and mechanisms of nanosonosensitizer to augment ROS production by ultrasound stimulation. The PMSN was composed of the silicate network by conjugating Cyclo (Arg-Gly-Asp-D-Tyr-Cys) (c(RGDyC))- Polyethylene glycol (PEG)-, PEG550- and Pc-silane. b Particle diameter of final PMSN sample via dynamic light scattering (DLS) measurement. c X-ray diffraction (XRD) pattern of PMSN from 2 theta of 15 to 90 degrees. d Isotherm Log Plot showing nitrogen absorption and desorption in relation to the relative pressure. e Pore size measurement by determination of the specific surface area of solids by gas adsorption using porosimeter. f PMSN morphology and dispersion with transmission electron microscopy (TEM) and scanning TEM (STEM) observation (Scale bar: 20 nm). g X-ray photoelectron spectroscopy (XPS) of the PMSN showed the elemental composition including both Aluminum (Al) of Pc and Silicon (Si) of silicate networks. h-i ROS yield efficiency via measuring Singlet Oxygen Sensor Green (SOSG) fluorescence intensity over 2 min in PMSN compared to various controls. PMSN contains phthalocyanine (Pc) and a mesoporous structure, Pc is the phthalocyanine molecule, MSN contains a mesoporous structure but without phthalocyanine conjugation, and PSNP contains phthalocyanine but with a solid structure. PBS: phosphate buffered saline.

2. Results and Discussion

2.1. Synthesis and characterization of PMSN nanosonosensitizers

The sonosensitizer of PMSN was synthesized by combining tetramethyl orthosilicate (TMOS) and aluminum phthalocyanine tetrasulfonate hydroxide (AlS4Pc-OH). The AlS4Pc-OH was converted from aluminum phthalocyanine tetrasulfonate chloride (AlS4Pc-Cl), which is a common hydrophilic photosensitizer[25], and would serve as comparison of sonodynamic efficacy in this study (Figure S1). For tumor targeting and cellular internalization, the surface of PMSN was modified with c(RGDyC)-PEG-silane (Figure S2-3). The doxorubicin (DOX)-loaded PMSN (DPMSN) was prepared by combining TMOS, AlS4Pc-OH and DOX similarly. The as-prepared PMSN and DPMSN with sub-10 nm size and c(RGDyC)-PEGylated modification are favorable for renal excretion, specific tumor targeting and altering sonosensitizer bio-distribution[26].

We verified the morphological properties of PMSN. Figure 1b and Figure S4 show average hydrodynamic diameters of 7.9 nm for PMSN, and 8.0 nm for DPMSN by dynamic light scattering (DLS); these measurements are slightly bigger than those obtained from the transmission electron microscope (TEM) technique, since TEM is insensitive to the PEG layer and the surrounding water layer, in contrast to the DLS measurements. Mesoporous silicate nanoparticles without Pc conjugation (MSN) and Pc-conjugated silicate nanoparticles with solid structure (PSNP) were also prepared and characterized (Figure S5). X-ray powder diffraction (XRD) patterns at a two-theta degree of 22° revealed the relatively uniform and mesostructured silicate particle of PMSN (Figure 1c). Based on Brunauer-Emmett-Teller (BET) analysis, the PMSN surface area was 844 m²/g, 7.5-fold greater than that of PSNP (113 m²/g). The Barrett-Joyner-Halenda’s (BJH) plot presented the type IV isotherm and type H1 hysteresis loop, indicating the connectivity of the pores in PMSN (Figure 1d). The average PMSN pore size was 3.4 nm (Figure 1e). TEM and STEM images demonstrated the ultrasmall size and mesoporous structure of PMSN (Figure 1f). These results demonstrate that PMSN has a greater potential for gas storage in hydrophobic cavities through hydrophilic PEG modification. When exposed to ultrasound, the preserved gas forms echogenic bubbles that undergo cavitation. The oscillating bubbles generate shock waves, microjets and intense shear forces that can break chemical bonds and enable sonochemical reactions [27, 28]. Therefore, PMSN is more likely to induce cavitation and improve the sonodynamic effect.

X-ray photoelectron spectroscopy (XPS) was used to analyze the PMSN elemental composition and Pc loading capability. Both Al of Pc and Si of silicate networks were observed in PMSN (Figure 1g). Additionally, ²⁷Al and ²⁹Si solid-state nuclear magnetic resonance (NMR) spectra showed the Al-O-Si bond linkage site (Figure S6), providing powerful evidence for the efficient conjugation of Pc on the silicate network [29]. Based on thermogravimetric analysis (TGA) and calculation of molecular weights, the Pc-loading capacity and conjugation efficiency of PMSN was evaluated to be 3.7% and 96%, respectively (Figure S7). There were about 4.6 Pc molecules and 4.0 c(RGDyC)-ligands in each PMSN particle. As displayed in Supplementary Table 1, the zeta potential of both PMSN and DPMSN was approximately neutral, and the lyophilized samples could be stored at room temperature for weeks with good stability.

2.2. Optimization of ultrasound parameters for augmenting SDT

The efficiency and extent of sonochemical effects are often influenced by ultrasound parameters (frequency and intensity) and the presence of sonosensitizers (Pc). Cavitating bubbles synergize with sonosensitizers to enhance sonochemical reactions. Thus, by controlling these factors, we can optimize the sonochemical reactions and enhance their outcomes.

To optimize the ultrasound conditions, we investigated the energy transfer to the PMSN aqueous solution from ultrasound transducers with different center frequencies (0.5, 1, 1.5 and 2.25 MHz). Acoustic attenuation is the energy loss when an ultrasound beam passes through a sample (i.e., tissues or sonosensitizers), resulting in a reduction in the intensity and amplitude. The attenuation amplitude depends on attenuation coefficient of the sample [27, 30]. When the acoustic pressure increased, the attenuation resulting from transmission through Pc and mesoporous silicate nanoparticles (MSN) increased, indicating that attenuation resulted from both cavitation and sonosensitizer activation. In particular, cavitation played a more important role in acoustic attenuation of MSN. With increased acoustic cavitation, the sonosensitizer is expected to produce more efficient sonochemical reactions (Figure S8-9). Cavitation dynamics of air bubble nuclei developed from the mesopores of PMSN could be explained by the Gilmore model[31]. Thus, the sonochemical reaction via absorbing sound energy was more sensitive to the ultrasound waves at lower fundamental frequency (500 kHz).

The ROS production yields of PMSN were investigated upon insonation at different ultrasound parameters. It was found that PMSN exhibited the highest ROS production yield in the condition of 400 kPa and 25% duty cycle (Figure S10-11). The temperature elevation was monitored during the SDT via infrared (IR) thermometer. From the temperature change, the energy used on ROS production was less. Mechanical force could improve the cavitation and amplify the energy absorption and electron transfer in sonosensitizers, while thermal effect could facilitate the reaction. If the acoustic power is beyond a certain safety threshold, tumor cells are directly impacted rather than inducing a sonodynamic effect. High temperature could cause cell deaths in both the surrounding tissues and the tumors. Hence, there was a plausible tradeoff between ROS production and heat release in order to efficiently kill tumor cells by SDT instead of by overheating during ultrasound treatment. Less than 40% of HepG2 cancer cells were viable after the PMSN-mediated SDT treatment (frequency = 400 kPa, duty cycle = 25%). Yet, cell damage was also observed upon stronger insonation without PMSN incubation, indicating undesired side effects from ultrasound treatment with higher PNP and duty cycles (Figure S12). To avoid such side effects, the optimized ultrasound parameters (frequency = 500 kHz, PNP = 400 kPa, and duty cycle = 25%) were chosen for the subsequent study.

