Triple-negative breast cancer (TNBC) accounts for around 20% of all types of breast cancers, and young women, in particular, are more susceptible to TNBC [1]. Unfortunately, TNBC has extensive heterogeneity, complexity, and severe invasiveness that are responsible for increased mortality in patients with breast cancer [2,3]. In addition, the negative expression of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 hinders responses of TNBC to normal hormone therapy or human epidermal growth factor receptor 2 receptor targeting therapy [4,5]. PD-1/PD-L1 immune checkpoint blockade is a novel and promising anticancer strategy in comparison with traditional treatments [6,7]. Because of unexpected immunogenicity and the high expense of anti-PD-1/PD-L1 antibodies, the development of organic small-molecule inhibitors as alternatives holds promise in the field of PD-1/PD-L1 immune checkpoint blockade [8]. Because BMS-202 can fill a deep hydrophobic pocket at the cross-section of the PD-L1 homodimer with a cylindrical-like shape [9], BMS-202 is a potent and nonpeptidic PD-1/PD-L1 immune checkpoint inhibitor (ICI). Unfortunately, most clinical outcomes are still unsatisfactory in TNBC [10,11]. In contrast, photodynamic therapy (PDT) can improve the treatment efficacy in a variety of solid tumors and reprogram the tumor microenvironment to be more susceptible to ICI treatments with few adverse effects [12,13]. Chlorin e6 (Ce6), as a well-investigated and effective small-molecule photosensitizer, has been widely used in PDT for various solid tumors [14]. Once Ce6 enters tumor intracellular environments, high concentrations of reactive oxygen species (ROS) can be generated in tumor cells under 660 nm of laser irradiation, which induces tumor cell apoptosis or necrosis and stimulates the host immune system by inducing the rapid release of antigens from tumor cell fragments, a process that leads to dendritic cell (DC) maturation, and activation of cytotoxic T lymphocytes and their infiltration into tumors [15]. Nevertheless, because of immunization checkpoints, it generally cannot be applied as an independent treatment for solid tumors either [16].

Therefore, the general management of monotherapy cannot fully achieve the goal of a radical cure for solid tumors, especially for TNBC. A combination of these emerging strategies can compensate for the deficiencies, integrate the advantages, produce a synergistic effect, and improve efficiency [17,18]. Nevertheless, the hydrophobicity and aggregation properties of ICI and Ce6 in aqueous media limit their application, resulting in insufficient tumor localization and thereby reducing their efficacy. To achieve simultaneous delivery of 2 small molecules, ICI and Ce6, to tumor sites, nanoparticles can generally increase the solubility of hydrophobic therapeutics and provide appropriate size and surface properties to prolong systemic circulation, allowing them to penetrate solid tumors through enhanced permeability and retention effect [19,20]. As a traditional drug delivery system, micellar nanoparticles automatically formed through amphiphilic copolymers have been applied to encapsulate hydrophobic drugs to increase their solubility [21]; however, there are still some disadvantages of micelles, including instability in vivo and limited drug-loading capacity than liposomes [22]. As another traditional drug delivery system, liposomes can change their pharmacokinetic (PK) behaviors in vivo, reduce toxic side effects, and improve efficacy after the drug is carried by liposomes [23,24]; however, fabrication of uniform liposomes usually requires a series of complicated procedures, and general liposomes can hardly maintain the initial size and structure for a period of time [25,26]. Moreover, other organic solvents involved in the fabrication process can hardly be removed completely. Mesoporous silica nanoparticles have been proven that they have a high surface area, pore volume, and high controllability; however, attributing to the silanol groups on their surface, they may bind with the phospholipids on the cell membranes of the red blood that may lead to hemolysis [27,28].

