Conjugation of haematopoietic stem cells and platelets decorated with anti-PD-1 antibodies augments anti-leukaemia efficacy
结合表达抗 PD-1 抗体的造血干细胞与血小板可增强抗白血病疗效

The traditional treatment for acute myeloid leukaemia (AML)—a clonal malignancy comprising an increase in myeloblasts in the bone marrow^1,2,3—includes anthracycline and cytarabine-based chemotherapy regimens^4,5. However, the efficacy of traditional chemotherapy for AML is far from satisfactory, as most patients who achieve complete remission will ultimately relapse due to the incomplete elimination of leukaemia cells^6,7,8,9. The prognosis of patients with relapsed leukaemia is dismal^10,11,12. Although relapsed leukaemia could be potentially cured by haematopoietic stem cell (HSC) transplantation, the cost of such transplantation is often associated with high mortality induced by infections or graft-versus-host disease^13,14. The emerging technologies of engineering T cells provide a new approach to treat AML¹⁵. T cells from patients themselves could be removed from the circulation and genetically modified to express an artificial T-cell receptor (designated as a chimeric antigen receptor) in vitro that is designed to specifically recognize the tumour-associated antigens^16,17,18. Chimeric antigen receptor-modified T cells enable the redirection of T-cell specificity and achieve impressive treatment outcomes against blood cancers in the clinic^19,20,21. However, alleviation of the side effects, such as cytokine storm and B-cell aplasia, remains clinically challenging^15,20. The development of new treatment approaches that can effectively eliminate leukaemia cells and avoid side effects is therefore highly desirable to enhance the therapeutic efficacy and prognosis of patients with AML.
传统的急性髓性白血病（AML）治疗方法——一种以骨髓中髓系原始细胞增多为特征的克隆性恶性肿瘤 ^1,2,3 ——包括基于蒽环类药物和阿糖胞苷的化疗方案 ^4,5 。然而，传统化疗对 AML 的疗效远未令人满意，因为大多数达到完全缓解的患者最终会因白血病细胞未被彻底清除而复发 ^6,7,8,9 。复发性白血病患者的预后极差 ^10,11,12 。尽管复发性白血病有可能通过造血干细胞（HSC）移植得到治愈，但此类移植往往伴随高感染率或移植物抗宿主病导致的高死亡率 ^13,14 。新兴的 T 细胞工程技术为治疗 AML 提供了新途径 ¹⁵ 。可以从患者自身循环中提取 T 细胞，并在体外进行基因改造，使其表达一种人工设计的 T 细胞受体（称为嵌合抗原受体），该受体旨在特异性识别肿瘤相关抗原 ^16,17,18 。 嵌合抗原受体修饰的 T 细胞能够重定向 T 细胞特异性，并在临床上对血液癌症取得显著的治疗效果。然而，减轻诸如细胞因子风暴和 B 细胞发育不全等副作用在临床上仍面临挑战。因此，开发既能有效清除白血病细胞又可避免副作用的新治疗策略，对于提升急性髓性白血病患者治疗效果和预后极为重要。

Programmed death-1 (PD-1) is an immune inhibitory co-receptor expressed on a variety of immune cells such as T cells, B cells and natural killer cells²². When bound by its ligands, PD-L1 and PD-L2, PD-1 functions by inhibiting an activated T-cell response^23,24. Tumour cells upregulate PD-L1 in response to inflammation, thereby suppressing an anti-tumour immune response²⁵. Blockade of PD-1 using monoclonal anti-PD-1 antibodies (aPD-1) inhibits tumour-mediated immune suppression and has been demonstrated to improve outcomes in a variety of cancers²⁶. Preclinical studies suggest that blocking the PD-1 pathway may improve outcomes in AML^27,28,29. Thus, the use of aPD-1 represents a promising strategy in the therapeutic armamentarium for AML. Here, we describe a HSC–platelet cellular combination delivery system that can facilitate transport of aPD-1 to the bone marrow and subsequent release of aPD-1 by in situ platelet activation (Fig. 1a). The construction of HSC–platelet assembly is mediated by conjugation of platelets with the HSC plasma membrane through a click reaction (Supplementary Fig. 1). The immune checkpoint inhibitor aPD-1 is covalently decorated on the surface of platelets. Furthermore, the release of aPD-1 can be promoted through the potential generation of platelet-derived microparticles (PMPs) after activation of platelets³⁰, which further enhances the binding of aPD-1 to T cells. After intravenous injection, we have demonstrated that HSC–platelet–aPD-1 assembly (designated as S–P–aPD-1) could effectively accumulate in the bone marrow, where the residual leukaemia cells locate after traditional treatment³¹. Using C1498 and WEHI-3 leukaemia-bearing mice as AML models, we found that S–P–aPD-1 could significantly inhibit leukaemia growth by inducing a potent immune response through the activation of T cells and generation of multiple cytokines and chemokines. Furthermore, such an immune response is durable as it can induce resistance to re-challenging leukaemia cells.
程序性死亡-1（PD-1）是一种免疫抑制性共受体，表达于多种免疫细胞如 T 细胞、B 细胞和自然杀伤细胞上。当与其配体 PD-L1 和 PD-L2 结合时，PD-1 通过抑制激活的 T 细胞反应发挥功能。肿瘤细胞在炎症反应中上调 PD-L1 表达，从而抑制抗肿瘤免疫反应。使用单克隆抗 PD-1 抗体（aPD-1）阻断 PD-1，可抑制肿瘤介导的免疫抑制，并已被证明能改善多种癌症的预后。临床前研究提示，阻断 PD-1 通路可能改善急性髓性白血病（AML）的预后。因此，aPD-1 的应用代表了 AML 治疗中一种有前景的策略。在此，我们描述了一种造血干细胞（HSC）与血小板细胞组合递送系统，该系统能促进 aPD-1 向骨髓的转运，并通过局部血小板激活实现 aPD-1 的释放（图 1a）。HSC 与血小板组件的构建是通过点击反应将血小板与 HSC 细胞膜连接实现的（补充图 1）。 免疫检查点抑制剂 aPD-1 通过共价修饰的方式被装饰在血小板表面。此外，在血小板激活后，可能生成血小板来源的微粒（PMPs），进而促进 aPD-1 的释放，从而增强 aPD-1 与 T 细胞的结合能力。经静脉注射后，我们证实了造血干细胞-血小板-aPD-1 复合体（简称 S–P–aPD-1）能够有效聚集在骨髓中，即传统治疗后残留白血病细胞的所在位置。以携带 C1498 和 WEHI-3 白血病的小鼠作为急性髓系白血病模型，我们发现 S–P–aPD-1 能通过激活 T 细胞并产生多种细胞因子和趋化因子，显著抑制白血病生长。此外，这种免疫反应具有持久性，因为它能诱导对再次挑战的白血病细胞产生抗性。

