Intravenous infusion of engineered megakaryocytes to produce oncolytic platelets in vivo for enhanced cancer immunotherapy

Significance

Oncolytic viruses (OVs) are emerging as promising cancer therapeutics that selectively kill tumor cells and induce antitumor immunity. However, therapeutic effectiveness of OVs via systemic administration is limited by antibody-mediated neutralization, premature drug clearance, off-target infection, and systemic inflammatory response. Inspired by the thrombopoiesis process and SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) spread mechanism, we herein engineered mature megakaryocytes (MKs) with oncolytic adenovirus type 5 (designated as M-Ad5) that enabled systemic administration for targeted viral immunotherapy. Megakaryocytes harboring virus can generate oncolytic platelets to propagate the virus to tumors and elicit systemic antitumor immunity. Intravenous infusion of M-Ad5 into mice has been demonstrated to effectively suppress the growth of lung cancer as well as prevent tumor recurrence and metastatic spread.

Abstract

Tumor-targeted delivery of oncolytic viruses (OVs) via systemic administration could not only expand virotherapy beyond primary tumors to widespread metastases, but also improve clinical adherence and convenience. We here engineer megakaryocytes encapsulating oncolytic adenovirus type 5 (M-Ad5) to produce oncolytic platelets in vivo by leveraging the thrombopoiesis process. Upon intravenous administration, M-Ad5 travels through the lungs, where it can release OVs-harbored therapeutic platelets into circulation under pulmonary turbulence microenvironments. Under the shelter of platelets, OVs resist inactivation by neutralizing antibodies and actively target widespread noninjectable cancer lesions. Intravenous infusion of M-Ad5 to mice with A549 lung cancer could significantly inhibit tumor growth and prolong survival. In multiple mouse tumor models, M-Ad5 induced a robust antitumor immune response by reprogramming the immunosuppressive tumor microenvironment, and potentiated the response to immune checkpoint inhibitors by recruiting more immune cells. We demonstrated that M-Ad5 in combination with PDL1 inhibitors activated tumor antigen-specific CD8⁺ T cells and memory T cells, thereby suppressing the growth of CT26 colorectal cancer metastasis, and preventing postsurgical B16F10 cancer recurrence and metastatic spread, as well as providing long-term immune protection against the rechallenged tumors.

Oncolytic viruses (OVs) preferentially lyse tumor cells as well as induce antitumor immune responses, emerging as a promising cancer therapeutic, especially for those exhibiting resistance to traditional therapies (1, 2). To date, four OVs products have been approved for marketing and a large number of OVs are also progressing into clinical trials (3, 4). However, administration of OVs usually requires intratumoral injection along with complex localization, which severely restricts the clinical application of OVs (4, 5). Intravenous administration is a more acceptable and preferable route to clinicians and enables improved access of OVs to the noninjected occult lesions, expanding the indications from solid tumors to metastases and hematological cancers (6–11). Notably, systemic delivery of OVs induces antibody-mediated neutralization and premature drug clearance, off-target infection, and systemic inflammatory response (1). Micro–nano formulations based on biomaterials, such as polymers (12), lipids (10), silica (13), and cell membranes (14), could protect OVs from neutralization during circulation, but are prone to rapid clearance and nonspecific uptake by healthy tissues, resulting in limited tumor accumulation. In contrast to synthetic materials, cells are biocompatible as natural carriers with active chemotaxis (15). For example, T cells and mesenchymal stem cells have been exploited as targeted carriers for OVs; but viral infection and intracellular replication may damage the physiologic activity of the carrier cells (16, 17). Due to the lack of nuclei for viral replication and destruction, erythrocytes are promising carriers that can improve blood circulation of OVs, but suffer from their poor tumor-targeting capacity (18). Strategies that enable effective systemic delivery of OVs to multiple tumor sites are still lacking.

Despite the presence of neutralizing antibodies in circulation, viruses have been known to be shielded from systemic clearance through transportation by blood cells (19). For instance, replication-competent virions in HIV-infected individuals reside within circulating megakaryocytes (MKs), the precursor cells of platelets, thereby protecting viruses and perpetuating infection (20). Recent studies also indicated that SARS-CoV-2 infected pulmonary MKs contributed to the systemic virus spread and lung inflammation through the release of virus-containing platelets into circulation in patients with COVID-19 (21, 22). Inspired by the natural mechanisms of viral infection, herein we developed oncolytic adenovirus type 5 (Ad5)-engineered MKs (designated as M-Ad5) for in vivo production of oncolytic platelets to achieve systemic virotherapy (Fig. 1A).

Fig. 1.

Schematic illustration and characterization of the M-Ad5. (A) Schematic illustration of the infusion of engineered M-Ad5 to yield oncolytic platelets in vivo for the treatment of lung cancer as well as tumor metastasis and recurrence. (B) Representative microscope image of mature MKs from the differentiation of L8057 cells after incubation with PMA for 5 d without purification. (Scale bar, 25 μm.) (C) Representative confocal laser scanning microscopy images of bone marrow–derived MK and L8057 cell-derived MK. Cells were stained with iFluor 594-labeled wheat germ agglutinin (WGA, red) and FITC-labeled anti-CD41 antibody (green), respectively. Nuclei were stained with Hoechst (blue). (Scale bar, 20 μm.) (D) Representative confocal laser scanning microscopy images of L8057 cell-derived MK after incubation with Cy5-labeled Ad5 (red) in the presence or absence of polybrene. Cells were stained with iFluor 488-labeled WGA (green). Nuclei were stained with Hoechst (blue). (Scale bar, 25 μm.) (E) Viral loading of M-Ad5 after incubation with Ad5 in the presence or absence of polybrene. Results are presented as means ± SD (n = 4 biologically independent samples). (F) Representative transmission electron microscopy image of M-Ad5. The red arrows indicate the internalized OVs. (Scale bar, 500 nm.)