To further improve the long-term stability, uniformity, sturdiness, and drug-loading space of nanocarriers, we designed hollow core–shell carbon spheres with multifunctional integration. In the past decade, the technologies for the fabrication of core–shell nanoparticles or carbon spheres have been well established for multiple applications [29,30]. In particular, hollow core–shell structures can be constructed having large empty capacities, which can be used for loading drugs. Although the resulting hollow product via hard templates can maintain a stable structure and a well-defined morphology, they need to possess channels with suitable sizes on their surface for drug access. In particular, a channel of small size might hinder drugs from accessing, while a channel of large size might hinder the outer shell from protecting the loading drugs and cause deformation of the outer shell. Nevertheless, the sizes are generally difficult to be controlled [31,32]. Regarding other carbon-sphere-based antitumor strategies, a representative study published in Nano Today successfully developed an innovative type of metal-free hollow porous nanozyme that significantly improved the peroxidase-like activity of the nanozyme, thereby achieving efficient activation of the selective prodrug in tumor sites [33]. Differently, we established a simple delivery system only suitable for loading and releasing hydrophobic small molecules in this study. In particular, we have fabricated a hollow carbon sphere having 1 to 2 channel holes ranging in a diameter size from 20 to 40 nm on each surface, which is conducive for the entry of drugs and maintains the whole structure and morphology at the same time. In addition, this hollow carbon sphere carries a strongly adsorbent silicon dioxide sphere inside to further attract the drugs after they access through the channel holes. Besides the advantages of the hollow core–shell carbon spheres for loading ICI and Ce6 above, another advantage of this study was performed by a fasting-mimicking diet (FMD) treatment. Recent research has revealed that FMD is a safe, applicable, and low-cost adjuvant treatment via dietary intervention that can further boost antitumor immunity. A representative study published in Cancer Discovery demonstrates several bioeffects of FMD, including the reversion of peripheral blood immunosuppressive myeloid cells and regulatory T cells, the promotion of T helper 1 differentiation, the increase in interferon-γ (IFN-γ) expressing cytotoxic lymphocytes, and other related immune markers in the circulation and breast tumor tissue. Therefore, we believe that FMD might be suitable in combination with other immunotherapies to achieve the most ideal therapeutic effect on breast tumors [34–36].

Overall, we used a series of simple methods, including precipitation, carbonization, calcination, etching reaction, and hydrothermal reaction, and the final structure includes a strongly adsorbent silicon dioxide sphere, sufficient hollow space for loading drug, a nanoscale hole on its surface, a robust shell, and an outer bilayer, which could efficiently encapsulate PD-1/PD-L1 small-molecule ICI and Ce6 (LEH-CSPC), without multiple chemical modifications. Notably, the shell of the carbon spheres can avoid excessive swelling in comparison to general liposomes. After systemic administration, LEH-CSPC could protect ICI and Ce6 in the circulation of blood and deliver them to tumor sites. Afterward, Ce6 released from LEH-CSPC can generate high concentrations of ROS in tumor cells and destroy tumor cells under laser irradiation. Meanwhile, the released ICI from apoptotic tumor cells can combine with PD-1 or PD-L1, ensure the function of immune cells such as T cells, and regain the clearance effect of T cells on tumors. Additionally and importantly, recent studies have shown that an FMD treatment can further enhance the cellular uptake of nanoparticles and the activities of immune cells. Thus, we initially performed a more innovative approach with the addition of FMD treatment based on the combination therapy above, established a synergistic triad treatment “PD-1/PD-L1 ICI + PDT + FMD”, and eventually achieved the dramatic TNBC-eradicating effect in 4T1-tumor-bearing mice.

