Cancer immunotherapy which drives patients’ immune system to combat tumors has made great progress recently1,2,3. However, the anticancer efficacy of current immunotherapy is impeded by the immunosuppressive microenvironment in “cold” triple-negative breast cancer (TNBC), which typically characterizes deficient T cell infiltration and low immunogenicity4. Recently, more and more research works on phototherapy and nanotechnology have concentrated on affecting the immune system for cancer treatment5. Phototherapy, including photodynamic therapy (PDT) and photothermal therapy (PTT), has attracted increasing attention in cancer immunotherapy as it creates a tumor-associated antigen (TAA) pool and contributes to dendritic cell (DC) maturation6,7. Nonetheless, local phototherapy monotherapy is not enough to evoke strong and durable tumor immunogenicity for inhibiting tumor metastasis and rechallenge, it urgently needs to combine phototherapy and immunostimulation strategy to realize synergistic outcomes for photoimmunotherapy of poorly immunogenic TNBC8.

Gas-based therapy, such as carbon monoxide (CO), nitric oxide (NO) and hydrogen sulfide (H2S), is an emerging green therapeutic strategy for cancer treatment9,10. CO and H2S gases are critical endogenous biosignaling molecules11. The delivery of exogenous CO/H2S aims to induce mitochondrial membrane potential (MMP) depolarization, which would further damage mitochondria and lead to the release of mitochondrial DNA (mtDNA) into cytoplasm12,13. Here, we hypothesize that the existence of cytoplasmic mtDNA induced by gas therapy could activate cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING) pathway to exert immunoadjuvant effect14. The cGAS-STING pathway activation can trigger proinflammatory cytokine expression and type I interferon (IFN) response for immune stimulation and further augment the therapeutic effects15. Besides, manganese (Mn2+) has been reported to be responsible for the activation of the cGAS-STING pathway in various aspects, from promoting cyclic GMP-AMP (cGAMP) production to enhancing cGAMP-STING binding affinity16,17. In this work, we explore the synergy of immunity enhancement between gas therapy and Mn2+-based cGAS-STING promotion in both tumor cells and DCs.

In view of the above considerations, a versatile aggregation-induced emission (AIE)-active luminogen (AIEgen) is tactfully synthesized as phototheranostic agent. It shows robust reactive oxygen species (ROS) and heat generation after laser exposure, and near-infrared-II (NIR-II) fluorescence signal, which could realize PDT/PTT synergistic therapy and NIR-II fluorescence imaging (FLI)/photoacoustic imaging (PAI)/photothermal imaging (PTI) multimodal imaging. Nevertheless, the therapeutic outcome of PDT is usually restricted by intracellular antioxidant glutathione (GSH)18,19. Therefore, engineering a nanoassembly with the capability of ROS production and endogenous GSH depletion can simultaneously magnify oxidative stress for more effective tumor treatment20. Moreover, owing to the restricted active radius and short lifespan of ROS, it is important to promote the internalization efficiency of nanoassembly for more effective PDT21. The design of surface topography presents a perspective to increase the cell entry of nanoparticles. Recently, some researches have shown that the virus-like surface is beneficial in enhancing the nanoparticle-cell adhesive interaction and nanoparticle uptake efficiency compared to sphere-like surface22,23.

In this work, the GSH/NIR sequentially initiated gas nanoadjuvant with AIEgen-mediated PDT/PTT synergistic therapy and tumor-specific amplified H2S/CO/Mn2+ generation is constructed to effectively invade cancer cells and significantly activate cGAS-STING pathway for poorly immunogenic TNBC treatment. Specifically, the tetrasulfide-bridged virus-mimicking hollow mesoporous silica (tvHMS) is engineered to load AIEgen and manganese carbonyl (Mn2(CO)10, abbreviated as MnCO) (Fig. 1a). Following tumor accumulation, the virus-like surface helps gas nanoadjuvant invade cancer cells more effectively through spike surface-assisted adhesion. After entering the cancer cell, the overexpressed GSH could break the tetrasulfide bond to initiate the disintegration of nanoassembly, resulting in tumor-specific and controllable drug release, amplified PDT via GSH consumption, and H2S production. Under NIR laser irradiation, the AIEgen-based PDT/PTT synergistic therapy could in situ activate the prodrug MnCO to generate Mn2+ and CO24. Both H2S and CO stimulate mitochondrial dysfunction of tumor cells to induce the intracellular release of mtDNA, demonstrating the immunoadjuvant property of engaging cGAS-STING pathway. Moreover, Mn2+ is a powerful cGAS activator to enhance STING-mediated type I IFN response in both tumor cells and DCs, involving STING, tank-binding kinase 1 (TBK1), and interferon regulatory factor 3 (IRF-3) phosphorylation (Fig. 1b)25,26. Afterwards, the IFN-β expression level is improved, which contributes to DC maturation and cross-primes anticancer T cells for adaptive immune responses. As a result, TNBC suppression, potent distant tumor inhibition, and effective resistance to tumor metastasis and rechallenge are clearly manifested in various mouse tumor models. Integrating AIEgen-based phototherapy and gas-mediated immunostimulation strategy of cGAS-STING pathway activation into the virus-inspired gas nanoadjuvant illustrates a paradigm for photoimmunotherapy of poorly immunogenic tumor.

Fig. 1: The fabrication and biological functions of gas nanoadjuvant.
figure 1

a Schematic illustrating preparation routes for gas nanoadjuvant. b Schematic diagram of gas nanoadjuvant-based cGAS-STING pathway-dependent antitumor immune responses. Following tumor accumulation, the virus-like surface helps gas nanoadjuvant effectively invade cancer cell through spike surface-assisted adhesion. After entering cancer cell, the overexpressed GSH could break tetrasulfide bond that enables H2S generation and drug release. Upon NIR laser irradiation, the AIEgen-based phototherapy could activate MnCO to produce CO and Mn2+. Both H2S and CO stimulate the intracellular release of mtDNA to exert immunoadjuvant property by cGAS-STING pathway activation. Meanwhile, Mn2+ is a powerful cGAS activator to enhance STING-mediated type I IFN response. The gas nanoadjuvant can engage cGAS-STING pathway in both tumor cells and DCs, leading to DC maturation and potent antitumor immune responses.


