“Artificial platelet injection system”: a plug-and-play platelet-based lysosome-targeting chimera for targeted protein degradation

Abstract

Targeted protein degradation technology via lysosomal machinery has attracted extensive attention due to its capability of degrading extracellular and membrane-associated proteins. Conventional lysosome-based degraders are established by conjugating bifunctional chimaeras, which requires complex multi-step synthesis and time-consuming purification. Isolated activated platelets have been found to target and be engulfed by tumor cells, raising the possibility of being harnessed for intracellular payload injection. Here, inspired by injection systems of endosymbiotic bacteria and bacteriophages, we develop an “artificial platelet injection system”, a plug-and-play platelet-based lysosome-targeting chimera (PLT-TAC) via a facile microfluidic approach. PLT-TACs are meticulously manufactured by anchoring protein ligands and lysosome-sorting signal peptides onto activated platelet membrane, and can be easily optimized by tuning the peptide ratio on platelets to pursue the efficient protein degradation. Once PLT-TACs adhere to the tumor cytomembrane, they can be internalized into cells, and lysosome-sorting peptides initiate the “injection” of targeted proteins into lysosomes, thereby promoting the lysosomal degradation process. To test our concept, we use a peptide ligand targeting the immune checkpoint PD-L1 to engineer PLT-TACs. Efficient degradation of PD-L1 is confirmed both in vitro and in vivo. Furthermore, PLT-TACs can also degrade other proteins by simply installing alternative ligands. Our study proposes a new autologous cell-based protein degrader, with broad applicability in research and clinical practice.

Graphical abstract

Caption: “Artificial platelet injection system”: a plug-and-play platelet-based lysosome-targeting chimera (PLT-TAC), a whole cell-based protein degrader, constructed by microfluidic technology and achieve effective membranous protein degradation via lysosomal proteolysis.

Highlights

• •“Artificial platelet injection system”, the first whole cell-based degrader, inspired by injection systems of endosymbiotic bacteria and bacteriophages.
• •Use a plug-and-play method to rapidly screen the optimal platelet-based lysosome-targeting chimera (PLT-TACs) via a facile microfluidic approach.
• •Elucidate the mechanism that PLT-TACs efficiently degrade membranous PD-L1 protein through enhanced endo/autophagosome pathway and the biogenesis of lysosomes.

Targeted protein degradation (TPD) has emerged as a promising therapeutic modality in the past two decades, which effectively degrades pathogenesis-related or pathogenetic proteins by hijacking the cellular degradation system [1,2]. Proteolysis targeting chimeras (PROTACs) have blazed a trail in the realm of TPD. Most of TPDs, for example, PROTACs [3], molecular glues [4], and Trim-away [5], utilize the ubiquitin–proteasome system (UPS) to target and eliminate intracellular proteins. Theoretically, a PROTAC molecule comprises heterobifunctional small molecules: a ubiquitin E3 ligase ligand, a protein of interest (POI) ligand, and a flexible linker connecting these two ligands. Meanwhile, the PROTAC drags the POI spatially close to an E3 ligase, and forms a POI-PROTAC-E3 ternary complex, leading to ubiquitination of the POI and subsequent degradation via the UPS.

However, PROTACs are restricted to intracellular proteins because of the location of UPS within the cytoplasm. Lysosome-targeting chimeras (LYTACs) have provided a method for the membranous or extracellular protein degradation [[6], [7], [8]]. These approaches make use of lysosome-trafficking receptors (LTRs) to shuttle the POI into endocytosis and eventually lead to lysosomal degradation. The first LYTAC takes advantage of a cation-independent mannose-6-phosphate receptor (CI-M6PR) ligand, M6Pn glycol-polypeptide, to guide the POI into lysosomes for proteolysis. Additionally, asialoglycoprotein receptors [7], integrin α_vβ₃ [9], autophagosomal proteins [10] and other LTRs have been proved to be effective in navigating secreted or membranous proteins to lysosomes for degradation. However, these LYTACs composed of bifunctional chimaeras require complex chemical synthesis steps or an arduous protein recombination process, which limits their scalability and applicability. Simultaneously, the non-selectivity of LTRs may render off-target effects owing to their widespread expression in different types of cells.

Inspired by the contractile injection systems of endosymbiotic bacteria and bacteriophages, we propose a novel “biomimetic” platelet-based lysosome-targeting chimera (PLT-TAC) by engineering platelets with PD-L1-binding peptides and lysosome-sorting signal motifs to circumvent these limitations (Fig. 1). These contractile injection systems of endosymbiotic bacteria and bacteriophages are generally composed of three parts: recognition proteins, payloads, and spikes [[11], [12], [13]]. Once the recognition components target cell membrane, the spikes are driven through the membrane and payloads are simultaneously punched into cells along with the spikes [14,15] (Supplementary Fig. 1). In light of this structure, we fabricate an “artificial platelet injection system”, taking advantage of the targeted uptakeing of platelets by cancer cells [[16], [17], [18], [19]], using protein-binding peptides as payloads to bind targeted proteins, activated platelets as the recognition modules to adhere to cancer cell membrane, and lysosome-sorting signal motifs as the spikes to guide this system into lysosomes for protein degradation. According to the literature, lysosome-sorting signals can enhance the lysosomal trafficking of membrane components [20,21]. Once the signals are identified intracellularly, the cargoes can be transported into lysosomes after endocytosis or phagocytosis process. Some studies have successfully conjugated targeted protein ligands with specific lysosome sorting peptides and subsequently achieved efficient lysosomal degradation [22,23]. Thus, the co-installation of lysosome-sorting signals on platelets might facilitate lysosomal trafficking and enhance degradation efficiency.

Fig. 1.

│Schematic illustration of platelet-based lysosome-targeting chimeras (PLT-TACs) for PD-L1 degradation. PLT-TACs are fabricated by anchoring PD-L1 ligands (anti-PD-L1) and lysosome-sorting peptides (P1) onto activated platelet membrane through microfluidic technology. Microfluidic technology provides a facile fabrication process. The plug-and-play PLT-TACs exhibit the intrinsically favorable binding capacity with tumor cell membrane, and P1 promotes the “injection” of membranous PD-L1, leading to its enhanced trafficking into lysosomes, all of which ultimately induce PD-L1 degradation via lysosomal proteolysis.

Platelets are anucleated cellular fragments derived from megakaryocytes and readily available from blood. Transfused platelet therapy has been widely applied and demonstrated safety in clinical practice. For example, platelet-rich plasma (PRP) is used as a surgical adjuvant or regenerative medicine preparation in many medical fields, particularly in sports medicine and orthopaedic surgery [[24], [25], [26], [27], [28], [29]]. Meanwhile, activated platelets demonstrate not merely high affinity with malignant cells, but also enhanced engulfment by tumor cells through membrane fusion and phago/endocytosis [16,18], which makes activated platelets a potential platform for membrane protein degradation. It is reasonable to predict that the fusion of lysosome-sorting motifs and protein ligands with platelets could promote lysosome-targeting degradation of proteins of interest in tumor cells.

To test our hypothesis, we utilized microfluidic technology to form a plug-and-play PLT-TAC, the “artificial platelet injection system”, by facilely inserting a PD-L1-binding peptide and a CI-M6PR-related lysosome-sorting peptide onto platelet membrane. PLT-TACs show efficient internalization and enhanced lysosome biogenesis. Consequently, PLT-TACs effectively degrade PD-L1 in vitro and in vivo, and demonstrate safety after systemic administration. Meanwhile, PLT-TACs can also degrade other proteins by simply installing their respective ligands. PLT-TACs demonstrate broad applicability and scalability, providing a new strategy in the targeted protein degradation landscape.

1. Results

1.1. Design and construction of PLT-TACs

Activated platelets have been identified with higher affinity to malignant cells, also the enhanced phago/endocytosis by tumor cells, whose characteristics have been harnessed for targeted therapies or the fabrication of targeted drug delivery systems [[30], [31], [32]]. In this study, we hypothesized that using these intrinsic properties of platelets and installing a POI binder along with a lysosome-sorting peptide onto the platelet surface could trigger the internalization and the eventual lysosomal degradation of POIs.

To test our concept, we chose PD-L1, a transmembrane glycoprotein protecting cancer cells from T-cell mediated immune surveillance, as the POI [33], and installed a lysosome-sorting signal peptide, P1, known involved in the internalization/lysosomal sorting of CI-M6PR, for inducing lysosome biogenesis and PD-L1 degradation [34,35]. First, we synthesized a PD-L1 antagonist peptide (CdNdYdSdKdPdTdDdRdQdYdHdF, anti-PD-L1) and a CI-M6PR related lysosome-sorting peptide (CRRRRKSFHDDSDEDLLHI, P1) with a terminal thiol group in cysteine residue via standard Fmoc solid-phase peptide synthesis. Then, these two peptides were incubated with DSPE-PEG2K-Mal to obtain lipid-modified peptides, DSPE-PEG2K-anti-PD-L1 and DSPE-PEG2K-P1, via maleimide-thiol click chemistry. The structures of peptides and DSPE-PEG2K conjugated peptides were verified by ¹H-NMR and ESI-MS (Supplementary Figs. 2–5). For fluorescence labeling, CdNdYdSdKdPdTdDdRdQdYdHdFC-Cy5 and CRRRRKSFHDDSDEDLLHIC-Cy3 were synthesized and purified (Supplementary Figs. 6 and 7).