The initial materials, carbon spheres filled with silicon dioxide, were synthesized, etched, subjected to a high-temperature hydrothermal reaction to puncture holes on the surface of the carbon spheres, modified with (3-aminopropyl) triethoxysilane (APTES) (EH-CS) that were applied to encapsulate 2 small molecules, and coated with liposome bilayers (LEH-CSPC) (Fig. 1A). As shown in Fig. S1, the size of the silica dioxide inside the carbon sphere is closely related to the etching time. To determine the optimal composition, we investigated the loading efficiency of carbon spheres with different etching times and the same hydrothermal reaction time on 2 small molecules. The loading contents of ICI and Ce6 were detected by ultraviolet (UV) spectrophotometry and fluorescence spectrophotometry, respectively. According to the encapsulation efficiency and loading content (Tables S1 and S2 and Fig. S2), we adopted the etching time of 36 h to carry out further studies, which indicates that nanocarriers for encapsulating small molecules need not only sufficient capacity but also sufficient adsorption force that silicon-dioxide-APTES could provide. We could also confirm the importance of the wider pores with diameters of 20 to 40 nm on the surface for encapsulating small molecules. Previous reports demonstrated that smaller pores exhibited a notably lower capacity for encapsulating molecules [37,38]. Obviously, the transmission electron microscope (TEM) images of the CS (etching time, 0 h) and EH-CS (etching time, 36 h) exhibit regular circular shapes and uniform diameters of less than 200 nm (Fig. 1B and C), suggesting that the benefit of this size could lay the firm foundation for further modifications and biomedical applications and the etching process might not have any effect on the size and whole morphologies of the materials. Notably, after reacting with 2 small molecules and being coated with liposome bilayers, the hollow space in the carbon sphere transferred into a state filled with molecules (Fig. 1D), preliminarily confirming that this material can encapsulate small molecules. Besides, the scanning electron microscope (SEM) image of the CS (etching time, 0 h) exhibited regular spherical morphology, uniform particle size, and surface integrity (Fig. 1E). Pores with diameters of 20 to 40 nm on the surface of EH-CS (etching time, 36 h) could be observed from either high magnification (Fig. 1F) or low magnification (Fig. S3). Moreover, the N₂ adsorption–desorption isotherms were further subjected to the Brunner–Emmet–Teller (BET) analysis as shown in Fig. S4. The typical type IV isotherm was observed for this carbon-coated silica dioxide, indicating a mesoporous structure. After coating liposome bilayers, there were no pores with diameters of 20 to 40 nm on the surface (Fig. 1G), ensuring a complete coating. The diameter and volume of LEH-CSPC and the inner silicon dioxide spheres were quantificationally calculated as shown in Fig. S5, and the inner hollow space indicated a large encapsulating capacity. Compared with the CS without coating liposome bilayers, LEH-CSPC showed an intense negative zeta potential (Fig. S6), further ensuring the successful coating. Furthermore, the wide-scan x-ray photoelectron spectrometer (XPS) spectra of the EH-CS surface shown in Fig. S7A uphold the existing elements of Si, C, N, and O, while the wide-scan XPS spectra of the LEH-CSPC surface shown in Fig. S7B uphold the existing elements of P, C, N, and O. By contrast, the binding energy at around 133 eV was attributed to the P 2p peaks that might be from liposome bilayers. As shown in Fig. S8, diffraction of x-rays indicates that the structure of EH-CS is amorphous, whereas the wide peaks at about 22° and 42.5° are attributed to the carbon layer coated on silica dioxide nanoparticles. According to the results above, carbon spheres in different stages and final LEH-CSPC showed well-defined morphologies, and the outermost lipid bilayer showed a negative charge. To evaluate the stability of LEH-CSPC in serum, we preserved LEH-CSPC in phosphate-buffered saline (PBS) containing 10% serum for 24 h and washed LEH-CSPC. The SEM image showed that LEH-CSPC still maintained the original morphology (Fig. S9A), illustrating the stability of the overall structure. In addition, there was no significant difference in the surface zeta potential of LEH-CSPC compared with that for 0 h (Fig. S9B), continuing to illustrate the stability of the outermost layer of LEH-CSPC. To evaluate the long-term stability of LEH-CSPC, we preserved LEH-CSPC at 4 °C under a dark environment for 10 d. The SEM image showed that LEH-CSPC still maintained the original morphology (Fig. S10A), illustrating the long-term stability of the overall structure. In addition, there was no significant difference in the surface zeta potential of LEH-CSPC compared with that for 0 h (Fig. S10B), continuing to illustrate the long-term stability of the outermost layer of LEH-CSPC.

In this study, we adopted a PDT strategy; however, carbon-based materials also have a photothermal effect [39] that could lead to an auxiliary effect on the release of internally encapsulated small molecules, including ICI and Ce6 (Fig. 1H). Overall, in the aqueous solution, the temperature increased with the increase in the power (0.2, 0.5, and 1.0 W/cm² of 660-nm laser irradiation) under the same concentration of LEH-CSPC and with the increase in the concentration of the LEH-CSPC under the same power (Fig. 1I to K), demonstrating the sensitive responsiveness of the carbon materials to laser irradiation. Because of the heat generated by laser irradiation and the composition of dipalmitoylphosphatidylcholine (DPPC), the outer liposome layer structure could slightly change, and the 2 small molecules inside might spill out. In LEH-CSPC solution (345 μg/ml) containing Ce6 (50 μg/ml), the release of ICI and Ce6 was extremely limited without executing laser irradiation, while the burst release of them occurred when the laser irradiation was applied for 5 min (average temperature, 42.2 °C) (Fig. 1L and M), suggesting that they will realize their functions after they are delivered to the tumor sites.