We then investigated the ROS production yields of PMSN in comparison with phosphate buffered saline (PBS), Pc, MSN with no Pc loading, and phthalocyanine-conjugated silicate nanoparticles (PSNP) with no mesopores. Upon ultrasound stimulation, the ROS yield of PMSN was 2.52, 2.51, 1.96 and 1.46 times that of the PBS, MSN, Pc and PSNP control treatments, respectively (Figure 1h-i). The air preserved in pores can form bubbles and serve as nuclei for cavitation with appropriate ultrasound conditions[28]. Cavitation can reinforce the sonodynamic effect of Pc and elevate ROS generation. Further, silica particles covalently encapsulating Pc can contribute to the shield of the silicate matrix, preventing energy dissipation from sensitizer aggregation and augmenting ROS generation consequently[32]. Therefore, PMSN-based SDT offered an unprecedented opportunity to cause excessive oxidative stress due to their excellent capability of ROS production as novel sonosensitizers.

2.3. Biodistribution, biocompatibility and SDT efficacy of PMSN

An ideal sonosensitizer should precisely accumulate at tumor sites after blood circulation with limited distribution in other healthy tissues (Figure 2a). Because integrin αvβ3 has been reported to be overexpressed in cancer and to mediate invasion and metastasis, ligand modification of c(RGDyC) allowed targeting of PMSN to integrin αvβ3-positive tumors and facilitated internalization [33]. We have also proven that with the modification of c(RGDyC), PMSN had higher cellular uptake in the liver cancer cell lines with high integrin αvβ3 expression (HepG2 and Hepa1–6) than those with low expression (WRL-68) (Figure S13). PMSN biodistribution and elimination were studied via in vivo fluorescence imaging and quantitative measurement of Pc fluorescence intensity in tumor and urine (Figure 2b-c). The PMSN-treated mice presented strong Pc fluorescence in tumors that increased with time, reaching a maximum intensity at 4 h after i.v. injection and then declining with blood clearance from the tumors. Alternatively, the free Pc-treated mice presented faint Pc fluorescence signals in the tumors, due to the short circulation time and rapid blood clearance of Pc in vivo. Figure S14a displayed the PMSN renal clearance via live imaging, and PMSN quantification in urine at 4 h was found to be 58% of the injected dose (%ID). The urine was collected at 4 h, and the particles were extracted for further TEM sample preparation. The TEM results with energy-dispersive X-ray spectroscopy (EDS) analysis presented the intact PMSN structures with retentive Pc conjugation (Figure S14b-d). The PMSN biodistribution in tumors was quantitatively measured ex vivo at 4 h after administration (Figure 2d-e). We found that PMSN achieved 15.47-fold accumulation at the tumor site compared to the free Pc molecules (p <0.005), indicating that PMSN could pass through the glomerular capillary walls and be eliminated from the kidneys. Meanwhile, hematoxylin and eosin (H&E)-stained sections did not show detectable organ damage or inflammation, and the blood biochemical indices indicated the functional capacity of liver and kidneys was within the standard value range (Figure S15). Therefore, PMSN-based SDT could be an efficient therapeutic regimen with limited damage to normal bodily function.

Figure 2: SDT efficacy of PMSN in an orthotopic liver cancer primary tumor mouse model.

a Schematic illustration of targeted ROS generation in murine liver tumor by applying sonodynamic (SDT) therapy. b Live imaging over 24 h of orthotopic tumor-bearing mice after i.v. administration of Pc and PMSN, c quantification of Pc fluorescence intensity showing the targeting ability and accumulation of PMSN in the liver tumor. d Ex vivo imaging of dissected tumors and major organs and e the quantification of their corresponding Pc fluorescence intensity. (Abbreviations: He: heart; Li: liver; Sp: spleen; Lu: lung; Ki: kidneys; Tu: tumors; Mu: muscles; Sk: skin.) f Therapeutic regimen of PMSN-mediated SDT in orthotopic liver cancer mice. SDT treatment began when bioluminescence intensity reached 10⁵ p/s/cm²/sr), approximately 10 days after tumor inoculation. g Representative bioluminescence images showed the orthotopic liver tumor growth on Day 0 (first-treatment time) and Day 35. h Tumor growth curves via measuring bioluminescence of the groups of mice treated with different drugs with and without ultrasound application. Statistical analysis was performed by one-way ANOVA followed by Tukey’s multiple comparisons test (n=3 for each group). i Corresponding hematoxylin and eosin (H&E) stained tissue slices showing the orthotopic liver tumors dissected on Day 35 (Scale bar: 500 μm).

Furthermore, comparative SDT treatments were performed in orthotopic liver cancer mouse models to assess tumor growth suppression (Figure 2f). By measuring the bioluminescent signal of HepG2 liver cancer cells, tumor growth was recorded over 35 days following the different treatments in each group (n=3) (Figure 2g). Figure 2h shows decreased bioluminescence right after SDT and dim bioluminescence signal in the PMSN-SDT treated mouse group. As shown in the tumor growth curves, PMSN +US significantly suppressed tumor progression compared with PBS, Pc, MSN and PSNP cohorts (p<0.005). H&E staining of the dissected tumors from each group on Day 35 also showed notable tumor size reduction after SDT (Figure 2i). In the group of PMSN +US, the remaining tumors were negligible. Meanwhile, for all groups of mice, there was no obvious bodyweight loss (Figure S16). These findings suggested that PMSN-based SDT significantly inhibited liver cancer growth in orthotopic murine tumor models, and neither the sonosensitizers nor ultrasound had adverse effects on mice.

2.4. Induction of targeted pyroptosis with PMSN-augmented SDT

Increased intracellular ROS can cause DNA damage, cell cycle arrest and mitochondria damage, and then trigger apoptosis, pyroptosis and other types of cancer cell deaths (Figure 3a). Besides, multiple types of cell deaths can be observed simultaneously after exposure to the same stimulus. Apoptosis is generally non-inflammatory and does not trigger an inflammatory response, whereas pyroptosis is highly inflammatory and associated with the release of pro-inflammatory cytokines, such as IL-1β, leading to an immune response.

Figure 3: PMSN enhanced sono-pyroptosis in HepG2 tumor cells and a murine orthotopic liver cancer model with optimized ultrasound parameters.

Ultrasound pressure and duty cycle were optimized for 500 kHz treatment. a Illustration of pyroptosis induction after ROS generation with SDT treatment. b Western blot results and c-f quantification of relative pyroptosis-related protein expression (Caspase 3, GSDME-N, Caspase 4 and GSDMD-N) in HepG2 tumor cells treated with PMSN in combination with different ultrasound pressures (200–600 kPa) and duty cycles (10–50%). g Western blot results and h-k quantification of pyroptosis-related protein expression in HepG2 tumor cells treated with PMSN or control treatments (PBS, Pc, MSN and PSNP) with and without ultrasound (400 kPa, 25% duty cycle). The PMSN contains phthalocyanine (Pc) and a mesoporous structure, Pc is the phthalocyanine molecule, the MSN contains a mesoporous structure but without phthalocyanine conjugation, and the PSNP contains phthalocyanine but with a solid structure. Data are presented as mean ± SD, and statistical analysis was performed by one-way ANOVA followed by Tukey’s multiple comparisons test for c-f and h-k (n=3 for each group). l In the HepG2-implanted orthotopic liver cancer mouse model, mice were treated with optimized ultrasound conditions (400 kPa, 25% pf duty cycle). Heatmap of pyroptosis-related gene expression in HepG2 orthotopic tumors collected from the non-treated control (NTC) mouse group (NTC) and the PMSN-mediated SDT mouse group, presented as normalized z-score of raw gene counts. Data are presented as mean ± SD. Statistical analyses were performed using Student’s t test (two-tailed). Asterisks (*) in l represent different levels of significance between groups: *p <0.05, **p<0.01, *** p<0.001 and **** p< 0.0001.