To engineer the platelet surface, microfluidic technology was introduced here because of its advantages of precise manipulation, low batch-to-batch variation, and rapid fabrication [36,37]. The herringbone structure was designed to improve turbulent flow for up-regulating mixing efficiency of the activated platelet phase and the lipid-peptide phase (Fig. 2A). Different ratios of platelets and DSPE-PEG2K-anti-PD-L1/DSPE-PEG2K-P1 (molar ratio = 1/1) were prepared for tuning the density of peptides on the platelet membrane. Meanwhile, as shown in transmission electron microscopy (TEM) results (Fig. 2B and C, and Supplementary Fig. 10), lipid-peptides did not noticeably change the morphology and the size of platelets (Supplementary Table 1). All PLT-TACs showed representative micro-antennas on the surface, and dense granules within the organelle zone. Scanning electron microscopy (SEM) demonstrated their three-dimensional morphology and the details of membrane surface (Fig. 2D and E). Both unmodified platelets and modified platelets showed canonical platelet morphology with rugose surface and open canalicular system (red rectangles regions), and no aggregates from lipid-peptides were found on cell surface. Fluorescence-labeled peptides were employed to validate the co-installation of P1 and anti-PD-L1. Both P1-Cy3 and anti-PD-L1-Cy5 were observed co-mounted on the platelet membrane (labeled with Dio) via confocal microscopy (Fig. 2F). Meanwhile, to calculate the drug loading efficiency, Cy3-and Cy5-labeled PLT-TACs were measured by fluorescence spectrophotometry. Formulations via microfluidic method reached >80% drug loading efficiency, higher than those constructed by dropwise mixing of DSPE-PEG2K-peptides with activated platelets (Supplementary Fig. 10). Dropwise mixing method showed more fluorescence intensity in the supernatant after 3000 rpm centrifugation and more green aggregates surrounding platelets visualized by Dio staining (Supplementary Fig. 11). Then, we calculated the number of anti-PD-L1 per platelet by measuring the number of platelets per mL via cell counting and determining the concentration of anti-PD-L1 through fluorescence spectrophotometry (Supplementary Table 2). After formulation process, the average number of platelets was 1.3 × 10⁸ per mL, and the number of anti-PD-L1 was equal to Avogadro's constant multiplied by the molar concentration. The number of anti-PD-L1 in platelets of PLT-TAC_l, PLT-TAC_m and PLT-TAC_h was 4.5 × 10⁷, 1.35 × 10⁸ and 3.8 × 10⁸ per platelet respectively. Flow cytometry (FCM) and 3D confocal imaging also verified the successful co-installation of anti-PD-L1 and P1 on activated platelets (Fig. 2G). To determine the presence and integrity of proteins on PLT-TACs, SDS-PAGE and FCM were performed. We tested the total protein composition from activated platelets and PLT-TAC_m. As shown in Fig. 2H, PLT-TAC_m exhibited similar protein bands to activated PLTs, indicating the integrity of protein in PLT-TACs. At the same time, it was observed that PLT-TACs exhibited the extracellular epitopes of CD41, CD47, and CD62P, indicating the retention of membrane proteins after formulation processes (Fig. 2I). Next, we detected the morphology and Young's moduli of PLTs and PLT-TAC_m by atomic force microscopy (AFM) using a spherical probe. PLT-TAC_m showed no significant difference in topography images compared to PLTs, whereas the Young's modulus of PLT-TAC_m slightly decreased from 2.18 kPa to 1.76 kPa as a result of lipid-peptide modification (Fig. 2J). These results above confirmed the structure integrity of platelets after the microfluidic processing, and enabled the successful modification of anti-PD-L1 and P1 in activated platelets. Furthermore, it was essential to verify the integrity of anti-PD-L1 and P1 in PLT-TACs prior to lysosomal trafficking. We incubated Cy3-and Cy5-co-labeled PLT-TAC_m with IFN-γ pre-treated 4T1 cells for 3h, and stained the lysosomes with lysotracker Green DND26 to observe the distribution of anti-PD-L1 and P1 in cytoplasm. As we expected, P1 and anti-PD-L1 tightly were co-located in green lysosome puncta (Fig. 2K). The fluorescence intensity profiles of anti-PD-L1 and P1 were overlapped closely with those lysosomes from confocal images (Fig. 2L). In conclusion, these results clarified that PLT-TACs had potential to become an effective lysosome-targeting protein degrader.

Fig. 2.

│Construction and characterization of PLT-TACs. A, Fast fabrication process of PLT-TACs via microfluidic device with a herringbone structure chip. B, TEM images and DLS size distribution of activated platelets (PLTs). C, TEM images and DLS size distribution of engineered activated platelets with medium density of anti-PD-L1 and P1 (PLT-TAC_m). D, SEM images of PLTs, E, SEM images of PLT-TAC_m, white squares indicate zoom-in regions, red squares indicate morphology features on the platelet surface. F, Confocal images of Cy3-, Cy5-and Dio-labeled PLT-TACs. G, FCM analysis and 3D confocal visualization for verifying the co-installation of anti-PD-L1 and P1 in PLT-TACs. H, SDS-PAGE analysis of protein expression profiles in activated platelets and PLT-TAC_m. I, FCM analysis of key platelet membrane proteins, CD41, CD47 and CD62P, expressed on PLT-TACs. J, AFM images of topography, height and Young's modulus mapping of PLTs and PLT-TAC_m. Scale bars, 5 μm (n = 5 technical replicates). K, CLSM visualization of the localization of anti-PD-L1 and P1 in 4T1 cells after internalization for 3h, anti-PD-L1 was labeled with Cy5, P1 was labeled with Cy3, nuclei were stained with Hoechst 33342, and lysosomes were stained with lysotracker Green DND26. Scale bar, 10 μm. L, Fluorescence intensity profiles along the line drawn in K (n = 3 biological replicates).

1.2. Internalization and PD-L1 degradation

To determine the factors underlying the degradation effect, we tested the trafficking route of PLT-TACs (Fig. 3A). The binding of POI ligand plays a critical role in the degradation of membrane proteins. The spatial distance or organization pattern between ligands can affect the conformational state or oligomerization of the receptors, thereby determining internalization efficiency [[38], [39], [40]]. For instance, the molecular size of small-molecule PD-L1 degraders affected PD-L1 dimerization, internalization, and degradation in the lysosomal vesicles to some extent [41]. Interferon-γ (IFN-γ) secreted by the microenvironment can upregulate PD-L1 on the surface of tumor cells [42]. PD-L1 expression was dramatically upregulated by pre-treatment with recombinant IFN-γ (50 ng/mL) (Supplementary Fig. 12). IFN-γ pre-treated cells were treated with free anti-PD-L1 and PLT-TACs at an equivalent amount of anti-PD-L1 at low temperature (4 °C) to inhibit cellular internalization. Flow cytometry analysis showed that PLT-TAC_l with low peptide density exhibited 7.6-fold and 1.5-fold binding intensity higher than free anti-PD-L1 and PLT-anti-PD-L1 (only carrying anti-PD-L1 without P1) respectively, which implied that both PLTs and P1 contributed to the enhanced binding capacity of PLT-TACs. Meanwhile, the binding fluorescence intensity decreased with increasing density of inserted peptides, suggesting that the high density of peptides had a negative effect on the binding between 4T1 cells and PLT-TACs (Fig. 3B). As for internalization study, all PLT-TACs showed higher internalization than PLT-anti-PD-L1 at different time points. Nevertheless, PLT-TAC_m with the middle density manifested the most efficient internalization, which was inconsistent with the surface binding result (Fig. 3C). Similarly, the highest number of fluorescence-labeled particles were observed in the cytoplasm of PLT-TAC_m treated 4T1 cells (Fig. 3D). The lack of correlation between cargo internalization and surface binding suggested that the underlying mechanism of PLT-TAC internalization was not only connected with the binding capacity but also might be related with the suitable deployment of POI ligands and lysosome-sorting signals.

Fig. 3.