Fabrication and characterization of gas nanoadjuvant

As a class of aggregation-induced emission (AIE) luminogens (AIEgens), 2, 2’-(2-((5’-(4-(diphenylamino) phenyl)-[2, 2-bithiophen]−5-yl) methylene)−1H-indene-1, 3(2H)-diylidene) dimalononitrile (TSSI) with exceptional ROS and heat generation capability has been prepared and thoroughly discussed in our previous work for photodynamic and photothermal therapy27. TSSI was purified with prominent yields of 80-85%, and the structure was verified using 1H NMR13,C NMR, and high-resolution mass spectra (HRMS) (Supplementary Fig. 13). Herein, we synthesized tetrasulfide-functionalized virus-like hollow mesoporous silica (tvHMS) nanoparticles for TSSI and manganese carbonyl (Mn2(CO)10, abbreviated as MnCO) encapsulation (MTHMS) (Fig. 2a, Supplementary Table 1). The tvHMS was fabricated via post-co-condensation of bis[3-(triethoxysilyl) propyl] tetrasulfide (TESPT, a tetrasulfide-functionalized silica precursor) with tetraethyl orthosilicate (TEOS) (v/v, 1:3) through the chemical homology principle on the solid silica surface. Firstly, the tetrasulfide-functionalized mesoporous silica-encapsulated solid silica (solid silica@tMS) nanoparticles were fabricated through post-condensation of TESPT and TEOS (v/v, 1:3) on solid silica surface, utilizing cetyltrimethylammonium bromide (CTAB) as the structure-directing agent28. Then, sodium hydroxide (NaOH) was used as the surface-morphological guide agent and etching agent for removing solid silica template and producing tvHMS. The pore-size of tvHMS was approximately 8.65 nm, determined by nitrogen adsorption-desorption isotherm, indicating the porous skeleton frameworks for cargo loading (Supplementary Fig. 4, 5)29. The underlying mechanism of the conversion of core-shell structure to the virus-mimicking hollow mesoporous structure by NaOH etching is as follows: (1) NaOH rapidly etches the solid silica to produce dissolved silicate species, the core is thus removed to yield the hollow structure; (2) NaOH slowly etches a part of mesoporous silica to produce dissolved silicate species, leaving behind the disordered, pot-holed surface layer while generating the spike structure. To further confirm the components of tvHMS, energy-dispersive X-ray spectroscopy (EDX) was used for elemental mapping study. As depicted in Fig. 2b, the Si, O, S, C, and N elements were present on the shell. The morphology of MTHMS was confirmed through the transmission electron microscope (TEM) (Fig. 2c), which showed a specific virus-like hollow core-porous shell structure with a spiky surface, with spike radial distance of ~20 nm. The dynamic light scattering (DLS) analysis demonstrated that mean hydrodynamic diameters of tvHMS and MTHMS were about 124.9 ± 5.2 nm and 128.1 ± 5.9 nm, respectively. The zeta potential of tvHMS and MTHMS were −17.8 ± 2.3 mV and −19.3 ± 3.2 mV, respectively (Fig. 2d). Moreover, MTHMS showed characteristic peaks of MnCO and TSSI in the UV-vis absorption spectrum (Fig. 2e), confirming the successful drug loading. The entrapment efficiency was calculated to 27.4 ± 2.8% and 32.6 ± 4.6%, respectively, according to the calibration curves of absorbance intensity from the UV-vis spectrum. Additionally, the photoluminescence (PL) spectra and photoacoustic signal further indicated the possibility of MTHMS for in vivo NIR-II FLI and PAI (Fig. 2f, g)30,31. The solution of MTHMS was then incubated in PBS or PBS + 10% FBS. At designated time point, aliquot of the solution was taken and monitored through DLS. As shown in Supplementary Fig. 6, the mean diameter of MTHMS in PBS or 10% FBS solution displayed negligible change after storage at ambient condition for 24 h, suggesting good in vitro stability of the MTHMS.

Fig. 2: Preparation and characterization of MTHMS.
figure 2

a Synthesis routes of MTHMS. b Representative EDS element mapping of tvHMS (n = 3 independent experiments). Scale bar, 100 nm. c Representative transmission electron microscope (TEM) images of (i) solid silica, (ii) solid silica@tMS, (iii) tvHMS, and (iv) MTHMS (n = 3 independent experiments). Scale bar, 100 nm. d Diameter distribution and zeta potential of tvHMS and MTHMS through DLS (n = 3 independent experiments). Data represent the mean ± SD. e Representative UV-vis spectra of tvHMS, MnCO, TSSI, and MTHMS (n = 3 independent experiments). f Representative PL spectra of MTHMS (1 × 10−5 M TSSI) (n = 3 independent experiments). g Representative photoacoustic images of MTHMS (n = 3 independent experiments). h Representative UV-vis spectra of DTNB for the detection of GSH depleting ability of tvHMS (n = 3 independent experiments). i H2S production from tvHMS dispersed in GSH solution (n = 3 independent experiments). j Representative detection results of H2S production from tvHMS in GSH solution by Pb(NO3)2 soaked circular paper (n = 3 independent experiments). k Time-dependent TSSI release from MTHMS in GSH solution (n = 3 independent experiments). l Representative TEM images of MTHMS under GSH stimulus (n = 3 independent experiments). Scale bar, 100 nm. m Representative measurement of ROS production of MTHMS (1 × 10−6 M TSSI) under NIR irradiation (n = 3 independent experiments). n Representative thermographic images and (o) temperature change curves of H2O, MHMS, THMS, and MTHMS (1 × 10−4 M TSSI) after NIR exposure (660 nm, 0.3 W cm−2) (n = 3 independent experiments). p Representative measurement of CO generation at 37 °C after various treatments (n = 3 independent experiments). Source data underlying df, h, i, k, m, o, p are provided as a Source Data file.

