Platelet-engineered CAR-T cells as adjuvant therapy after cancer surgery

Significance

Adjuvant therapies can inhibit postsurgical tumor recurrence, while the applications of current therapeutic modalities are restricted by side effects after systemic infusion, repetitive dosing requirements, and limited efficacy. To address these limitations, we developed platelet-conjugated chimeric antigen receptor (CAR)-T cells as a safe and effective adjuvant therapy. Due to the wound-homing nature of platelets, the systemically administered CAR-T cells can be navigated to the postsurgical wounds to clear the residual tumor cells. Additionally, we showed that platelet activation could strengthen the persistence of the CAR-T cells to inhibit tumor recurrence and metastases. This approach can enhance the biosafety and efficacy of CAR-T cells as adjuvant therapy after cancer surgery.

Abstract

Surgery remains the mainstay treatment for many kinds of solid tumors, while tumor recurrence frequently occurs. Adjuvant therapy can reduce the risk of recurrence and improve the prognosis of surgery. Chimeric antigen receptor (CAR)-T cell therapy can be leveraged as an alternative adjuvant therapy to clear residual cancer cells and prevent tumor recurrence. However, systemic administration of CAR-T often results in insufficient tumor infiltration and side effects to normal organs. Given that platelets can preferentially accumulate at postsurgical wounds, we proposed that conjugating platelets to CAR-T cells may enhance the accumulation of the CAR-T cells within the surgical bed after the resection of solid tumors. In this study, we conjugated platelets to B7-H3.CAR-T cells via click chemistry. In postsurgical human pancreatic cancer mouse models, platelet-CAR-T cells showed enhanced tumor infiltration and elevated antitumor cytokine levels, resulting in superior suppression effects on tumor recurrence, compared with CAR-T cells. Additionally, platelet-CAR-T cells showed enhanced efficacy in inhibiting metastasis and prolonging the survival time of the mice in postsurgical triple-negative breast cancer (TNBC) models. Mechanistic studies revealed that platelet activation could improve the CAR-T cell activity and persistence, as evidenced by an upregulation of genes associated with T cell infiltration and a downregulation of genes related to T cell exhaustion. Finally, we further validated the biosafety profile and efficacy of platelet-CAR-T in a postsurgical patient-derived xenograft TNBC-bearing humanized mouse model. The results suggested that the CAR-T cell strengthened by platelet engineering is a promising adjuvant therapy against postsurgical tumor recurrence.

Despite advancements in surgical techniques, the high rate of postsurgical recurrence remains a significant challenge that causes poor prognosis and limited survival for patients with certain solid tumors, such as pancreatic cancer or triple-negative breast cancer (TNBC) (1). The residual tumor cells at the resection margin and micrometastases that failed to be removed can eventually develop into new tumors, resulting in local recurrences and metastases spreading (2). Although adjuvant therapy has proven effective in reducing the risk of postsurgical recurrence of multiple types of cancer, current adjuvant treatment options, such as chemotherapy, radiation, endocrine therapy, and immune checkpoint inhibition are often limited in their application due to inadequate specificity and the lack of personalization, which may cause incomplete tumor elimination and potential side effects (3–5). Additionally, most of these adjuvant treatments require repeated dosing, which can result in cumulative toxicity and secondary resistance to the therapy (6). There is an unmet clinical need for an alternative postsurgical adjuvant therapy to eradicate tumor cells after incomplete excision.

Having demonstrated remarkable success in treating hematologic malignancies, chimeric antigen receptor (CAR)-T cell therapy, emerging as a paradigm-shifting therapeutic modality, has huge potential as a new option for adjuvant therapy with several advantages (7). First, since the CAR-T cells are living therapeutics that retain the ability to proliferate, a single infusion of these cells provides continuous surveillance on tumor recurrence, potentially improving patient compliance and reducing cumulative toxicity. In addition, CAR-T cells can be designed and manufactured in a personalized manner to recognize specific tumor antigens, which enables antigen-specific trafficking and immune response, thus minimizing damage to healthy tissues compared to conventional adjuvant therapies (8). Moreover, CAR-T cell therapy fosters the generation of immune memory, which can offer durable protection against tumor recurrence and metastasis (9). However, the investigation of this approach as an adjuvant therapy against postsurgical tumor recurrence is limited due to numerous challenges. In solid tumors, the complex tumor microenvironment with dense extracellular matrix and immunosuppressive factors presents significant barriers to CAR-T cell therapy (10–14). Limited infiltration of CAR-T cells into the tumor area leads to suboptimal efficacy in eliminating residual tumor cells after surgery (15). In addition, CAR-T-related toxicity, such as cytokine-release syndrome (CRS), neurotoxicity, and “on-target, off-tumor” toxicities, is another concern that hinders the clinical application of CAR-T cell therapy (16, 17). The overactivation of CAR-T cells and lack of differentiation in tumor and normal tissues cause massive release of cytokines and damage to normal organs, which can consequently lead to life-threatening toxicity events. Moreover, higher doses of CAR-T have been associated with severe toxicity events (18). Addressing these limitations is crucial for improving efficacy and minimizing toxicity of CAR-T cells as a new type of adjuvant therapy.

