Design and characterization of MNPs
In order to construct a multifunctional treatment platform able to co-encapsulate small chemical drugs, nucleic acid, peptide and/or protein, we synthesized a polyethyleneimine-oleic acid polymer complex (PEI-OA) by conjugating OA to the carboxyl groups of PEI (molecular weight 1800 Da) (Fig. 2a). Nanoparticles 1 were then prepared by the thin-film hydration method using PEI-OA, HS15, and Soluplus (Fig. 1a). Next, nanoparticles 1 were added to a solution of CpG and OVA in 5% glucose to prepare nanoparticles 2. The positive charge of nanoparticles 1 was partially neutralized by the negative charge of CpG and OVA (Fig. 2b). The nanoparticle solution was then added to a CS solution to obtain completely negatively charged nanoparticles 2, which gave nanoparticles 3 after reaction with atezolizumab.
Scanning electron microscopy indicated that nanoparticles 2 and 3 were homogeneously spherical, but nanoparticles 3 had a rougher surface and a remarkably different size distribution and morphology (Fig. 2c), confirming that a layer of antibodies had adsorbed to the MNP surface. To verify that MNPs were generated through electrostatic interactions between nanoparticles 1 and CpG, nanoparticles 1 were incubated with fluorescein amidite-labeled CpG (FAM-CpG) and analyzed by confocal microscopy imaging. DiD was encapsulated in MNPs (DiD-N/A) and showed red fluorescence. FAM-CpG was loaded in MNPs and showed green fluorescence. The particles co-loaded with DiD and FAM-CpG (DiD+FAM-CpG-N/A) showed obvious green and red fluorescence as well as obvious colocalization of the two types of fluorescence (Fig. 2d), implying the successful integration of nanoparticles 1 and CpG. These results were confirmed by agarose gel electrophoresis (Fig. 2e), which indicated the complete encapsulation of CpG into the nanoparticles.
To examine whether the described nanoplatform could immobilize large-molecular-weight proteins, MNPs were incubated with FITC-labeled OVA and analyzed by Western blotting and flow cytometry (Fig. 2f, g). The appearance of both FITC and DiD signals in the DiD+FITC-OVA-N/A group confirmed that MNPs could immobilize simultaneously a nuclei acid and a protein, serving as a universal platform for chemo-immunotherapy-based combination therapy.
Next, we examined whether MNPs could serve as a versatile platform for the immobilization of atezolizumab. Dynamic light scattering measurements showed that the average hydrodynamic diameter of nanoparticles increased by approximately 11 nm after incubation with atezolizumab at 4 °C for 1 h, suggesting the successful immobilization of the antibody (Fig. 2h). The nanoparticle size also increased after incubation, with nanoparticles 3 having the largest size, followed by nanoparticles 2 and 1. In addition, nanoparticles 1 showed a positive surface charge, while nanoparticles 2 and 3 had a negative surface zeta potential due to the anionic CS polymer. The encapsulation efficiency of atezolizumab-coated MNPs co-loaded with PTX, CQ, CpG, and OVA (CpG+OVA+PTX+CQ-N/A) was 98.75 ± 1.6% for PTX and 96.8 ± 2.1% for CQ, while the total drug loading efficiency was around 4.30% for PTX and 0.86% for CQ.
Cellular uptake, cytotoxicity, and apoptosis of MNP-treated tumor cells
To examine the effect of the developed MNPs on the cellular uptake of small and macromolecular drugs, we used the 4T1 cell line that overexpresses both the CD44 and PD-L1 receptors [26]. Confocal microscopy showed that the fluorescence intensity of DiD+FAM-CpG-N/A was stronger than that of a solution containing free DiD and FAM-CpG (DiD+FAM-CpG-S) (Fig. 3a), and that the intensity increased with incubation time from 2 to 4 h. Moreover, the number of DiD-positive cells in the DiD+FITC-OVA-N/A group at 2 h post-incubation was 18 times higher than the number in the DiD+FITC-OVA-S group, and the number increased with incubation time (Fig. 3b). Similar results were observed for FITC-OVA-positive cells in both groups, while the intensity of DiD+FITC-OVA-N/A was found to be nine times higher than that of DiD+FITC-OVA-S (Fig. 3c). These findings indicate that cellular uptake was time-dependent and that MNPs promoted drug penetration, consistent with our previous study [22].