a, Schematic of HSC–platelet assembly-assisted aPD-1 delivery. After intravenous delivery, the S–P–aPD-1 could home to the bone marrow and the platelets could be locally activated and release aPD-1 to bind T cells for an enhanced immune response. MHC, major histocompatibility complex; TCR, T-cell receptor. b, Confocal microscopy (top) and SEM characterization (bottom) of S–P–aPD-1 conjugates. The platelets were labelled with rhodamine B for confocal observation. White arrows indicate the presence of platelets. c, Transmission electron microscopy (TEM) characterization of PMPs from S–P–aPD-1 after activation by 0.5 U ml⁻¹ thrombin. d, Fluorescence imaging of S–P–aPD-1 with different ratios of platelets to HSCs. e, Quantitative analysis of conjugated platelets on HSCs (the number of bound platelets per cell is indicated as 0, 1, 2, 3.). The quantification was based on counting HSCs (numbers = 200) under the confocal microscope. The experiments were repeated three times. Data are presented as means ± s.d.
a, HSC-血小板组装辅助 aPD-1 递送示意图。静脉给药后，S-P-aPD-1 可归巢至骨髓，血小板在局部被激活并释放 aPD-1，以结合 T 细胞，增强免疫反应。MHC，主要组织相容性复合体；TCR，T 细胞受体。b, S-P-aPD-1 偶联物的共聚焦显微镜（上）和扫描电镜（下）表征。血小板用罗丹明 B 标记以进行共聚焦观察。白色箭头指示血小板的存在。c, 经 0.5 U ml ⁻¹ 凝血酶激活后，S-P-aPD-1 的 PMPs 透射电子显微镜（TEM）表征。d, 不同血小板与 HSC 比例下 S-P-aPD-1 的荧光成像。e, HSC 上偶联血小板的定量分析（每细胞结合的血小板数量表示为 0、1、2、3）。定量基于在共聚焦显微镜下计数 HSC（数量=200）。实验重复三次。数据以均值±标准差表示。

Full size image

Platelets were collected from the whole mouse blood and treated with prostaglandin E1 (PGE1) to inhibit platelet activation³². aPD-1 was conjugated to the platelets using sulfosuccinimidyl-4-(N-maleimidomethyl)-cyclohexane-1-carboxylate (Sulfo-SMCC) as a linker using a covalent conjugation method (Supplementary Fig. 2). Both flow cytometry analysis and confocal images confirmed the successful decoration of aPD-1 on the platelets (Supplementary Figs. 3 and 4). Furthermore, we quantified the conjugation amount of aPD-1 on the platelets using an enzyme-linked immunosorbent assay (ELISA), which showed that the maximum conjugation amount of aPD-1 that could be achieved was ~0.3 pg per platelet (Supplementary Fig. 5a). Such coupling showed negligible cytotoxicity and did not induce the lysis of platelets, which was confirmed by the integrity of platelets after 24 h (Supplementary Fig. 5b). In addition, we characterized the functionality of platelets via collagen binding and platelet aggregation. As shown in Supplementary Fig. 6, P–aPD-1 effectively bound to collagen, with insignificant difference compared with native platelets. Both platelets and P–aPD-1 aggregated after activation, indicating that aPD-1 conjugation did not alter platelet functionality. Furthermore, we validated that after activation, a remarkable amount of aPD-1 was released potentially due to the generation of PMPs, which was significantly higher than that of the non-activated platelets, indicating the well-preserved biofunctionality of platelets after decoration of aPD-1 (Supplementary Fig. 5c)³³.
从全鼠血液中采集血小板，并用前列腺素 E1（PGE1）处理以抑制血小板激活。通过使用磺基琥珀酰亚胺基-4-（N-马来酰亚胺甲基）环己烷-1-羧酸酯（Sulfo-SMCC）作为连接剂，采用共价结合方法将 aPD-1 与血小板结合（补充图 2）。流式细胞术分析和共聚焦图像均证实了 aPD-1 成功修饰于血小板上（补充图 3 和 4）。此外，我们利用酶联免疫吸附测定（ELISA）量化了血小板上 aPD-1 的结合量，结果显示每血小板可达到的最大结合量为~0.3 pg（补充图 5a）。这种结合显示出可忽略的细胞毒性，并未引发血小板溶解，这通过 24 小时后血小板的完整性得到确认（补充图 5b）。此外，我们通过胶原蛋白结合和血小板聚集来表征血小板的功能。如补充图 6 所示，P-aPD-1 有效地结合了胶原蛋白，与天然血小板相比差异不显著。 血小板和 P–aPD-1 在激活后均发生聚集，表明 aPD-1 的结合并未改变血小板的功能。此外，我们验证了在激活后，大量 aPD-1 被释放，这可能是由于血小板微粒（PMPs）的生成，其量显著高于未激活血小板所释放的 aPD-1，表明血小板经 aPD-1 修饰后生物功能得以良好保留（补充图 5c）。