To test the hypothesis of implementing CAR-T cells as an adjuvant therapy, in this study, we selected B7-H3.CAR-T cells and characterized their efficacy in preventing postsurgical cancer recurrence. B7-H3 (CD276) is a type I transmembrane protein that has been found overexpressed in various types of cancers, such as lung adenocarcinoma, glioma, pancreatic cancer, and breast cancer, serving as a promising target of CAR-T therapy (19, 20). However, B7-H3 is not exclusively expressed in tumors and is also detected in certain normal tissues, raising concerns of potential off-tumor on-target toxicity. Platelets have emerged as promising drug delivery carriers that target postsurgical wounds due to their natural role in wound healing and their ability to home to the injury sites (21–23). Notably, in the case of cancer surgery, platelets could serve as a navigation signal to precisely transport therapeutic cargo to postoperative cancer sites for enhanced treatment outcomes. Collectively, to maintain the potency of CAR-T cells as adjuvant therapy and overcome the potential nonspecific biodistribution-associated toxicity, in this study, we present a platelet-based engineering approach to redirect systemically administered CAR-T cells to the postsurgical tumor site as alternative adjuvant therapy (Fig. 1A). The platelets are conjugated to the B7-H3.CAR-T cell surface via a facile click chemical approach to result in platelet-CAR-T cell conjugates, which could be able to navigate CAR-T cells to postsurgical cancer sites to clear the residual tumor cells after cancer surgery. Besides this important function of platelets in redirecting CAR-T cells, we also demonstrated that contents released from the activated platelets, which are triggered by the inflammatory environment in the postsurgical cancer site, could potentially strengthen the antitumor activity and persistence of the CAR-T cells (24, 25). In human pancreatic cancer and TNBC-bearing NSG mouse models, we validated that platelet-CAR-T cell conjugates efficiently infiltrate and expand in the residual tumor tissues, effectively inhibiting the recurrence of primary tumors and the spread of metastases following incomplete surgery. Notably, due to the significant contribution of platelets in enhancing the selective accumulation of CAR-T cells at the cancer site and improving the persistence of CAR-T cells through platelet activation, we were able to achieve similar efficacy in preventing postsurgical tumor recurrence on a TNBC patient-derived xenograft (PDX)-bearing humanized mouse model with a half-dose of platelet-CAR-T cell conjugates compared to the full dose of CAR-T cells, while with less concern of potential side effects. Collectively, this study offers an adjuvant therapy by using emerging CAR-T cells as auxiliary therapeutics after cancer surgery and further enhances the treatment efficacy and biosafety profiles through platelet engineering.

Fig. 1.

Design and characterization of platelet-CAR-T as postsurgical adjuvant therapy. (A) Schematic illustration of the conjugating platelets to CAR-T cells to enhance their accumulation in the postsurgical wounds to suppress tumor recurrence. (B) Representative images of B7-H3 expressions in different normal tissues, breast, and pancreatic cancers. (Scale bar, 50 μm.) (C) Cell viability of untreated and GalNAz-treated T cells. Data are shown as mean ± SD (n = 3) and analyzed with an unpaired t-test. ns, not significant. (D) Cell viability of untreated and DBCO-NHS-treated platelets. Data are shown as mean ± SD (n = 3) and analyzed with an unpaired t-test. ns, not significant. (E) Confocal images of platelet-CAR-T prepared at different reaction ratios. CAR-T cells were labeled with CFSE (green), and platelets were labeled with Rhodamine-B (red). (Scale bar, 5 μm.) (F) SEM characterization of platelet-CAR-T. (Scale bar, 1 μm.) (G) Quantitative analysis of conjugated platelets on CAR-T cells. The number of platelets conjugated to 100 CAR-T cells was counted under a confocal microscope (n = 3). Data are presented as means ± SD.

Results

Engineering B7-H3.CAR-T Cells with Platelets.

Given the high expression of B7-H3 in various cancer types and its relatively limited expression in normal tissues, B7-H3-targeted immunotherapies are under investigation in many clinical trials (26). To evaluate the potential and safety profiles of B7-H3.CAR-T cells as an alternative adjuvant therapy after cancer surgery, we investigated the expression landscape of B7-H3 in tumors and normal human tissues by performing immunohistochemical (IHC) staining on a tissue microarray (TMA) consisting of samples derived from various tissues. As shown in Fig. 1B, pancreatic cancers and breast cancers showed overall higher expression of B7-H3, while most major organs, such as the heart, lung, and kidney, express no or low levels of the protein, supporting the concept of applying emerging B7-H3.CAR-T cells as a potential adjuvant therapy to prevent postsurgical tumor recurrence. However, several normal tissues exhibited different expression levels of B7-H3, with moderate to high expression in tissues such as the liver, spleen, intestines, and adrenal gland, which generate concerns of “on-target, off-tumor” toxicity against these tissues in postsurgical cancer patients, highlighting the need to improve the tumor specificity of B7-H3.CAR-T cells (Fig. 1B and SI Appendix, Figs. S1 and S2).

To generate B7-H3.CAR-T cells, we transduced T cells with the B7-H3.CAR encoding the single-chain variable fragment (scFv) derived from the B7-H3 376.96 mAb, human CD8α hinge and transmembrane domain, CD28 costimulatory domains, and CD3ζ intracellular signaling domain. The transduction efficiency was around 84.4 ± 9.4% (SI Appendix, Fig. S3). Given the fact that platelets can home to the injured area and then adhere to wounds via various glycoproteins (27), we hypothesized that a platelet-based engineering strategy can redirect the B7-H3.CAR-T cells to residual tumor tissues following surgery, minimizing the nonspecific distribution in normal organs. To demonstrate this hypothesis, we utilized click chemistry between azide groups and dibenzylcyclooctyne (DBCO) groups to tether platelets to the B7-H3.CAR-T cells (28). To functionalize B7-H3.CAR-T cells with azide groups, the cells were cultured in a medium supplemented with an azide sugar, N-azidoacetylgalactosamine tetraacylated (GalNAz), which can label the mucin-type O-linked glycoproteins on the cell surface through glycometabolic engineering. After 48 h of metabolic labeling, we confirmed the presence of azide groups on the surface of B7-H3.CAR-T cells by staining them with DBCO-Cy5.5 (SI Appendix, Fig. S4A). We also examined the viability of CAR-T cells after this glycometabolic engineering, which did not show obvious toxicity to the B7-H3.CAR-T cells (Fig. 1C). Then, we further labeled the surface of platelets with DBCO groups using DBCO-PEG₄-NHS, which can react with the amine groups of the platelets. The successful modification was validated using flow cytometry (SI Appendix, Fig. S4B). Besides, the introduction of the DBCO group did not induce toxicity to platelets, as evidenced by no significant viability change after functionalization with DBCO-NHS (Fig. 1D). In addition, DBCO-modified platelets retained their collagen-binding ability and thrombin-induced activation, showing no significant differences compared to unmodified platelets (SI Appendix, Figs. S5 and S6).