Upon intravenous administration, M-Ad5 are entrapped in the lung and generate virus-containing oncolytic platelets in situ in pulmonary microenvironments along with shear stress (23, 24), which subsequently enter the systemic circulation. OVs concealed within platelets can evade the host immune system and spread via circulating platelets. Notably, engineered MKs enable to achieve high-sufficient viral cargo packaging and yield oncolytic platelets, in contrast to the poor particle loading within donor platelets. Meanwhile, platelets play a broad range of roles from hemostasis to cancer metastasis, allowing for accumulation in the surgical wound and adhering circulating tumor cells (CTCs) (25, 26). With the assistance of platelet carriers, OVs can target to CTCs in the blood, nascent metastatic lesions, and residual microtumors after surgery. Once engaging with the tumor cells, oncolytic platelets become activated and release OVs to selectively infect tumor cells. Systemic administration of M-Ad5 enhanced antitumor immune responses and sensitized tumors to PD-1/PD-L1 blockade by modulating immunosuppressive microenvironments and inducing T cell infiltration. Our results substantiated that intravenous injection of M-Ad5 to mice with disseminated cancer or metastasis elicited tumor-specific T cells and memory T cells, thereby inhibiting tumor growth and providing long-term immune protection.

Results

Engineering of M-Ad5.

Mature MKs were first obtained by treating murine megakaryocytic progenitor cell line L8057 with phorbol-12-myristate13-acetate (PMA). After induction, L8057 cells displayed decreased proliferation and developed into high ploidy large cytoplasmic MKs (Fig. 1B). We also isolated bone marrow cells from C57BL/6 mice to induce into primary MKs by treatment with thrombopoietin (TPO). During megakaryopoiesis, cells differentiated from both bone marrow precursors and L8057 cells undergo morphological changes, with exceptionally large sizes and highly polyploid nuclei (Fig. 1C). The expression of integrin CD41, a platelet glycoprotein receptor, was identified by both immunofluorescence and western blot, further indicating the successful differentiation of MKs (Fig. 1C and SI Appendix, Fig. S1). Flow cytometry analysis showed that the expression of CD41 and CD42 increased in both L8057-derived cells and bone marrow–derived cells after 7 d of differentiation (SI Appendix, Fig. S2), indicating the successful induction of MKs. Additionally, the proportion of L8057-derived cells with ploidy levels greatly increased after differentiation for 7 d (SI Appendix, Fig. S2F), which is consistent with the observed increase in their cell size (SI Appendix, Fig. S3). Confocal images also observed that L8057-differentiated MKs displayed similar polyploid structure to bone marrow–derived MKs (SI Appendix, Fig. S2H). The numbers of generated platelets from MK gradually increased to 130 platelets per MK after 9 d of differentiation (SI Appendix, Fig. S2G). Because L8057 cells were efficiently differentiated into mature MKs with comparable characteristics similar to bone marrow cells, they were employed for subsequent experiments. As previously reported (27, 28), the process of proplatelet extension and platelet biogenesis from L8057-derived MKs was confirmed by microscopy techniques (SI Appendix, Fig. S3). We also substantiated the generation of anucleate platelets and platelet-like particles from L8057 cells-derived MKs by confocal laser scanning microscopy images, and characterized them by Calcein AM cell viability assay (SI Appendix, Fig. S4) and the expression of CD41 marker (SI Appendix, Fig. S1).

Platelets are small cytoplasmic pieces shed from mature MKs. Prior studies also substantiated that MKs could package macromolecules into the released platelets (29). Therefore, M-Ad5 were further engineered by enriching terminally differentiated MKs and incubating with Ad5. Confocal images of MKs after incubation with Cy5-labeled Ad5 demonstrated that OVs could be efficiently uptake by MKs (Fig. 1D). In contrast, few Cy5-labeled Ad5 signal was observed when directly coculture donor platelets with Ad5 (SI Appendix, Fig. S5A). Of note, anucleate platelets are challenging to actively uptake large particles for cytoplasmic loading and transfection compared with nucleated mammalian cells (30). Additionally, internalization of viruses potentially leads to platelet activation via receptor-mediated binding (31), as indicated by the transformation of resting platelets into activated platelets (SI Appendix, Fig. S5B). Alternatively, engineered MKs enable the efficient generation of functional platelets with high cargo encapsulation efficiency by leveraging the thrombopoiesis process. Internalization of Ad5 into cells occurs upon binding of the adenoviral motif to the cell surface integrins αvβ3 and αvβ5 (32), which also present on the surface of megakaryocytes (33) (SI Appendix, Fig. S6). To enable efficient binding of the penton base to cell surface integrins in vitro, Ad5 particles need to attach to the cell surface. Therefore, the addition of polybrene, a cationic polymer that can neutralize the electrostatic repulsion between viral particles and cell surfaces (34), further enhanced the internalization of OVs by MKs (Fig. 1D) and improved the drug loading of M-Ad5 (Fig. 1E). Flow cytometry analysis showed that the percentage of Cy5-positive MKs accounted for up to 98.5% (SI Appendix, Fig. S7A), suggesting that the majority of MKs successfully loaded with Ad5. After the addition of polybrene, the percentages of Cy5-Ad5 positive MKs remained unchanged (SI Appendix, Fig. S7A), while the fluorescence intensity significantly improved (SI Appendix, Fig. S7 B and C), indicating that the addition of polybrene mainly increases the viral load with few impacts on the number of MKs loaded with Ad5. Transmission electron microscopy (TEM) images also observed a large number of viral particles within the M-Ad5 (Fig. 1F) in contrast to native MKs (SI Appendix, Fig. S8). To assess whether Ad5 could be replicated in MKs, we treated L8057 cells and MKs with GFP-encoded Ad5. After incubation for 24 h, we did not observe the GFP expression in MKs (SI Appendix, Fig. S9) due to the selective replication of Ad5 in tumor cells. The platelet biogenesis function of M-Ad5 was also substantiated by a large number of small nonnuclear platelets detached from MKs in confocal images (SI Appendix, Fig. S10). Meanwhile, Calcein-AM fluorescent assays further validated the cell viability of M-Ad5 and the released platelets (SI Appendix, Fig. S11).