Therefore, we hypothesized that SDT enhanced with a nanosensitizer could result in ROS levels sufficient for inducing biomolecular changes related to cell death. To answer this hypothesis, we studied the impact of the nanosonosensitizer-augmented SDT on the expression levels of the cell-death-related caspases (i.e., Caspase 3/4/9), cleaved gasdermin (i.e., GSDME-N/GSDMD-N) and poly-ADP-ribose polymerase (PARP) in HepG2 cells. The above optimized ultrasound conditions in the sonochemical reaction were used for efficient pyroptosis induction. After incubation of cancer cells with PMSN for 3 h, the cell samples were treated by ultrasound with different PNP (200–600 kPa) and duty cycles (10–50%). The western blot results of pyroptosis-related protein expression revealed that the ultrasound thresholds for efficient pyroptosis induced by PMSN were 400 kPa PNP and 25% duty cycle (Figure 3b-f). As a sonosensitizer, PMSN displayed significantly higher expression of pyroptosis-related proteins compared to the PBS, Pc, MSN and PSNP control treatment groups (Figure 3g-k). In the PMSN+US group, Caspase 3/4 was more significantly elevated and more GSDME/GSDMD were activated and cleaved than in the other treatment groups tested. Meanwhile, the pyroptotic proteins of Caspase 4/GSDME exhibited higher dependency on the situations induced by PMSN-mediated SDT. This process could be mediated by the generation of mitochondrial ROS [34, 35]. It was noted that Caspase 3 was also involved in Caspase9/PARP-mediated apoptotic pathways (Figure S17a-b). Therefore, our study showed both apoptosis and pyroptosis occurred after SDT in HepG2 cells, supported by the cleavage of both gasdermin and PARP, both GSDMD and GSDME participated in pyroptotic effect, and PMSN-mediated SDT caused mitochondria ROS to induce GSDMD cleavage.

In many studies, pyroptosis has shown anticancer activity through suppressing the tumor growth in liver cancers[36], and the pyroptosis pathway is driven by intracellular sensor proteins, including NLRP1, NLRP3, NLRC4, AIM2, and Pyrin[37]. The activation of the NLRP3 inflammasome was integral for the activation of pyroptosis, further leading to caspase-1 recruitment, gasdermin cleavage, and stimulating the release of IL-1β and IL-18. A comparison of the gene RNA levels in HepG2 tumors within the non-treated control group and the PMSN+US treated group are presented in Figure S19a. To clarify the specific gene function, we analyzed the enriched genes in the gene ontology (GO), including biological process, molecular function, and cellular components. We found that relevant genes from PMSN-mediated SDT were mainly involved in interleukin-1 production, positive regulation of superoxide anion generation, positive regulation of inflammasome complex regulation of inflammatory response, pyroptosis, cysteine-type endopeptidase activity involved in apoptotic process, cysteine-type endopeptidase activator activity, and cytokine receptor binding in GO analysis (Figure S19b). Additionally, among the genes, we selected 23 pyroptosis- (Casp3, Casp4, Casp1, Casp9, Casp8, Gsdme, Gsdmd, Gsdmc, Gsdma, Nlrc4, Nlrp3, Pycard, Naip2, Aim2 Pjvk and Nlrp1) and pro-inflammatory cytokine- (Il18, Il1b, Tnf, Il6, Ila, Csf2 and Csf1) related genes[38] from identified differentially expressed genes (DEGs). After PMSN+US treatment, Casp3, Gsdme, Naip2 and Il18 had significant upregulation (p <0.005), Casp4, Casp1, Casp9, Casp8, Gsdmd, Nlrc4, Nlrp3, Il1b and Tnf were also enriched in tumors (p<0.05), while Pjvk and Nlrp1 were downregulated (p<0.05) when comparing to the NTC group (Figure 3i). The execution of necroptosis is mediated by mixed lineage kinase domain-like (MLKL) oligomers in the plasma membrane, whereas gasdermin mediates pyroptosis. However, we did not find significant changes in Mlkl and necroptosis/necrosome-related genes (Ripk1/Ripk3)[39]. We explored the signatures directly related to pyroptosis and detected that the NLRP3/CASP4/GSDMD pathway-related pyroptosis was activated, implying that pyroptosis participated in the mechanism of PMSN-mediated SDT treatment, which is capable of inducing pyroptosis and elevating the inflammatory level in tumors (Figure 4a).

Figure 4: PMSN-enhanced SDT resulted in cell death through both pyroptosis and apoptotic cell death in ex vivo tumor cells and in the orthotopic HepG2 murine tumor model.

a Schematic illustration of SDT inducing apoptosis and pyroptosis. b Images showing ROS production (Scale bar: 100 μm) and c pyroptosis (Scale bar: 30 μm) in HepG2 cells treated with PMSN-mediated SDT. HepG2 cells were treated by PMSN or other control treatments (PBS, Pc, MSN, PSNP) with or without ultrasound. PMSN contains phthalocyanine (Pc) and a mesoporous structure, Pc is the phthalocyanine molecule, MSN contains a mesoporous structure but without phthalocyanine conjugation, and PSNP contains phthalocyanine but with a solid structure. d Quantification of cell viability in HepG2 cells treated with PMSN or control treatments with or without ultrasound isonation. e Annexin V-FITC- and PI-positive cell populations in each sample were measured by flow cytometry. PI (propidium iodide) signal is displayed on the Y-axis and Annexin-FITC signal is on the X-axis. f PI enhancement in cells treated with different drugs and with ultrasound application. g Percentage of Annexin V⁺/PI⁺ cells after different drug treatments with ultrasound. h Schematic illustration of pyroptotic cell population enhancement in tumors after SDT treatment. i Representative immunofluorescence images (Scale bar: 100 μm) showing TUNEL (green) and Caspase 4 (red) positive cells in HepG2 orthotopic liver tumor slices stained with DAPI as a nucleus indicator (blue); the images in the second row are the magnified areas indicated by the white-dashed squares in the corresponding images above (Scale bar: 25 μm). Quantification of TUNEL-positive (j) and Caspase 4-positive (k) cell populations in HepG2 liver tumors treated by SDT with different drugs. Data are presented as mean ± SD, and statistical analysis was performed by one-way ANOVA followed by Tukey’s multiple comparisons test for d, g and j-k (n=3 for each group).

The PMSN-based SDT had remarkably higher ROS production than the Pc-, MSN- or PSNP-based SDT (Figure 4b, Figure S17c). The outcomes of cell morphologies, forms of cell death and cell viability after SDT played a decisive role in evaluating the SDT performance of PMSN. Observing the difference in cellular morphology after treatment of different drugs with or without ultrasound, we found severe pyroptosis occurred in the PMSN+US-treated cells, whereas the control groups exhibiting low ROS generation had negligible pyroptotic cell death (Figure 4b-c). The phenomenon demonstrated the correlation of ROS and pyroptosis, which was consistent with the expression of Caspases and gasdermin. We also characterized LDH and IL-1β in the HepG2 cell supernatants. As shown in Figure S18, LDH release was induced after SDT with PMSN (15.36%), whereas the response was not significantly changed in other control groups. In agreement with the LDH level, the measured IL-1β release in PMSN+US-treated HepG2 cells was also substantially induced (35.26 pg/mL), compared to only 0.97, 10.95, 4.27 and 14.50 pg/mL in the PBS+US, Pc+US, MSN+US and PSNP+US control groups, respectively.

Furthermore, CCK-8 indicated the decreased cell viability after SDT. Only 10.4% of cancer cells were viable in PMSN+US treatment group, much lower than that found with other treatments (p<0.005) (Figure 4d). It is also critical to determine the cellular fraction undergoing pyroptosis in the tumor after SDT treatment (Figure 4h). The in vitro SDT-treated cells were stained by Annexin V-FITC and propidium iodide (PI) for flow cytometry. Based on the observation of the morphological changes in the cancer cells, we determined that most of the cells were undergoing apoptosis (cell shrinkage and membrane blebbing) and pyroptosis (cell swelling with protrusions)[39]. Thus, we used Annexin V-FITC and PI double staining to identify pyroptotic cell death (Annexin V-FITC⁺/PI⁺ cells[4, 11, 40]) (Figure 4e-g) versus apoptotic cells (Annexin V-FITC⁺/PI⁻ cells [41]) (Figure S20). The total percentage of dead cells (Annexin V-FITC⁺ cells) was about 77.7% after the PMSN-based SDT treatment. The percentage of pyroptotic cells was 18.7% in the PMSN +US group, significantly larger than that of PBS (1.04%), Pc (6.79%), MSN (1.39%) and PSNP (8.32%) with ultrasound treatment, and PMSN (5.83%), PBS (0.28%), Pc (0.65%), MSN (1.97%) and PSNP (2.67%) without ultrasound treatment (p<0.005). Besides pyroptosis, SDT could also cause cancer cell apoptosis. The percentage of apoptotic cells was 54.7% in the PMSN+US group, significantly higher that of PBS (0.96%), Pc (12.1%), MSN (1.98%) and PSNP (16.5%) with ultrasound treatment, and PMSN (2.33%), PBS (0.18%), Pc (1.71%), MSN (1.39%) and PSNP (2.14%) without ultrasound treatment (p <0.005).