To conjugate platelets to B7-H3.CAR-T cells, azide-modified B7-H3.CAR-T cells and DBCO-modified platelets were coincubated for 1 h at room temperature. Afterward, mPEG₄-azide was added to quench the excessive DBCO groups on the platelets to prevent potential cell aggregation. We further tested different reaction ratios (platelet: CAR-T from 8:1 to 1:1) to optimize the platelet-CAR-T conjugations. As shown in Fig. 1 E and F, the successful conjugation of the platelets to the surface of CAR-T cells was observed for all reaction ratios under the confocal microscope and scanning electron microscopy (SEM). Colocalization of platelets and CAR-T cells was also verified using flow cytometry (SI Appendix, Fig. S7). Besides, the number of platelets conjugated to each CAR-T cell increased along with the increase in the initial reaction ratios of platelets and CAR-T cells (Fig. 1G and SI Appendix, Fig. S8). Specifically, we observed an average of 1, 2, and 3 platelets bound to the CAR-T cells with the reaction ratio of platelets to CAR-T cells at 2:1, 4:1, and 8:1, respectively.

In Vitro Tumor Cell Killing Effects of Platelet-CAR-T Cell Conjugates.

To further demonstrate the impact of platelet conjugation on the CAR-T cell activity, we chose human pancreatic cancer cell Panc-1 and breast cancer cell MDA-MB-231 as the target cell lines, which showed high expression of B7-H3 (SI Appendix, Fig. S9), to test the tumor cell killing effects of B7-H3.CAR-T cells after conjugation with platelets. Nontransduced T (NT), CAR-T, or platelet-CAR-T cells were cocultured with Panc-1 or MDA-MB-231 cells at an effector: target (E: T) ratio of 1:3 (Fig. 2A). After an incubation of 72 h, we collected all the cells for flow cytometry analysis. NT cells did not have any impact on the proliferation of either tumor cell. In contrast, both CAR-T and platelet-CAR-T cells killed over 90% of the tumor cells with comparable efficiency, suggesting the conjugation of platelets did not affect the tumor cell killing activity of the CAR-T cells (Fig. 2B and SI Appendix, Fig. S10). Besides, the levels of IFN-γ and IL-2, critical biomarkers indicating CAR-T cell activation, in the supernatant from the coculture system were measured (29). As shown in Fig. 2 C and D, CAR-T and platelet-CAR-T cell groups released comparable and large amounts of IFN-γ and IL-2, suggesting the potent CAR-T cell activity. In addition, T cell proliferation was assessed using the carboxyfluorescein succinimidyl ester (CFSE) dilution study. As shown in Fig. 2E, both CAR-T and platelet-CAR-T cells showed similar proliferation profiles when cocultured with target tumor cells. Notably, the conjugation ratio of platelets to CAR-T cells, ranging from 2:1 to 8:1, did not show obvious impacts on the CAR-T cell activity and proliferation capability, highlighting the feasibility of applying platelet-CAR-T cells for anticancer treatment.

Fig. 2.

Tumor-killing activity and proliferation of platelet-CAR-T. (A) Schematic illustration of tumor cell killing and T cell proliferation assay after coculturing tumor cells and B7-H3.CAR-T cells. (B) Killing activity of CAR-T cells or platelet-CAR-T against MDA-MB-231 and Panc-1 cells examined by flow cytometry at 72 h after coculture. E: T ratio = 1:3 (n = 4). (C and D) ELISA analysis of levels of IFN-γ (C) and IL-2 (D) released by NT, CAR-T, and platelet-CAR-T in the supernatant after 24 h of coculture with the tumor cells (n = 3). Data are mean ± SD and analyzed with one-way ANOVA followed by Dunnett’s multiple comparison test. (E) Proliferation of platelet-CAR-T after coculturing with the tumor cells detected by flow cytometry. CAR-T cells were prelabeled with CFSE (n = 3). (F) Bioluminescence imaging of mice at different time points after postsurgical injection of CAR-T and platelet-CAR-T prepared with different ratios at the dose of 5 × 10⁶ GFP-FFLuc-B7-H3.CAR-T cells. (n = 3 per group). The white circle indicates the tumor site. (G) Region-of-interest analysis of bioluminescence in (F). Data are mean ± SEM and analyzed with two-way ANOVA followed by Tukey’s multiple comparison test. (H) Immunofluorescence staining of T cell infiltration in residual tumors collected from mice treated with CAR-T or platelet-CAR-T at 96 h. n.s., not significant; *P < 0.05, ***P < 0.001.