Production of Oncolytic Platelets from M-Ad5.

To investigate whether virus-containing platelets could be released from M-Ad5, Cy5-labeled Ad5 was used for immunofluorescence imaging. Following the differentiation of M-Ad5, numerous Cy5-labeled Ad5 were observed in the proplatelets and platelets (SI Appendix, Fig. S11). The numbers of generated platelets per MK gradually increased with prolonged differentiation time, and the platelet yield from M-Ad5 is comparable to that from MKs (SI Appendix, Fig. S12 A and B), indicating that drug loading did not impair thrombopoiesis capacity. Flow cytometry analysis showed that approximately 90.2% of Cy5-Ad5 positive platelets were produced by M-Ad5 (SI Appendix, Fig. S12C), demonstrating a high proportion of virus-loaded platelets. The fluorescence microscopy images showed that M-Ad5-derived platelets exhibited similar collagen adhesion ability compared with MK-derived platelets and mouse donor platelets (SI Appendix, Fig. S13A), indicating that viral loading did not compromise the collagen adhesion function. We further measured the platelet size using dynamic light scattering and found that the average diameter of the M-Ad5 and MK-derived platelets was approximately 1.48 ± 0.42 µm, compared to 1.57 ± 0.16 µm of donor platelets (SI Appendix, Fig. S13B). Flow cytometry analysis showed that both M-Ad5 and MK-derived platelets exhibited platelet markers CD41 and CD42 comparable to those of mouse donor platelets (SI Appendix, Fig. S14). We also found that there was no significant difference in the levels of proinflammatory mediators PF4 and serotonin released by the M-Ad5 and MK-derived platelets (SI Appendix, Fig. S15 A and B). Moreover, viral loading did not elicit platelet activation, as evidenced by the unchanged expression levels of CD62P before and after drug loading (SI Appendix, Fig. S15C). Platelets can bind with CTCs and form aggregates, which protect from shear stress of the bloodstream and immune elimination. Like native platelets, MK-derived platelets also adhered to tumor cells (Fig. 2A). Western blot analysis showed that P-selectin (CD62P) was expressed on MKs and released platelets (Fig. 2B), while CD44 protein was detected on tumor cells (Fig. 2C). Upon activation, P-selectin as an adhesion molecule on the surface of platelets binds with CD44, thus increasing the interaction between platelets and tumor cells (Fig. 2D).

Fig. 2.

Viral infection of tumor cells and oncolytic platelet production. (A) Representative confocal laser scanning microscopy images showing the interaction between cancer cells and M-Ad5-derived platelets. Cancer cells were stained with WGA (orange), and platelets were labeled with Cy5 (pink). Nuclei were stained with Hoechst (blue). (Scale bars, 10 μm.) (B) Western blot showing the expression of P-selectin in MKs and platelets. (C) Western blot showing the expression of CD44 in different cancer cells. (D) The ligand–receptor interaction between cancer cells and platelets. (E) Schematic depicting the M-Ad5 and tumor cells coincubation in a transwell system. (F) Representative confocal laser scanning microscopy images of A549 tumor cells after coincubation with M-Ad5 in above mentioned transwell system. Cell nuclei were stained with Hoechst (blue) and Ad5 was encoded with GFP (green). (Scale bars, 100 μm.) (G) Representative flow cytometry showing the proportion of A549 cancer cells infected by GFP-encoded Ad5. (H) Quantification of frequencies of Ad5-infected A549 tumor cells. Data are mean ± SD (n = 3). P value was performed by one-way ANOVA multiple comparison test. (I) Representative flow cytometry showing the apoptosis of A549 cancer cells after treatment with MK, Ad5, or M-Ad5, respectively. (J) Quantification of apoptosis of A549 tumor cells after treatment with MK, Ad5, or M-Ad5. Data are mean ± SD (n = 3). P value was performed by one-way ANOVA multiple comparison test. (K) Quantitative analysis of lung accumulation from mice that received Cy5-labeled MK, donor platelet, L8057 cell, and MK-derived platelet, respectively. Data are mean ± SD (n = 3). P value was performed by one-way ANOVA multiple comparison test. (L) Representative immunostaining images of mouse lungs. M-Ad5 were stained with anti-CD41 antibody (green) and Ad5 was labeled with Cy5 (pink) before injection, respectively. Nuclei were stained with DAPI (blue). (Scale bar, 10 μm.) The white arrows indicate merged signals of CD41-specific anucleate cytoplasmic discs (proplatelets and platelets) derived from M-Ad5 with Cy5-labeled Ad5.