Assessment of H2S generation and drug release

Because the tetrasulfide bond is sensitive to the reductive microenvironment, it makes a difference in enhancing PDT of TSSI by depleting GSH so as to induce more reactive oxygen species (ROS) generation and accumulation32,33. In order to ascertain the GSH consumption ability, the GSH level was investigated using 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB), which can interact with free sulfhydryl34. As illustrated in Fig. 2h, UV-vis absorption spectrum exhibited a time-dependent loss of absorbance at 412 nm in tvHMS, indicating the clearance of GSH based on tetrasulfide cleavage. Given that the reaction between tetrasulfide bond and GSH could generate H2S for gas therapy, we evaluated the yield of H2S from tvHMS. As depicted in Fig. 2i, 144.5 × 10−6 M of H2S could be detected from the mixture of 1 mg mL−1 tvHMS and GSH at 48 h, and it increased to 199 × 10−6 M when the tvHMS concentration reached 2 mg mL−1. Moreover, Pb(NO3)2 soaked circular paper visualized the production of H2S from tvHMS. The specific reaction process was listed hereafter:


As shown in Fig. 2j, the color of test paper turned black and became deeper over time, suggesting sustained production of H2S. In contrast, the disulfide bond incorporated ones (dvHMS) did not change the color of Pb(NO3)2 soaked circular paper (Supplementary Fig. 7)35.

Subsequently, the drug release profile of MTHMS was investigated. As shown in Fig. 2k, the cumulative release of TSSI was as low as 14.6 ± 4.6% after 48 h incubation without GSH, while the release was increased to 57.2 ± 7.4% and 88.6 ± 7.6% when GSH concentration was 5.0 mM and 10.0 mM, respectively. Moreover, TEM images revealed that an apparent change of MTHMS was found when incubated with pH 7.4 PBS containing 10 mM GSH (Fig. 2l). The particles appeared to swell and even collapse, manifesting MTHMS achieved rapid decomposition in tumor microenvironment (TME).

Photothermal performance and production of ROS and CO

The ROS production capability of MTHMS was further assessed with 2,7-dichlorodihy-drofluorescein diacetate (DCFH-DA), which was converted to DCF to emit green fluorescence in response to ROS36. As shown in Fig. 2m, fluorescence signal at 525 nm of DCF increased rapidly with MTHMS under NIR irradiation, demonstrating the high ROS production efficiency. The heat production of MTHMS was subsequently evaluated under NIR irradiation (660 nm, 0.3 W cm−2). As shown in Fig. 2n, o, a rapid rise of temperature was observed, which reached a plateau at 54 °C within 5 min, attributable to the active intramolecular motion of TSSI. The heat-producing ability makes MTHMS a desirable candidate for PTT. Meanwhile, the elevated temperature is beneficial to the breaking of Mn-CO coordination bond to release extra CO.

As a CO releasing molecule (CORM), MnCO is widely explored37. To verify the phototherapy-triggered CO burst from MTHMS, the CO release was further quantified by hemoglobin (Hb) assay. It is worth noting that the combination of hydrogen peroxide (H2O2) and NIR irradiation (0.3 W cm−2, 10 min) led to an obviously amplified production of CO in MTHMS solution (Fig. 2p). This was ascribed to superior ROS and heat generation of MTHMS upon laser irradiation.

Cellular uptake

The cellular uptake behavior of virus-like MTHMS was investigated in 4T1 cells, and the sphere-like MTHMS with similar surface charge and mean hydrodynamic diameter was chosen as a control (Fig. 3a, b, Supplementary Fig. 8). From confocal laser scanning microscope (CLSM) observation, the uptake efficiency of MTHMS with virus-mimicking rough surfaces was much higher than that of sphere-like ones (Fig. 3c, d, Supplementary Fig. 9), confirming augmented entry of virus-mimicking MTHMS into 4T1 cells. The flow cytometry was employed for further quantitation analysis. As illustrated in Fig. 3d, Supplementary Fig. 9, virus-like MTHMS displayed 2.8 times higher uptake efficiency than spherical ones after 4 h incubation, suggesting that the virus-mimicking surface endowed MTHMS with superior cellular adhesion and internalization38,39. Compared to the smooth surface of sphere-like MTHMS, the rough surface endowed virus-like MTHMS with more contacting chances and elevated adhesion interaction with cytomembrane to improve the cellular internalization efficiency39,40.

Fig. 3: Evaluation of MTHMS in vitro.
figure 3

a Schematic diagram of fabrication and b TEM and scanning electron microscope (SEM) imaging of virus-like and sphere-like MTHMS (scale bar = 50 nm). CLSM observation and flow cytometric analysis of 4T1 cells incubated with virus-like and sphere-like MTHMS for 1 h c and 4 h d (scale bar = 15 μm). e WSP-1 (H2S), f FL-CO-1 (CO), g JC-1 (red for aggregate and green for monomer), and h DCFH-DA (ROS) fluorescence imaging with various treatments (scale bar = 15 μm). For bh, experiment was repeated three times independently with similar results.

Intracellular H2S/CO generation and mitochondrial dysfunction

The intracellular consumption of GSH and production of H2S after various treatments were monitored with ThiolTracker Violet and Washington State Probe-1 (WSP-1, H2S probe). As seen in ThiolTracker Violet fluorescence images (Supplementary Fig. 10), tvHMS-incubated 4T1 cells showed attenuated green fluorescence due to GSH depletion. As depicted in Fig. 3e, no apparent green fluorescence of WSP-1 was visible from cells incubated with PBS and disulfide bond incorporated virus-like hollow mesoporous silica (dvHMS). However, tvHMS treated group showed a notable enhancement of green fluorescence, indicating that the introduction of tetrasulfide bonds could initiate the GSH-induced intracellular generation of H2S.