HSCs were isolated from the femur and tibia of C57BL/6J mice (Supplementary Fig. 7) and cultured in 40 µM N-azidoacetylgalactosamine-tetraacylated (Ac₄GalNAz) containing medium for 72 h. Ac₄GalNAz has been identified to label many cells through N-acetylgalactosamine metabolism and incorporate into mucin-type O-linked glycoproteins^34,35. The cell viability study confirmed insignificant cytotoxicity of Ac₄GalNAz at the studied concentration (Supplementary Fig. 8). The presence of azide groups on the surface of HSCs was determined by the addition of an alkynyl-based probe, fluorescein (FAM) alkyne³⁵. The flow cytometry analysis showed increased fluorescence signal on Ac₄GalNAz-treated HSCs when reacted with FAM alkyne through a click reaction, which was significantly higher than that of the HSC control (Supplementary Fig. 9). Furthermore, the Ac₄GalNAz-treated HSCs displayed obvious fluorescence under confocal observation, while negligible fluorescence signal was found on the HSC control (Supplementary Fig. 10).
从 C57BL/6J 小鼠的股骨和胫骨中分离出造血干细胞（HSCs）（补充图 7），并在含有 40 µM N-叠氮乙酰半乳糖胺四酰化物（Ac ₄ GalNAz）的培养基中培养 72 小时。已发现 Ac ₄ GalNAz 通过 N-乙酰半乳糖胺代谢标记多种细胞，并整合到粘蛋白型 O-连接糖蛋白中 ^34,35 。细胞活力研究证实，Ac ₄ GalNAz 在研究浓度下无显著细胞毒性（补充图 8）。通过添加基于炔基的探针，即荧光素（FAM）炔烃 ³⁵ ，确定 HSCs 表面存在叠氮基团。流式细胞术分析显示，Ac ₄ GalNAz 处理的 HSCs 与 FAM 炔烃通过点击反应作用后，荧光信号增强，显著高于 HSC 对照组（补充图 9）。此外，Ac ₄ GalNAz 处理的 HSCs 在共聚焦观察下显示出明显的荧光，而 HSC 对照组几乎未见荧光信号（补充图 10）。

Next, we conjugated the platelets with dibenzocyclooctyne-PEG₄-N-hydroxysuccinimidyl ester (DBCO-PEG₄-NHS ester)³⁶, which reacts with the amine groups on the platelet surface. The decoration of DBCO-PEG₄-NHS ester was determined by the covalent attachment of an azide-based fluorescence probe, azide-fluor 488. The flow cytometry analysis showed brighter fluorescence signals in DBCO-PEG₄-NHS ester-treated platelets than in non-treated platelets and platelets physically mixed with azide probe (Supplementary Fig.11).
随后，我们将血小板与二苯并环辛炔-聚乙二醇-N-羟基琥珀酰亚胺酯（DBCO-PEG ₄ -NHS 酯） ³⁶ 进行共轭，该化合物能与血小板表面的氨基发生反应。通过共价连接基于叠氮的荧光探针——叠氮荧光素 488（azide-fluor 488），确认了 DBCO-PEG ₄ -NHS 酯的修饰效果。流式细胞术分析显示，经 DBCO-PEG ₄ -NHS 酯处理的血小板荧光信号明显强于未处理血小板及仅物理混合叠氮探针的血小板（补充图 11）。

To decorate the HSCs with platelets through the click reaction, DBCO-PEG₄-NHS ester-treated platelets were subjected to the azide-incorporated HSCs and incubated for 45 min. To avoid the aggregation of platelet–HSC assembly, excessive azide-PEG was added to block the free DBCO groups on the platelets³⁷. As observed from the confocal and scanning electron microscopy (SEM) imaging, a direct conjugation at the cell–cell interface was verified and the cellular decoration was tuned at a 1:1 ratio (platelets:HSCs) (Fig. 1b). Additionally, both HSC and platelet morphologies were well maintained after conjugation. The platelet function was preserved, as evidenced by preservation of several key proteins and generation of PMPs after activation (Supplementary Fig. 12 and Fig. 1c). To further investigate the effect of numbers of conjugated platelets on the surface of HSCs, we increased the reaction ratio of platelets to HSCs and quantified the numbers of platelets bound to the HSCs. At a ratio of 1:1 (platelets:HSCs), the majority of assemblies were one platelet on one HSC (about 66%). By increasing platelet numbers, the conjugated platelets on the surface of HSCs were increased correspondingly. At a ratio of 8:1, over 80% of the HSCs were bound with more than three platelets (Fig. 1d,e). However, the HSC viability decreased along with the increased numbers of conjugated platelets (Supplementary Fig. 13). We therefore selected a reaction ratio of 1:1 (platelets:HSCs) for the following studies.
通过点击反应将 HSCs 与血小板装饰，使用 DBCO-PEG ₄ -NHS 酯处理的血小板与包含叠氮基团的 HSCs 混合并孵育 45 分钟。为防止血小板-HSC 复合物的聚集，额外添加了叠氮-PEG 以阻断血小板上未反应的 DBCO 基团 ³⁷ 。从共聚焦显微镜和扫描电子显微镜(SEM)成像观察，证实了细胞间界面的直接结合，并且细胞装饰比例调整为 1:1（血小板:HSCs）(图 1b)。此外，结合后 HSC 和血小板的形态均保持良好。血小板功能得以保留，表现为关键蛋白的保留及激活后血小板微粒（PMPs）的生成（补充图 12 和图 1c）。为进一步探究 HSC 表面结合血小板数量的影响，我们增加了血小板与 HSC 的反应比例，并量化了附着于 HSC 上的血小板数量。在 1:1（血小板:HSCs）比例下，大多数复合物为每个 HSC 上附着一个血小板（约 66%）。 通过增加血小板数量，与造血干细胞（HSCs）表面结合的血小板数量也相应增加。在 8:1 的比例下，超过 80%的 HSCs 结合了三个以上的血小板（图 1d、e）。然而，随着结合血小板数量的增加，HSC 的活性逐渐下降（补充图 13）。因此，我们在后续研究中选择了 1:1（血小板:HSCs）的反应比例。