│Surface binding, internalization, and PD-L1 degradation. A, Schematic illustration of the trafficking route of PLT-TACs. B, Fluorescence intensity analysis and histogram plots of PD-L1 binding to 4T1 cell surface after treatment with free anti-PD-L1, PLT-anti-PD-L1 or different PLT-TACs for 30min at 4 °C, measured by FCM (n = 3 biologically independent samples). C, Fluorescence intensity analysis and histogram plots of internalized PLT-anti-PD-L1 or different PLT-TACs in IFN-γ pre-treated 4T1 cells for 1, 2, or 4 h at 37 °C, measured by FCM (n = 3 biologically independent samples). D, Visualization of internalization of PLT-anti-PD-L1 or different PLT-TACs in IFN-γ pre-treated 4T1 cells after treatment for 2 h, observed by 40 × CLSM. Scale bar, 10 μm. E, Redistribution of PLT-TAC_m membrane after treatment for 2 h or 4 h, observed by oil 63 × CLSM. Platelets were stained with Dio. Scale bar, 5 and 20 μm. F, Visualization and fluorescence intensity profiles along the line of colocalization of PLT-anti-PD-L1 or different PLT-TACs with lysosomes in IFN-γ pre-treated 4T1 cells after treatment for 4 h, observed by oil 63 × CLSM. Lysosomes were stained with lysotracker Green DND26. Scale bar, 10 μm. G, Mean fluorescence intensity of levels of PD-L1 in IFN-γ pre-treated 4T1, measured by FCM after treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) for 36 h (n = 3 biologically independent samples). H, Western blot analysis of PD-L1 in IFN-γ pre-treated 4T1 cells after treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) for 36 h (n = 3 biologically independent samples). I, Western blot analysis of PD-L1 in IFN-γ pre-treated 4T1 after treated with PLT-TAC_m with different concentration (0, 0.025, 0.25, 1.25, 2.5, 5 μM) for 36 h (n = 3 biologically independent samples). J, Western blot analysis of PD-L1 in IFN-γ pre-treated 4T1 after treated with PLT-TAC_m (5 μM) with different time (0, 6, 12, 24, 36, 48 h) (n = 3 biologically independent samples). K, Immunofluorescence of surface PD-L1 in IFN-γ pre-treated 4T1 cells treated with PLT-TAC_m (5 μM) for 36 h. Scale bar, 20 μm. L, Visualization of PD-L1 degradation in IFN-γ pre-treated 4T1 after treated with PLT-TAC_m (5 μM) for 0 h or 24 h. Scale bar, 5 μm. M, Western blot analysis of CD24, CD71 and CD47 in IFN-γ pre-treated 4T1 after treated with PLT-TAC_m (5 μM) for 36 h (n = 3 biologically independent samples). Data are presented as mean ± SD. p values were determined by one-way ANOVA test. ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Meanwhile, the platelet membrane was stained with Dio to detect the redistribution of PLT-TAC_m after cellular uptake. As depicted in Fig. 3E, the majority of anti-PD-L1 and P1 were co-located in the cytoplasm at both 2 h and 4 h. Dio fluorescence was observed in both the cytoplasm and cell membrane at 2 h, and most Dio signal had been redistributed to cell membrane and nuclear membrane at 4 h. This redistribution of platelet membrane validated not only the internalization of platelet membrane along with anti-PD-L1 and P1, but also the fusion or recycle within plasm and nuclear membrane. These phenomena indicated that the classical cycling process of endo/lysosome system was involved in the trafficking route of PLT-TACs. Given both the surface binding and internalization results, it was necessary to explore the intracellular trafficking route of PLT-TACs. Following incubation with PLT-anti-PD-L1 or PLT-TACs, the highest anti-PD-L1 fluorescence intensity and largest overlap area with lysosomes were observed in PLT-TAC_m treated cells. Moreover, we also found that the majority of PLT-TAC_m were distributed throughout the cytoplasm region (Fig. 3F). In stark contrast, an obvious fraction of PLT-anti-PD-L1 aggregated in the peripheral cytoplasm and membranous regions, and the overlap area with lysosomes was much smaller than that of other PLT-TACs, all of which revealed PLT-TACs contributed to the stronger internalization and the shuttling of cargoes into lysosomes than PLT-anti-PD-L1 without P1 modification. Of note, the colocalization regions of PLT-TAC_h were obviously less than that of PLT-TAC_l and PLT-TAC_m. We considered that the shorter spatial distance between anti-PD-L1 and P1 compromised the internalization of PLT-TAC_h owing to the steric hindrance effect weakening the affinity between ligands/sorting signals and their receptors [40,43].

To investigate whether the PLT-TACs could degrade membranous PD-L1 and its relationship with internalization, we evaluated the degradation efficacy of PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m, and PLT-TAC_h. IFN-γ pre-treated 4T1 cells were incubated with these degraders for 36 h. We found that PLTs or free anti-PD-L1 barely degraded PD-L1 (Supplementary Fig. 13). Then, we compared the degradation levels of different PLT-TACs. PLT-TAC_m decreased fluorescence intensity by approximately 68% in FCM and reduced approximately 79% PD-L1 expression in Western blot, both of which were superior to other PLT-TACs and PLT-anti-PD-L1 treatment (Fig. 3G and H). Meanwhile, PLT-TAC_l showed more PD-L1 degradation than PLT-TAC_h and all PLT-TACs demonstrated higher degradation efficiency than PLT-anti-PD-L1, in accordance with the internalization results, suggesting the degradation efficiency was positively correlated with the internalization efficiency. As the literature reports, the high density of ligands would lead to steric hindrance, affecting the binding of proteins, whereas the low density of ligands would increase the spatial distance between the POI and lysosome-targeting receptors, causing the loss of degradation efficacy [40,43]. We inferred that PLT-TAC_m possessed the appropriate spatial distance between P1 and anti-PD-L1, which contributed to higher PD-L1 degradation compared with PLT-TAC_l and PLT-TAC_h. In addition, we examined whether the degradation was concentration- and time-dependent. With concentrations ranging from 0 to 5 μM, an increasing trend of degradation efficiency of PD-L1 was observed via FCM and Western blot, and maximum degradation was achieved at a concentration of 5 μM (Fig. 3I). Meanwhile, PD-L1 degradation efficiency was enhanced as time increased (Fig. 3J and Supplementary Fig. 15). For visualizing PD-L1 degradation, membranous PD-L1 was stained with the anti-PD-L1 antibody and imaged by confocal microscopy. PD-L1 was discernibly observed much less compared with saline, validating the degradation capability of PLT-TAC_m, consistent with results from FCM and Western blot (Fig. 3K). Moreover, to investigate the degradation mechanism, we incubated cells with 5 μM PLT-TAC_m in presence of a lysosome inhibitor (chloroquine, 100 μM) or a proteasome inhibitor (MG132, 5 μM). The results showed that PD-L1 degradation were significantly attenuated by chloroquine, rather than by MG132 (Supplementary Fig. 14). After 24 h of treatment, we observed the obvious depletion of surface PD-L1 in cells and most of PD-L1 proteins were colocalized in lysosomal puncta (Fig. 3L). These results indicated that PLT-TAC_m-induced PD-L1 degradation was dependent on lysosomal pathway. In addition, to explore the specificity of protein degradation, we measured the degradation of the other membranous proteins, CD24, CD71 and CD47, after PLT-TAC_m treatment (Fig. 3M). The results indicated that PLT-TAC_m did not exert a significant degradation effect on these non-targeted proteins. Meanwhile, PLT-TAC_m containing anti-PD-L1 degraded significantly more PD-L1 than PLT-P1 lacking anti-PD-L1, ruling out potentially non-specific degradation effects mediated by P1 itself (Supplementary Fig. 16). These results described above confirmed the degradation specificity of PLT-TAC_m.