To further validate whether virus-bearing platelets could release from the M-Ad5 and infect tumor cells, M-Ad5 was incubated with tumor cells in a transwell coculture system (Fig. 2E). As a result, oncolytic platelets were generated from M-Ad5 in the upper compartment and migrate through the micropores, further infecting tumor cells in the lower compartment (Fig. 2F). We also collected oncolytic platelets through a gradient centrifugation, and incubated them with tumor cells to validate their infection ability toward A549 tumor cells SI Appendix, Fig. S16). To investigate whether M-Ad5 can directly release Ad5, we have detected the amount of Ad5 in the release medium and platelets after thrombopoiesis of M-Ad5. The results showed that M-Ad5 generated Ad5-loaded oncolytic platelets, along with a small percentage of free Ad5 (SI Appendix, Fig. S17). Confocal images revealed that tumor cells after infection with OVs displayed significant changes in the cell morphology (SI Appendix, Fig. S18). Flow cytometric detection further substantiated that both free Ad5 and M-Ad5 effectively infected A549 tumor cells, resulting in 94.0% and 96.8% of GFP-positive cells, respectively (Fig. 2 G and H). Consequently, Ad5 and M-Ad5 induced comparable tumor killing effects on A549 cells (Fig. 2 I and J), in contrast to the large number of viable cells in the control group.

In Vivo Virus Delivery of M-Ad5.

To examine the biodistribution of MKs, cells were conjugated with Cy5 and tracked by IVIS spectrum imaging. After infusion, MKs exhibited a tendency to become entrapped in the lung due to their large size compared to native platelets and undifferentiated L8057 cells (Fig. 2K and SI Appendix, Fig. S19). To investigate whether infused MKs home to the bone marrow, the spread of infused MKs in the bone marrow and other organs was further examined. The maximum fluorescence intensity was observed in the lungs while no fluorescence was detected in the bone marrow of mice (SI Appendix, Fig. S20), indicating that infused MKs preferred to accumulate in the lungs instead of the bone marrow. Importantly, there is no significant damage observed on the histological studies of the lung from mice infused with MKs (SI Appendix, Fig. S21). Emerging evidence has shown that the lung is a MK reservoir and large amounts of platelets are generated from pulmonary MKs via the shear forces of blood stream in the lungs (24). To investigate whether infusion of M-Ad5 into mice could yield oncolytic platelets within the lung, Ad5 and MKs were fluorescently labeled with Cy5 and CD41 for immunofluorescence imaging. After intravenous administration, CD41-labeled and nonnucleated fragmentation (unstained by DAPI) were observed within the lungs of mice (Fig. 2L), demonstrating the production of proplatelets and platelets from MKs. Moreover, the fluorescence of Cy5-Ad5 detected in CD41-positive platelets suggested that M-Ad5 could shed virus-containing platelets in vivo (Fig. 2L), consistent with in vitro release of oncolytic platelets.

Upon intravenous injection, OVs would undergo rapid neutralization by preexisting antiviral antibodies and liver sequestration, thus decreasing the oncolytic potency (35). To evaluate whether M-Ad5 protects OVs from neutralization, we incubated cancer cells with GFP-encoded Ad5 or M-Ad5 in the presence of antiviral antibodies. The use of M-Ad5 to shield viruses from neutralizing antibodies enabled to efficiently infect tumor cells with a high GFP-positive cell population compared to free Ad5 (SI Appendix, Fig. S22). In addition, in vivo pharmacokinetics showed that intravenously administered M-Ad5 significantly prolonged the systemic circulation of OVs in contrast to the rapid clearance of naked Ad5 (SI Appendix, Fig. S23). The serum levels of OVs in mice administered with M-Ad5 were 4.03-fold and 6.01-fold higher than those of free Ad5 at 2 h and 4 h, respectively, indicating the extended systemic exposure to OVs.

M-Ad5 Systemic Virotherapy for Lung Cancer.

To evaluate the tumor-targeting ability and in vivo antitumor efficacy of M-Ad5, an orthotopic A549 lung cancer model was established in nude mice (Fig. 3A). Ad5 was labeled with Cy5 to indicate in vivo delivery efficiency of M-Ad5. Ex vivo fluorescence imaging showed that M-Ad5 significantly increased the accumulation of Ad5 within the lung tumors, while few fluorescence signals were detected post intravenous injection of the naked Ad5. (Fig. 3 B and C). Furthermore, fluorescence images of lung slices also validated the improved tumor delivery efficiency of Ad5 when hitchhiked in MKs (Fig. 3D).

Fig. 3.