The intracellular generation of CO was further estimated through a CO detection system (FL-CO-1 + PdCl2)24,37,41,42. As illustrated in Fig. 3f, MnCO encapsulated tvHMS (MHMS) treated cells showed higher fluorescence intensity compared with free MnCO treated cells, which may be attributed to the GSH depletion of tetrasulfide bond that facilitated H2O2-mediated MnCO activation. By contrast, more remarkable green fluorescence was emitted from MTHMS-treated cells after laser irradiation, confirming that the photodynamic and photothermal effect of TSSI could accelerate the breaking of coordination bonds in MnCO to release more CO.

H2S/CO was reported to cause severe damage to cells with reduced mitochondrial membrane potential (MMP) concomitantly. Therefore, we monitored the mmP with JC-1 assay43. JC-1 polymerized to produce red fluorescence at high mmP but dissembled to a monomer yielding green fluorescence at low mmP. Compared to cells treated with TSSI + L (“ + L” represents NIR laser irradiation), TSSI encapsulated tvHMS (THMS) + L treated cells displayed distinctly increased green fluorescence and clearly declined red fluorescence, suggesting significant mitochondrial dysfunction with the assistance of H2S. Moreover, MTHMS + L treatment resulted in the highest green fluorescence intensity, which was ascribed to the synergetic effect of H2S and CO (Fig. 3g).

The DCFH-DA was utilized as an intracellular oxidant probe to measure ROS generation44. As shown in Fig. 3h, THMS + L group presented significantly enhanced green fluorescence compared with TSSI + L, implying that the tetrasulfide bonds in tvHMS contributed to the depletion of GSH for improving the PDT efficiency under NIR exposure with enhanced ROS accumulation. Additionally, the fluorescence intensity in MTHMS + L group further elevated, which was attributed to the activation of mitochondrial ROS signaling pathway by CO-mediated gas therapy45,46.

mtDNA release and cGAS-STING pathway stimulation

To investigate whether H2S/CO could induce mitochondrial DNA (mtDNA) release and further facilitate the cGAS-STING pathway activation in a synergistic manner, the cytosolic and extracellular mtDNA was investigated through quantitative reverse transcription polymerase chain reaction (RT-PCR)47. As seen in Fig. 4a, THMS + L group showed a higher level of cytosolic and extracellular mtDNA than TSSI + L group, confirming the H2S-mediated mtDNA release. Notably, MTHMS + L group displayed much more mtDNA release due to the mitochondrial dysfunction synergistically induced by released CO. Furthermore, the cGAS-STING pathway existed in dendritic cells (DCs) and cancer cells. So, the levels of IFN-β, C-X-C motif chemokine 10 (CXCL10), and IL-6 in cancer cells after various treatments and bone marrow-derived dendritic cells (BMDCs) incubated with pretreated cancer cells were measured. As seen in Fig. 4b, c, the IFN-β, CXCL10, and IL-6 expression in MTHMS + L group elevated significantly both in cancer cells and BMDCs, evidencing the effective cGAS-STING pathway activation. The frequency of mature DCs induced by tumor cells with different treatments was evaluated because of cGAS-STING activation. 4T1 cells with different treatments were lysed and co-cultured with BMDCs. As expected, MTHMS + L treated cancer cells significantly boosted DC maturation (Fig. 4d, e, Supplementary Fig. 11), which was driven by the striking IFN-β release after cGAS-STING pathway activation. To further confirm the cGAS-STING activation-induced DC maturation, the highly potent and selective inhibitors (RU.521, C-178, or C-176) of the cGAS-STING pathway were separately added when incubating BMDCs with MTHMS + L treated cancer cells48,49,50. As shown in Supplementary Fig. 12, the cGAS-STING pathway inhibition significantly decreased the IFN-β, CXCL10, and IL-6 expression and reduced the frequency of mature DCs, evidencing the cGAS-STING pathway-dependent DC maturation.

Fig. 4: Evaluation of gas nanoadjuvant-induced mtDNA release and antitumor immune response through cGAS-STING activation.
figure 4

a RT-PCR detection of relative mtDNA copy number in cytosol and supernatant of 4T1 cancer cells after various treatments (n = 3 independent experiments). The detection of cytokines (IFN-β, CXCL10, and IL-6) in culture supernatants of b 4T1 cells following the indicated treatments and c BMDCs incubated with pretreated cancer cells (n = 3 independent experiments). b the p values of MTHMS + L to THMS + L in the detection of IFN-β, CXCL10, IL-6 are 0.0052, 0.0143, and <0.0001, respectively. And the p values of THMS + L to TSSI + L in the detection of IFN-β, CXCL10, IL-6 are 0.0133, 0.0307, and 0.0015, respectively. d Flow cytometric assessment images and e relative quantification of DC maturation (CD11c+CD80+CD86+) triggered by cancer cells with different treatments (n = 3 independent experiments). f Inhibitory effects of formulations on proliferation of 4T1 cancer cells investigated via CCK-8 assay (n = 3 independent experiments). h Flow cytometric assessment images and g relative quantification of apoptosis of 4T1 cancer cells after receiving the indicated treatments, cells were stained with Annexin V-FITC/PI (n = 3 independent experiments). g The p values of MTHMS + L to THMS + L and MTHMS + L to MnCO+TSSI + L are 0.0002 and <0.0001, respectively. i Calcein-AM/PI staining of 4T1 cells with various treatments (scale bar = 40 μm). Experiment was repeated three times independently with similar results. Data represent the mean ± SD. Statistical significance was calculated through one-way ANOVA using a Tukey post hoc test. Source data underlying ac, eg are provided as a Source Data file.