At the cellular level, FMD could enhance the cellular uptake of nanoparticles containing Ce6 that could induce high levels of ROS under laser irradiation and cause apoptosis or necrosis of tumor cells (Fig. 2A). For in vitro experiments, we selected a triple-negative mouse breast cancer cell line 4T1 as the main research target and a murine colorectal carcinoma cell line CT26 as the control. CT26 and 4T1 are homologous tumor models that are adopted to estimate novel therapeutic approaches. The 4T1 tumor model is generally recalcitrant to most therapeutic agents [40]. Considering the homologous tumor models, we explored whether 4T1 differs from its homologous tumor model CT26 in the cellular uptake of LEH-CSPC. To some extent, the cellular uptake of most kinds of nanoparticles (over 50 nm) is cell-type-dependent [41]. Cellular uptake of LEH-CSPC should probably occur in nonhomogeneous tumor cell lines with different efficiencies. We speculate that it is better to select the homologous tumor models to investigate the difference in cellular uptake. Regardless of laser application, by tracking Ce6 with fluorescent properties, both 4T1 and CT26 cells in the FMD-applied group showed strong fluorescence intensity under confocal microscopy (Fig. 2B and Fig. S11A, respectively). The quantitative fluorescence intensity was investigated via flow cytometry (Fig. 2C and Fig. S11B). Under norm conditions, the fluorescence intensity in 4T1 cells is apparently lower than that in CT26 cells. Interestingly, the fluorescence intensity in 4T1 cells was comparable to CT26 cells under FMD conditions, and the fluorescence intensity in FMD-applied cells was dramatically different from cells under norm conditions. On this basis, we could observe that the LEH-CSPC with simultaneous application of laser irradiation and FMD condition (LEH-CSPC + laser + FMD) could induce the strongest intensity of ROS in both 4T1 and CT26 cells under confocal microscopy (Fig. 2D and Fig. S12). The composition without Ce6 can hardly induce ROS in both 4T1 and CT26 cells, while free Ce6 can freely enter cells and generate a certain degree of ROS. ROS can achieve therapeutic purposes by accelerating the apoptosis of tumor cells. Currently, drugs aimed at increasing the level of ROS in tumor cells are gradually being applied in clinical trials. Regarding cytotoxicity of carbon sphere and ICI to tumor cells, the cell viability dramatically reduced when the concentration of ICI reached over 10 μg/ml (Fig. 2E and Fig. S13A), verifying that ICI could also assist in inhibiting the tumor cell viability. Furthermore, the LEH-CSPC + laser + FMD group containing ICI (10 μg/ml) showed remarkable cytotoxic effects at the determined benchmark (Fig. 2F and Fig. S13B), which might be caused by the combined effects of ROS, ICI, and FMD. Importantly, the LEH-CSPC + laser + FMD group could induce comparable toxicity in 4T1 and CT26 cells, demonstrating efficient suppression of TNBC by our strategy. DC maturation is the major biomarker of PDT-elicited antitumor immune response. 4T1 cells underwent various treatments, and the culture supernatants of 4T1 cells in these groups were adopted to stimulate DC2.4 cells. As costimulatory molecules, levels of CD80 and CD86 of DC2.4 in the LEH-CSPC + laser group showed significant enhancements compared with those of LEH-CSPC alone, indicating the importance of the PDT effect on antigen release from tumor cells and subsequent DC maturation. Interestingly, FMD conditions could further promote these effects (Fig. S14).