The cytotoxicity of the developed nanoplatform was measured in vitro with MTT assay. Among the tested groups, cells treated with blank nanoparticles showed the highest viability, suggesting that the carrier was non-toxic and caused negligible side effects (Fig. 3d). Moreover, the IC50 value was 33.21 μg/mL for free PTX and 22.85 μg/mL for a solution containing free CpG, OVA, and PTX (CpG+OVA+PTX-S). However, after loading onto nanoparticles, the IC50 value of PTX-loaded MNPs (PTX-N) and CpG/OVA/PTX-loaded MNPs (CpG+OVA+PTX-N) decreased to 22.18 and 18.12 μg/mL, respectively, reflecting the beneficial effect of the nanoplatform on cytotoxicity. Further loading of CQ reduced the IC50 value of CpG+OVA+PTX+CQ-N to 5.54 μg/mL, while atezolizumab immobilization (CpG+OVA+PTX+CQ-N/A) resulted in an IC50 value of 5.23 μg/mL (Fig. 3d). These results indicate that the synergistic effect of CQ and PTX along with atezolizumab can significantly enhance the killing effect of individual drugs in cancer cells.
Cells treated with the same formulations as in the cell viability study were then stained with FITC-Annexin V and propidium iodide (PI). Flow cytometry showed that most cells were in early-stage apoptosis or were already dead (Fig. 3e, f). Compared to the PTX-S and CpG+OVA+PTX-S groups, PTX-N and CpG+OVA+PTX-N showed 1.29 and 1.55 times higher ability to induce apoptosis, respectively, implying that MNPs can greatly promote cellular uptake. CpG+OVA+PTX+CQ-N led to higher apoptosis or necrosis rates than CpG+OVA+PTX-N, indicating that CQ can strongly promote cell apoptosis. Furthermore, 67.06% of cells in the CpG+OVA+PTX+CQ-N/A group were in the apoptotic or necrotic stage, which was 1.1 times higher than in the CpG+OVA+PTX+CQ-N group (60.87%) (Fig. 3e, f), suggesting that atezolizumab favors MNP uptake, leading to increased apoptosis.
Effect of MNPs on autophagosome formation
The effect of MNPs on autophagosome formation in 4T1 cell lines was evaluated using an immunofluorescence assay with an anti-LC3B antibody, followed by confocal microscopy. Compared to the control group, the number of LC3B-positive puncta increased after treatment with CpG+OVA+PTX-N/A (Fig. 4a, b), indicating that this formulation promoted autophagosome formation. This result was confirmed by transmission electron microscopy of HeLa cells treated with CpG+OVA+PTX-N/A (Fig. 4c). Given that nanomaterial-induced autophagy may promote cancer cell survival [27], cells were then treated with CpG+OVA+PTX+CQ-N/A, which resulted in significant autophagosome puncta accumulation due to the inhibitory effect of CQ (Fig. 4b).
To investigate the mechanism of nanoparticle-associated autophagy, we examined autophagy flux in HeLa cells treated with different formulations using a tandem fluorescent indicator, mCherry-GFP-LC3B (GFP, green fluorescent protein) [28]. Green fluorescence is very sensitive to the acidic environment of lysosomes and quickly quenched in autolysosomes, so red fluorescence was attributed to autolysosomes. CpG+OVA+PTX-N/A significantly promoted autophagosome formation, but even greater autophagosome accumulation was observed in the CpG+OVA+PTX+CQ-N/A group, as CQ inhibited autophagosome–lysosome fusion (Fig. 4d, e).
HeLa cells in DMEM were treated with different formulations to test their ability to induce autophagy. Western blot analysis showed that LC3B-II level was significantly upregulated in the CpG+OVA+PTX-N/A group compared to controls, while its downstream substrate SQSTM1 was downregulated. LC3B-II and SQSTM1 were significantly accumulated in the CpG+OVA+PTX+CQ-N/A group compared to CpG+OVA+PTX-N/A, suggesting that CQ was effective in reducing the flux as evidenced by accumulation of LC3B-II and SQSTM1 (Fig. 4f, g).