1.3. Internalization mechanism analysis

Based on the binding and internalization studies, PLT-TAC_m exhibited superior internalization and colocalization with lysosomes. We set out to investigate the trafficking process of PLT-TAC_m. First, we used deep learning–based protein sequence design approach to generate the mutated peptide (M1) with a similar structure with P1 using ProteinMPNN (Fig. 4A and Supplementary Fig. 8) [44,45]. M1 lacked the lysosome-sorting sequence (DDSDEDLL) present in P1 [23,46] and was not supposed to stimulate lysosomal trafficking pathway. PLT-TAC_m (M1) was formed with M1 instead of P1, showing significantly less internalization than PLT-TAC_m (Fig. 4B). Concurrently, PLT-TAC_m (M1) demonstrated markedly lower PD-L1 degradation efficiency than PLT-TAC_m, which validated that P1 played an indispensable role in PD-L1 degradation (Fig. 4C). Next, we observed the entire internalization process of PLT-TAC_m by confocal microscopy. In the video (Fig. 4D), the Cy5-labeled PLT-TAC_m adhered to the cell surface and then immediately penetrated the cell membrane to enter the cytoplasm. Then, we examined the colocalization and overlap coefficients with early endosomes (marked by EEA1), lysosomes (marked by LAMP1) to verify whether the lysosome trafficking process was involved. As for 2 h of incubation, substantially more PLT-TAC_m overlapped with EEA1 than with LAMP1 (Fig. 4E), alongside many endosome puncta. As for 4 h of incubation, it was obvious that most of PLT-TAC_m were colocalized with LAMP1, whereas only minimal overlap with EEA1 remained, which confirmed the transformation of endosomes into lysosomes. To validate whether the internalization of PLT-TACs depend on the CI-M6PR, we incubated CI-M6PR ligand, mannose-6-phosphate (M6P, 5 mM) together with PLT-TAC_m, and analyzed the PD-L1 degradation. The degradation efficacy of PD-L1 reduced approximately from 81% to 20%, which demonstrated the coexistence of M6P could compete with PLT-TAC_m for the lysosomal trafficking pathway (Supplementary Fig. 17). These phenomena reflected the canonical process of endo/lysosome trafficking pathway, and proved that the lysosomal trafficking was closely related to the internalization of PLT-TAC_m. Next, we questioned whether the PLT-TAC_m could induce lysosome biogenesis. We found more LAMP1 puncta after PLT-TAC_m treatment compared with PLT-anti-PD-L1 treatment at both 2 h and 4 h, which also verified that P1 played a key role in lysosome biogenesis (Fig. 4F and G). Simultaneously, PLT-TAC_m promoted higher LAMP1 expression than PLT-anti-PD-L1, analyzed by Western blot (Fig. 4H). These results confirmed that PLT-TAC_m induced lysosome biogenesis. Of note, we observed increased autophagosome biogenesis and colocalization puncta of PLT-TAC_m and LC3-II than that after PLT-anti-PD-L1 treatment (Fig. 4I), and the ratio of LC3-II/LC3-I increased after the treatment with PLT-TAC_m (Fig. 4J), both of which demonstrated that autophagy was stimulated by PLT-TAC_m. Transcription factor EB (TFEB) is the master regulator of lysosome-related biogenesis, and the nuclear translocation of TFEB stimulates the expression of lysosome-related genes [47]. We found that more nuclear translocation of TFEB was observed after PLT-TAC_m treatment, which indicated the lysosome-related biogenesis was stimulated by PLT-TAC_m (Fig. 4K). These data above suggested that PLT-TAC-based degraders facilitate autophagosome biogenesis and eventual lysosomal degradation of cargoes.

Fig. 4.

│Internalization mechanism analysis of PLT-TACs. A, Mutated P1 peptide (M1) designed using ProteinMPNN. B, Fluorescence intensity analysis of internalized M1-modified PLT-TAC_m(M1) for 1, 2, or 4 h at 37 °C, measured by FCM (n = 3 biologically independent samples). C, Western blot analysis of PD-L1 in IFN-γ pre-treated 4T1 cells after treated with PLT-TAC_m(M1) and PLT-TAC_m (at an equivalent amount of 5 μM) for 36 h (n = 3 biologically independent samples). D, Dynamic internalization process of PLT-TAC_m at different time points, observed by confocal microscopy. Scale bar = 5 μm. E, Visualization and fluorescence intensity profiles along the line of the colocalization of PLT-TACs with endo/lysosomes in IFN-γ pre-treated 4T1 cells after treated for 2 h or 4 h, observed by oil 63 × CLSM (Early endosomes were stained by EEA1, lysosomes were stained by LAMP1, anti-PD-L1 was labeled with Cy5, nuclei were stained with Hoechst 33342.). Scale bar, 10 μm (n = 3 biologically independent samples). F, Visualization of the amount of lysosome biogenesis in IFN-γ pre-treated 4T1 cells after treated with PLT-anti-PD-L1 or PLT-TACs for 2 h or 4 h, observed by oil 63 × CLSM. Lysosomes were stained by LAMP1. Scale bar, 10 μm. G, Quantification of LAMP1 positive puncta in saline and PLT-TAC_m treated cells after 4 h treatment. H, Western blot analysis of LAMP1 after treated with saline and PLT-TAC_m for 36 h. I, Visualization of the colocalization of PLT-anti-PD-L1 and PLT-TAC_m with LC3-II-labeled autophagosomes (yellow box) in IFN-γ pre-treated 4T1 cells after treatment for 4 h, observed by oil 63 × CLSM. Autophagosomes were stained by LC3-Ⅱ, anti-PD-L1 was labeled with Cy5, nuclei were stained with Hoechst 33342. Scale bar, 10 μm. J, Western blot analysis of LC3-I and LC3-II in after treated with PLT-anti-PD-L1 and PLT-TAC_m for 36 h. K, Visualization of localization of TFEB and statistical analysis of cells with nuclear localization of TFEB after treated with PLT-anti-PD-L1 and PLT-TAC_m for 36 h. Scale bar, 20 μm. (n = 3 biologically independent samples). L, Fluorescence intensity of internalized PLT-anti-PD-L1 and PLT-TAC_m in IFN-γ pre-treated 4T1 cells after treatment for 2 h in presence of chlorpromazine, nystatin, colchicine, and cytochalasin D, measured by FCM (n = 3 biologically independent samples). M, Visualization of internalization of PLT-TACs and lysosome biogenesis in IFN-γ pre-treated 4T1 cells after treatment for 2 h in presence of chlorpromazine, nystatin, colchicine, and cytochalasin D, observed by oil 63 × CLSM. Scale bar, 10 μm. N, SEM images of internalization of PLT-TAC_m by IFN-γ pre-treated 4T1 cells. Scale bar, 10 μm. Data are presented as mean ± SD. p values were determined by one-way ANOVA test. ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

As previous reports indicated that phagocytosis plays a vital role in the uptake of platelets by different cell types, we utilized different inhibitors (chlorpromazine, clathrin-mediated endocytosis inhibitor; nystatin, caveolae-mediated pathway inhibitor; colchicine, macropinocytosis inhibitor; cytochalasin D, phagocytosis inhibitor) to identify the internalization pathways of PLT-TAC_m [18,48]. Cells were pre-treated with the inhibitors (37 °C) for 1 h and incubated with PLT-TAC_m for 4 h. FCM analysis showed that chlorpromazine, colchicine, and cytochalasin D significantly reduced PLT-TAC_m uptake by 1.5-fold, 1.4-fold, and 1.8-fold, respectively, whereas nystatin did not significantly reduce the uptake (Fig. 4L and M), which suggested that clathrin-mediated endocytosis, macropinocytosis and phagocytosis were involved in the uptake process of PLT-TAC_m. Intriguingly, only colchicine and cytochalasin D showed 1.5-fold and 2.0-fold reduced uptake efficiency in PLT-anti-PD-L1 group, respectively. We reckoned the introduction of P1 provoked the clathrin-mediated endocytosis of PLT-TACs. PLT-anti-PD-L1 lacking P1 induced only macropinocytosis and phagocytosis. Meanwhile, we used SEM to observe the internalization process of PLT-TAC_m. Phagocytic cups, pseudopods and endocytic invaginations were observed while PLT-TAC_m was engulfed by cells (Fig. 4N), which also verified that endocytosis, macropinocytosis and phagocytosis were involved in the internalization trafficking process of PLT-TAC_m. These results provided a unique view of internalization trafficking mechanism of PLT-TACs.

1.4. Cell apoptosis, invasion and migration after incubation with PLT-TACs

The external domain of PD-L1 interacts with PD-1 on the surface of immune cells, resulting in an inhibitory effect on immune cells. Meanwhile, the cytoplasmic region of PD-L1 can stimulate an intracellular signaling pathway (PD-L1/PTK/STAT3) to promote the proliferation of tumor cells. Therefore, degrading PD-L1 can inhibit tumor cell growth and relieve the immune escape of cancer cells [49].

Previous studies have validated a correlation between PD-L1 and signal transducer and activation of transcription 3 (STAT3, a known driver of oncogenesis) that promotes the proliferation and metastasis potential of tumor cells (Fig. 5A). PLT-TAC_m showed the highest apoptosis rate among the PLT-TAC degraders, and induced 4.0-fold increase of apoptosis rate compared to the saline group (Fig. 5B and C). Wound healing assay was conducted to evaluate the anti-migration efficacy. As shown in Fig. 5D and E, PLT-TAC_m reduced 4T1 migration by 4.8-fold, exhibiting a stronger anti-migration effect than other groups. In addition, we investigated the anti-invasion effect via transwell invasion assays. The minimum number of 4T1 cells were found invading into the underside of transwell membrane after treated with PLT-TAC_m (Fig. 5F and G). These phenotypic changes were in accordance with PD-L1 degradation levels, which validated the superior applicability of PLT-TAC_m in PD-L1 related modalities.

Fig. 5.