To further optimize the platelet-CAR-T cell conjugates, we established a postsurgical subcutaneous Panc-1 model on NSG mice to evaluate the accumulation efficiency of platelet-CAR-T cells with the hypothesis that more platelet conjugations on the CAR-T cell surface could improve the tumor-specific delivery, given the capability of platelets to home to postsurgical tumor sites. After the tumor was surgically removed, CAR-T or platelet-CAR-T cells were administered via i.v. injection. Here, B7-H3.CAR-T cells expressing luciferase were used to study the biodistribution through in vivo imaging. As shown in Fig. 2 F and G, bioluminescence signals from all luciferase-expressing B7-H3.CAR-T cell groups gradually accumulated at the postsurgical tumor area over time, attributed to the targeting effects of B7-H3.CAR-T cells toward B7-H3-expressing Panc-1 tumors. Notably, the bioluminescence signal was significantly stronger in all platelet-CAR-T groups than that in the CAR-T cell group at the tumor site, highlighting the substantial contribution of platelet conjugation in redirecting more CAR-T cells to the postsurgical cancer site. In addition, the platelet-CAR-T cells (8:1) showed more accumulation at the tumor site compared to platelet-CAR-T cells at the ratios of 4:1 and 2:1, even though there is no significant difference among these groups. Quantitatively, the bioluminescence signals of the postsurgical tumor sites in the platelet-CAR-T cell group (8:1) were 1.81-fold higher than those in the CAR-T group. Finally, immunofluorescence staining of residual tumor tissues also confirmed that more human CD3⁺ T cells infiltrated the postsurgical tumors in the platelet-CAR-T cell group (8:1) than the CAR-T group (Fig. 2H). Combining the unimpacted CAR-T cell activity and more accumulation at the postsurgical tumor site, we selected platelet-CAR-T cells (8:1) at the optimal conjugation ratio for the following studies.

Platelet-CAR-T Cells Effectively Infiltrate Residual Tumors and Prevent Pancreatic Cancer Recurrence.

Before testing the efficacy of platelet-CAR-T cells in preventing postsurgical cancer recurrence, we further validated the B7-H3 expression level in human pancreatic cancer. The bioinformatic analysis of the expression of B7-H3 in pancreatic adenocarcinoma with the data reported from the Gene Expression Profiling Interactive Analysis (GEPIA) public database showed a significant upregulation of B7-H3 expression in pancreatic adenocarcinoma compared to normal pancreatic tissue (Fig. 3A). IHC staining of human pancreatic tumor TMA further substantiated a high expression of B7-H3 in pancreatic cancer tissues, with an average OD value 4.14-fold higher than that of normal pancreas as analyzed by ImageJ (Fig. 3 B and C).

Fig. 3.

Infiltration and efficacy of platelet-CAR-T in the postsurgical pancreatic cancer model. (A) Clinical analysis of B7-H3 expressions in normal pancreas and pancreatic cancers using GEPIA. (B and C) Representative immunohistochemistry images (B) and quantification (C) of B7-H3 expressions in normal pancreas and pancreatic cancers. (Scale bar, 50 μm.) Data are analyzed with an unpaired t-test. *P < 0.05. (D) Schematic illustration of the establishment and treatment in postsurgical recurrent pancreatic cancer model to evaluate efficacy. The Panc-1-luc tumor-bearing mice received 5 × 10⁶ CAR-T or platelet-CAR-T via i.v. injection after surgery. (E) Representative tumor bioluminescence images of Panc-1-luc tumor recurrence after treated with saline, NT, CAR-T, platelet-CAR-T, or half dose of platelet-CAR-T. (F) Region-of-interest analysis of bioluminescence intensity in IVIS images (n = 7 mice per group). Data are mean ± SEM and analyzed with two-way ANOVA followed by Dunnett’s multiple comparison test. (G) Survival curve of the mice in the postsurgical recurrent Panc-1-luc model after different treatments (n = 7). Data are analyzed with the log-rank (Mantel-Cox) test. (H) Study design of RNA sequencing of CAR-T cells after coculture with activated platelets. (I) Volcano plot of the RNA contents in CAR-T cells cocultured with activated platelets compared with untreated CAR-T cells. (J) Heatmap analysis of the related gene expression differences (n = 4). *P < 0.05, ***P < 0.001.

Based on the validation of B7-H3 as a promising target for pancreatic cancer treatment, we established a postsurgical subcutaneous luciferase-expressing Panc-1 pancreatic cancer mouse model and treated the mice with saline, NT, CAR-T, or platelet-CAR-T cells (T cell dose: 5 × 10⁶) via i.v. injection after the surgical removal of the majority of the tumor mass (Fig. 3D). Given that platelet-CAR-T cells showed a much higher accumulation (~1.8 fold) at the tumor site as redirected by platelet conjugation compared to CAR-T cells, we hypothesize that a lower dose of platelet-CAR-T cells might generate a similar antitumor effect as CAR-T cells, while with fewer safety concerns. Thus, a half-dose group (½ platelet-CAR-T cells) was included in the experiment, where the mice were administered platelet-CAR-T cells with a dose of 2.5 × 10⁶ T cells. The postsurgical tumor recurrence was monitored by the bioluminescence of luciferase-expressing Panc-1 cells using IVIS (Fig. 3 E and F). Compared with the saline group which displayed quick tumor relapse, NT cell treatment failed to delay the growth of tumors either. In contrast, CAR-T, platelet-CAR-T, and half-dose platelet-CAR-T cells effectively suppressed the recurrence of the Panc-1 tumor. Notably, platelet-CAR-T cells showed more potent efficacy against tumor recurrence than CAR-T cells, as demonstrated by much weaker bioluminescence signals at the postsurgical tumor sites. More importantly, mice treated with platelet-CAR-T cells exhibited significantly prolonged survival, highlighted by four out of seven mice with tumor-free status on Day 120, compared to all other treatment groups (Fig. 3G). Besides, half-dose platelet-CAR-T cells achieved a comparable efficacy with the full-dose CAR-T cell treatment, as evidenced by a similar tumor regrowth curve and mouse survival (Fig. 3 F and G). In addition, all the treatment groups did not induce obvious toxicity in the mice as assessed by the body weight during the treatment course (SI Appendix, Fig. S11).