Retention of residual leukaemia cells in the bone marrow is one of the main reasons for AML relapse³⁸. Thus, selective accumulation of anti-leukaemia drugs in the bone marrow is crucial to enhance the therapeutic index^39,40. To investigate the bone marrow homing capability of S–P–aPD-1, we first evaluated the in vivo pharmacokinetics of S–P–aPD-1. As shown in Fig. 2a, S–P–aPD-1 displayed a significantly longer half-life than free aPD-1, which could be ascribed to the longer persistence in the circulation of HSCs with platelets. We further demonstrated that aPD-1 and Cy5.5-aPD-1 showed an insignificant difference in clearance in vivo (Supplementary Fig. 14). We then tested the bone marrow homing capability of S–P–aPD-1. aPD-1 was labelled with Cy5.5 and conjugated to HSC, platelets and HSC–platelet, respectively. Free aPD-1, HSC–aPD-1 (designated as S–aPD-1), platelet–aPD-1 (designated as P–aPD-1), HSC + platelet–aPD-1 mixture (designated as S + P–aPD-1) and S–P–aPD-1 were intravenously injected into C57BL/6J mice at the equivalent aPD-1 dose and the leg bones were taken out for imaging after 6 h. Both S–aPD-1 and S–P–aPD-1 showed higher fluorescence signals in the bone marrow than P–aPD-1, S + P–aPD-1 mixture and free aPD-1 groups, which suggested superior bone marrow accumulation capability of HSCs (Fig. 2b). The quantitative results showed over 25-fold greater fluorescence signal from bone tissue in S–aPD-1- and S–P–aPD-1-treated mice than the other groups (Fig. 2c). In addition to the bone marrow, the S–P–aPD-1 conjugates could also accumulate in the liver, spleen and lungs (Supplementary Fig. 15). Moreover, the simple blend of HSC and P–aPD-1 did not enhance the accumulation of aPD-1 in the bone marrow. The bone marrow accumulation of HSC and HSC–platelet assembly was further confirmed by fluorescence imaging (Supplementary Fig. 16a–c), and demonstrated the preservation integrity of the membranes of both HSC and platelets localized within the bone marrow. Furthermore, the fluorescence signals of PMPs were found in the bone marrow treated with HSC–platelets, whereas insignificant PMPs were observed in the platelet and PMPs groups (Supplementary Figs. 16d and 17). The existence of PMPs suggested the in situ potential generation of PMPs in the bone marrow after treatment with HSC–platelets, which might be triggered by the leukaemia microenvironment in the bone marrow^41,42,43,44.
骨髓中残留白血病细胞的滞留是 AML 复发的主要原因之一 ³⁸ 。因此，选择性增强抗白血病药物在骨髓中的积累对于提高治疗指数至关重要 ^39,40 。为了研究 S–P–aPD-1 的骨髓归巢能力，我们首先评估了 S–P–aPD-1 的体内药代动力学。如图 2a 所示，S–P–aPD-1 的半衰期显著长于游离 aPD-1，这可归因于与血小板结合的造血干细胞在循环中的持久性更长。我们进一步证明，aPD-1 和 Cy5.5-aPD-1 在体内的清除率无显著差异（补充图 14）。随后，我们测试了 S–P–aPD-1 的骨髓归巢能力。将 aPD-1 用 Cy5.5 标记，并分别与造血干细胞、血小板及造血干细胞-血小板复合物结合。分别将游离 aPD-1、造血干细胞-aPD-1（标记为 S–aPD-1）、血小板-aPD-1（标记为 P–aPD-1）、造血干细胞+血小板-aPD-1 混合物（标记为 S + P–aPD-1）以及 S–P–aPD-1 以等效 aPD-1 剂量静脉注射到 C57BL/6J 小鼠体内，6 小时后取出腿骨进行成像。 S–aPD-1 和 S–P–aPD-1 在骨髓中的荧光信号均高于 P–aPD-1、S + P–aPD-1 混合物及游离 aPD-1 组，这表明造血干细胞（HSC）具有更优越的骨髓富集能力（图 2b）。定量结果显示，与其它组相比，接受 S–aPD-1 和 S–P–aPD-1 处理的小鼠骨组织荧光信号高出 25 倍以上（图 2c）。除骨髓外，S–P–aPD-1 偶联物还能在肝脏、脾脏和肺部积累（补充图 15）。此外，单纯混合 HSC 与 P–aPD-1 并未增强 aPD-1 在骨髓中的积累。HSC 及 HSC-血小板复合体的骨髓富集情况通过荧光成像进一步得到确认（补充图 16a-c），并证实了定位在骨髓内的 HSC 和血小板膜结构的完整性。此外，在经 HSC-血小板处理的骨髓中检测到 PMPs 的荧光信号，而在血小板和 PMPs 组中未见显著的 PMPs（补充图 16d 和 17）。 PMPs 的存在表明，在 HSC-血小板处理后，骨髓中可能存在 PMPs 的原位潜在生成，这可能是由骨髓中的白血病微环境触发的 ^41,42,43,44 。