In vivo targeting ability of MNPs
The in vivo targeting ability of CpG+OVA+PTX+CQ-N/A was examined by living imaging. At 2 h post-injection, DiD-loaded CpG+OVA+PTX+CQ-N/A (N) showed a stronger fluorescence intensity at the tumor site than a solution of free DiD (S), suggesting that atezolizumab and CS greatly promoted nanoparticle accumulation (Fig. 5a). However, at 24 and 48 h post-injection, the fluorescence intensity decreased in both groups due to body scavenging, although the residual fluorescence intensity of the N group remained higher than that of the S group. Ex vivo imaging of major organs also showed that fluorescence was mainly distributed in the liver, lung, and spleen (Fig. 5b), as the liver and spleen are rich in reticuloendothelial cells, such as macrophages, and express the CD44 receptor, favoring the engulfment of the CD44-responsive MNPs. In addition, the fluorescence intensity at 48 h post-injection was 3.05 times higher in the N group than in the S group, confirming that atezolizumab and CS synergistically promoted the targeting ability of nanoparticles toward 4T1 tumor cells (Fig. 5c). Further staining of tumor slices with CD44 and PD-L1 showed that CpG+OVA+PTX+CQ-N/A emitted stronger fluorescence than the free DiD solution (Fig. 5d), consistent with the in vivo imaging results. These results suggest that the enhanced targeting and penetration ability of the developed MNPs was due mainly to the tumor-homing effect of atezolizumab and CS.
Anti-tumor effect of MNPs
To evaluate the in vivo antitumor effect of the different preparations, we used a subcutaneous 4T1 breast cancer model of Balb/c mice that usually show decreased long-term survival, even after various treatments [5, 29]. On day 0, mice bearing a large breast tumor were subcutaneously injected with 4T1 cells into the right flank. At 5 days post-injection, the tumor volume reached ~50 mm3. On days 5 and 10, mice were injected intravenously with different formulations and the antitumor efficacy, immune response, and autophagy response were analyzed (Fig. 6a).
Estimation of the tumor volume in the different treatment groups indicated that CpG+OVA+PTX-S could not suppress tumor growth due to the poor targeting capacity of free drugs (Fig. 6b). In contrast, a stronger suppressive effect was observed in the CpG+OVA+PTX-N group, suggesting that MNPs effectively targeted the tumor site. Although further loading of the nanoplatform with CQ enhanced the tumor-suppressive effect and inhibited autophagy, CpG+OVA+PTX+CQ-N did not completely block tumor growth when compared to CpG+OVA+PTX-N. Conversely, CpG+OVA+CQ+PTX-N/A significantly delayed tumor progression, indicating the beneficial synergistic effect of chemotherapy and immune-checkpoint blockade therapy.
The strongest suppressive effect of CpG+OVA+CQ+PTX-N/A compared to the other nanoformulations was also confirmed in measurements of tumor weight, which was < 0.3 g only for the CpG+OVA+CQ+PTX-N/A group (Fig. 6c). Based on these results, we also calculated the inhibition rates of the different preparations, which revealed several treatments that were ineffective based on poor tumor inhibition rate: PTX-N (16.03%), CpG+OVA-N (35.72%), CpG+OVA+PTX-S (21.71%), and CpG+OVA+PTX-N (30.24%). In contrast, CpG+OVA+PTX+CQ-N reduced tumor volume by 69.45%, which was 2.3 times higher than the reduction achieved by CpG+OVA+PTX-N, indicating that CQ greatly enhanced the therapeutic effect of MNPs. After atezolizumab immobilization (CpG+OVA+PTX+CQ-N/A), the inhibition rate increased further to 79.45%, i.e., 1.15 times higher than CpG+OVA+PTX+CQ-N, highlighting the therapeutic effect of the anti-PD-L1 antibody. This conclusion was confirmed by the tumor inhibition rate of CpG+OVA+PTX-N/A (59.97%), which was 1.98 times higher than that of CpG+OVA+PTX-N (30.24%). Consistent with these results, analysis of the tumor morphology suggested that CpG+OVA+PTX+CQ-N/A had the strongest suppressive effect (Additional file 1: Fig. S1).
The body weight of mice in different groups did not change significantly during treatment (Fig. 6d), suggesting that the developed MNPs were safe and biocompatible. Staining of all major tissues (heart, liver, spleen, lung, kidney) with hematoxylin and eosin (H&E) indicated no obvious toxicity for MNPs in Balb/c mice (Additional file 1: Fig. S2). In addition, hemolysis assay (Additional file 1: Fig. S3), routine blood examination (Additional file 1: Fig. S4), apoptosis (Additional file 1: Fig. S5), and autophagy (Additional file 1: Fig. S6) characterization in the liver and spleen of mice further proved the safety of CpG+OVA+PTX+CQ-N/A for normal organs. Among the treatment groups, CpG+OVA+PTX+CQ-N/A achieved the highest survival rate (Fig. 6e) and caused severe nuclei damage and cytosol degradation in tumor cells, as determined by co-staining with terminal deoxynucleotidyl transferase dUTP nick end-labeling (TUNEL) and with 4′,6-diamidino-2-phenylindole (DAPI) (Fig. 6f). Thus, the administration of an autophagy inhibitor along with immunoadjuvants and an immune-checkpoint inhibitor is a more effective cancer treatment strategy than the individual therapies.