│Efficacy and versatility of PLT-TACs. A, Molecular mechanism underlying PD-L1 degradation-induced apoptosis. B, C, FCM analysis and quantification of cell apoptosis after treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) for 24 h, followed by annexin V-FITC/PI staining (n = 3 biologically independent samples). D, E, Images and statistical analysis of wound healing assay in IFN-γ pre-treated 4T1 cells after treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) for 24 h. Scale bar, 200 μm (n = 3 biologically independent samples). F, G, Images and statistical analysis of transwell cell invasion assay in IFN-γ pre-treated 4T1 cells after treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) for 24 h. Scale bar, 200 μm (n = 3 biologically independent samples). H, Structures of CD71 and T7 peptide and the binding site of CD71/T7 complex, calculated by AlphaFold 3. I, Construction and characterization of PLT-TAC^T7 by TEM and confocal microscopy. J, Mean fluorescence intensity of levels of surface CD71 in 4T1 cells after treated with saline, PLT-anti-PD-L1, or different PLT-TAC^T7 (at an equivalent amount of 10 μM) for 36 h, measured by FCM (n = 3 biologically independent samples). K, Western blot analysis of CD71 in IFN-γ pre-treated 4T1 cells after treated with saline, and different PLT-TAC^T7 (at an equivalent amount of 10 μM) for 36 h (n = 3 biologically independent samples). Data are presented as mean ± SD. p values were determined by one-way ANOVA test. ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Next, we asked whether the PLT-TAC platform could effectively degrade targets other than PD-L1. We selected CD71 as an additional membrane-associated protein for degradation. CD71 (transferrin receptor-1) mediates iron uptake by binding to the iron-carrying protein transferrin and is overexpressed in proliferating cells [50]. T7 peptide (HAIYPRH) was identified via phage display library for high affinity to CD71 [51], and AlphaFold 3 predicted the binding site of T7 peptide in the central cavity of CD71 structure (Fig. 5H and Supplementary Fig. 9). We formed PLT-TAC^T7 using the same procedure as above. PLT-TAC^T7 showed an intact platelet structure and the colocalization of peptides with membrane (Fig. 5I). We tested the degradation efficacy of PLT-TAC^T7. After incubation with different PLT-TACs^T7 at the equivalent concentration of 10 μM for 36 h, all different PLT-TACs^T7 showed CD71 degradation effects, and PLT-TAC^T7_m achieved the maximum degradation of CD71 among all the treatment, consistent with PD-L1 degradation results (Fig. 5J and K), demonstrating the versatility of PLT-TAC platform.

1.5. In vivo antitumor study and degradation efficacy

We next investigated targeted PD-L1 degradation and tumor suppression in the 4T1 orthotopic mouse model. Saline and degraders (anti-PD-L1 dosage of 5 mg/kg) were administered intratumorally to tumor-bearing mice (n = 6 per group) every 3 days for 5 times (Fig. 6A). When the tumor volume in the control group exceeded approximately 1500 mm³, the mice were euthanized, and the tumors and major organs were collected for analysis. Treatment with PLT-TAC_m significantly inhibited tumor growth (>60% tumor volume inhibition), while only relatively moderate tumor suppression was found in PLT-anti-PD-L1 treatment (Fig. 6B, C, 6D, 6G and Supplementary Fig. 21). Moreover, the PLT-TAC_m treatment significantly prolonged the median survival time from 29 to 41 days, whereas PLT-anti-PD-L1 treatment only extended the median survival time to 31 days (Fig. 6F).

Fig. 6.

│In vivo antitumor and degradation efficacy by PLT-TACs. A, Schematic schedule of the treatment and evaluation in BALB/c mice with 4T1 orthotopic tumor model. B, Tumor growth curves, C, images of dissected tumors, D, tumor weights, E, survival curves, F, body weight changes, G, tumor growth kinetics from the control and treatment groups (saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h) at a dosage of 5 mg/kg (n = 6 biologically independent samples). H, tSNE maps of different phenotypes of tumor-infiltrating lymphocytes after treatment, analyzed by FCM (n = 4 biologically independent samples). I, J, Quantitative analysis of matured dendritic cells and natural killer cells in draining lymph nodes by FCM (n = 4 biologically independent samples). K. Immunohistochemical images of PD-L1 in dissected tumors after treatment. L, Mean fluorescence intensity of PD-L1 in CD45-negative tumor cells from dissected tumors after treatment (n = 4 biologically independent samples). Data are presented as mean ± SD. p values were determined by one-way ANOVA test. ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Next, to investigate tumor-specific immune responses elicited by the eradication of PD-L1 following PLT-TAC treatment, we analyzed tumor-infiltrating lymphocytes (TILs) in tumor tissue and carried out t-distributed stochastic neighbor embedding (tSNE) analysis. The infiltration of total T cells (CD3⁺) significantly increased in PLT-TAC_m group, in which more cytotoxic T cells (CTLs, CD8⁺), activated CD8⁺ cells (CD69⁺CD8⁺) and IFN-γ-secreting CTLs (IFN-γ+CD8⁺) were found (Fig. 6H and Supplementary Fig. 18). In addition, more matured dendritic cells (mDCs, CD80⁺CD86⁺ in CD11c) and natural killer cells (NKs, NK1.1+CD3⁻) were detected in draining lymph nodes by FCM, reflecting elevated systemic anti-tumor immunity following PLT-TAC_m treatment (Fig. 6I and J and Supplementary Figs. 19 and 20).

Immunohistochemical analysis showed that PLT-TAC_m induced significantly higher degradation of PD-L1 compared with the control and the other groups (Fig. 6K). Also, PD-L1 degradation was verified in CD45-negative tumor cells (Fig. 6L), all of which demonstrated the degradation efficacy of PLT-TACs and showed good relevance with the in vitro degradation study. Body weight assessment showed no significant differences among all the treatment, showing that PLT-TACs were well-tolerated in vivo (Fig. 6F). Meanwhile, histological analysis showed PLT-TACs did not cause obvious damage in main organs, and blood biochemical analysis manifested no statistically significant difference in three parameters (AST, ALT, and CREA), indicating PLT-TACs had no significant toxicity in vivo (Supplementary Figs. 22 and 23). In general, PLT-TACs had a good degradation capacity and manifested the applicability of targeted protein degradation in vivo.

Then, we evaluated whether the degradation efficacy could be achieved in another cell line. Melanoma B16F10 cells were incubated with the same formulations used in PD-L1 degradation study in 4T1 cells. After incubation with degraders, we observed a consistent result that PLT-TAC_m achieved the greatest PD-L1 degradation (approximately 85% in FCM and 86% in Western blot) among all treatment (Fig. 7A and B), which confirmed that the density of anti-PD-L1 and P1 played a pivotal role in protein degradation. Next, B16F10 tumor model was established to evaluate the in vivo degradation efficacy (Fig. 7C). As expected, PLT-TAC_m inhibited B16F10 tumor growth much more effectively than other treatment (Fig. 7D, E, 7F and Supplementary Fig. 24), and significantly prolonged the survival time (Fig. 7G). Importantly, FCM and immunohistochemical analysis showed that PLT-TAC_m promoted evident degradation of PD-L1 (Fig. 7H and I). All results above validated the effective degradation capacity of PLT-TACs in different cell lines.

Fig. 7.

│PD-L1 degradation efficacy in vitro and in vivo by PLT-TACs in B16F10 cells. A, Western blot analysis of PD-L1 in IFN-γ pre-treated B16F10 cells after treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) for 36 h (n = 3 biologically independent samples). B, Mean fluorescence intensity of levels of surface PD-L1 in IFN-γ pre-treated B16F10 after treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h, measured by FCM (at an equivalent amount of 5 μM) for 36 h (n = 3 biologically independent samples). C, Schematic schedule of the treatment and evaluation in C57BL/6 mice with B16F10 homografting tumor model. D, Tumor growth curves, E, tumor weights, F, images of dissected tumors, G, survival curves from treatment groups (saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h) at a dosage of 5 mg/kg (n = 6 biologically independent samples). H, Immunohistochemical images of PD-L1 in dissected B16F10 tumors after treatment. I, Mean fluorescence intensity of PD-L1 in CD45-negative tumor cells from dissected tumors after treatment (n = 4 biologically independent samples). Data are presented as mean ± SD. p values were determined by one-way ANOVA test. ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

1.6. Biodistribution and antitumor study by intravenous treatment

After verifying the in situ tumor inhibition and PD-L1 degradation effects of PLT-TACs, we further examined their targeting ability, efficacy, and safety after systemic administration. According to the literature, the targeted degradation platform based on lipid nanoparticles (LNP) exhibits favorable passive targeting and in vivo protein degradation efficacy [52]. We constructed an LNP-TAC to compare the its biodistribution, antitumor, and degradation effects with those of PLT-TAC_m (Fig. 8A and B). At 24 h after intravenous administration of Dir-labeled PLT-TAC_m and LNP-TAC, the major organs and tumors were dissected for imaging analysis. The results showed that both LNP-TAC and PLT-TAC_m displayed greater tumor-targeting efficacy compared with free Dir (Fig. 8C). Moreover, PLT-TAC_m exhibited greater accumulation in the tumor tissue and reduced accumulation in the liver than LNP-TAC, which showed the active targeting ability of PLT-TACs to the tumor tissue. Additionally, PLT-TAC_m and LNP-TAC had longer circulation times than free Dir (Fig. 8D), suggesting that PLT-TAC_m could be used as a targeted platform for protein degradation in systemic drug delivery. Additionally, we intravenously administered Cy3-and Cy5-labeled PLT-TACs, and collected it from blood to observe PLT-TAC morphology after 48 h post-injection. Cy3-labeled P1 and Cy5-labeled anti-PD-L1 remained co-located in platelets, which showed the circulatory stability of PLT-TACs (Supplementary Fig. 25).