In vitro cytotoxicity and apoptosis assay

We first utilized the Chou-Talalay method to analyze the combination. The combination index (CI) was determined referring to the equation: CIn = (D)a/(Dn)a + (D)b/(Dn)b, where (Dn)a and (Dn)b are the dosages of drugs a and b which suppress the cell growth by n%. And (D)a and (D)b express the respective drug concentrations of a and b in which the drug combo suppresses the growth of cells by n%. CI > 1, CI = 1, and CI < 1 represent antagonistic, additive, and synergistic effect, respectively. The growth inhibition effect (fraction affected, Fa) was determined referring to the equation: Fa = 1-(%growth/100). The plots of CI at various Fa levels were constructed. As depicted in Supplementary Fig. 13, mass ratio 1:3 of TSSI to MnCO exhibits the best synergic effect under laser irradiation. The cytotoxicity of various treatments was then evaluated via CCK-8 assay. As shown in Fig. 4f and Supplementary Table 2, MHMS showed limited toxicity, the cell-killing rate was nearly 44.76 ± 5.66%, even with a high concentration of MnCO at 120 μg mL−1. While MnCO+TSSI + L and THMS + L exhibited apparently enhanced cytotoxicity against 4T1 cells with the survival rate of 31.17 ± 3.02% and 21.35 ± 2.09%, respectively. Note that MTHMS + L treated cells displayed the lowest survival rate of 12.06 ± 2.46%, which was owing to the synergistic antitumor efficacy of phototherapy and CO/H2S gas therapy. The CI50 value of MTHMS + L against 4T1 cancer cells was 0.32, demonstrating the strong synergism and good sequential MnCO activation. Moreover, the pretreated 4T1 cells were further incubated for 12 h, flow cytometry was adopted to test the cell apoptosis ratio after annexin V-FITC/PI staining (Supplementary Fig. 14). Up to 97.50 ± 0.91% of PBS treated cells were in the viable area (Fig. 4g, h). Nevertheless, the proportion of apoptotic and necrotic cells increased to 20.79 ± 2.99% and 39.83 ± 4.43% in MHMS and THMS + L treated groups. More evidently, extensive cell apoptosis and necrosis ratio (58.91 ± 3.36%) could be detected in MTHMS + L treated 4T1 cells, confirming the satisfactory therapeutic effect. In addition, the cell apoptosis and live/dead staining of various time points (0, 1, 3, 6, and 12 h) after laser irradiation were assessed to provide a time course for the cell death following MTHMS + L treatment. As shown in Supplementary Fig. 15a, b, the apoptosis ratio and cell death were gradually increased with prolonged incubation time after laser irradiation. Meanwhile, the live/dead staining further indicated the excellent antitumor potential of MTHMS + L (Fig. 4i).

Multimodal imaging

TSSI possessed multimodal imaging capabilities including NIR-II FLI, PAI, and PTI, allowing to acquire rich and accurate tumor data51. As illustrated in Supplementary Fig. 16, a distinct NIR-II fluorescence signal was observed at the tumor site in MTHMS-treated mice compared with free TSSI-treated mice. The signal intensity gradually increased and peaked at 12 h post-injection. To further explore the biodistribution of MTHMS, the mice were euthanized to harvest tumors and main organs at 24 h post-injection for fluorescence imaging and quantitative analysis. As shown in Supplementary Fig. 17a, b, the tumor displayed higher fluorescence intensity than normal organs, probably driven by the enhanced permeability and retention (EPR) effect and favorable tumor adhesion as well as internalization based on the unique virus-like structure. Inspired by the exceptional efficiency of MTHMS to convert light to heat, the same tumor model was applied for PAI. As shown in Supplementary Fig. 18a, the photoacoustic signal in the tumor region significantly strengthened with time, and reached maximum at 12 h, which agreed well with the NIR-II FLI results. Afterwards, the PTI was assessed by monitoring the real-time temperature with an infrared thermal camera. As shown in Supplementary Fig. 18b, after being irradiated by 660 nm laser for 10 min, the temperature rised rapidly from 37.2 °C to 55.7 °C, reflecting the desirable potential of MTHMS for PTT application.

In vivo anticancer study

We further evaluated the in vivo anticancer efficacy of tetrasulfide-doped virus-mimicking and GSH/NIR sequentially initiated gas nanoadjuvant. 4T1 tumor model was employed via inoculation of 4T1 cells into the mammary fat pad of BALB/c mice. On day 10, 4T1 cells were intravenously (i.v.) injected to simulate malignant tumor invasion and hematogenous metastasis (Fig. 5a)52. The additional i.v. injection of 4T1 tumor cells into tumor-bearing mice to simulate hematogenous metastasis has been widely applied as an artificial whole-body spreading tumor model53,54,55,56. Compared with spontaneous lung metastasis, the whole-body metastasis model was more aggressive and challenging, which was suitable for specialized anti-metastasis evaluation. The circulating cancer cells can invade various organs, especially the lung. The treatment of the orthotopic tumor was done twice before the i.v. inoculation of the tumor cells. The tumor growth and body weight changes were recorded every two days. As seen in Fig. 5b–d, the tumor volume of saline group with/without laser irradiation presented a remarkable increase and reached ~1550 mm3 on day 20. By contrast, MHMS exhibited a slight delay of tumor growth (~1160 mm3). Moreover, THMS + L and MnCO+TSSI + L treatments showed distinct inhibition of tumor progression compared to the saline control, but still reached ~720 mm3 and ~910 mm3, respectively. Notably, the tumor growth was completely suppressed after MTHMS + L treatment (~260 mm3). To evaluate the ROS level after MTHMS + L treatment, the DCFH-DA staining of tumor slides was performed. As seen in Supplementary Fig. 19, MTHMS + L treatment induced strong DCF fluorescence intensity in tumor tissue, implying intratumoral ROS burst. We then investigated the intratumoral CO level with FL-CO-1 fluorescence probe. As seen in Supplementary Fig. 20, MTHMS + L treatment provoked considerable CO generation in the tumor site. The survival curve was also recorded (Fig. 5e). As expected, mice in MTHMS + L group had a dramatically prolonged lifespan, displaying the highest survival rate (83%) within 40 days. Additionally, no pronounced reduction of body weight was detected in MTHMS + L group (Supplementary Fig. 21), implying no severe systemic toxicity. Meanwhile, the biosafety of MTHMS + L treatment was further validated by hematoxylin and eosin (H&E) staining of major organs and hepatorenal function measurement. As shown in Supplementary Fig. 22, 23, there was no visible organ damage in H&E staining and no noticeable abnormality in hepatorenal function. To confirm immunological memory initiated by MTHMS + L treatment, 2 × 105 4T1 cells were injected into the left mammary fat pad of the survived mice in MTHMS + L group on day 40. Meanwhile, 4T1 cells were inoculated into naive mice as a control. As shown in Fig. 5f, g, the survived mice in MTHMS + L group showed effective resistance against cancer rechallenge, and the survival time was significantly prolonged, confirming that the gas nanoadjuvant could generate long-lasting antitumor immunity to suppress cancer relapse. Moreover, H&E-stained tumor slices of MTHMS + L treated mice showed serious injury. The Ki67 and terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining of tumor slices also confirmed the widespread apoptosis induced by MTHMS + L treatment (Fig. 5h). We next excised lungs for evaluating the degree of metastasis using Bouin’s fluid staining. Compared with the saline group, no visible lung metastasis nodule could be observed in MTHMS + L group (Fig. 5i, Supplementary Fig. 24), demonstrating efficient suppression of lung metastasis. In addition, lung and liver metastases of tumor cells were assessed with H&E staining. As shown in Fig. 5i, the MTHMS + L treatment effectively inhibited the progression of tumor metastasis.