M-Ad5 against A549 lung cancer. (A) Experimental timeline of M-Ad5 for the treatment of A549 lung cancer. (B) Ex vivo images and (C) quantitative analysis of mouse lung that received Cy5-labeled Ad5 and M-Ad5. Data are mean ± SD (n = 3). Statistical significance was performed by unpaired two-tailed Student t tests. (D) Representative immunostaining images of lungs from A549 tumor-bearing mice that received Cy5-labeled Ad5 and M-Ad5. Ad5 was labeled with Cy5 (pink) and nuclei were stained with Hoechst (blue). (Scale bar, 500 μm.) (E) In vivo bioluminescence images of A549 lung cancer in BALB/c nude mice after receiving PBS as a control group, Ad5, or M-Ad5. (F) Quantitative analysis of A549 tumor bioluminescence signals. Data are mean ± SEM. (n = 7). Statistical significance was performed by one-way ANOVA with multiple comparisons. (G) Tumor bioluminescence intensity changes in A549 tumor-bearing mice with different treatments. (H) Survival curves for mice treated with different formulations (n = 7). (I) Representative H&E staining of lungs from A549 lung tumor-bearing mice. (Scale bar, 2 mm.)

We next assessed the therapeutic effects of M-Ad5 in the human A549 lung cancer model by monitoring the tumor growth via the bioluminescence of A549 cells (Fig. 3A). Tumor growth was significantly inhibited in the mice treated with M-Ad5 and tumor bioluminescence signals were not detectable in four out of seven mice even at day 23 (Fig. 3E). In contrast, increased tumor bioluminescence was observed in all mice that received free Ad5 (Fig. 3 F and G). Consequently, M-Ad5 prolonged the mouse survival rate compared with free Ad5, with least 28.5% of the animals surviving more than 50 d (Fig. 3H). Hematoxylin and eosin (H&E) staining of lung sections further substantiated those mice treated with M-Ad5 showed less tumor lesions in contrast to free Ad5 treatments (Fig. 3I).

M-Ad5 Therapy Sensitizing Metastatic Tumors to Immune Checkpoint Blockade.

OVs exert their therapeutic effects mainly by directly lysing tumor cells as well as triggering antitumor immune (1). Of note, OVs exhibited a more potent tumor killing effect on human A549 lung carcinoma epithelial cells compared to murine CT26 colon carcinoma cells and B16F10 melanoma cells (SI Appendix, Fig. S24). Given that OVs induce antitumor immunity, M-Ad5 may potentially recruit T cells into the tumor and potentiate the effects of immune checkpoint inhibitors (36). We observed that viral infection greatly increased the expression of PDL1 on CT26 cells compared with the control group (Fig. 4A). Although virotherapy induced PDL1 expression in cancer cells, OVs can sensitize immune checkpoint blockades by enhancing immune cell infiltration (SI Appendix, Fig. S25). Meanwhile, flow cytometry of CT26 cells post Ad5 or M-Ad5 treatments also showed that PDL1 expression was remarkably upregulated compared to the untreated cancer cells (Fig. 4 B and C). Furthermore, B16F10 melanoma tumor cells were similarly susceptible to viral infection and upregulated the expression of PDL1 proteins when treated with Ad5 or M-Ad5 (SI Appendix, Fig. S26).

Fig. 4.

M-Ad5 combining with aPDL1 for the treatment of CT26 lung metastasis. (A) Western blot analysis of the expression of PDL1 protein in CT26 cancer cells. (B) Representative flow cytometry plots of the expression of PDL1 protein in CT26 cancer cells. (C) Quantitative analysis of the expression of PDL1 protein in CT26 cancer. Data are mean ± SD (n = 5). (D) Schematic of metastasis monitoring and survival assessment after different treatment regimens. (E) Representative in vivo bioluminescence images of CT26 lung metastases in BALB/c mice. (F) Quantitative analysis of CT26 metastasis bioluminescence signals (n = 5). (G) Survival curves of CT26 lung metastasis-bearing BALB/c mice (n = 7). (H) Body weight of CT26 lung metastasis-bearing BALB/c mice (n = 7). (I) Representative flow cytometry plots of the percentage of CD8⁺ T cells in the peripheral blood. (J) Quantitative analysis of CD8⁺ T cells in the peripheral blood. Data are mean ± SD (n = 6). P value was performed by one-way ANOVA multiple comparison test. Quantitative analysis of (K) CD8⁺ T and (L) CD4⁺ Foxp3⁺ T cells in the metastatic lungs. Data are mean ± SD (n = 5). P value was performed by one-way ANOVA multiple comparison test. (M) IFN-γ in the metastatic lungs. Data are mean ± SD (n = 4). P value was performed by one-way ANOVA multiple comparison test. (N) Representative immunofluorescence staining of CD8⁺ T cells (pink) in lungs of mice received different treatments. Nuclei were stained with DAPI (blue). (Scale bar, 50 μm.)