Having observed the superior antitumor recurrence efficacy of platelet-CAR-T cells, we further investigated additional underlying mechanisms beyond enhanced tumor accumulation attributed to the platelet homing capability toward postsurgical tumor sites. Platelets have been widely reported to be activated under various physiological conditions, including postsurgical tumor microenvironment, to release multiple chemokines and cytokines that can impact the immune cell functionality (30–32). Therefore, we investigated the impact of platelet activation on the function of CAR-T cells using RNA sequencing technology (Fig. 3H). As depicted in Fig. 3I, platelet activation clearly affected the expression of a variety of genes in CAR-T cells, with 259 genes downregulated and 102 genes upregulated. Compared with untreated CAR-T cells, CAR-T cells impacted by platelet activation showed a down-expression of exhaustion-related genes (such as LAG3 and CD38) and a low basal activation level (CD69) (Fig. 3J) (17, 33, 34). In addition, we observed an upregulation of genes that are reported to be associated with immune cell infiltration, such as FUT7, ITGB3, and AQP3 (35–37).

Platelet Conjugation Boosted Proliferation and Immune Function of CAR-T Cells.

To investigate the in vivo expansion of T cells at the tumor site, the remaining pancreatic tumors after the surgery were harvested on Days 7 and 14 after treatment for quantification of T cells via flow cytometry. On Day 7, the number of CD3⁺CD8⁺ human T cells in the residual tumor tissues in CAR-T, platelet-CAR-T, and half-dose platelet-CAR-T cell groups was ~6.6-, 11.7-, and 5.9-fold higher than the NT group, respectively (Fig. 4 A and B). On Day 14, the number of CD3⁺CD8⁺ T cells in CAR-T, platelet-CAR-T, and half-dose platelet-CAR-T cell groups was ~6.4-, 14.2-, and 7.8-fold higher than the NT group, respectively (Fig. 4 A and B). Besides, platelet-CAR-T cells displayed a substantial proliferation on Day 14 compared to that on Day 7, with a 1.4-fold increase in the density of CD3⁺CD8⁺ T cells; while the T cell density in both CAR-T and half-dose platelet-CAR-T cells did not change much. Notably, the highest density of CD3⁺CD8⁺ T cells was found in the platelet-CAR-T cell group, which was ~1.8-fold higher than the CAR-T group on Day 7 and ~2.2-fold on Day 14, respectively. In addition, significantly elevated levels of cytokines, including IFN-γ and TNF-α, were detected in the residual tumor in the platelet-CAR-T cell group than the other groups (Fig. 4 C and D). On Day 14, the platelet-CAR-T cell group showed 1.9-fold and 1.4-fold higher levels of IFN-γ and TNF-α than the CAR-T cell group, respectively. The mouse spleens were collected on Day 14 after different treatments for enzyme-linked immunosorbent spot (ELISpot) analysis of IFN-γ production. As shown in Fig. 4 E and F, splenocytes collected from the platelet-CAR-T cell group showed the highest IFN-γ production ability after restimulation with target tumor cells. Collectively, these results demonstrated that the platelet-CAR-T cell conjugates could facilitate the proliferation and activity of CAR-T cells and generate robust antitumor effects to prevent postsurgical cancer recurrence. To further evaluate the long-term memory T cell response, we euthanized the surviving mice from the CAR-T, platelet-CAR-T, and half-dose platelet-CAR-T cell groups on Day 100 after the treatment and examined the memory T cell subsets in the spleen (Fig. 4G). A higher percentage of central memory (CM) and effector memory (EM) T cells was found in the platelet-CAR-T cell treatment group compared to those in the CAR-T cell group, which suggested a long-term protection effect against tumor recurrence.

Fig. 4.

In vivo proliferation and immune response of platelet-CAR-T in the postsurgical pancreatic cancer model. (A) Representative plots showing CD3⁺CD8⁺ T cell numbers detected by flow cytometry. (B) Data analysis of the number of CD3⁺CD8⁺ cells normalized by tumor weight. n = 4 mice per group. Data are mean ± SD and analyzed with one-way ANOVA followed by Dunnett’s multiple comparison test. (C and D) ELISA analysis of the concentrations of IFN-γ (C) and TNF-α (D) in the tumor supernatant after different treatments (n = 4). Data are mean ± SD and analyzed with one-way ANOVA followed by Dunnett’s multiple comparison test. (E) ELISpot analysis of IFN-γ production by splenocytes restimulated with Panc-1 cells. (F) Quantitative analysis of spots in (E) (n = 3). Data are mean ± SD and analyzed with one-way ANOVA followed by Dunnett’s multiple comparison test. (G) Flow cytometry analysis of memory CD3⁺ T cells in the spleens of the mice (n = 3). Naïve T cells (CCR7⁺/CD45RO⁻); CM, central memory T cells (CCR7⁺/CD45RO⁺); EFF, terminally differentiated effector T cells (CCR7⁻/CD45RO⁻); EM, effector memory cells (CCR7⁻/CD45RO⁺). Data are mean ± SD and analyzed with one-way ANOVA followed by Dunnett’s multiple comparison test. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001.

Platelet-CAR-T Cells Suppress Postsurgical TNBC Recurrence and Metastasis.