Fig. 5: Assessment of therapeutic efficiency of MTHMS.
figure 5

a Schematic depicting 4T1 tumor model, 4T1 cells were intravenously (i.v.) administrated into the tumor-bearing mice on day 10 to simulate the hematogenous metastasis (WB: western blotting assay; FC: flow cytometry analysis; IF: immunofluorescence; ELISA: enzyme-linked immunosorbent assay; TUNEL: terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling; H&E: hematoxylin and eosin). b Tumor growth curve (n = 5 mice), (c) tumor weight variations (n = 5 mice), (d) tumor photographs (n = 5 mice), and e survival curve (n = 6 mice) following different treatments. For (c), the p value of MTHMS + L to THMS + L is <0.0001. f Tumor growth curve and (g) survival curve of mice rechallenged with 4T1 cancer cells (n = 5 mice). h Representative images of H&E, TUNEL, and Ki67 staining of tumor slices collected from mice receiving various treatments (n = 3 mice). i Representative photos of lung stained with Bouin’s fluid and representative images of H&E staining of lung and liver following different treatments (n = 3 mice). Dashed outlines indicate lung and liver metastases in H&E staining. Scale bar = 100 μm. Data represent the mean ± SD. Statistical significance was calculated through one-way ANOVA using a Tukey post-hoc test (b), log-rank (Mantel–Cox) test (e, g), or two-tailed student’s t test (f). Source data underlying b, c, e–g are provided as a Source Data file.

To investigate whether the observed antitumor effect is dependent on the immune system (on CD8 T cells), anti-CD8α antibody was intraperitoneally injected into the mice for in vivo lymphocyte depletion. The results showed that CD8+ T cell depletion significantly impaired tumor suppression, led to severe lung metastasis, and shortened the survival time of mice treated with MTHMS + L (Supplementary Fig. 25a–e, 26, 27), confirming the central role of the immune system in gas nanoadjuvant-assisted photoimmunotherapy. Meanwhile, the survived mice in MTHMS + L group were also rechallenged with 4T1 cells with/without CD8+ T cell depletion. As shown in Supplementary Fig. 25f, the tumor volume of CD8+ T cell-depleted mice exhibited an evident increase and reached ~780 mm3 on day 20, validating the effect of the immune system in resistance to tumor relapse.

We also evaluated the effect of MTHMS on tumor growth and lung metastasis without laser irradiation. The lack of laser irradiation resulted in the failure of PDT/PTT therapy and insufficient MnCO activation, which hindered the sequentially initiated stimulation of gas nanoadjuvant. As shown in Supplementary Fig. 25a–e, 26, 27, MTHMS treatment displayed a slight delay of tumor progression (~1060 mm3) and a large number of lung metastasis nodules, indicating a significantly attenuated therapeutic effect.

cGAS-STING pathway activation and antitumor immune responses

The released Mn2+ in cytosol could stimulate cGAS and sensitize cGAS to leaked mtDNA induced by gas nanoadjuvant, promoting the activation of STING pathway and maturation of DCs (Fig. 6a)57,58. We further determined the cGAS-STING pathway-related protein expression and cytokine secretion in tumor tissue through western blotting assay and enzyme-linked immunosorbent assay (ELISA)59,60. As shown in Fig. 6b, in contrast to saline group with/without laser irradiation, the expression levels of phosphorylated STING (PSTING), TBK1(pTBK1), and IRF-3 (PIRF-3) in MHMS group were much increased, implying that tumor microenvironment-driven H2S/CO/Mn2+ release could augment STING-mediated type I interferon production to some extent. Notably, MTHMS + L induced higher expression of STING pathway-related phosphorylated proteins in comparison with THMS + L, indicating that phototherapy-amplified MnCO activation provided the huge potential for cGAS-STING pathway activation. Meanwhile, the secretion of cGAS-STING-related cytokines such as IFN-β, CXCL10, and proinflammatory cytokines involving TNF-α, IFN-γ, IL-6, IL-12 were substantially increased in MTHMS + L group compared to the other groups (Supplementary Fig. 28), supporting the cGAS-STING pathway activation and the potential to switch the “cold” tumor to more sensitive “hot” tumor. We supposed that the MTHMS + L induced cGAS-STING pathway activation related to IFN-β secretion may contribute to DC maturation, which induced the subsequent T cell activation in tumor-draining lymph nodes (TDLNs). To test the above hypothesis, we explored the immune responses elicited by various treatments. As shown in Fig. 6c, Supplementary Fig. 29, the expression of mature DCs (CD11c+CD80+CD86+) in TDLNs significantly boosted after MTHMS + L treatment in contrast with the other groups. Mature DCs could present antigens to T lymphocytes to initiate tumor-specific adaptive immunity. Encouragingly, the frequency of tumor-infiltrating CD8+ T cells showed remarkable increase in MTHMS + L group (Fig. 6d, Supplementary Fig. 30). Consistently, the rate of tumor-infiltrating immunosuppressive regulatory T cells (Tregs, CD4+Foxp3+) in MTHMS + L group declined obviously in comparison with the other groups (Fig. 6e, Supplementary Fig. 30). Moreover, MTHMS + L treatment showed a marked limitation to anti-inflammatory M2-like tumor-associated macrophages (TAMs, CD206hiCD11b+F4/80+), while the population of proinflammatory M1-like TAMs (CD80hiCD11b+F4/80+) increased evidently (Fig. 6f, g, Supplementary Fig. 31, 32). Immunofluorescence staining visualized the noticeable increase of CD8+ T cell and macrophage infiltration in tumor tissue after MTHMS + L treatment (Supplementary Fig. 33), confirming the induction of robust immune responses. Furthermore, we monitored the level of CD8+ effector memory T cells (TEM, CD3+CD8+CD62LlowCD44hi) of mice after different treatments. As seen in Fig. 6h, Supplementary Fig. 34, TEM proportion in spleen was enhanced substantially after treatment with MTHMS + L, indicating a durable immunological memory built by the gas nanoadjuvant. To assess the tumor-specific T cell responses triggered by gas nanoadjuvant-assisted photoimmunotherapy, the percentage of gp70 tetramer-specific CD8+ T cells in the spleen was monitored by flow cytometry analysis. Following MTHMS + L treatment, the proportion of tumor-reactive gp70 tetramer-specific CD8+ T cells rose to a high level of ~4.5% (Supplementary Fig. 35, 36), indicating the induction of tumor-specific T-cell immunity.