To verify whether M-Ad5 combined with aPDL1 can increase therapeutic outcomes, we established the CT26 mouse lung metastatic model (Fig. 4D). Tumor growth was recorded by bioluminescence imaging and mice received treatments following the development of lung metastases on day 5 (SI Appendix, Fig. S27). Rapid growth of tumor metastasis occurred in the untreated mice that caused the death of all the mice within 18 d (Fig. 4E). Although mice treated with M-Ad5 showed a moderate therapeutic benefit, all mice developed significant tumor progression (SI Appendix, Fig. S28). In contrast, M-Ad5 combined with aPDL1 (aPDL1+M-Ad5) induced a potent antitumor effect and led to the lowest tumor bioluminescence signals (Fig. 4F). Consequently, aPDL1+M-Ad5 treatments significantly increased the survival rate of mice, with 42.8% of mice alive after 30 d (Fig. 4G). Meanwhile, aPDL1+M-Ad5 treatment attenuated the weight loss in tumor metastatic mice (Fig. 4H). The lungs were also collected on day 14 for H&E staining. Histological analysis showed few metastatic foci observed in lungs of mice that received M-Ad5 and aPDL1 treatments compared with that in the other groups (SI Appendix, Fig. S29). Importantly, no noticeable damages or abnormal changes were detected on histological analysis of main organs (SI Appendix, Fig. S30), biochemical tests, and blood cell counts (SI Appendix, Fig. S31). Additionally, the histological analysis of the brain of mice after administration with M-Ad5 also confirmed the safety (SI Appendix, Fig. S32), potentially due to the difficulty of Ad5 in crossing the blood–brain barrier (SI Appendix, Fig. S33). To examine whether these engineered platelets might have dysfunctions affecting hemostasis in vivo, we performed a tail coagulation assay. The proportion of platelets in the blood was measured after 90 min following MK infusion, as previous studies have shown that platelets derived from infused MKs reach their peak at this time point (37). The total platelet count in mice did not show significant changes after MKs injection for 90 min, consistent with the finding that platelets derived from the infused cells accounted for only 1.49% of the total platelet count in recipient mice (SI Appendix, Fig. S34 A and B). Although mice treated with MKs and M-Ad5 exhibited a slight reduction in bleeding time compared to the control group, no statistically significant differences were observed among the different treatment groups (SI Appendix, Fig. S34C), indicating that M-Ad5 treatment had no effect on tail bleeding time in mice compared to the control group.

M-Ad5 Combined with aPDL1 to Induce Robust Antitumor Immune Response.

To further elucidate the immune mechanism of M-Ad5 therapy, in vivo antitumor immune response was analyzed in murine CT26 lung metastatic model by flow cytometry (SI Appendix, Figs. S35 and S36). The CD8⁺ T cell population in peripheral blood was markedly increased in mice treated with the aPDL1+M-Ad5 compared with other groups (Fig. 4 I and J). Meanwhile, aPDL1+M-Ad5 therapy also induced a significantly high frequency of CD8⁺ T cells in lung metastatic tumors (Fig. 4K and SI Appendix, Fig. S37A). Immunofluorescence staining further substantiated that the percentage of CD8⁺ T cell in lung metastatic tumors elevated upon dual treatments with M-Ad5 and aPDL1 (Fig. 4N). Moreover, we also observed a significant reduction in tumor-infiltrating CD4⁺Foxp3⁺ T cells in mice treated with M-Ad5 and its combination, but not in the free Ad5-treated group (Fig. 4L and SI Appendix, Fig. S37B). Altogether, these data suggested that MKs could effectively deliver Ad5 to the metastatic tumor sites and transform resistant tumors more sensitive to aPDL1, potentiating antitumor immune response. In addition, the concentrations of interferon-γ (IFN-γ) in tumors significantly increased after treatment of cancer cells with M-Ad5 (Fig. 4M). Of note, IFN-γ is able to upregulate the expression of PDL1 in tumor cells (38), and thus the combination of M-Ad5 with aPDL1 therapy could augment the objective response rate of synergistic therapy.

M-Ad5 Therapy Combined with aPDL1 for Postsurgical Recurrence and Metastasis.

Given that intravenously injected MKs could release oncolytic platelets into the circulation, M-Ad5 virotherapy also holds potential to target the tumor cell dissemination and surgical wounds induced by surgery. To investigate whether M-Ad5 could treat CTCs and postsurgical tumor recurrence, we further assess the potential of this synergistic virotherapy in B16F10 mouse melanoma with surgical resection and metastasis (Fig. 5A). After surgical removal of the primary tumor, the fluorescence of Cy5-labeled Ad5 was 3.34-fold higher around the surgical cancer wound in mice treated with M-Ad5 than that of free Ad5 (Fig. 5 B and C), indicating the enhanced delivery efficiency to residual tumors. Furthermore, administration of free Ad5 showed a low lung metastasis-targeting delivery efficiency (SI Appendix, Fig. S38), accompanied by a low lung/liver ratio of accumulation. In contrast, systemic delivery of M-Ad5 significantly increased the accumulation of Ad5 within the lung tumor metastasis (SI Appendix, Fig. S38), potentially maximizing therapeutic efficacy and minimizing off-target side effects.

Fig. 5.

M-Ad5 combining with aPDL1 suppressed lung metastasis and local recurrence of melanoma. (A) Schematic illustrating the antitumor assessment in a mouse melanoma model of postsurgical resection and metastasis. (B) Representative in vivo fluorescence imaging of surgical wounds of mice after intravenous administration with Ad5 or M-Ad5 at an equivalent dose. (C) Quantitative analysis of Ad5-Cy5 fluorescence intensity in the surgical bed. Data are mean ± SD (n = 5). Statistical significance was performed by unpaired two-tailed Student t tests. (D) Representative in vivo bioluminescence imaging of postsurgical B16F10 tumors. (E) Tumor growth curves of B16F10 tumor-bearing mice with different treatments. (F) Quantitative analysis of B16F10 tumor growth (n = 7). (G) Survival curves of B16F10 tumor-bearing mice with different treatments. (H) Representative in vivo bioluminescence imaging of B16F10 metastasis on day 14. (I) Representative lung photographs of B16F10 metastasis-bearing mice on day 14. Black arrows indicate the lung metastatic sites. (J) Quantitative analysis of CD3⁺ T cells in CD45⁺ cells in the residual tumors. Data are mean ± SD (n = 4). (K) Quantitative analysis of CD8⁺ T cells in CD45⁺ cells in the residual tumors. Data are mean ± SD (n = 4). (L) Quantitative analysis of CD4⁺ T cells in CD45⁺ cells in the residual tumors. Data are mean ± SD (n = 4 biologically independent samples). (M) Quantitative analysis of CD86⁺ cells in F4/80⁺ cells in the residual tumors. Data are mean ± SD (n = 4). (J–M) P value was performed by one-way ANOVA multiple comparison test.