a, In vivo pharmacokinetics of free aPD-1 and S–P–aPD-1 at the aPD-1 dose of 1 mg per kg (n = 3). Data are presented as means ± s.d. b, Fluorescence images of bone tissues from mice treated with saline (1), Cy5.5-labelled free aPD-1 (2), P–aPD-1 (3), S–aPD-1 (4), S + P–aPD-1 (5) and S–P–aPD-1 (6). The experiments were repeated three times. c, Region-of-interest analysis of fluorescent intensities from bone tissues. Data are presented as means ± s.d. (n = 3). ***P < 0.0001, one-way ANOVA, followed by Tukey’s HSD post hoc test. d, Schematic of building a C1498 leukaemia model and treatment plan. e, Bioluminescence images of mice treated with saline, free aPD-1, S–aPD-1, P–aPD-1, S + P–aPD-1 and S–P–aPD-1 (HSCs/platelets: 5 × 10⁷ cells in 100 µl PBS; aPD-1: 0.5 mg per kg). The experiments were repeated three times. f, Region-of-interest analysis of bioluminescence intensities from whole mouse bodies. g, Flow cytometry analysis of the number of C1498 cells in peripheral blood. The experiments were repeated three times. h, Survival curves for treated and control mice (n = 8). Statistical significance was calculated by log-rank test (***P < 0.0001). i, Morphologies of spleens from mice receiving different treatments (saline (1); free aPD-1 (2); S–aPD-1 (3); P–aPD-1 (4); S + P–aPD-1 (5); and S–P–aPD-1 (6)). The experiments were repeated three times. j, Weights of spleens after the various treatments. Data are presented as means ± s.d. (n = 8). **P = 0.0037, one-way ANOVA, followed by Tukey’s HSD post hoc test.
a, 体内药代动力学研究：在 1 mg/kg aPD-1 剂量下，游离 aPD-1 与 S–P–aPD-1 的药代动力学特性（n = 3）。数据以均值±标准差表示。b, 经生理盐水（1）、Cy5.5 标记的游离 aPD-1（2）、P–aPD-1（3）、S–aPD-1（4）、S + P–aPD-1（5）及 S–P–aPD-1（6）处理的小鼠骨组织荧光图像。实验重复三次。c, 骨组织荧光强度感兴趣区域分析。数据以均值±标准差表示（n = 3）。***P < 0.0001，采用单因素方差分析（ANOVA）后进行 Tukey’s HSD 事后检验。d, 构建 C1498 白血病模型及治疗方案示意图。e, 经生理盐水、游离 aPD-1、S–aPD-1、P–aPD-1、S + P–aPD-1 及 S–P–aPD-1 处理的小鼠生物发光图像（HSCs/血小板：5 × 10^6 细胞溶于 100 µl PBS；aPD-1：0.5 mg/kg）。实验重复三次。f, 小鼠全身生物发光强度感兴趣区域分析。g, 外周血中 C1498 细胞数量流式细胞术分析。实验重复三次。h, 治疗组与对照组小鼠生存曲线（n = 8）。统计显著性通过 log-rank 检验计算（***P < 0.0001）。 i, 接受不同处理的小鼠脾脏形态（生理盐水（1）；游离 aPD-1（2）；S–aPD-1（3）；P–aPD-1（4）；S + P–aPD-1（5）；以及 S–P–aPD-1（6））。实验重复三次。j, 经过各种处理后的脾脏重量。数据以均值±标准差表示（n = 8）。**P = 0.0037，单因素方差分析，随后进行 Tukey’s HSD 事后检验。

Full size image

To investigate the treatment efficacy of S–P–aPD-1 against AML, the C1498 cell line was intravenously injected into C57BL/6J mice⁴⁵. The surface expression of PD-L1 on C1498 cells after administration was confirmed by flow cytometry (Supplementary Fig. 18). The C1498 leukaemia-bearing mice were then treated with three doses of saline, HSCs, platelets, S–P–Rat-immunoglobulin G (IgG), free aPD-1, S–aPD-1, P–aPD-1, S + P–aPD-1 or S–P–aPD-1 every other day after one week at the aPD-1 dose of 0.5 mg per kg (Fig. 2d). In addition, another group of mice were treated with daily administration of aPD-1 for 6 d (aPD-1 dose: 0.25 mg per kg). The development of leukaemia was monitored by the bioluminescence of C1498 cells in vivo. As shown in Fig. 2e,f, the mice treated with S–P–aPD-1 displayed decreased bioluminescence signals after two weeks, and the bioluminescence signals could barely be detected after three weeks. Furthermore, seven of the eight mice in this group displayed a strong immune response with insignificant detectable leukaemia cell signals. In contrast, the mice treated with aPD-1 and daily administration of aPD-1 did not show significant response. The modest treatment efficacy of the P–aPD-1 groups could be attributed to the lack of bone marrow homing capability of the platelets. Furthermore, the lack of immune responses in the S–aPD-1 group could be ascribed to the inefficient activation of T cells due to the steric hindrance of cell–cell interaction⁴⁶, internalization of aPD-1 by HSCs (Supplementary Fig. 19) and/or difficulty of aPD-1 release. In contrast, S–P–aPD-1 effectively accumulated in the bone marrow and released aPD-1 to unleash leukaemia-specific T cells. The survival rate was about 87.5% after 80 d in mice receiving S–P–aPD-1. However, no mice survived beyond 40 d for all other treatments with aPD-1, or beyond 30 d for the saline, HSC and platelet control groups (Fig. 2h and Supplementary Fig. 20c). We also analysed the presence of C1498 cells in the peripheral blood using flow cytometry. As shown in Fig. 2g and Supplementary Figs. 20b and 21, mice receiving the S–P–aPD-1 treatment displayed lower numbers of C1498 cells compared with the other treatment groups with aPD-1, as well as the saline, HSC and platelets control groups. To further demonstrate that the enhanced immune response observed in mice treated with S–P–aPD-1 was not mainly due to the long circulation of aPD-1 and not secondary to an immune reaction against the rat-derived aPD-1 antibody, S–P–Rat-IgG and daily aPD-1 treatments were also tested. As shown in Supplementary Fig. 20a–c, the S–P–Rat-IgG group and treatment groups receiving a daily administration of aPD-1 clearly did not have inhibited leukaemia growth.
为了研究 S–P–aPD-1 对急性髓系白血病（AML）的治疗效果，将 C1498 细胞系通过静脉注射方式注入 C57BL/6J 小鼠体内。注射后，通过流式细胞术确认了 C1498 细胞表面 PD-L1 的表达（补充图 18）。随后，携带 C1498 白血病的小鼠在一周后每隔一天接受三次盐水、造血干细胞（HSCs）、血小板、S–P–大鼠免疫球蛋白 G（IgG）、游离 aPD-1、S–aPD-1、P–aPD-1、S + P–aPD-1 或 S–P–aPD-1 的治疗，aPD-1 剂量为每公斤 0.5 毫克（图 2d）。此外，另一组小鼠每日接受 aPD-1 治疗，持续 6 天（aPD-1 剂量：每公斤 0.25 毫克）。通过 C1498 细胞体内生物发光监测白血病进展。如图 2e、f 所示，接受 S–P–aPD-1 治疗的小鼠在两周后生物发光信号减少，三周后几乎无法检测到生物发光信号。此外，该组八只小鼠中有七只显示出强烈的免疫反应，且几乎检测不到白血病细胞信号。相比之下，接受 aPD-1 及每日 aPD-1 治疗的小鼠未表现出显著反应。 P–aPD-1 组治疗效果较为有限，可能归因于血小板缺乏骨髓归巢能力。此外，S–aPD-1 组缺乏免疫反应，可能是由于细胞间相互作用的空间阻碍、aPD-1 被造血干细胞内化（补充图 19）以及 aPD-1 释放困难所致。相比之下，S–P–aPD-1 在骨髓中有效积累并释放 aPD-1，从而激活白血病特异性 T 细胞。接受 S–P–aPD-1 治疗的小鼠在 80 天后存活率约为 87.5%。然而，所有其他使用 aPD-1 的治疗方案中，小鼠均未能在 40 天后存活，而生理盐水、HSC 和血小板对照组的小鼠则在 30 天后全部死亡（图 2h 及补充图 20c）。我们还通过流式细胞术分析了外周血中 C1498 细胞的存在情况。如图 2g 及补充图 20b、21 所示，接受 S–P–aPD-1 治疗的小鼠体内 C1498 细胞数量明显低于其他使用 aPD-1 的治疗组，以及生理盐水、HSC 和血小板对照组。 为了进一步证明在小鼠中观察到的 S–P–aPD-1 处理后增强的免疫反应并非主要由于 aPD-1 的长循环效应，也并非继发于对鼠源 aPD-1 抗体的免疫反应，还测试了 S–P–Rat-IgG 和每日 aPD-1 给药。如补充图 20a–c 所示，S–P–Rat-IgG 组和每日接受 aPD-1 治疗的组别均未显示出抑制白血病生长的效果。