Fig. 8.

│Biodistribution and antitumor study by intravenous treatment. A, Schematic schedule of the intravenous treatment in BALB/c mice with 4T1 orthotopic tumor model. B, TEM image and DLS size distribution of LNP-TAC. C, The images of biodistribution at 24 h after treatment (n = 3 biologically independent samples). D, Hemodynamics analysis (n = 3 biologically independent samples). E, Tumor growth curves, F, images of dissected tumors, G, tumor weights, H, tumor growth kinetics after treatment (saline, LNP-TAC, and PLT-TAC_m) at a dosage of 5 mg/kg (n = 6 biologically independent samples). I, Immunohistochemical images of PD-L1 in tumors after treatment. J, Mean fluorescence intensity of PD-L1 in CD45-negative tumor cells from dissected tumors after the indicated treatment (n = 4 biologically independent samples). K, Hemolysis test (n = 3 biologically independent samples). L, Blood biochemical analysis (ALT, AST and CREA) after treatment (n = 4 biologically independent samples). Data are presented as mean ± SD. p values were determined by one-way ANOVA test. ns, no significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

The tumor inhibition study revealed that both PLT-TAC_m and LNP-TAC had a certain level of tumor inhibitory effect, and the inhibitory effect of PLT-TAC_m was more prominent than that of LNP-TAC (Fig. 8E, F, 8G, 8H). Simultaneously, PD-L1 immunohistochemistry (Fig. 8I) and flow cytometry analysis (Fig. 8J) demonstrated that both PLT-TAC_m and LNP-TAC could degrade PD-L1 in tumors, and the degradation effect of PLT-TAC_m was more potent than that of LNP-TAC, possibly attributable to its higher accumulation in tumors. Subsequently, we evaluated the in vivo safety of PLT-TAC_m. The hemolysis test indicated that PLT-TAC_m did not induce obvious hemolysis (Fig. 8K). The blood biochemical test showed that PLT-TAC_m did not cause significant changes in ALT, AST, and CREA (Fig. 8L) and did not cause obvious toxicity in the main organs (Supplementary Fig. 26). The above experiments suggest that PLT-TAC_m had the excellent tumor targeting and targeted protein degradation performance after intravenous administration, without causing obvious adverse effects after systemic administration, indicating its potential in clinical practice.

In summary, the PLT-TACs platform successfully achieves targeted degradation of membrane-associated proteins via lysosomal degradation. We found PLT-TACs exhibited targeted affinity to tumor cells, and the lysosome-sorting peptide promoted its endocytosis/phagocytosis by tumor cells. Once PLT-TACs were internalized into cells, lysosome-sorting peptide promoted the “injection” of targeted protein into lysosomes, thereby initiating the lysosomal degradation. We demonstrated that inserting anti-PD-L1 or T7 peptide in PLT-TACs enables efficient degradation of membranous PD-L1 or CD71 protein. PLT-TACs allowed rapid optimization of degradation efficacy by adjusting the ratio of platelets to peptides. PLT-TACs showed the efficient PD-L1 degradation in vitro and in vivo. Moreover, PLT-TACs possessed the tumor-targeting ability and demonstrated efficient PD-L1 degradation following systemic administration without obvious systemic side effects. We anticipate that PLT-TACs can be a new autologous cell-based protein degrader with broad applicability in research and clinical practice.

2. Method

2.1. Materials and reagents

All the AR-grade solvents were purchased from Aladdin Bio-Chem Technology Co., Ltd. (Shanghai, China). Anhydrous solvents (dimethylsulfoxide and N, N-dimethylformamide) were dehydrated by calcium hydride. Reagent-grade 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[maleimide (polyethylene glycol)-2000] (DSPE-PEG2K-Mal), chloroquine, MG132 were purchased from Macklin Technology Co., Ltd. (Shanghai, China). Sulfo-Cyanine 3-Mal and sulfo-Cyanine 5-Mal were purchased from Duofluor Inc. (Wuhan, China). HSPC (1-palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine) and cholesterol were purchased from AVT Inc. (Shanghai, China). Chlorpromazine, nystatin, and colchicine were purchased from Aladdin Bio-Chem Technology Co., Ltd. 4,6-diamidino-2-phenylindole (DAPI), Hoechst 33342 and Dir (1,1′-Dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide) were purchased from Beyotime Biotechnology Co. Ltd. (Shanghai, China). Apoptosis kits, collagenase IV, hyaluronidase and DNase I were purchased from Yeasen Biotechnology Co., Ltd. (Shanghai, China). Lysotracker Green DND26, Lysotracker Red DND99, cytochalasin D, recombinant IFN-γ were purchased from Meilunbio, Inc. (Dalian, China). PD-L1, CD71 primary antibodies, secondary antibodies HRP-labeled goat anti-rabbit IgG and Alexa Fluor 488-goat anti-rabbit IgG (H&L) were purchased from Cell Signaling Technology (Boston, USA), All fluorescent antibodies for flow cytometry were purchased from BD Inc. (New Jersey, USA). DMEM, RPMI-1640 medium, penicillin–streptomycin, fetal bovine serum (FBS), and trypsin were purchased from Gibco (Thermo Fisher Scientific^TM, USA).

D-peptide dNdYdSdKdPdTdDdRdQdYdHdF (anti-PD-L1), CdNdYdSdKdPdTdDdRdQdYdHdF, CRRRRKSFHDDSDEDLLHI (lysosome-sorting signal peptide, P1), CRRRRKSFTVGDEETLAQL (mutated peptide, M1), and CHAIYPRH (T7) were synthesized by standard Fmoc-based solid phase method. Fluorescently-labeled anti-PD-L1-Cy5, P1-Cy3, and T7-Cy5 were conjugated via click chemistry reaction between a terminal thiol group and a maleimide group of sulfo-modified Cy3 or Cy5. Reverse phase high performance liquid chromatography (RP-HPLC) was used for peptide purification. The molecular weights were identified by electrospray ionization mass spectrometry (ESI-MS).

Breast tumor cell line 4T1 and melanoma tumor cell line B16F10 were purchased from the American Type Culture Collection (ATCC), and cultured in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. All cell lines were verified free of mycoplasma, and cultured in a humidified incubator at 37 °C under 5% CO₂. High PD-L1 expression on 4T1 or B16F10 cells was induced by pre-culturing in medium with 50 ng/mL IFN-γ for 24 h before experiments.

BALB/c mice (female, 5-6 weeks) and C57BL/6 mice (female, 5-6 weeks) were obtained from Shanghai JSJ bio-science Co., Ltd. (China) and bred under specific pathogen-free (SPF) condition. All the animal experiments were approved and consented by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University (SYXK(Hu) 2018-0028).

2.2. Fabrication of the microfluidic chip

The polydimethylsiloxane (PDMS) microfluidic chips were fabricated using conventional UV photolithography and soft-lithography methods. Geometries of the microfluidic devices were designed in AutoCAD (Autodesk, San Rafael, CA) and patterned on a silicon plate using conventional photolithography method. The microfluidic chips were composed of a herringbone structure and S-type channels with a depth of 80 μm and a width of 200 μm. The detailed parameters of the herringbone were depicted in Fig. 2A. Base layers were made of PDMS and inlet/outlet holes were punched on it. Afterwards, a PDMS layer and a glass slide were sealed by oxidizing the surfaces using an oxygen plasma cleaner so as to obtain the microfluidic chip.

2.3. Platelet isolation and activation

Isolated platelets were immediately formulated after isolation. In detail, blood collected from the orbital sinus was transferred to a 15 mL centrifuge tube pre-coated with EDTA. Afterwards, the blood sample was mixed at a volume ratio of 1:1 with anticoagulant buffer (2 μM prostaglandin E1, 1 mM EDTA). Mixture was centrifuged at 150 g for 10 min at 4 °C, then the lower layer of red blood cells was discarded. After red blood cells were discarded one more time, the transparent supernatant was centrifuged at 800 g for 20 min at 4 °C and washed one more time to collect platelets. Platelet pellets were suspended in 20 μM adenosine 5′-diphosphate (ADP) contained PBS buffer and incubated for 30 min at 37 °C to activate platelets. Activated platelet protein concentration was measured by BCA assay. Dio was used to stain the platelet membrane in necessity.