After the in vivo SDT treatment in orthotopic HepG2 tumor mice, pyroptosis in tumors was evaluated by immunofluorescence (IF) staining of Caspase 4 and Terminal-deoxynucleotidyl Transferase Mediated Nick End Labeling (TUNEL) (Figure 4i, Figure S21). Caspase 4 is the key component associated with pyroptosis[42, 43], and we also found the dependency of Caspase4/GSDMD for PMSN-mediated SDT in vitro. In the IF images and the quantitation of positive cells, it was found that Caspase 4⁺ pyroptotic cells occupied 14.3% of the cells in the PMSN+US group, much higher than that of the PBS (0%), Pc (0.5%), MSN (0.24%) and PSNP (0.14%) groups with ultrasound insonation, and the PMSN (0.75%), PBS (0%), Pc (0%), MSN (0%) and PSNP (0.22%) groups with no ultrasound insonation (p<0.01). TUNEL results showed programmed cell death. TUNEL is an assay that mainly detects apoptosis; some pyroptotic cells with less DNA fragmentation will also be TUNEL⁺. The percentage of TUNEL⁺ cells in tumors was, on average, 85.7% in the PMSN+US group, remarkably higher than that of the PBS (0.43%), Pc (4.26%), MSN (1.16%) and PSNP (7.64%) groups receiving ultrasound, and the PMSN (2.05%), PBS (0.18%), Pc (0%), MSN (0%) and PSNP (0.72%) groups not receiving ultrasound (p<0.005) (Figure 4j-k, Figure S21).Therefore, the PMSN-mediated SDT resulted in notable cell death through pyroptosis and apoptosis, showing potential therapeutic opportunities for further liver cancer suppression. As expected, both pyroptosis and apoptosis were observed during PMSN-augmented noninvasive SDT, with more involvement of pyroptosis in cell-death pathway.

2.5. SDT-induced pyroptosis to enhance aPD-L1 efficacy in mice

It has been proven that pyroptosis-mediated inflammation can induce anticancer immunity[4, 9]. The anticancer efficacy and ability of PMSN-mediated SDT to initiate an abscopal immune response were next investigated here by performing SDT in mice bearing subcutaneously transplanted bilateral Hepa1–6 liver tumors, where ultrasound was only applied to one tumor (Figure 5a). The tumors were harvested for H&E and immunohistochemical (IHC) staining one day after SDT treatment. As shown in Figure S22, the expression of PD-L1 was found to be higher in the tumor subjected to SDT compared to the tumor not receiving ultrasound. Nuclear deletions were clearly found upon analysis of H&E-stained tumor slices in the PMSN+US group, even in the distant tumors (Figure 5b).

Figure 5: PMSN-enhanced SDT increased immune cell infiltration in tumors in the bilateral subcutaneous Hepa1–6 mouse model.

a Schematic illustration of cell pyroptosis improved immune cell infiltration in tumor sites. Following PMSN-mediated SDT, tumors were harvested for immunohistochemistry (IHC) or flow cytometry after 24 h. b Representative hematoxylin and eosin (H&E) images and c-e representative IHC images (Scale bar: 200 μm) of treated and distant Hepa1–6 tumors. f-g Quantified percentages of immune cell populations by flow cytometry analysis, including CD45⁺ (tumor-infiltrating lymphocytes) TILs, CD4⁺ T cells, CD8⁺ T cells, CD19⁺ B cells, F4/80⁺ tumor-associated macrophages (TAMs), CD64⁺ TAMs, NK1.1⁺ natural killer (NK) cells and CD11c⁺ dendritic cells (DCs), in both the treated and distant tumors of mice in the different treatment groups. Data are presented as mean ± SD. Statistical analysis was performed by one-way ANOVA followed by Tukey’s multiple comparisons test (n=3 for each group).

Pyroptotic cells release many cytokines and provide danger signals to affect tumor proliferation and recruit immune cells infiltration into tumors. As assessed by enzyme linked immunosorbent assay (ELISA), the expression of corresponding cytokines in mouse serum in the PMSN+US mouse group was upregulated compared to the PBS+US mouse group and showed a 5.38-, 2.94- and 5.02-fold increase of TNF-α, IFN-γ and IL-6, compared to the PBS+US group (p<0.0001), respectively (Figure S25). IHC staining revealed typical immune-cell infiltration (CD8⁺, CD11b⁺, CD68⁺ cells) (Figure 5c-e, Figure S23a). Furthermore, flow cytometry analysis (Figure 5f-g) using the gating strategy shown in Figure S24 for tumor-infiltrating lymphocytes (TILs) showed that, in PMSN+US-treated mice, 5.09% and 2.37% of TILs were CD8⁺, and 20.54% and 3.27% of TILs were CD4⁺ T cells in the ultrasound-treated and distant tumors, respectively. In the ultrasound-treated tumors, B-cell populations were also increased to 3.99% after SDT with PMSN, compared to the PBS+US control group (0.08%). After PMSN+US treatment, TAMs in both the ultrasound-treated and distant tumors were 33.02% and 20.59% of TILs, respectively, compared to only 5.51% and 9.87% of TILs in the PBS+US control group. M1 TAMs (CD64⁺/F4/80⁺) in TILs in treated tumors were also increased from 0.19% (PBS+US group) to 3.28% (PMSN+US group). Meanwhile, natural killer (NK) cells and dendritic cells (DCs) were also recruited to the ultrasound-treated tumors (0.13% and 0.52% of TILs, respectively) in the PMSN+US group. NK cells and DCs were both scarce in the PBS+US control group. CD8⁺ T cells showed significantly increased recruitment (2.36%, p<0.05) in the distant tumor with the PMSN+US treatment compared to the PBS+US control (0.28%). Because of upregulated ROS levels by the PMSN-mediated SDT treatment, tumor cells possessing an immunosuppressive environment would most likely evolve into non-inflammatory apoptosis and even pro-inflammatory pyroptosis or other forms of programmed cell death associated with the inflammasome response [44]. Therefore, PMSN could serve as a powerful nanosonosensitizer for SDT with enhanced immune cell infiltration.

The anticancer efficacy of PMSN-based SDT was then investigated in mice bearing bilateral Hepa1–6 liver tumors (Figure 6a), and the tumor growth was recorded (Figure 6b-f, Figure S27). The PMSN+US treatment efficiently suppressed tumor growth, showing an 85.93% and 77.09% inhibition rate for the treated and distant tumors, respectively, on Day 31. These inhibition rates were higher than those found in the US (19.63%, 28.99%), Pc+US (59.31%, 48.46%) or MSN+US (51.23%, 45.14%) treatment groups (Figure 6g-h). Additionally, aPD-L1 intraperitoneal (i.p.) injection was applied for the combinational SDT regimen (Figure 6i). Tumors receiving PMSN+US+aPD-L1 treatment were reduced significantly, even in the distant tumor, whereas tumors in the aPD-L1 (p<0.005) and other control groups grew aggressively (Figure 6j-l, Figure S23b). The PMSN+US+aPD-L1 treatment resulted in an 88.19% and 87.12% inhibition rate for the treated and distant tumors, respectively, which was much higher than that found for the aPD-L1 group (10.93%) and other control groups. The PMSN-mediated SDT not only killed the cancer cells by excessive oxidative stress, but also modulated the tumor microenvironment by inducing a pro-inflammatory response and promoting tumor infiltration of immune cells for anti-tumor effects. In conjunction with aPD-L1 checkpoint blockade, PMSN-mediated SDT suppressed tumor growth and significantly enhanced aPD-L1 efficacy in both the treated and distant tumors through acute inflammation with targeted pyroptosis, promoting immune cell infiltration and achieving robust abscopal effect. Further, no noticeable change in the body weights of the mice receiving the SDT monotherapy or aPD-L1 combinational therapy was observed (Figure S26a-b). These findings showed that PMSN-mediated SDT can synergize with aPD-L1 due to its capability to induce targeted pyroptosis, leading to better therapeutic outcomes to treat liver cancer efficiently with no harm to normal hepatic functions.