Furthermore, the spleens of the mice receiving different treatments were removed and imaged (Fig. 2i and Supplementary Fig. 20d). The spleen from S–P–aPD-1-treated mice displayed a normal morphology, whereas the spleens from other treatment groups were enlarged^47,48. The quantitative results showed that the spleens from the mice treated with S–P–aPD-1 were 1/2 ~ 1/3 the weight of other spleens (Fig. 2j and Supplementary Fig. 20e). We then used haematoxylin and eosin staining to investigate the development of leukaemia in the main organs. As shown in Supplementary Fig. 22, leukaemia cells were found in the bone marrow, liver, spleen and lung tissues in mice treated with saline. In contrast, the mice treated with S–P–aPD-1 displayed negligible numbers of leukaemia cells in the main organs.
此外，将接受不同治疗的小鼠的脾脏取出并进行成像（图 2i 及补充图 20d）。S–P–aPD-1 治疗组小鼠的脾脏呈现正常形态，而其他治疗组的脾脏则出现肿大现象。定量结果显示，S–P–aPD-1 治疗小鼠的脾脏重量约为其他组脾脏的 1/2 至 1/3（图 2j 及补充图 20e）。随后，我们采用苏木精-伊红染色法探究主要器官中白血病的发展情况。如补充图 22 所示，接受生理盐水治疗的小鼠在骨髓、肝脏、脾脏和肺组织中均发现了白血病细胞。相比之下，S–P–aPD-1 治疗的小鼠在主要器官中白血病细胞数量极少。