Having tested the treatment outcomes of platelet-CAR-T cells in postsurgical pancreatic cancer models, we then attempted to extend their application to other B7-H3-overexpressing tumors, such as TNBC. We first analyzed B7-H3 expression data from the public database, which showed a significantly higher expression on human breast tumors compared to normal breast tissues (Fig. 5A). We further performed IHC staining in TNBC TMAs to validate the high B7-H3 expression levels, which displayed similar results on TNBC patient samples (Fig. 5 B and C). We then tested the antirecurrence treatment efficacy of platelet-CAR-T cells in an orthotopic MDA-MB-231 TNBC-bearing NSG mouse model. Rapid recurrence and metastasis were observed in mice treated with saline and NT cells. In contrast, CAR-T and half-dose platelet-CAR-T cells could delay the relapse and metastasis of postsurgical TNBC tumors, as evidenced by the suppressed tumor growth curve compared to saline and NT groups (Fig. 5 D and E). Notably, the TNBC tumor growth was markedly reduced in the platelet-CAR-T cell treatment group, as evidenced by the lowest bioluminescence signals from the luciferase-expressing MDA-MB-231 cells compared to all other treatment groups (Fig. 5 D and E). Survival curve analysis revealed that platelet-CAR-T cell treatment significantly prolonged median survival time from 29 d (saline group), 39 d (CAR-T), 40 d (half-dose platelet-CAR-T cell) to 57 d, highlighting the superior treatment efficacy (Fig. 5F). Finally, as TNBC metastasis is the major reason for the dismal prognosis for clinical TNBC treatment, we assessed the efficacy of platelet-CAR-T cell conjugates in inhibiting tumor metastasis. As shown in Fig. 5G, the platelet-CAR-T cell treatment group showed the weakest luminescence signals in the lung tissues among all treatment groups, which reflects the efficacy in preventing TNBC lung metastasis (Fig. 5G). Moreover, the lungs collected from the mice treated with the platelet-CAR-T cell treatment group have an overall smooth surface with only scattered metastatic nodules compared to the obvious lung metastasis in all other treatment groups (Fig. 5H), which was also corroborated by the hematoxylin and eosin (H&E) images, where the platelet-CAR-T treatment showed minimal metastasis in the lungs and livers among all the groups (Fig. 5I). Besides, no obvious weight loss of mice was observed during the course of treatment (SI Appendix, Fig. S12).

Fig. 5.

Antitumor recurrence efficacy of platelet-CAR-T in postsurgical breast cancer model. (A) Clinical analysis of B7-H3 expressions in normal breast and breast cancers using GEPIA. (B and C) Representative immunohistochemistry images (B) and quantification (C) of B7-H3 expressions in normal breast and breast cancers. Images were obtained from tumors at different malignant stages (IA–IIIC). (Scale bar, 50 μm.) Data are analyzed with an unpaired t-test. (D) Representative tumor bioluminescence images of MDA-MB-231-luc tumor recurrence after treated with saline, NT, CAR-T, platelet-CAR-T, or half dose of platelet-CAR-T. (E) Region-of-interest analysis of bioluminescence intensity in IVIS images (n = 7 mice). Data are mean ± SEM and analyzed with two-way ANOVA followed by Dunnett’s multiple comparison test. (F) Survival curve of the mice in the postsurgical recurrent MDA-MB-231-luc model after different treatments (n = 7). Data are analyzed with the log-rank (Mantel-Cox) test. (G) Representative bioluminescence of MDA-MB-231-luc tumor metastasis in lungs and liver areas. (H) Representative photographs of lungs resected from mice after different treatment. (Scale bar, 20 mm.) (I) H&E staining of metastases in the lungs and livers after treatment. Arrows indicate metastatic nodules. n.s., not significant; *P < 0.05, **P < 0.01.

Platelet-CAR-T Cells Show a Great Biosafety Profile and Suppress Postsurgical Tumor Recurrence in a PDX TNBC-Bearing Humanized Mouse Model.

To assess the biosafety profile and antirecurrence efficacy against patient tumors in the context of human immune systems, we established a PDX TNBC-bearing humanized mouse model (Fig. 6A). Toxicities, particularly “on-target, off-tumor” toxicities, and CRS, are significant concerns associated with CAR-T cell therapy in the clinic. To evaluate the biosafety profile of platelet-CAR-T cells in a humanized mouse model, postsurgical TNBC-bearing mice were intravenously injected with saline, CAR-T, platelet-CAR-T, and half-dose platelet-CAR-T cells at a CAR-T full dose of 5 × 10⁶ cells/mouse; the major health indicators, including behavior scores, blood pressure, body weight, and body temperature, were closely monitored and recorded throughout the treatment course. As shown in Fig. 6B, the body temperature remained steady for all treatment groups; however, the blood pressure for CAR-T and platelet-CAR-T cell groups dropped in the first week after the treatment and then bounced back (Fig. 6C). Notably, the half-dose platelet-CAR-T cell group did not show any substantial changes in body temperature as well as blood pressure. We further applied a comprehensive scoring system to fully evaluate the health condition of the mice based on behaviors and activities (SI Appendix, Table S1). Specifically, after the treatment of CAR-T or platelet-CAR-T cells, the mice showed ruffled fur, reduced activity, and slow behavior, and slightly decreased body weight within the first week of the treatment, and returned to normal condition afterward (Fig. 6 D and E). Notably, lowering the dose of platelet-CAR-T cells to ½ could substantially mitigate this acute toxicity. Collectively, the platelet-CAR-T cell treatment strategy generally displayed a good biosafety profile for in vivo applications.

Fig. 6.

Safety and efficacy of platelet-CAR-T in postsurgical TNBC PDX model in humanized mice. (A) Schematic illustration of the establishment and treatment in postsurgical recurrent TNBC PDX model in humanized mice. (B and C) Mouse body temperature (B) and mean blood pressure (C) after different treatments (n = 3 mice). (D) Health scoring chart of mouse behaviors and activities. (E) Mouse body weight after different treatments. (F–H) Mouse plasma levels of IL-6 (F), IL-10 (G), and IL-1β (H) (n = 3). Data are shown as mean ± SD and analyzed with one-way ANOVA followed by Dunnett’s multiple comparison test. (I) Biochemical analysis of mouse plasma (n = 3). (J) Representative H&E staining of major organs in different groups (n = 3). (Scale bar, 100 µm.) (K) Image of recurrent tumors in different groups collected on Day 35. (Scale bar, 1 cm.) Black circles indicate tumor-free status. (L) Tumor weight of the collected tumor tissues after different treatments (n = 3). Data are shown as mean ± SEM. and analyzed with one-way ANOVA followed by Dunnett’s multiple comparison test. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001.