Figure 6: PMSN-enhanced SDT suppressed bilateral subcutaneous Hepa1–6 murine liver tumor growth and increased aPD-L1 efficacy.

a Therapeutic regimen of SDT in bilateral Hepa1–6 tumor-bearing mice. Treatment began when tumors reached 3–4 mm in diameter. Following the intravenous (i.v.) administration of PMSN and other control treatments, ultrasound was applied to one tumor after 3 h. The treatment was repeated on Day 2 and Day 4. b-f Tumor growth curves of the treated and distant tumors from mice in the no-treatment control (NTC), ultrasound (US) only, and PMSN-mediated SDT (PMSN+US) groups. Fold changes of tumor volumes on Day 31 of g US-treated tumors and h distant tumors from mice in the different treatment groups tested. i Therapeutic regimen of SDT with aPD-L1 in bilateral Hepa1–6 tumor-bearing mice. Treatment began when tumors reached 3–4 mm in diameter. SDT was applied on Day 0 and Day 4, and aPD-L1 was injected intraperitoneally (i.p.) on Day 2 and Day 6. j-l Tumor growth curves of each mouse in the aPD-L1 and PMSN-mediated SDT with aPD-L1 treatment groups. m Fold changes of tumor volumes on Day 31 of ultrasound-treated tumors (Tr) and distant (Dis) tumors from mice in the different treatment groups tested. Statistical analysis was performed by one-way ANOVA followed by Tukey’s multiple comparisons test (n=5 or 6 for each group).

2.6. Chemo-SDT-induced Pyroptosis to Boost aPD-L1 Efficacy

To further enhance pyroptotic cell death and improve therapeutic efficacy, the chemotherapeutic DOX, which has been reported to show caspase3/GSDME-dependent pyroptosis [45], was loaded into PMSN to obtain DPMSN (Figure 7a). DOX is an anthracycline antibiotic and is used as an effective chemotherapeutic for a wide spectrum of tumors. DOX interacts with DNA and can also generate ROS to damage cell components, and cause tumor oxidative stress[46], and also pyroptosis; ultrasound could also enhance the ROS generation by DOX in many studies[47, 48]. As shown in Figure S27, DPMSN synergized with SDT and chemotherapy by releasing DOX into cell nuclei from PMSN after ultrasound insonation. ROS generation in cells was further enhanced by 1.45-fold after the DPMSN+US treatment compared to the PMSN+US treatment (p<0.05) (Figure S12c). Additionally, obvious pyroptosis was observed with optical microscopy (Figure 7b). The DPMSN+US treatment resulted in about 83.3% dead cells (the populations of Annexin V-FITC⁺) (Figure S27d) and 25.4% Caspase 4⁺ pyroptotic cells (Figure 7c, Figure S28a) (p<0.05), much higher than that seen with the PMSN+US treatment. Additionally, the levels of released LDH and IL-1β were also enhanced in DPMSN+US-treated cells (20.03% release of LDH, and 37.70 pg/mL IL-1β) compared to PMSN+US-treated cells (Figure S18). More severe inhibition of liver cancer cell viability was achieved after Chemo-SDT of DPMSN (approx. 4.3%, p<0.05, Figure S12c), which was much higher than that seen with the mono-SDT with PMSN. These results demonstrate the capability of DPMSN to induce acute pyroptosis and the potential of the Chemo-SDT to further liver-cancer suppression.

Figure 7: PMSN-enhanced SDT anticancer efficacy was further augmented by loading doxorubicin (DOX).

a Illustration of DOX-loaded PMSN (DPMSN) and DOX release after ultrasound insonation. b Images showing ROS production (Scale bar: 100 μm) and pyroptosis (Scale bar: 30 μm) in HepG2 cells treated with chemo-SDT of DPMSN. c Representative immunofluorescence images (Scale bar: 100 μm) show TUNEL (green) and caspase 4 (red) positive cells in tumor slices stained with DAPI (blue) in the murine HepG2 orthotopic liver tumor model. d Therapeutic regimen of chemo-SDT in HepG2 orthotopic liver cancer mice. SDT treatment began when HepG2 tumor bioluminescence intensity reached 10⁵ p/s/cm²/sr), approximately 10 days after tumor cell inoculation. Following the intravenous (i.v.) administration of DPMSN, ultrasound was applied to the tumor after 3 h. The treatment was repeated three times. e Representative bioluminescence images showing HepG2 liver tumor growth on Day 0 (first-treatment time) and Day 35. f Corresponding H&E-stained tumor slices showing the orthotopic HepG2 liver tumors dissected on Day 35 (Scale bar: 500 μm). g Therapeutic regimen of chemo-SDT with aPD-L1 in Hepa1–6 bilateral tumor-bearing mice. Treatment began when tumors reached 3–4 mm in diameter. After DPMSN-mediated SDT, tumors were harvested for histology and flow cytometry after 24 h. For the tumor growth study, DPMSN-mediated SDT was applied on Day 0 and Day 4, and aPD-L1 was injected intraperitoneally (i.p.) on Day 2 and Day 6. h Representative H&E and IHC images of the ultrasound-treated and distant tumors in the DPMSN group (Scale bar: 200 μm). i Flow cytometry plots showing the CD4⁺ and CD8⁺ T cell populations following CD3⁺ T cells gating in the ultrasound-treated and distant tumors of the DPMSN+US treatment group. j Quantified percentages of immune cell populations by flow cytometry analysis, including CD45⁺ tumor infiltrating lymphocytes (TILs) and the major immune cells in both the ultrasound-treated and distant tumors following DPMSN+US administration. k Tumor growth curves in Hepa1–6 bilateral tumor-bearing mice receiving the different drug treatments tested. l Fold changes of tumor volumes on Day 31 of ultrasound-treated (Tr) and distant (Dis) tumors in Hepa1–6 bilateral mice receiving the different drug treatments tested. Statistical analysis was performed by one-way ANOVA followed by Tukey’s multiple comparisons test (n=5 or 6 for each group).

After the application of DPMSN-based SDT in bilateral Hepa1–6 tumor-bearing mice following the treatment regimen outlined in Figure 7g, PD-L1 levels were enhanced in both the ultrasound-treated and distant tumors (Figure S22). Additionally, the cytokines TNF-α, IFN-γ, and IL-6 were also upregulated in the serum of DPMSN+US-treated mice (Figure S25). IHC staining demonstrated enhanced infiltration of CD8⁺, CD11b⁺ and CD68⁺ immune cells (Figure 7h). Furthermore, in Figure 7i-j, with the analysis of tumors by flow cytometry, TILs were increased to 47.07% and 44.03% in the ultrasound-treated and distant tumors, respectively, of the DPMSN treated mouse group. Among these immune cells, CD8⁺ and CD4⁺ T cells were increased by 108- and 36.67-fold in the DPMSN+US-treated mouse group, respectively, compared to the PBS+US group (Figure 5f-g). The populations of TAMs and CD64⁺ M1 TAMs were increased 5.39- (p=0.0224) and 11.44-fold (p=0.0356), respectively, in the ultrasound-treated tumor after DPMSN+US treatment and were also significantly increased by 4.35- (p=0.0311) and 17.61-fold (p<0.001), respectively, in the distant tumors, compared to the PBS+US mouse group. Meanwhile, NK cells and DCs made up 0.4% and 0.52% of the TILs in the ultrasound-treated tumors of the DPMSN+US-treated mice, and 0.37% and 1.57% in the distant tumors, respectively. As the infiltration of immune cells increased not only in the ultrasound-treated tumors but also in the distant tumors as well, the DPMSN-mediated SDT showed a better abscopal effect than the SDT with PMSN as sonosensitizers.