M-Ad5 combined with aPDL1 efficiently inhibited the local tumor recurrence and three of seven mice did not show markedly tumor progression even on day 30 (Fig. 5 D and E). In contrast, rapid tumor growth was observed in mice after the treatment with free Ad5 and aPDL1, evidenced by the increased tumor volume changes and bioluminescence signals of the B16F10 cells (Fig. 5 D–F). As a result, the mice receiving M-Ad5 and aPDL1 synergistic therapy exhibited a prolonged survival time than those treated with either M-Ad5 or free Ad5 therapy (Fig. 5G). Furthermore, aPDL1+M-Ad5 synergistic therapy also effectively suppressed the lung metastasis growth and reduced the lung metastatic foci, in contrast to obvious metastatic foci that occurred in the other groups (Fig. 5 H and I). To further elucidate the underlying immune mechanism, tumor-infiltrating lymphocytes from the recurrence tumors were analyzed. In line with the antitumor efficacy, aPDL1+M-Ad5 treatments significantly enhanced the infiltration of CD3⁺ T cells in the tumor (Fig. 5J). Similarly, the percentages of CD8⁺ T cells and CD4⁺ T cells elevated in residual tumors after aPDL1+M-Ad5 treatments as compared to other control groups (Fig. 5 K and L). Moreover, the proportion of M1-type macrophages significantly increased in the residual tumors after M-Ad5 therapy (Fig. 5M and SI Appendix, Fig. S39). These results indicated that M-Ad5 combined with aPDL1 effectively promoted intratumoral immune cell infiltration and circumvented immune suppression, thus eliciting robust antitumor response.

T Cell–Mediated Antitumor Efficacy and Immune Memory for M-Ad5 and aPDL1 Therapy.

CD8⁺ T cells are important for antigen-specific immune responses and antitumor immunotherapy. To investigate the contribution of CD8⁺ T cells in M-Ad5 and aPDL1 combination therapy, we deplete CD8⁺ T cells by intraperitoneal injection of CD8 antibody before treatments. The elimination of CD8⁺ T cells greatly abolished the therapeutic efficacy of M-Ad5 and aPDL1 in mice, leading to a rapid tumor progression and poor survival (Fig. 6 A–C). It further indicated that CD8⁺ T cells are required to exert antitumor effects for M-Ad5 and aPDL1 immunotherapy. To explore whether tumor antigen-specific CD8⁺ T cells, a subpopulation of cells capable of effectively killing tumor cells, were activated by M-Ad5 combination therapy, we established B16F10 expressing ovalbumin (B16F10-OVA) tumor model and harvested tumors and spleens from mice receiving different treatments for flow cytometry (SI Appendix, Fig. S40). The combined administration of M-Ad5 and aPDL1 efficiently activated tumor-specific CD8⁺ T cells, resulting in a significant increase in the level of OVA Tetramer-specific CD8⁺ T cells within tumors (Fig. 6D and SI Appendix, Fig. S41). Also, as observed in tumors, a higher proportion of tumor antigen-specific CD8⁺ T cells were detected in the spleens of mice treated with M-Ad5 and aPDL1 compared to other groups (Fig. 6E and SI Appendix, Fig. S41). Collectively, these results suggest that M-Ad5 enhanced the infiltration of immune cells to potentiate the aPDL1 blockade and synergistically activated tumor-specific CD8⁺ T cells to delay tumor progression.

Fig. 6.

M-Ad5 combining with aPDL1 inhibited rechallenged tumors and induced immune memory protection. (A) Tumor growth curves of B16F10 tumor-bearing mice with different treatments. Data are mean ± SEM (n = 6). (B) Tumor volume of B16F10 tumor-bearing mice with different treatments on day 18. Data are mean ± SD (n = 6). P value was performed by one-way ANOVA multiple comparison test. (C) Survival curves of B16F10 tumor-bearing mice with different treatments. (D) Quantitative analysis of tumor antigen-specific CD8⁺ T cells in the B16F10-OVA tumors. (E) Quantitative analysis of tumor antigen-specific CD8⁺ T cells in the spleens. Data are mean ± SD (n = 5). P value was performed by one-way ANOVA multiple comparison test. (F) Tumor growth curves of mice with rechallenged B16F10 tumors. Data are mean ± SEM (n = 6). (G) Survival curves of mice with rechallenged B16F10 tumors. (H) Quantitative analysis of CD8⁺ effector memory T cells and CD8⁺ central memory T cells in the spleens. Data are mean ± SD (n = 6). Statistical significance was performed by unpaired two-tailed Student t tests. (I) Representative flow cytometry plots of the percentage of CD8⁺ memory T cells in the spleens.