To understand the cellular mechanisms underlying the observed therapeutic effects of S–P–aPD-1, upon treatment, the T cells in the peripheral blood were collected and analysed by flow cytometry. An approximately fourfold increase in the total number of CD3⁺ T cells was observed in mice receiving the S–P–aPD-1 treatment compared with the saline, HSC and platelet control groups (Fig. 3b and Supplementary Fig. 23). CD3⁺ T cells in the S–P–aPD-1 treatment group showed a 1.9 ~ 2.4-fold increase compared with the other treatment groups with aPD-1 (Fig. 3b). Moreover, mice receiving S–P–aPD-1 treatment displayed a 1.5-fold increase in CD8⁺ T cells compared with the saline, HSC and platelet control groups, and an approximately 1.3-fold increase compared with the other treatment groups with aPD-1 (Fig. 3a,c and Supplementary Fig. 23). Furthermore, IFNγ⁺CD8⁺ T cells in the S–P–aPD-1 treatment group increased by 1.9 ~ 2.6-fold compared with the other treatment groups with aPD-1, and 6.4 ~ 17.8-fold compared with the saline, HSC and platelet control groups (Supplementary Fig. 24). Increases in CD3⁺, CD8⁺ and IFNγ⁺CD8⁺ T cells were suggestive of boosted T-cell immune responses in the S–P–aPD-1-treated mice. We further analysed T-cell subsets in the bone marrow. PD-1-expressing T cells were detected in the bone marrow of leukaemia-bearing mice by flow cytometry (Supplementary Fig. 25). Mice receiving the S–P–aPD-1 treatment displayed 1.8 ~1.9-fold increases in CD8⁺ T cells compared with the saline, HSC and platelet control groups, and approximately 1.3-fold increases compared with the other treatment groups with aPD-1 (Supplementary Fig. 26). Moreover, the CD8⁺ T cells in the bone marrow of mice after treatment with S–P–aPD-1 were about 44% Granzyme B positive (Supplementary Fig. 27), which was significantly higher than the other treatment groups, suggesting that effector T-cell number was increased after S–P–aPD-1 treatment. Besides this, the number of effector T cells exhibiting early activation status in the S–P–aPD-1 treatment group was higher than in the other treatment groups, as evidenced by higher percentages of CD8⁺CD44⁺CD69⁺ (Supplementary Fig. 28) and CD8⁺CD44⁺CD25⁺ (Supplementary Fig. 29) T cells. IFNγ⁺CD8⁺ T cells in the S–P–aPD-1 treatment group showed 2.6 ~ 2.9-fold increases compared with the other aPD-1 treatment groups, and 5.3 ~ 21.2-fold increases compared with the saline, HSC and platelet control groups (Supplementary Fig. 30). In contrast, CD8⁺ and IFNγ⁺CD8⁺ T-cell subsets in the bone marrow of non-leukaemia-bearing mice did not show significant increases after treatment with S–P–aPD-1 (Supplementary Fig. 31). Luminex-based quantification of cytokines and chemokines⁴⁹ revealed four clusters of co-regulated proteins, with pro-inflammatory factors increased in the peripheral blood after treatment with aPD-1 (Fig. 3d). Furthermore, the majority of cytokines and chemokines were upregulated in the S–P–aPD-1 group compared with the other treatment groups with aPD-1. The increased serum cytokine levels might reflect the alteration of other immune cell subsets, including monocytes and myeloid cells, as AML is characterized by prevention of maturation of monocytes and myeloid cells^3,50. The development of leukaemia could potentially cause abnormalities of monocytes and other myeloid cells, including granulocytes in AML patients, as shown by previous studies^51,52.
为了理解 S–P–aPD-1 所观察到的治疗效果背后的细胞机制，在治疗后，采集了外周血中的 T 细胞并通过流式细胞术进行分析。与生理盐水、造血干细胞和血小板对照组相比，接受 S–P–aPD-1 治疗的鼠中，CD3+ T 细胞总数大约增加了四倍（图 3b 及补充图 23）。在 S–P–aPD-1 治疗组中，CD3+ T 细胞相较于其他含有 aPD-1 的治疗组显示了 1.9 至 2.4 倍的增加（图 3b）。此外，与生理盐水、造血干细胞和血小板对照组相比，接受 S–P–aPD-1 治疗的鼠中，CD8+ T 细胞数量增加了 1.5 倍，相较于其他含有 aPD-1 的治疗组，则大约增加了 1.3 倍（图 3a、c 及补充图 23）。进一步地，S–P–aPD-1 治疗组中的 IFNγ+ CD8+ T 细胞相较于其他含有 aPD-1 的治疗组增加了 1.9 至 2.6 倍，相较于生理盐水、造血干细胞和血小板对照组则增加了 6.4 至 17.8 倍（补充图 24）。 CD3 ⁺ 、CD8 ⁺ 及 IFNγ ⁺ CD8 ⁺ T 细胞的增加提示在 S–P–aPD-1 治疗的小鼠中 T 细胞免疫反应得到了增强。我们进一步分析了骨髓中的 T 细胞亚群。通过流式细胞术检测到白血病小鼠骨髓中存在表达 PD-1 的 T 细胞（补充图 25）。接受 S–P–aPD-1 治疗的鼠中，CD8 ⁺ T 细胞数量相较于生理盐水、HSC 及血小板对照组增加了 1.8 至 1.9 倍，与其他使用 aPD-1 的治疗组相比也增加了约 1.3 倍（补充图 26）。此外，经 S–P–aPD-1 治疗后小鼠骨髓中的 CD8 ⁺ T 细胞约有 44%为 Granzyme B 阳性（补充图 27），这一比例显著高于其他治疗组，表明 S–P–aPD-1 治疗后效应 T 细胞数量有所增加。除此之外，S–P–aPD-1 治疗组中表现出早期激活状态的效应 T 细胞数量也高于其他治疗组，表现为 CD8 ⁺ CD44 ⁺ CD69 ⁺ （补充图 28）及 CD8 ⁺ CD44 ⁺ CD25 ⁺ （补充图 29）T 细胞比例更高。 在 S–P–aPD-1 治疗组中，IFNγ ⁺ CD8 ⁺ T 细胞相较于其他 aPD-1 治疗组显示出 2.6 至 2.9 倍的增加，相较于生理盐水、造血干细胞和血小板对照组则增加了 5.3 至 21.2 倍（补充图 30）。相比之下，非白血病小鼠骨髓中的 CD8 ⁺ 和 IFNγ ⁺ CD8 ⁺ T 细胞亚群在接受 S–P–aPD-1 治疗后并未显示出显著增加（补充图 31）。基于 Luminex 的细胞因子与趋化因子 ⁴⁹ 定量分析揭示了四个协同调控的蛋白质簇，其中外周血中的促炎因子在 aPD-1 治疗后有所增加（图 3d）。此外，与接受 aPD-1 的其他治疗组相比，S–P–aPD-1 组中大多数细胞因子和趋化因子均上调。血清细胞因子水平的升高可能反映了其他免疫细胞亚群，包括单核细胞和髓系细胞的变化，因为急性髓性白血病以阻止单核细胞和髓系细胞的成熟为特征 ^3,50 。 白血病的发展可能会导致单核细胞及其他髓系细胞（包括急性髓性白血病患者的粒细胞）的异常，此结论已由先前研究 ^51,52 所证实。