We additionally executed DPMSN-based SDT in the orthotopic HepG2 liver cancer mouse model every other day for three repetitions (Figure 7d). The bioluminescent signals of the HepG2 liver cancer cells after 35 days were 3.42×10⁴ and 8.95×10⁴ p/s/cm²/sr for the DPMSN+US (Figure 7e) and PMSN+US (Figure 2g) treatment groups, respectively. These data illustrate that DPMSN+US treatment significantly suppressed tumor progression, compared to DPMSN without US (Figure S28b) and further compared to the PMSN+US treatment (Figure 2h) (p<0.01). H&E staining of dissected tumors on Day 35 showed notable tumor volume reduction after DPMSN-based SDT (Figure 7f). Meanwhile, for all groups of mice, there was no obvious loss of body weight (Figure S28c).

The immune response was observed in bilateral tumor models treated by DPMSN+US. SDT-mediated pyroptosis with DPMSN promoted the inflammation level of liver tumors to facilitate immune-cell infiltration and the rescue of the innate and adaptive immune systems. Tumor growth was then recorded over 31 days after DOX, DPMSN-mediated SDT monotherapy and SDT-aPD-L1 combinational therapy. The ultrasound-treated tumors (p<0.01) and the distant tumors (p<0.05) in the DPMSN+aPD-L1 mouse group were significantly inhibited compared to treatment groups without aPD-L1 treatment. Both the monotherapy (74.5% and 59.9% inhibition rate in the treated and distant tumors, p<0.001 and p<0.005, respectively) and the combinational therapy (86.4% and 71.5% inhibition rate in the treated and distant tumors, p<0.001 and p<0.001, respectively) showed significantly better anticancer efficacy than DOX with no obvious bodyweight change (Figure 7k-l, Figure S26c). Due to Chemo-SDT-induced robust pyroptosis, the in-situ conversion of the DPMSN+US treated tumor to an immunological hotbed effectively facilitated immune cell recruitment in both the treated and distant tumor sites and promoted excellent therapeutic outcomes in murine liver tumor models when combined with aPD-L1.

3. Conclusion

Despite the discovery of possible mechanisms in SDT, pyroptosis and the relationship between SDT efficacy and pyroptosis have not previously been investigated. Here we present a novel, versatile and synergistic SDT strategy for inducing precise and acute pyroptotic tumor cell death, which, in turn, promotes efficient cancer suppression and the increased immunotherapeutic efficacy of aPD-L1. Further, a safe and efficacious sonosensitizer, PMSN, was developed for amplifying the sonochemical reaction in optimized ultrasound conditions to augment oxidative stress in the tumor.

The well-designed PMSN nanostructure with sub-10 nm size and c(RGDyC)-PEGylated modification facilitated an enhanced tumor-targeting and retention capability, as well as enhanced penetration into tumors. The unique design of PMSN promoted regional ROS generation in tumors and significantly increased ROS production yield during PMSN-mediated SDT treatment on account of the cavitation effect facilitated by the porous architectures of PMSN. Consequently, expression of Caspase 3/GSDME and Caspase 4/GSDMD was significantly enhanced in tumors, indicating that robust tumor-specific pyroptosis was induced in avoidance of normal tissue damage. Due to the excellent pyroptosis-inducing capacity of PMSN-mediated SDT treatment, the inflammatory status of tumors in murine liver cancer models was upregulated. Therefore, PMSN-mediated SDT treatment could not only efficiently mobilize and decrease tumor bulk but could also suppress residual tumors through the reactivation and restoration of antitumor immune responses by PD-L1 blockade via tumor-targeted pyroptosis. More importantly, loading of the chemotherapeutic DOX into PMSN (DPMSN) intensified the SDT-pyroptotic process and presented much higher anticancer effectiveness when combined with aPD-L1 therapy. Our in vivo data, derived from both an orthotopic and a bilateral murine liver cancer model, provides a strong rationale for further clinical assessment of aPD-L1 blockade after inducing robust tumor-targeted pyroptosis with DPMSN-mediated chemo-SDT treatment.

In conclusion, the use of PMSN as a sonosensitizer to enhance SDT provided a new way to leverage oxidative stress and innovatively provides available thoughts for further improvement in inducing an anticancer immunity by immunogenic cell deaths such as pyroptosis in orthotopic liver cancer. Further optimizing SDT by using DPMSN could be an efficacious, safe and amenable choice with higher therapeutic relevance and fewer toxic complications than conventional regimens for liver cancer treatment. DPMSN-triggered chemo-SDT combinational therapy could prove an efficient and novel strategy for treating tumors and boosting anti-PD-1 efficiency via its robust pyroptosis-inducing capability in the future. Although a limitation of this work is that the murine cancer model does not fully recapitulate the development of clinical liver cancer, this work displayed adequate rationale that such a paradigm does show the potential to treat cancers efficiently and safely.

4. Experimental Section/Methods

Materials and chemical agents:

Tetramethyl orthosilicate (TMOS), CTAB and 7 M ammonium hydroxides were purchased from Sigma Aldrich. Sulfonated aluminum phthalocyanine chloride was purchased from MedKoo Biosciences. DOX and ALZ were purchased from TCI Inc. Acetic acids were purchased from Santa Cruz Biotechnology. Methoxy-terminated poly(ethylene glycol) chains (PEG-silane, molar mass of ∼550 g/mol) were purchased from Nanocs Inc. Maleimide-PEG-silane chains (MAL-PEG-silane, molar mass of approximately 800 g/mol) were purchased from Medbio Inc. Cyclo (Arg-Gly-Asp-D-Tyr-Cys), an abbreviation of c(RGDyC), was purchased from Shanghai Apeptide Co, Ltd.

Cell lines and animals:

Liver cancer cell lines, including HepG2 (human species) and Hepa1–6 (mouse species) were obtained from the American Type Culture Collection (ATCC), tested for mycoplasma contamination, and cultured in Dulbecco’s modified eagle medium (DMEM) (Gibco), supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Athymic BALB/C male nude mice (6-weeks-old) and C57BL/6 male mice (6-weeks-old) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. All the animal experiments in this work were approved by the Institutional Animal Care and Use Committee (IACUC) of Peking University.

Characterization of silicate nanoparticles:

Hydrodynamic particle sizes, size distributions and zeta potential of particles were measured by using a dynamic light scattering (DLS) analyzer (Brookhaven Instruments). Each DLS sample was diluted with DI water and measured three times. The average diameter of each sample was calculated by averaging the mean diameters of the number percentage. To study the long-term stability of particles, samples were stored in the fridge for weeks, during which the hydrodynamic sizes of particles were monitored. Each data point was measured three times, and the results were averaged.

For transmission electron microscopy (TEM), the particles were suspended in a 90% ethanol solution, and a drop of particle solution (10 μL) was added on Ultra-thin micro-grid copper mesh. TEM images were taken using a TECNAI F30 (FEI) microscope operated at an acceleration voltage of 200 kV.

For X-ray powder diffraction (XRD), the particles were concentrated and left under 80 °C for 3 hours to dry. The samples were ground to a fine powder and placed on the center of a glass container. The container was put into the diffractometer (X-Pert3 Powder, PANalytical) and the samples were measured from 2 theta of 15 to 90 degree at 4 degree/min. Data were collected and analyzed via HighScore Plus software. Then, the recycled samples were collected to measure the elemental composition via X-ray Photoelectron Spectrometer (XPS, AXIS Supra, Kratos Analytical Ltd.).