The major advantages of nanoparticles include improvement of targeted delivery efficiency for small molecules at tumor sites and reduction of nonspecific distribution in other major organs, and PK and tissue distribution of nanoparticles are crucial in therapeutics efficacy and toxicity in vivo [42]. All animal experiments here were performed in accordance with the relevant laws and followed the institutional guidelines approved by the Institutional Animal Care and Use Committee of the National Center for Nanoscience and Technology (NCNST21-2109-0404). In our study, we adopted the 2-compartment model to analyze the in vivo PK and biodistribution of LEH-CSPC (Fig. 3A). Mice were intravenously injected with different formulations through the tail vein. We collected serums at different time points and detected the Ce6 concentration to analyze major PK parameters. As shown in Fig. 3B, the Ce6 concentration in serum of the free ICI + Ce6 group exhibited a sharper decrease than that of the LEH-CSPC group within 5 h, and Ce6 from the LEH-CSPC group could still be detected in the serum over 10 h. In comparison to the free ICI + Ce6 group, Ce6 from LEH-CSPC showed a 70% increase in both the area under the curve and the mean residence time in the systemic availability. Accordingly, the area under the moment curve value for Ce6 from LEH-CSPC was nearly 3 times than that in the free ICI + Ce6 group. Similarly, half-life (t_1/2) was found to be nearly 2.5-fold as compared with that of the Ce6 in the free ICI + Ce6 group. Moreover, the systemic clearance of Ce6 from LEH-CSPC (0.68 ml/h) revealed a nearly 2-fold decrease in comparison to the free ICI + Ce6 group (1.16 ml/h). The aforementioned PK parameters including the volume of distribution verified that LEH-CSPC can prolong the circulation of small molecules in vivo (Fig. 3C). Meanwhile, the average Ce6 fluorescence intensity in tumors in the LEH-CSPC group under norm or FMD conditions was much higher than that of other organs, and the tumor accumulation was significantly improved compared with the free PD-1 and Ce6 group after 24 h (Fig. 3D to F), verifying that this delivery system could enhance the accumulation of free drugs in tumor sites; however, this result indicates that the FMD condition cannot enhance the tumor accumulation of LEH-CSPC nanoparticles. The main advantage of the FMD condition is the promotion of cellular uptake of LEH-CSPC nanoparticles as discussed before.

On the basis of the effective tumor cytotoxicity and tumor tissue accumulation of LEH-CSPC, we eventually explored the effect on 4T1 tumor cell apoptosis or necrosis in vivo. Regarding this case (Fig. 4A), LEH-CSPC reached the tumor site and was taken up by tumor cells. After laser irradiation, Ce6 could generate ROS in the intracellular environment of tumor cells, promote apoptosis of tumor cells, release tumor antigens to DCs, and further induce infiltration of tumor antigen-specific cytotoxic T cells to tumor sites. Meanwhile, these apoptotic tumor cells could also release the ICI that could bind with PD-L1 of other tumor cells or PD-1 of these cytotoxic CD8⁺ T cells. Eventually, these cytotoxic CD8⁺ T cells above further destroy other tumor cells. In addition, former results in this study also demonstrated that FMD conditions could promote the cellular uptake of LEH-CSPC by tumor cells, which might lead to a large amount of ROS. The synergistic effect via these multiple pathways might accelerate the apoptosis and even necrosis of tumor cells. A total of 2 cycles of treatment were performed on the tumor in vivo, and the brief schedule is shown in Fig. 4B. By measuring the body weights of the mice throughout the whole treatment, only the mice in the LEH-CSPC + laser + FMD group experienced momentary weight losses during the 2 administration intervals, and the body weights of the mice could rapidly recover after finishing each administration. The body weights of the other groups could remain steady within the normal range (Fig. 4C). Moreover, there was no obvious abnormality in the hematoxylin and eosin (H&E) staining sections of the main organs in each group (Fig. S15), ensuring that all formulations might not lead to marked systemic toxicity in vivo. The intratumoral temperature after the administration of LEH-CSPC followed by the laser application was 42.5 °C on average. The tumor volume on the 20th day was smaller than that on the 10th day from the LEH-CSPC + laser + FMD group that achieved the most significant tumor suppression. In contrast, the tumors in the other control groups were not notably inhibited; especially, the tumors in the control and EH-CS + laser groups were nearly 2,000 mm³ on the 14th day. For the ICI + Ce6 group (with laser application), ICI and Ce6 as free small molecules caused a higher proportion of nonspecific distribution in other organs, and few small molecules could travel to the tumor site. For the LEH-CSPC group (without laser application), after LEH-CSPC entered tumor cells, the outer liposome membrane in steady state was destroyed in the acidic environment of lysosomes, and the released ICI from LEH-CSPC via lysosomal escape could lead to certain toxicity to tumor cells; therefore, LEH-CSPC group (without laser application) can cause slight tumor inhibition. In the end point, the LEH-CSPC + laser + FMD group had an apparent tumor-suppressive effect with many areas of tumor necrosis in comparison to other groups (Fig. 4D to F and Fig. S16). There was a positive correlation between tumor weight and final tumor volume in all groups. To further investigate apoptosis of tumor cells in the tumor tissues, H&E staining sections of tumors clearly presented a large reduction of normal cells and many abnormal nuclei in the LEH-CSPC + laser + FMD group compared to other controls (Fig. 4G and Fig. S17). Anyhow, these results could firmly confirm the advantages of LEH-CSPC + laser + FMD formulation for tumor therapeutic efficacy and safety.