Since a dramatic elevation of interleukin-6 (IL-6) in the peripheral blood is a key hallmark of CRS (38), we also measured the IL-6 levels in the mouse plasma before and after various treatments. IL-6 levels increased in all treatment groups containing CAR-T cells on Day 9 but significantly dropped on Day 16 (Fig. 6F). Other cytokines, such as the anti-inflammatory IL-10 and the proinflammatory IL-1β, also showed a temporary elevation, followed by a return to levels similar to those observed before the surgery (Fig. 6 G and H). Blood samples were collected and subjected to biochemical and blood count analysis (Fig. 6I and SI Appendix, Fig. S13). All treatment groups, including CAR-T, platelet-CAR-T, and half-dose platelet-CAR-T cells, showed similar levels of AST, ALT, and BUN to those in the saline group, which suggested undamaged liver and kidney functions (Fig. 6I). No significant difference was found in complete blood count analysis among all treatment groups (SI Appendix, Fig. S13). In addition, H&E staining of the major organs also demonstrated the good safety profile of the platelet-CAR-T cell treatment group (Fig. 6J).

We further evaluated the treatment efficacy of platelet-CAR-T cells in PDX TNBC-bearing humanized mice. The tumors were harvested on Day 35 and weighed. Although we did not find a significant difference between CAR-T and platelet-CAR-T cell groups, probably due to the small sample size, tumors in the platelet-CAR-T cell treatment group showed the smallest sizes and lowest weight, suggesting a promising suppression effect against tumor recurrence on a humanized mouse model (Fig. 6 K and L). Collectively, these in vivo evaluations in humanized mice demonstrated that the platelet-CAR-T cell treatment strategy is a safe and effective adjuvant therapy for preventing postsurgical tumor recurrence.

Discussion

To develop an adjuvant therapy as an alternative to current auxiliary therapies, including chemo- and radiotherapy, to prevent postsurgical cancer recurrence, in this proof-of-concept study, we devised B7-H3.CAR-T as an adjuvant therapy after cancer surgery and further equipped CAR-T cells with platelets to facilitate their selective accumulation at the postsurgical cancer site to maximize the therapeutic outcomes and reduce the nonbiospecific distribution-associated side effects. The initial accumulation (in 96 h) and the subsequent proliferation (in 14 d) of systemically infused platelet-CAR-T cells at the residual tumors were assessed, which showed increased infiltration at the postsurgical cancer site than CAR-T cells and contributed to higher levels of antitumor cytokines. In human pancreatic cancer and breast cancer models, platelet-CAR-T cells showed superior therapeutic efficacy in inhibiting tumor recurrence and prolonging survival time compared with CAR-T cells. Mechanistic studies revealed an upregulation of genes associated with T cell infiltration and a downregulation of genes related to T cell exhaustion and activation, impacted by the platelet activation, suggesting an additional role of platelets in improving T cell function besides the primary contribution in redirecting CAR-T cells to selectively accumulate at the postsurgical cancer site. Finally, we validated that the platelet-CAR-T cell treatment strategy is a safe and effective adjuvant therapy for preventing postsurgical tumor recurrence in a postsurgical TNBC PDX-bearing humanized mouse model. Moreover, the enhanced therapeutic efficacy of platelet-CAR-T enables the use of a lower dose, which may further mitigate the risk of potential CAR-T-related toxicity. Collectively, these results highlight the potential of platelet-CAR-T cells as an adjuvant therapy after cancer surgery.

Current adjuvant treatment options, such as chemotherapy, radiation, and immune checkpoint inhibitors, are often limited by the requirement of repeated dosing, inadequate specificity, and a lack of personalization, which may result in incomplete tumor elimination and potential severe side effects. In contrast, CAR-T cells are living, proliferative therapeutics that can be customized to specifically recognize tumor antigens, offering a more targeted and long-lasting therapeutic approach. Furthermore, we demonstrated that conjugating platelets to CAR-T cells enhanced their accumulation at the postsurgical tumor site and improved their persistence through platelet activation. A single dose of platelet-CAR-T showed potent and durable efficacy against tumor recurrence without causing significant toxicity to other organs. These results support platelet-CAR-T cells as a promising adjuvant therapy with validated therapeutic efficacy and favorable biosafety profiles.

Platelets have been found to play multifaceted roles in tumor development and immunotherapy. In this study, we focused on B7-H3 as the target. Whether this platelet-based engineering approach has similar effects across different types of CAR-T cells requires further investigation. In addition, cases have been reported in which CAR-T cell therapy triggered thrombocytopenia in patients (39). The interaction between platelets and CAR-T cells has not been fully understood. Further investigation is needed to elucidate how platelets influence the activity, persistence, and toxicity of CAR-T cells, especially in immunocompetent models. Another potential challenge in translating this technology into clinical application is that cancer patients have a tendency to form thrombosis that could attract platelet accumulation, which may redirect platelet-CAR-T cells to other places rather than tumor sites (40). If that becomes a significant issue, pretreatment of cancer patients with low doses of thrombolysis drugs before cancer surgery could potentially solve this problem. However, the dose must be carefully calibrated to avoid bleeding issues during and after the cancer surgery. Finally, there is a key translational challenge of this platform in achieving scalable, sterile, and reproducible manufacturing suitable for clinical application. Recent advances in bioreactor technology offer promising solutions by enabling automated, standardized, and GMP-compliant production processes, accelerating their translation to clinical therapy.