Solid-state nuclear magnetic resonance (NMR):

The samples of PMSN and MSN (500 mg) were submitted to the Analytical Instrumentation Center of Peking University. 27Al and 29Si Cross-polarization (CP)/magic angle spinning (MAS) NMR (Bruker-400MHz) with a 2.5-mm probe were used for characterization.

Thermogravimetric Analysis (TGA):

The particles were concentrated and left under 80 °C for 3 hours to dry. The dried particle samples were applied to a TA Instruments Model Q600. For measuring, the temperature is increased from RT to 100 °C with a ramp of 10 °C/min and then remains at 100 °C for 2 h to entirely exclude any residual water. Afterward, the temperature is further increased to 600 °C with a ramp of 10 °C/min removing any organic moieties and leaving pure inorganic silicates behind. Thus, the percentage of PEG chains on particles was then estimated based on TGA results. According to the gel permeation chromatography (GPC) elugram, the PMSN particle was assumed as approximately 110 kD in molecular weight (Supplementary information about PMSN synthesis and characterization, and Figure S3c). TGA curves (Figure S7) shows that Pc was about 3.7% and the total organic composition (Pc, RGD and PEG) in PMSN was about 45.2%. Total conjugated Pc was approximately 4070 g/mol (3.7% of 110 kDa), and calculated by its molecular weight (870 g/mol), so there were 4.6 molecules in one particle. The ratio of RGD-PEG (1522 g/mol): PEG (500 g/mol) was supposed to be 1:20, and total PEG was approximately 45650 g/mol (41.5% of 110kDa), so there was 4.0 RGD-PEG and 80 PEG in one particle.

Construction of liver cancer models:

Liver cancer orthotopic models were established in BALB/c athymic nude male mice by inoculating 10⁶ HepG2 cells carrying luciferase genes in 1:1 serum-free medium/growth factor-reduced Matrigel at the large lobe of the liver. All mice developed tumors in the livers, and the tumor growth in the liver was visualized by bioluminescence live imaging after approximately 10 days (Max radiance efficiency was approximately 10⁵ p/s/cm²/sr).

The bilateral tumor model for immune response evaluation were established in C57BL/6 male mice by inoculating 3×10⁶ Hepa1–6 cells in 1:1 serum-free medium/growth factor-reduced Matrigel, subcutaneously into hind flank. The tumor volume was calipered, and the sizes of tumors were about 3–4 mm in diameters after a week.

In vivo fluorescence imaging and biodistribution study: The orthotopic HepG2 tumor-bearing mice were randomly divided into 3 groups (n =3 per group): PBS, Pc (1 μM) and PMSN at the equivalent Pc dosage. The mice received intravenous (i.v.) administration of the different drugs. Imaging was analyzed on an IVIS Imaging Spectrum System under specific parameters (Ex =640 nm, Em =710 nm, exposure time =1 s). The tumor-bearing mice (n =3 per group) were intravenously injected with Pc or PMSN, respectively. Fluorescence images were acquired at predetermined time points after the injection of samples.

Ultrasound attenuation measurement:

As shown in Figure S8, the hydrophone (HNR-0500, Onda) connected to an oscilloscope (DPO 2012B, Tektronix) received the signal from the ultrasound and performed the fast Fourier transform (FFT) analysis, and ultrasound attenuation was calculated:

$$A=20lo{g}_{10}\left[\frac{I_{Sample}}{I_{Water}}\right]$$

where A is the attenuation, I_PMSN and I_Water are FFT of respective samples and degassed DI water, respectively. The data was processed with the MATLAB program.

PMSN was filled in a sample holder (OptiCell, BioCrystal). The ultrasound transducers are in different central frequencies, including 500 kHz (H-107G, Sonic concept) with its impedance matching, 1 MHz (A303S-SU, Olympus), 1.5 MHz (IL01512HRF, Valpey Fisher), and 2.25 MHz (V305, Olympus). All of the transducers were calibrated prior to the experiment. The signal generator (33500B Series, KEYSIGHT) and power amplifier (2100L, E&I) were connected to the transducer in sequence. The signal generator was used to control the waveforms, output voltage and burst periods to provide different acoustic pressures and duty cycles. The acoustic attenuation was measured by detecting the changes in amplitudes. At each condition, the samples were treated for at least 30 seconds, and the received amplitudes were decreasing and maintained a stable value.

ROS yield measurement:

Singlet oxygen is widely recognized to be the primary cytotoxic component of ROS and has a longer intracellular half-life compared to other types[18, 49]. Singlet Oxygen Sensor Green (SOSG) (S36002, Invitrogen) is a commercially available and widely used fluorescent probe for in vitro detection and quantification of ROS, especially singlet oxygen (¹O₂) [50]. SOSG will appear to be fluorescent (Ex =488 nm, Em =525 nm) when the environmental ROS are increased, and the intensity is in positive correlation to ROS concentration. ROS quantum yield enhancement was calculated according to the equation:

$$\text{kt}=Ln\left(\frac{A}{A_0}\right)$$

$$\text{Φ}\Delta =\frac{k_{Sample}}{k_{Water}}$$

where kt was along ultrasound insonation time, A and A₀ were the SOSG fluorescence intensity and initial SOSG concentration measured using a microplate reader (SYNERGY HT, BioTek) or the rebuilt fluorescence imaging system (FOBI, NEO Science, Malaysia). The Log plot of SOSG concentration versus insonation time should be linear with a slope proportional to ΦΔ[51].

Western Blot:

HepG2 cells (10⁶ cells per sample) treated with different ultrasound conditions or drugs were incubated in 6-well plates overnight. The cells were lysed, and the protein samples were collected for analysis. Equal amounts of protein samples (quantified by BCA assay) were loaded per lane, separated by sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE) at a constant voltage of 120 V, and then transferred onto methanol-activated Polyvinylidene fluoride (PVDF) membranes at a constant current of 300 mA. The membranes transferred with proteins were blocked using 5% bovine serum albumin (BSA) tris-buffered saline with Tween 20 (TBST) for 1 h, and then incubated with primary antibodies of the targeted proteins, including Caspase 3 (ab32351, Abcam), Caspase 4 (ab238124, Abcam), PARP (D64E10, Cell Signaling Technology), GSDMD (39754, Cell Signaling Technology), GSDME (ab215191, Abcam) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (51332S, Cell Signaling Technology) overnight at 4 °C. The next day, membranes were washed with TBST and incubated at room temperature with the corresponding secondary antibodies, a goat anti-mouse IgG antibody with horseradish peroxidase (HRP) conjugation (91196S, Cell Signaling Technology), for 1.5 h. Chemiluminescence reagent (Beyotime Biotechnology, China) was used to detect signals. The results of the expressed proteins were analyzed using ImageJ.

Cell viability test:

The cytotoxicity of PMSN against HepG2 cells was evaluated by the Cell Counting Kit-8 (CCK-8) assay (Meilunbio, China). HepG2 cells were seeded on 96-well plates (10⁴ cells/well) in triplicate and treated with different drugs in DMEM medium for 48 h: PBS as control, Pc, MSN, PSNP, PMSN at the equivalent concentration of Pc (1 μM) or silicate (58.8 μM). After 3 h incubation, the cells were then treated with focused ultrasound (500 kHz, 400 kPa, 25% duty cycle, 2 min). The cells without ultrasound treatment served as controls. The cells were further incubated for 24 h and then replaced with the CCK-8 solution. The absorbance in each well was measured by a microplate reader at the wavelength of 450 nm, and the cell viability ratio was measured as a percentage of the control.

In addition, the cell apoptosis by Annexin V-fluorescein isothiocyanate (FITC)/ Propidium Iodide (PI) double staining (Meilunbio, China) with flow cytometry was measured. HepG2 Cells (5×10⁵ cells per sample) after each treatment were collected by centrifugation (1000 rpm, 4 min), washed with PBS and suspended with 500 μL Annexin V Binding Buffer. Then, the staining solution was added to tubes, and the mixture was incubated for 20 min at RT in the dark. Afterwards, the samples were analyzed within 1 h by flow cytometry (BD FACSVerse).