2.4. Preparation of lysosome-targeting chimeric platelets (PLT-TACs)

To obtain DSPE-PEG2K-anti-PD-L1, DSPE-PEG2K-P1, DSPE-PEG2K-T7 or DSPE-PEG2K-M1, DSPE-PEG2K-Mal was mixed with CdNdYdSdKdPdTdDdRdQdYdHdF or CRRRRKSFHDDSDEDLLHI, CRRRRKSFTVGDEETLAQL or CHAIYPRH (T7) and incubated in anhydrous solvents (DMSO/DMF, v/v = 1:1) at molar ratio of 1:1 for 24 h. Then, the mixture was dialyzed to remove the solvents and unreacted ingredients and lyophilized to get powdery DSPE-PEG2K-modified peptides. Fluorescence-labeled DSPE-PEG2K-peptides were synthesized via the same route. ¹H-NMR (Bruker, Ascend 400) and MALDI-TOF MS (Shimadzu, MALDI-TOF 7090) were used for characterization analysis (solvent: d⁶-DMSO, D₂O).

The activated platelets suspended in PBS solution (5% DMSO) and the mixture of DSPE-PEG2K-anti-PD-L1/DSPE-PEG2K-P1 (molar ratio, 1/1) in DMSO were loaded into sterile syringes, respectively. The two syringes were mounted on a dual-syringe pump and connected with the inlets of the microfluidic chip. The flow rate ratio of platelet solution to peptide mixture was adjusted to 20:1, and the total flow rate was set to 10 mL/min. After formulation through the microfluidic chip, the mixture was incubated for 30 min and Millipore ultrafiltration device with MWCO 30 kDa was deployed for removing unloaded substrates and organic solvents. PLT-TACs were collected by centrifugation at 800g for 20 min. The PLT-TACs with different ratios were fabricated under the same protocol (The amount of anti-PD-L1 per 1 mg of platelet protein = 1.5 nmol, 4.5 nmol, 13.5 nmol, the amount of anti-PD-L1 per 1 mg of platelet protein in PLT-anti-PD-L1 is 4.5 nmol).

The hydrodynamic diameters were determined by dynamic light scattering (DLS; Zeta plus, Brookhaven, USA). The morphology was imaged by transmission electron microscopy (TEM, 120 kV, JEM-1230, Japan) after negative staining with 1 wt% uranyl acetate. For the surface observation, platelets or PLT-TACs were fixed in 2.5% glutaraldehyde at 4 °C, followed by gradient dehydration in 50%, 70%, 90%, and 100% ethanol. The surface and 3D morphology of PLT-TACs were imaged by scanning electron microscope (SEM, JSM-7800F, Japan). The SEM images of internalization process by cells were acquired by the same procedure after incubating cells with PLT-TACs. The encapsulation efficiency of DSPE-PEG2K-anti-PD-L1 or DSPE-PEG2K-P1 was quantified by fluorescence spectrophotometry.

Morphology and modulus were characterized by atomic force microscopy (AFM). AFM imaging and force measurements were conducted using a scanning model by NanoWizard V (Bruker). AFM force measurements were conducted in liquid under ambient conditions. PLTs and PLT-TAC_m samples were deposited dropwise and fixed onto a poly-lysine-coated glass coverslip. Commercially available AFM ball probes with a radius of 6 μm were used. The AFM results were processed by JPK Data Processing software.

2.5. Protein validation on PLT-TACs

For flow cytometry, PLT-TACs and platelets were suspended in PBS at the density of 10⁶/mL, blocked with 1% BSA buffer for 30 min. Then, CD47, CD41 and, CD62P primary antibodies were incubated with samples at 4 °C for 1 h and washed with 5% BSA in TBST three times. After washing three times, samples were blocked with 1% BSA buffer for 30 min, and incubated with secondary antibody Alexa Fluor 488-goat anti-rabbit IgG (H&L) for 1 h at 4 °C. After rinsing three times with PBS, the fluorescence of samples was measured by FCM (Flow cytometer, Fortessa, BD, USA).

For SDS-PAGE assay, PLT-TACs and platelets were mixed with SDS–PAGE loading buffer (5 × ) and heated at 95 °C for 5 min. After cooling, equal amounts of samples were loaded on a polyacrylamide gel and electrophoresed. After samples were separated, a slice of gel was stained with Coomassie blue buffer for 2 h at room temperature and then destained with H₂O. After that, the gel was imaged.

2.6. Western blot

After treatment, cells were washed with cold PBS three times and lysed with RIPA buffer supplemented with 1 mM protease inhibitor PMSF. Then, cells were scraped, transferred to tubes and centrifuged at 13000 rpm for 20 min at 4 °C, subsequently followed by protein concentration assay by BCA protein assay. Protein samples were mixed with SDS–PAGE loading buffer (5 × ) and heated at 95 °C for 5 min. After that, equal amounts of loaded sample were analyzed by electrophoresis in a 4-12% polyacrylamide gel and then transferred into a polyvinylidene fluoride (PVDF) membrane. The PVDF membranes were blocked with 5% BSA in TBST buffer at room temperature for 1 h, then incubated with the primary antibody at 4 °C overnight and washed with 5% BSA in TBST three times. Later, PVDF membranes were incubated with the HRP-conjugated secondary antibody for 1 h at room temperature and washed with TBST buffer three times. The blots were immersed in an ECL solution for 1-2 min and imaged by Bio-Rad chemiluminescence imaging system. Band intensities were quantified with ImageJ.

2.7. Cell surface binding analysis by flow cytometry

To verify the binding profiles of PLT-TACs to the tumor cells. Cy5-labeled PLT-TACs or PLT-anti-PD-L1 were incubated with IFN-γ pre-treated 4T1 cells at 4 °C for 30 min. Then cells were scraped for detachment, washed three times with 1% BSA in PBS, and analyzed by flow cytometry.

2.8. In vitro internalization

IFN-γ pre-treated 4T1 cells were incubated with 5 μM PLT-TACs at 37 °C for 1, 2, or 4h. Cells were detached using an enzyme-free cell dissociation solution (BeyoTryp™ Express Enzyme, Beyotime), and washed twice with 0.2 M glycine buffer containing 0.15 M NaCl (pH 3.0) to remove membrane-bound samples, then followed by FCM analysis.

To further study the internalization mechanism, IFN-γ pre-treated 4T1 cells were pre-incubated with chlorpromazine hydrochloride (Clathrin-mediated pathway inhibitor, 20 μM), colchicine (Macropinocytosis inhibitor, 5 μM), nystatin (Caveolae-mediated pathway inhibitor, 10 μM) or cytochalasin D (Phagocytosis inhibitor, 10 μM) at 37 °C for 1h, then treated with PLT-TACs at 37 °C for 2h. After treatment, cells were detached using an enzyme-free cell dissociation solution, and washed twice with 0.2 M glycine buffer containing 0.15 M NaCl (pH 3.0) to remove membrane-bound samples, then followed by FCM analysis.

2.9. Confocal microscopy for colocalization of PLT-TACs

Cells were seeded into a confocal glass bottom dish (5 × 10⁴ cells per dish) and treated with medium supplemented with 50 ng/mL IFN-γ for 24 h before experiments, then incubated with Cy5-labeled PLT-TACs at 37 °C for different time points. After the removal of medium, cells were washed with 0.2 M glycine buffer containing 0.15 M NaCl (pH 3.0) to remove membrane-bound samples, then incubated with 50 nM LysoTracker Green DND26 for 30 min, and subsequently stained with Hoechst 33342 solution for 30 min at room temperature. After washing, cells were fixed with 4% paraformaldehyde for 20 min at room temperature and rinsed three times with PBS, then visualized by a confocal laser scanning microscope (CLSM, ZEISS LSM 900).

For visualization of colocalization with organelles, IFN-γ pre-treated 4T1 cells were incubated with Cy5-labeled PLT-TACs at 37 °C. The cells were immersed with Permeabilization and Wash Buffer (Beyotime), then incubated with the appropriate primary antibodies (EEA1, LAMP1, LC3-II) and secondary antibodies in PBS containing 5% BSA. After washing, cells were stained with Hoechst 33342 and fixed with 4% paraformaldehyde for 20 min at room temperature. Afterwards confocal images were captured by CLSM.

2.10. Lysosome biogenesis, autophagy activation, and TFEB translocation

To validate lysosome biogenesis and autophagy activation, IFN-γ pre-treated 4T1 cells were incubated with 5 μM PLT-TAC_m and PLT-anti-PD-L1 at 37 °C for 36 h. After treatment, the expression of LAMP1 and LC3 were analyzed via Western blot. For studying the nuclear translocation of TFEB, IFN-γ pre-treated 4T1 cells were incubated with 5 μM PLT-TAC_m at 37 °C for 36 h. After that, The cells were immersed with Permeabilization and Wash Buffer, then incubated with the primary antibody (TFEB) and the secondary antibody in PBS containing 5% BSA. Then, cells were stained with Hoechst 33342 and fixed with 4% paraformaldehyde for 20 min at room temperature. subsequently were observed by CLSM.