Materials and Methods

Generation of CAR-T Cells.

GFP-FFLuc retroviral vector and B7-H3.CAR have been previously described (19, 41). Specifically, the B7-H3.CAR was generated using the scFv derived from the B7-H3 376.96 mAb cloned in-frame with the human CD8α hinge and transmembrane domain, CD28 costimulatory domains, and CD3ζ intracellular signaling domain. Retroviral supernatants used to transduce human T cells were prepared as previously described (41). For the generation of CAR-T cells, peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats of deidentified healthy donors (Gulf Coast Regional Blood Center, Houston, TX) by Lymphoprep (Accurate Chemical and Scientific Corporation) density-gradient centrifugation. PBMCs were activated with agonistic CD3 (Miltenyi Biotec) and CD28 (Becton Dickinson) antibodies in complete media, which contains RPMI 1640 hyclone (Cytiva) 45%, Click medium (Irvine Scientific) 45%, supplemented with 10% FBS and 1% L-glutamine, and 1% penicillin/streptomycin, transduced, and expanded in the presence of IL-7 (10 ng/mL, PeproTech) and IL-15 (5 ng/mL, PeproTech) as previously described (42). Cells were expanded for up to 12 to 14 d and used for in vitro and in vivo experiments. The expression of B7-H3.CAR and GFP in T cells were detected by flow cytometry. CAR expression was analyzed using the recombinant human B7-H3-Fc chimeric protein (R&D) and the secondary goat αHuman IgG (H+L, Jackson Lab) Ab (43).

Cell Conjugation.

Mouse platelets were isolated from fresh whole mouse blood by centrifugation at 100 g for 20 min twice. The supernatant was collected, and the platelets were pelleted by centrifugation at 1,000 g for 15 min. Platelets were resuspended in PBS supplemented with 1 µM PGE1 to avoid activation. To functionalize platelets with DBCO groups, platelets were reacted with 20 μM DBCO-PEG₄-NHS ester for 45 min at RT. The platelets were washed three times with PBS and collected by centrifugation at 1,000 g. To examine the presence of DBCO groups on the platelets, the modified platelets were incubated with 20 μM azide-Rhodamine B for 15 min and then subjected to flow cytometry analysis. Platelet viability after modification was assessed using calcein acetoxymethyl ester (calcein-AM) staining.

To functionalize B7-H3.CAR-T cells with azide groups, the cells were cultured in complete T cell media containing 45 μM GalNAz for 48 h. T cells were collected by centrifugation at 350 g for 5 min. To detect the presence of azide groups on the surface of CAR-T cells, the T cells were incubated with DBCO-Cy5.5 for 15 min. The resulting cells were washed with PBS three times and subjected to flow cytometry analysis. CAR-T cell viability after modification was evaluated using the Cell Counting Kit-8 (CCK-8) assay.

To conjugate platelets to CAR-T cells, DBCO-functionalized platelets were incubated with GalNAz-treated CAR-T cells at different reaction ratios (platelet: CAR-T from 8:1 to 1:1) for 1 h at room temperature. Then, mPEG₄-azide (50 μM) was added and incubated for 15 min to quench residual DBCO. The cell conjugates were washed three times with PBS and centrifuged at 350 g for 5 min. The cell conjugates were observed using SEM imaging and confocal microscopy. To observe the cell conjugates under a confocal microscope, CAR-T cells were prelabeled with CFSE following the manufacturer’s instructions. After the incubation of CAR-T cells and platelets, azide-Rhodamine B instead of mPEG₄-azide was added to the mixture to label the platelets for 15 min. The cell conjugates were washed three times with PBS, centrifuged at 350 g for 5 min, and observed under a confocal microscope. The quantification was based on counting platelets conjugated on 100 CAR-T cells under the confocal microscope. To verify the colocalization of CAR-T cells and platelets, CAR-T cells were labeled with Cy5.5-NHS, and platelets were labeled with Rhodamine B-NHS. The fluorescence of platelet-CAR-T was detected using flow cytometry.

In Vivo Efficacy.

The animal study protocol was approved by the Institutional Animal Care and Use Committee at the University of Wisconsin–Madison. The subcutaneous pancreatic cancer model was established by injecting Panc-1-luc cells into the right flank of NSG mice. Once the tumor size reached about 250 mm³, surgery was performed to remove 95% of the tumor mass under a microscope. The mice were randomly allocated to five groups and received different treatments, including saline, NT cells (5 × 10⁶ T cells), CAR-T (5 × 10⁶ T cells), platelet-CAR-T (5 × 10⁶ T cells), and half-dose of platelet-CAR-T (2.5 × 10⁶ T cells) via intravenous injection (n = 7). On predetermined days, the mice were intraperitoneally injected with 150 mg/kg D-luciferin, and the bioluminescence signals were monitored by IVIS. The bioluminescence signals were analyzed using Living Image Software v.4.3.1 (Perkin Elmer). The body weight and survival of mice were monitored after the treatment. Once the tumor volume was greater than 1,500 mm³ (length × width² × 0.5), mice were euthanized following the University of Wisconsin–Madison guidelines for Death as Experimental End-points and Euthanasia.

The orthotopic breast cancer model was established by injecting 1 × 10⁶ MDA-MB-231-luc cells into mouse mammary fat pads. The mice received the same treatment as mentioned above. The bioluminescence signals, body weight, and survival of mice were monitored. To investigate whether platelet-CAR-T could prevent tumor metastasis, the lungs and livers of four mice in different treatment groups were harvested on Day 23. The lungs were fixed in Bouin’s solution and photographed. Then, H&E staining was performed to observe the metastases in the lungs and livers.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

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