Fig. 6: The cGAS-STING pathway activation and antitumor immunity after treatments.
figure 6

G1:Saline, G2:Saline+L, G3: MHMS, G4: MnCO+TSSI + L, G5: THMS + L, G6: MTHMS + L. a Schematic illustration of gas nanoadjuvant-mediated cGAS-STING promotion. b Western blotting assay of STING, TBK1, IRF-3, and phosphorylation of proteins in tumor. Samples derived from the same experiment and gels/blots were processed in parallel. Experiment was repeated twice independently with similar results. c Flow cytometric assay of mature DC (CD11c+CD80+CD86+) in TDLNs (n = 4 mice). The p values of G6 to G5 and G6 to G4 are 0.0003 and <0.0001. d Flow cytometric assay of tumor-infiltrating CD8+ in CD3+ T cells (n = 4 mice). The p values of G6 to G5 and G6 to G4 are 0.0001 and <0.0001. e Flow cytometric assay of tumor-infiltrating CD4+Foxp3+ Tregs (n = 4 mice). The p values of G6 to G5 and G6 to G4 are both <0.0001. Flow cytometric assay of tumor-infiltrating f M2-like macrophages (CD206hiCD11b+F4/80+) and g M1-like macrophages (CD80hiCD11b+F4/80+) (n = 4 mice). f p values of G6 to G5 and G6 to G4 are 0.0006 and 0.0001. g p values of G6 to G5 and G6 to G4 are 0.0002 and <0.0001. h Flow cytometric assay of CD3+CD8+CD62LlowCD44hi TEM in spleen (n = 4 mice).The p values of G6 to G5 and G6 to G4 are 0.0001 and <0.0001. Data represent mean ± SD. Data were compared through one-way ANOVA using Tukey post-hoc test. Source data underlying bh are provided as a Source Data file.

Abscopal effect in bilateral tumor model

We then constructed a bilateral tumor model and investigated whether gas nanoadjuvant could induce systemic immunity and suppress the distant tumor61,62. The bilateral tumor model was established by injection of 4T1 cells into right mammary fat pad as the primary tumor on day 0. And on day 6, 4T1 cells were inoculated into left mammary fat pad of the same mice as the distant tumor. On day 7, when primary tumor reached 80 mm3, mice were randomly assigned to two groups. Saline and MTHMS were intravenously injected on days 7 and 14. Only the primary tumor was irradiated directly at 12 h post-injection and the distant tumor was kept growing naturally (Fig. 7a). Encouragingly, MTHMS + L treatment displayed efficient suppression of both primary and unirradiated distant tumors (Fig. 7b–d), revealing an abscopal effect of gas nanoadjuvant-assisted photoimmunotherapy. To explore whether the tumor growth was affected by CD8+ T cells, anti-CD8a antibody was intraperitoneally injected into the mice for in vivo lymphocyte depletion. The results showed that CD8+ T cell depletion impaired the suppression of the secondary tumor after MTHMS + L treatment (Supplementary Fig. 37), confirming the critical role of CD8+ T cells in the reduced growth of the secondary tumor. To rule out the direct effects of the injected nanoparticles in the abscopal experiment, we evaluated the therapeutic outcome of MTHMS without laser irradiation. As shown in Supplementary Fig. 37, the volume of the secondary tumor increased significantly after MTHMS administration without laser irradiation, whereas MTHMS + L treatment effectively suppressed the secondary tumor, confirming the abscopal effect of gas nanoadjuvant-assisted photoimmunotherapy.

Fig. 7: Gas nanoadjuvant-induced systemic antitumor immunity.
figure 7

a Schematic illustration of bilateral tumor model. Tumor on the right side represents “primary tumor” with laser irradiation, whereas tumor on the left side was defined as “distant tumor” without any treatment (FC: flow cytometry analysis; ELISA: enzyme-linked immunosorbent assay). b Tumor photographs, c tumor growth profiles, d tumor weight variations after the indicated treatments (n = 5 mice). e Flow cytometric assay and relative quantification of tumor-infiltrating CD8+ in CD3+ T cells (n = 4 mice). f Flow cytometric assay and relative quantification of tumor-infiltrating CD4+Foxp3+ Tregs (n = 4 mice). cf The p values of MTHMS + L (right) to Saline+L (right) and MTHMS + L (left) to Saline+L (left) are all <0.0001. g Flow cytometric assay and relative quantification of CD3+CD8+CD62LlowCD44hi TEM in spleen (n = 4 mice). g The p value of MTHMS + L to Saline+L is <0.0001. h The secretion of cytokines (IL-6, TNF-α, and IFN-γ) in serum (n = 4 mice). h The p values of MTHMS + L to Saline+L in IL-6, TNF-α, IFN-γ secretion are <0.0001, 0.0019, <0.0001, respectively. Data represent the mean ± SD. Statistical significance was calculated through one-way ANOVA using a Tukey post-hoc test (cf) or two-tailed student’s t test (g, h). Source data underlying ch are provided as a Source Data file.