2.11. In vitro protein degradation by PLT-TACs

4T1 cells or B16F10 cells were plated into 6-well culture plates at the density of 2 × 10⁵ per well and pre-treated with IFN-γ of 50 ng/mL for 24 h. IFN-γ pre-treated 4T1 cells were incubated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) at 37 °C for 36 h; or treated with PLT-TAC_m with different concentration (0, 0.025, 0.25, 1.25, 2.5, 5 μM) at 37 °C for 36 h; or treated with PLT-TAC_m (5 μM) for different time (0, 6, 12, 24, 36, 48 h) at 37 °C. Cells were detached using an enzyme-free cell dissociation solution, and washed twice with 0.2 M glycine buffer containing 0.15 M NaCl (pH 3.0) to remove membrane-bound samples, then blocked with CD16/32 antibody for 30 min and stained with 1 μg PE-labeled anti-PD-L1 antibody at 4 °C for 30 min. After washing with PBS, FCM was used to detect the PD-L1 degradation on cells. The CD71 degradation assay was conducted using the same protocol as described above.

For analyzing degradation by Western blot, cells were treated as the same protocol as the above, Then, cells were washed three times and processed as indicated in Western blot section. For observing the localization of PD-L1 protein, IFN-γ pre-treated 4T1 cells were incubated with 5 μM PLT-TAC_m at 37 °C for 24 h, then were immersed with Permeabilization and Wash Buffer, incubated with PD-L1 primary antibody and the secondary antibody in PBS containing 5% BSA. Afterwards, cells were stained with LysoTracker Green DND99 and Hoechst 33342 at room temperature, and analyzed by CLSM. To further study the lysosome-dependent PD-L1 degradation, 100 μM chloroquine as lysosome inhibitor or 5 μM MG132 as proteasome inhibitor was added in cell culture medium. Then, cells were treated with the PLT-TAC_m at 37 °C for 36 h, and the level of PD-L1 was detected by FCM.

2.12. PLT-TACs-induced apoptosis in vitro

Annexin V/PI staining assay was used to study the effect of PD-L1 degradation-induced cell apoptosis. IFN-γ pre-treated 4T1 cells were incubated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) at 37 °C for 24 h, then cells were detached by scraping and collected. Next, cells were incubated with 5 μL of FITC-labeled Annexin V and PI staining kits (40302 ES, Yeasen) for 15 min at room temperature. After that, cells were immediately analyzed by FCM.

2.13. Cell migration and invasion assay in vitro

IFN-γ pre-treated 4T1 cells were plated into 24-well plates at a density of 2 × 10⁵ per well. After 90% confluence, the monolayer of cells was scratched with a 200 μL pipette tip and washed with PBS to remove the floating cells. The cells were then treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM) at 37 °C for 24 h. The images of the scratching wound were captured at 0 and 24 h. The scratching wound area was measured using by Image J software.

Transwell invasion assay was conducted to evaluate the anti-invasion effect of PLT-TACs. The upper chamber with 8 μm PET membrane was coated with 50 μL matrigel and incubated for 30 min for gelling. 2.5 × 10⁴ cells in serum-free medium were plated into the upper chamber and treated with saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h (at an equivalent amount of 5 μM). Complete medium was added to the lower chamber. After incubation for 24 h, the cells were fixed in methanol for 20 min and stained with 0.1% crystal violet for 20 min. Cells above the membrane were wiped off with cotton swabs, and the cells beneath the membrane were imaged by microscope.

2.14. In vivo antitumor study and degradation efficacy

IFN-γ pre-treated 1✕10⁶ 4T1 cells were inoculated subcutaneously in the second mammary pad of female BALB/c mice. When the tumor size reached 50 mm³, the mice were randomly divided into five groups (saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h) with six mice per group. The drugs were administrated at an anti-PD-L1 dosage of 5 mg/kg intratumorally. The same treatment schedule was performed in the survival study. The tumor volume and body weight were measured every two days. The tumor volume was calculated as the formula: V = L✕W✕W/2 (L: the longest dimension, W: the shortest dimension). Once the tumor exceeded 1500 mm³ in volume or the lengths over 15 mm, mice were euthanized. Tumor-infiltrating lymphocytes (TILs) were isolated from the dissected tumor tissues using lymphocyte separation medium, then stained with FCM antibodies and analyzed by FCM. The dissociated cells from the tumor tissues were stained with CD45 and PD-L1 antibodies to detect levels of PD-L1 degradation. Concurrently, tumors and main organs were fixed in 4% PFA solution for hematoxylin-eosin (H&E) staining and immunohistochemical analysis. The blood biochemical analysis was conducted using an automatic biochemistry analyzer (BS-360s, Mindray).

IFN-γ pre-treated 1✕10⁵ B16F10 cells were inoculated subcutaneously in the left flank of female C57BL/6 mice. When the tumor size reached 50 mm³, the mice were randomly divided into five groups (saline, PLT-anti-PD-L1, PLT-TAC_l, PLT-TAC_m or PLT-TAC_h) with six mice per group. The drugs were administrated at an anti-PD-L1 dosage of 5 mg/kg intratumorally. The measurement and analysis assays were implemented using the same protocol as described above.

2.15. Biodistribution and antitumor study by intravenous treatment

To compare the biodistribution and antitumor efficacy, liposome nanoparticles targeting chimeras (LNP-TAC) were prepared by microfluidic method. The mixture of HSPC/Cholesterol/DSPE-PEG2K/DSPE-PEG2K-P1/DSPE-PEG2K-anti-PD-L1(molar ratio, 50:40:8:1:1) was dissolved in N, N-dimethylformamide/ethanol (1/1, v/v), followed by preparation using microfluidic method (water/oil phase = 3: 1, total flow rate = 12 mL/min). The effluent was dialyzed within PBS to remove the solvents. The characterization of LNP-TAC was analyzed by TEM and DLS. The fluorescence labeled peptides were used to calculate the encapsulation efficiency via fluorescence spectrophotometry.

The 4T1 orthotopic tumor model was established as described above. Dir was incorporated with lipid-peptides in formulation process for the biodistribution study. When the tumor size reached 300 mm³, the mice were intravenously administered with free Dir and Dir labeled LNP-TAC or PLT-TAC_m. After 24 h post-injection, mice were euthanized to harvest major organs for ex vivo imaging, and the fluorescence signal was measured (IVIS spectrum, PerkinElmer). Blood samples at different time points were collected from the orbital cavity for hemodynamics analysis. To test the circulatory stability of PTL-TAC, Cy3 and Cy5-labeled PLT-TACs were intravenously administrated into mice. After 48 h post-injection, blood was collected and observed via CLSM.

For the antitumor study, the mice were randomly divided into three groups (saline, LNP-TAC or PLT-TAC_m) with six mice per group. The drugs were intravenously administrated at an anti-PD-L1 dosage of 5 mg/kg. The tumor volume was calculated as the formula: V = L✕W✕W/2 (L: the longest dimension, W: the shortest dimension). Once the tumor exceeded 1500 mm³ in volume or the lengths over 15 mm, mice were euthanized. The dissociated cells from the tumor tissues were stained with CD45 and PD-L1 antibodies to detect levels of PD-L1 degradation. Tumors and main organs were fixed in 4% PFA solution for H&E staining and immunohistochemical analysis. Blood samples were collected for biochemical analysis. For the hemolysis experiment, red blood cells (RBC) were separated from the anticoagulated whole blood of BALB/c mice. RBCs were dispersed in PBS to prepare a 4% RBC suspension. PLTs, LNP-TAC and PLT-TAC_m were mixed with 4% RBC suspension and incubated at 37 °C for 2 h. The supernatant was collected by centrifugation at 1500 rpm for 10 min, and then the absorbance was measured using Microreader at 545 nm. Saline and 0.2% Triton-X 100 were used as negative control and positive control, respectively. The hemolysis rate was calculated as the formula: Hemolysis rate = (A_sample – A_saline)/(A _{Triton-X 100} – A_saline) × 100%.

2.16. Statistical analysis

Experiments were replicated at least three times with biologically independent samples. Statistical analysis was performed by one-way ANOVA using GraphPad Prism 9.0 software. Data were considered statistically significant if p value was less than 0.05.

CRediT authorship contribution statement

Chenming Zou: Writing – original draft, Visualization, Validation, Software, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Haichao Zhu: Writing – original draft, Project administration, Methodology, Investigation, Formal analysis, Data curation. Yuepeng Tang: Validation, Methodology, Investigation. Shengrong Guo: Writing – review & editing, Supervision, Resources, Funding acquisition, Conceptualization.

Ethics approval and consent to participate

BALB/c mice (female, 5-6 weeks) and C57/BL6 mice (female, 5-6 weeks) used in this work were obtained from Shanghai JSJ bio-science Co., Ltd. (China) and bred under specific pathogen-free (SPF) condition. All the animal experiments were approved and consented by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University (SYXK(Hu) 2018-0028).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.