Effect of autophagy on the anticancer activity of MNPs
The effect of autophagy on the anticancer activity of MNPs in vivo was assessed by immunofluorescence analysis. Compared to the control group, CpG+OVA+PTX-N/A significantly promoted autophagosome formation, as indicated by the increased number of LC3B-positive puncta. In addition, autophagy inhibitor CQ was added into MNP created as CpG+OVA+PTX+CQ-N/A which led to significantly higher autophagosome accumulation than CpG+OVA+PTX-N/A, indicating that inhibition of autophagy flux in the tumor greatly favors autophagosome accumulation (Fig. 7a, b). Western blot analysis also confirmed that CpG+OVA+PTX-N/A significantly upregulated LC3B-II compared to the control group, while CpG+OVA+PTX+CQ-N/A promoted the accumulation of LC3B-II at the tumor site (Fig. 7c, d), confirming that autophagy inhibition enhances the anticancer activity of MNPs.
In vivo immune response
Cytotoxic T lymphocytes can directly kill cancer cells by releasing perforin, granzymes, and granulysin, while helper T lymphocytes act by regulating adaptive immunity [30]. Here, we determined the levels of CD8+ and CD4+ T cells by flow cytometry and performed immunohistochemical staining to assess the potential immune responses induced by the developed nanoparticles. Compared to the control group, CpG+OVA+PTX+CQ-N/A significantly increased the levels of both cell types (Fig. 8a, b), while immunohistochemistry on tumor biopsies revealed more extensive brown areas in CpG+OVA+PTX+CQ-N/A-treated mice compared to the other treatment groups (Fig. 8c, d, Additional file 1: Fig. S7). Moreover, CpG+OVA+PTX+CQ-N/A increased the number of CD8+ and CD4+ T cells in the spleen by 1.48 and 4.63 times, respectively, compared to the control group (Fig. 8e–g), suggesting that CD8+ and CD4+ T cells are the main effector cells of antitumor response in our animal model. CD3+ and CD8+ T-cell stimulation was confirmed by immunohistochemistry in the spleen (Additional file 1: Fig. S8, Fig. 8h). In contrast, CpG+OVA-N showed reduced immune responses, suggesting that PTX and CQ can lyse tumor cells upon irradiation, serving as tumor-associated antigens.
Our combination therapy not only activated immune cells such as T and natural killer cells, but it also promoted the maturation of dendritic cells and regulated the secretion of cytokines. CpG+OVA+PTX-N, CpG+OVA+PTX+CQ-N, and CpG+OVA+PTX-N/A upregulated the serum levels of tumor necrosis factor-α (TNF-α), which plays an important role in host defense and cellular immunity [31], as well as the levels of interferon-γ (IFN-γ), a mediator of Th1 cells that regulates cell-mediated immune responses [32]. Among these formulations, CpG+OVA+PTX+CQ-N/A nanoparticles led to the greatest increase of TNF-α and IFN-γ (Fig. 8i, j), while they could also considerably reduce the number of PD-L1+ cells compared to control (Additional file 1: Fig. S9), suggesting that they exerted their therapeutic effect by regulating cytokine expression and blocking PD-L1 in T lymphocytes.
Taken together, our experiments suggest that autophagy inhibition in tumors combined with immunoadjuvants and an anti-PD-L1 antibody can enhance immunotherapy of advanced breast cancer, delay tumor growth, and prevent recurrence. Our formulations appear to exert these effects by inhibiting immune evasion of tumor cells via immunosuppression reversal and inducing a T cell-mediated antitumor immune response through dendritic cell maturation, upregulation of tumor-associated antigens and cytokines, and generation of T and natural killer cells.
Immune-memory effect and metastasis inhibition
During T cell proliferation and differentiation, some memory cells are differentiated to exert long-term antitumor effects [33]. To verify the occurrence of long-lasting immune memory in this study, we established a mouse lung metastasis model after removing the primary tumor. On the 10th day of follow-up, severe lung metastasis was observed in the control group, while Ki67 staining confirmed the rapid proliferation of tumor cells in the lungs, which could then trigger systemic spread and lead to death (Additional file 1: Fig. S10a-b). Interestingly, CpG+OVA+PTX+CQ-N/A partially inhibited lung metastasis, reduced lung weight (Additional file 1: Fig. S10c) and the number of lung metastases (Additional file 1: Fig. S10d), and prolonged survival time compared to control. These results strongly suggest that the described MNPs are able to treat primary breast cancer as well as inhibit lung metastasis.