RNA sequencing analysis:

We obtained the RNA sequencing data of two groups of HepG2 orthotopic tumor mice (n=3 per group), the no-treatment control (NTC) and the PMSN-based SDT group. Following SDT treatment, we harvested the tumors after 24 h and stored them in liquid nitrogen for further processing and analysis. RNA sequencing (RNA-seq) of the tumors was run at Genewiz Inc. using standard RNA-seq services. We extracted 23 pyroptosis- or pro-inflammatory cytokines-related genes from prior studies[41, 52]. The expression data in both datasets were normalized to fragment per kilobase million (FPKM) values before comparison. We identified differentially expressed genes (DEGs) in tumor samples with the “limma” package in R (version 3.6.3). Gene Ontology (GO), including the biological process (BP), cellular component (CC), and molecular function (MF) categories, was conducted with the “ggplot2” package in R.

Anticancer efficacy in orthotopic liver cancer mice:

The mice were inoculated with HepG2 tumors at the orthotopic site before the experiment. After the measured volume of the tumor reached sufficient luciferase expression (Max radiance efficiency is approximately 10⁵ p/s/cm²/sr), the tumor-bearing mice were divided randomly into 12 groups (n=3 per group): PBS, Pc (3 mg/kg), MSN, PSNP, PMSN and DPMSN formulation at the equivalent dosage of Pc or silicates with or without ultrasound application. If with ultrasound, the treatment was applied 3 or 4 h after i.v injection of Pc or particles, respectively. The corresponding treatments were implemented every 2 days for totally three times, and the time of the first treatment was recorded as Day 0. The tumor growth with bioluminescence living imaging and the bodyweight of mice were monitored during the experiment. On Day 35, tumors were dissected and fixed overnight in 10% neutral buffered formalin and transferred to 70% alcohol the next day.

Anticancer efficacy in mice bearing bilateral tumors:

The mice were inoculated with Hepa1–6 tumors bilaterally before the experiment. After the measured diameters of the tumors reached about 3–4 mm, the mice were divided randomly into 12 groups (n =5 or 6 mice per group): PBS as NTC and DOX (4.05 mg/kg, equivalent dosage of DOX in DPMSN) as chemotherapy control, and Pc (3 mg/kg), MSN, PSNP, PMSN and DPMSN formulations at the equivalent dosage of Pc or silicates with ultrasound application in only one of the bilateral tumors per mouse. In groups receiving ultrasound, the treatment was applied 3 or 4 h after i.v injection of Pc or particles, respectively. The corresponding treatments were implemented every 2 days for a total of three times, and the time of the first treatment was recorded as Day 0. The tumor volumes were calculated by measuring the length (a) and width (b) of the tumors with calipers according to the equation:

$$Tumorvolume=\frac{a\times b^2}{2}$$

Fold change of tumor volumes were calculated as compared to the tumor volume on Day 0. The bodyweights of mice were also monitored over 31 days.

PD-L1 blockade efficacy in mice bearing bilateral tumors:

The mice were inoculated with Hepa1–6 tumors bilaterally before the experiment. After the measured diameters of the tumors reached about 3–4 mm, the mice were divided randomly into 10 groups (n =5 or 6 mice per group): aPD-L1 (12.5 mg/kg), Pc (3 mg/kg)+aPD-L1, MSN+aPD-L1, PMSN+aPD-L1 and DPMSN+aPD-L1 formulations at the equivalent dosage with ultrasound application in only one of the bilateral tumors per mouse. In the regimen, aPD-1 was injected intraperitoneally (i.p.) on Day 2 and Day 6, and SDT treatment was executed on Day 0 and Day 4. The tumor volumes were calculated by measuring the length and width of tumors with calipers, and the bodyweight of mice was monitored over 31 days.

Immunohistological staining:

For immunofluorescence (IF) staining, anti-PD-L1 (Ab213480, Abcam), TUNEL-FITC (GDP1041, ServiceBio, China) and Caspase 4 primary antibody, were used. The histological tumor sections were fixed with ice-cold 4% paraformaldehyde (PFA) for 10 min. Then, the sections were incubated with 5% BSA for 0.5 h at RT to block nonspecific antigen binding. Afterwards, the primary antibodies were used for incubation at 4°C overnight, then were removed. The secondary detection for PD-L1 or Caspase 4 was performed using a goat anti-rabbit IgG antibody (GB21303, ServiceBio) or a goat anti-mouse IgG antibody conjugated with Cy3 (GB21301, ServiceBio) at RT for 1 h. The sections were finally stained with DAPI and mounted for microscopic observation. Before and after each step, the sections were rinsed with TBST three times, respectively. The percentages of the positive cells in images were counted using ImageJ.

For immunohistochemistry (IHC) staining, anti-CD8 (GB13429, ServiceBio), anti-CD11b (GB11058, ServiceBio), anti-CD68 (GB113109, ServiceBio). CD8⁺ cells are cytotoxic T cells to induce anti-cancer activity. CD11b⁺ cells are myeloid cells and important for leukocyte recruitment into the site of inflammation. CD68⁺ cells in tumors are often tumor-associated macrophages (TAMs), which are macrophages that have been recruited or differentiated within the tumor microenvironment. TAMs can exhibit diverse phenotypes and functions, ranging from pro-inflammatory (M1-like) to anti-inflammatory (M2-like). M1-like TAMs are associated with an immune-stimulatory phenotype and can promote anti-tumor immune responses. CD68⁺ cells in tumors can also include monocyte-derived cells, such as infiltrating monocytes or monocyte-derived dendritic cells. These cells are recruited to the tumor site and can differentiate into macrophages or DCs within the tumor microenvironment. The histological sections were fixed with ice-cold 4% PFA for 10 min. Then, the sections were incubated with 5% BSA for 0.5 h at RT to block nonspecific antigen binding. Afterwards, the primary antibodies were used for incubation at 4°C overnight, then were removed. The secondary detection was performed using a goat anti-mouse IgG antibody conjugated with HRP at RT for 1 h. The sections were finally stained with hematoxylin and mounted for microscopic observation. Before and after each step, sections were rinsed with TBST three times.

Flow cytometry for immune cell phenotyping:

Tumors were dissociated using Liberase (05401020001, Roche) and red blood cells were removed using ACK lysis buffer (554656, BD Biosciences). The cells from tumor samples were suspended in Brilliant staining buffer (566349, BD Biosciences), and pretreated with anti–CD16/32-Fc monocyte blocking agent (14598, Biolegend) and stained with fluorophore-conjugated antibody solution according to the manufacturer-suggested dilutions on ice for 1 h. Antibodies used for the flow cytometry studies were anti-CD19-BUV737 (741829, BD Biosciences), anti-NK-1.1-BV510 (427, Biolegend), anti-CD64-BV605 (139323, Biolegend), anti-CD45-BV711 (10439, Biolegend), anti-CD8a-FITC (4071, Biolegend), anti-CD11c-PerCp-Cy5.5 (1816, Biolegend), anti-F4/80-PE-Cy7 (4070, Biolegend), anti-CD4-APC (245, Biolegend), anti-CD11b-AF700 (3388, Biolegend) and anti-CD3-APC-Cy7 (100221, Biolegend), as well as the viability dye Live/Dead Aqua (L34957, ThermoFisher). Beads were used for fluorescence compensation (01–2222-41, ThermoFisher). Flow cytometry samples were collected, and the data were acquired on an LSR Fortessa cytometer (BD Biosciences) and analyzed using FlowJo software.

Statistical analysis:

The data shown in the manuscript and in supplementary information are presented as mean ± standard deviation (SD), and the tests for significant differences between the groups were performed using Graphpad.

Data availability

All data generated or analyzed during this study are included in this published article (and its supplementary information files).