PDT can induce immune responses by the released tumor antigens from apoptotic tumor cells, carbon materials, or small chemical molecules, and this immune response as synergistic effects could further assist solid tumor treatment (Fig. 5A). We first investigated DC maturation in lymph nodes around subcutaneous 4T1 tumors in vivo after day 3 following the schedule in Fig. 4B. Levels of CD80 and CD86 of DCs in the LEH-CSPC + laser group showed enhancements compared with those of LEH-CSPC alone, indicating the importance of the PDT effect on antigen release from tumor cells in vivo and subsequent DC maturation. Interestingly, FMD conditions could significantly enhance these effects (Fig. S18). In addition, tumors in ICI + Ce6 + laser, LEH-CSPC, LEH-CSPC + laser, and LEH-CSPC + laser + FMD groups had high percentages of CD45⁺ lymphocytes (Fig. 5b). Importantly, tumors in the CSPC + laser + FMD group had the highest levels of CD8⁺ T cells and IFN-γ expressed CD8⁺ T cells (Fig. 5C to F and Fig. S19), demonstrating that the immune responses were in accordance with the tumor suppression and FMD could further enhance these immune responses. It was also confirmed that FMD enhanced the immune responses in vivo via detecting the level of IFN-γ expressing CD8⁺ T lymphocytes in lymph nodes (Figs. S20 and S21). Nevertheless, the mice in the LEH-CSPC + laser + FMD group could not produce remarkable levels of inflammatory factors in comparison to the LEH-CSPC + laser group, such as tumor necrosis factor α (TNF-α) and interleukin-6 (IL-6) (Fig. S22), although they could produce significant levels of inflammatory factors in comparison to the control groups, suggesting that the auxiliary effect on tumor suppression was mainly from cellular immune responses of activated CD8⁺ T cells. Moreover, mice in each group could not produce any immune storm via detecting immunoglobulin E (IgE) (Fig. S23), further ensuring the safe application of the immune responses. Furthermore, exposure to carbon-based nanomaterials including carbon spheres in vivo is highly related to potential health risks. Previous publications have proved that these carbon materials could be degraded by enzymatic oxidation with peroxidase enzymes, such as horseradish peroxidase, eosinophil peroxidase, myeloperoxidase expressed in neutrophils, and lactoperoxidase expressed in goblet cells. Importantly, the residues of carbon-based nanomaterials after biodegradation can hardly generate cytotoxicity or inflammatory responses [43–46].

In summary, regarding the challenges of TNBC therapy, we have established a combination therapy model. We designed and synthesized a robust carbon sphere on a nanoscale with pores of 20 to 40 nm in diameter on the surface, superabsorbent capacity for hydrophobic molecules, and large loading space inside, which could ensure efficient loading for ICI and Ce6. Then, they could be delivered to tumor sites in vivo, and high intensity of ROS was generated under laser irradiation, which posed a huge threat to the viability of tumor cells. Meanwhile, ICI could be freely released from these apoptotic tumor cells to block PD-1 or PD-L1 immune checkpoint to promote the suppression effect of cytotoxic T cells on other tumor cells. Importantly, we introduced FMD treatment that could improve the cellular uptake of nanoparticles and the immune effect to assist the therapy above to form a combination therapy of “PD-1/PD-L1 ICI + PDT + FMD”, thereby further inhibiting the development of the 4T1 tumor. Therefore, this combination therapy provides a novel opportunity for the clinical transformation of human TNBC therapy with great significance.