To evaluate the antitumor immune memory of M-Ad5 and aPDL1 combination, we rechallenged the cured mice with B16F10 cancer cells again to examine their long-term protective immunity. All naïve mice without treatments developed tumors and did not survive more than 23 d (Fig. 6 F and G). By contrast, the combination of M-Ad5 with aPDL1 significantly inhibited tumor growth, resulting in complete tumor rejection in 4 of 6 mice. To explain the mechanism of the robust immune protection, memory T cells collected from the spleens of the cured mice or naïve mice were analyzed by flow cytometry. Compared to naïve mice, we observed a significant increase in effector memory CD8⁺ T cells (Tem, CD8⁺CD62L⁻CD44⁺) and central memory CD8⁺ T cells (Tcm, CD8⁺CD62L⁺CD44⁺) in the spleens of mice cured by combination therapy with M-Ad5 and aPDL1 (Fig. 6 H and I), indication the induction of long-term immune memory. As such, intravenous administration of M-Ad5 and aPDL1 promoted robust systemic antitumor immune responses and effective long-term immunity.

Discussion

Toxicological evaluation of OVs is crucial for advancing their transition from laboratory research to clinical application, ensuring the provision of innovative therapies that are both effective and safe for cancer patients (1, 2). Currently available clinical results indicate that intratumorally injection of OVs offers acceptable tolerability and safety, with a low risk of long-term toxicity (2, 3). While systemic delivery of OVs enables broader in vivo distribution to treat inaccessible or metastatic tumors compared to intratumorally injection, it also presents greater challenges and complexity in toxicology studies (1). Future studies need to comprehensively evaluate potential long-term toxicological effects beyond the therapeutic target that may arise from repeated or long-term use of M-Ad5. Additionally, assessment of long-term protective immunity over multiple time points and its antitumor efficacy in humanized cancer models are expected to further advance the clinical translation potential of M-Ad5.

Despite the development of a variety of cells for OVs delivery, such as mesenchymal stem cells and T cells, the replication of OVs within cells may impair the functional activity of the carrier cells, thereby compromising their delivery capacity (16, 17). Distinct from other cell carriers, platelets offer unique advantages for OVs delivery because they lack the nucleus for viral replication and attack. Due to their key role in hemostasis and inflammatory immunity, platelets can not only efficiently deliver cargoes to postoperative tumor sites, but also precisely target tumor neovascularization and metastases (39). In addition, platelets could be selectively activated by tumor-associated signals to facilitate virus release and spread. However, anucleate platelets are challenging to actively uptake large particles for cytoplasmic loading and transfection in contrast to nucleated mammalian cells (30). Additionally, the internalization of viruses such as HIV and SARS-CoV-2 potentially leads to platelet activation via receptor-mediated binding (22, 31). In this study, we engineered M-Ad5 to produce oncolytic platelets in vivo to avoid rapid clearance of the virus by preexisting neutralizing antiviral antibodies and other immune responses, as well as enhance viral delivery to the tumor site by leveraging the physiologic interaction of platelets with tumors. Instead of existing donor-platelet delivery systems, engineering MKs, the precursors of platelets, provide an alternative strategy for efficient macromolecule delivery.

Importantly, in vivo production of functional platelets could avoid the preactivation of platelets after ex vivo-derivation from MKs (37, 40). Furthermore, MKs and proplatelets can be differentiated from stem cells to satisfy clinical-scale demand (41), thus promoting the translational potential of the M-Ad5. Despite the advantages of allogeneic cell immunotherapy, such as the possibility of multiple gene editing, robust scalability, and on-demand availability of off-the-shelf therapy, immunogenicity, and host rejection of allogeneic donor cells have hindered its widespread clinical application (42). Since allogeneic MKs can also cause transplantation rejection, an adequate source of autologous MKs is critical for clinical translation. With the development of gene editing technology, it is expected to reduce immune rejection by genetically engineering induced pluripotent stem cells (41). Alternatively, cells can be used as both carriers and therapeutic agents. Virotherapy has been reported to increase the expression of PDL1 in cancer cells, while OVs could sensitize tumors to PD-1/PD-L1 blockade by recruiting more immune cells (36). Therefore, MKs can be genetically engineered to express therapeutic molecules (28, 43), such as cytokines and checkpoint blockades, for further boosting the systemic antitumor immunity of M-Ad5. Beyond the use of MKs for viral delivery, this strategy also offers opportunities to engineer therapeutic cells for cancer, thrombocytopenia, and autoimmune diseases.

Materials and Methods

Preparation of M-Ad5.

OVs were thawed on ice and diluted with 2% FBS-containing culture medium to the desired concentration for use. Mature MKs were obtained by the 1.5 to 3% BSA two-step gradient. M-Ad5 was obtained by the coincubation of MKs with OVs at a multiplicity of infection (MOI) of 500 for 4 h at 37 °C in the presence of 10 μg/mL polybrene. Upon completion of incubation, cells were washed twice with PBS and collected by centrifugation at 100 g for 5 min. The drug loading and OVs copies were quantitated by qPCR analysis (44).

Detailed materials and methods are available in SI Appendix. This includes the following: in vitro differentiation of mature MKs, preparation and characterization of M-Ad5, in vitro viral infection and cell viability, quantitative PCR, western blot, in vivo biocompatibility and biodistribution, in vivo antitumor assay, cytokine detection, immunofluorescence staining, flow cytometry analysis of immune responses, tail bleeding assay, and statistics analysis.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data are included in the article and/or SI Appendix.