Immunosuppressive TME helps tumor progression and immune escape, which severely hinders antitumor activity. The inhibition of unirradiated distant tumor could demonstrate the activation of antitumor immune responses, which improves the recruitment of cytotoxic T lymphocytes and transforms immunologically “cold” TME to “hot” TME. The intratumoral infiltration of CD8+ T cells is a pivotal marker for defining “hot” and “cold” TME. As shown in Fig. 7e, MTHMS + L treatment caused much more CD8+ T cell infiltration in both primary and distant tumors, verifying the relief of immunosuppressive TME. Meanwhile, the remarkable decrease of Tregs in primary and distant tumors, as well as a significant boost of TEM in spleen, was found in MTHMS + L group (Fig. 7f, g). We also evaluated the activation of tumor-specific T cell responses in the abscopal model. As shown in Supplementary Fig. 38, the MTHMS + L evoked the peak frequency of ~4.1% gp70 tetramer-specific CD8+ T cells, manifesting the elicitation of robust tumor-specific T cell immunity. Furthermore, MTHMS + L treatment led to high expression of proinflammatory cytokines in serum including TNF-α, IFN-γ, and IL-6 (Fig. 7h). In addition, no significant body weight change was monitored in MTHMS + L group (Supplementary Fig. 39).


TNBC characterizing significant morbidity and mortality is a common solid tumor malignancy. The immunosuppressive microenvironment in TNBC severely limits the efficacy of current immunotherapy. In the past few years, an increasing number of studies on phototherapy and nanotechnology have focused on affecting the immune system for cancer treatment. Phototherapy is a hopeful modality to eliminate cancer upon laser irradiation and meanwhile stimulates immune response for improved anticancer immunity. However, local phototherapy monotherapy is insufficient for the induction of durable and strong tumor immunogenicity to resist tumor metastasis and rechallenge. Therefore, it is of urgent need to develop a synergistic strategy to amplify the phototherapy-mediated immune activation for photoimmunotherapy of poorly immunogenic TNBC63,64.

In recent years, diverse efforts have been devoted to eliminating tumors through the nanomaterial-based combination of phototherapy and gas therapy65,66. For example, Wan and coworkers loaded biocompatible L-arginine into PCN-224 as an NO donor to combine PDT and NO gas therapy67. Upon laser exposure, the donor L-arginine could react with ROS and H2O2 to produce NO with a wide diffusion range and long half-life. In the hypoxic microenvironment, NO could sensitize cancer cells to PDT-generated ROS and almost completely eradicated cancer. However, the immunostimulating property of gas therapy was rarely explored to assist photoimmunotherapy.

The development of nanomedicine-based strategies for cGAS-STING pathway activation profoundly revolutionized cancer immunotherapy68. Recently, some DNA-damaging drugs, such as teniposide, cisplatin, and olaparib (PARP inhibitor), have manifested the capability to activate cGAS-STING signaling in cancer cells, inducing robust antitumor immune responses. For example, Hou and coworkers designed a nanoactivator that could lead to the presence of DNA in cytosol and improve Mn2+ accumulation in cancer cells59. The nanoactivators were stable in the systemic circulation and activated to release doxorubicin (Dox) and Mn2+ to damage DNA and enhance cGAS-STING activity, which increased DC maturation and boosted intratumoral infiltration of cytotoxic T lymphocytes. Herein, we revealed the promising potential of tumor microenvironment-driven gas therapy in cGAS-STING pathway activation.

The prospective therapeutic implication of this work is the introduction of gas-mediated immunoadjuvant property into phototherapy to combat poorly immunogenic tumor. Gaseous molecule-based gas therapy shows great promise for antitumor treatment. Typically, both H2S and CO are critical physiological messenger molecules. The delivery of exogenous H2S/CO could exert therapeutic effect by decreasing mmP, which would further damage mitochondria, and release mtDNA into cytoplasm. Thus, we explored whether the exposure of cytoplasmic mtDNA could stimulate immunity through cGAS-STING pathway activation, and revealed the immunoadjuvant effect of tumor microenvironment-driven gas therapy. Moreover, mesoporous silica has been widely applied in biomedicine and exhibited great biocompatibility. It was demonstrated that mesoporous silica is excreted through feces and urine. He and coworkers reported that urinary excretion could account for 15–45% of the injected mesoporous silica nanoparticles at 0.5 h post-administration69. The excellent biodegradability guarantees safety for clinical application.

In summary, the GSH/NIR sequentially initiated gas nanoadjuvant was developed via encapsulating AIEgen and MnCO into tetrasulfide-doped virus-mimicking hollow mesoporous silica. Under intratumoral GSH and in situ NIR sequential stimulation, the gas nanoadjuvant could achieve tumor-specific amplified H2S/CO/Mn2+ release to trigger immune responses by cGAS-STING pathway activation, which effectively assisted AIEgen-mediated PDT/PTT therapy in poorly immunogenic TNBC treatment. The synergistic therapeutic effect was verified through the enhanced tumor regression, effective distant tumor suppression, and significant elimination of tumor metastasis and rechallenge. This tetrasulfide-doped virus-mimicking and GSH/NIR sequentially initiated gas nanoadjuvant paves a strategy to boost photoimmunotherapy potency for poorly immunogenic tumor treatment.