a, Flow cytometry analysis of CD8⁺ T cells (gated on CD3⁺ T cells) in peripheral blood. The experiments were repeated three times. b, Quantitative analysis of the number of CD3⁺ T cells. Data are presented as means ± s.d. (n = 8). **P = 0.0048, one-way ANOVA, followed by Tukey’s HSD post hoc test. c, Quantitative analysis of the percentage of CD8⁺ T cells. Data are presented as means ± s.d. (n = 8). ***P < 0.0001, one-way ANOVA, followed by Tukey’s HSD post hoc test. d, Luminex-based quantification of cytokines and chemokines (n = 8).
a, 外周血中 CD8 ⁺ T 细胞（基于 CD3 ⁺ T 细胞门控）的流式细胞术分析。实验重复了三次。b, CD3 ⁺ T 细胞数量的定量分析。数据以均值±标准差表示（n = 8）。**P = 0.0048，单因素方差分析，随后进行 Tukey’s HSD 事后检验。c, CD8 ⁺ T 细胞百分比的定量分析。数据以均值±标准差表示（n = 8）。***P < 0.0001，单因素方差分析，随后进行 Tukey’s HSD 事后检验。d, 基于 Luminex 的细胞因子与趋化因子定量分析（n = 8）。

Full size image

S–P–aPD-1 treatment caused a shift of native CD8⁺ T cells to the active phenotype and to the central memory CD8⁺ T-cell phenotype in vivo. The CD44^hiCD62L^hi central memory T-cell subset showed a 2.1-fold increase in the S–P–aPD-1-treated mice compared with the saline control (Fig. 4a). Similarly, the CD44^hiCD122^hi memory T-cell subset also showed an increase (Supplementary Fig. 32). The functionality of the memory subsets in S–P–aPD-1-treated mice was shown in re-challenged experiments. Mice showing control of leukaemia growth after treatment with S–P–aPD-1 were re-challenged with 1 × 10⁶ C1498 cells at day 80. Leukaemia grew rapidly in native mice, while S–P–aPD-1-treated mice remained leukaemia free at day 60 (Fig. 4b,c). To further confirm the key role of the T-cell-mediated immune response in S–P–aPD-1 treatment, T-cell-knockout mice (rag^−/−) were injected with C1498 cells and then administered with S–P–aPD-1. As shown in Fig. 4d, control mice, as well as mice treated with free aPD-1 and S–P–aPD-1 showed equal leukaemia growth. In addition, CD8 T-cell depletion experiments in mice bearing C1498 leukaemia showed that the therapeutic effects of S–P–aPD-1 were abrogated in the absence of CD8⁺ T cells (Supplementary Fig. 33 and Fig. 4e). We also investigated the treatment efficacy of S–P–aPD-1 in PD-1 knockout mice (PD^−/−). A significantly diminished anti-leukaemia effect was found in PD^−/− mice (Fig. 4f), which was associated with partial exhaustion of CD8⁺ T cells, as evidenced by the increase of TIM-3⁺ and LAG-3⁺ T cells (Supplementary Fig. 34).
S–P–aPD-1 治疗促使天然 CD8 ⁺ T 细胞在体内向活跃表型和中枢记忆型 CD8 ⁺ T 细胞表型转变。与生理盐水对照组相比，S–P–aPD-1 处理的小鼠中，CD44 ^hi CD62L ^hi 中枢记忆 T 细胞亚群数量增加了 2.1 倍（图 4a）。同样，CD44 ^hi CD122 ^hi 记忆 T 细胞亚群也表现出增加（补充图 32）。S–P–aPD-1 治疗小鼠的记忆亚群功能通过再挑战实验得以展示。在 S–P–aPD-1 治疗后表现出白血病生长控制的小鼠，于第 80 天再次接受 1 × 10 ⁶ C1498 细胞的挑战。白血病在天然小鼠中迅速生长，而 S–P–aPD-1 处理的小鼠在第 60 天仍保持无白血病状态（图 4b,c）。为进一步确认 T 细胞介导的免疫反应在 S–P–aPD-1 治疗中的关键作用，T 细胞缺失小鼠（rag ^−/− ）被注射 C1498 细胞后接受 S–P–aPD-1 治疗。如图 4d 所示，对照组小鼠、自由 aPD-1 治疗小鼠及 S–P–aPD-1 治疗小鼠的白血病生长情况相同。 此外，在携带 C1498 白血病的鼠中进行的 CD8 T 细胞耗竭实验显示，在缺乏 CD8 T 细胞的情况下，S–P–aPD-1 的治疗效果被取消（补充图 33 和图 4e）。我们还研究了 S–P–aPD-1 在 PD-1 基因敲除鼠中的治疗效果（PD）。在 PD 鼠中发现了显著减弱的抗白血病效应（图 4f），这伴随着 CD8 T 细胞的部分耗竭，表现为 TIM-3 和 LAG-3 T 细胞的增加（补充图 34）。

a, Flow cytometry analysis of CD44^hiCD62L^hi T cells (gated on CD8⁺ T cells) in saline- and S–P–aPD-1-treated mice (n = 3). b, Bioluminescence images of native and treated mice re-challenged with 1 × 10⁶ C1498 cells at 3 weeks. c, Survival curves for native and treated mice after re-challenge with C1498 cells (n = 7). ***P < 0.001, log-rank test. d–f, Survival curves for rag^−/− mice (d), CD8-depleted mice (e) and PD^−/− mice (f) treated with PBS, free aPD-1 and S–P–aPD-1 at an aPD-1 dose of 0.5 mg per kg (n = 8).
a, 流式细胞术分析在生理盐水和 S–P–aPD-1 处理的小鼠中，CD44 ^hi CD62L ^hi T 细胞（以 CD8 ⁺ T 细胞为门控）的情况（n = 3）。b, 原生及处理后小鼠在 3 周时再次接种 1 × 10 ⁶ C1498 细胞后的生物发光图像。c, 原生及处理后小鼠在再次接种 C1498 细胞后的生存曲线（n = 7）。***P < 0.001，对数秩检验。d–f, 接受 PBS、游离 aPD-1 和 S–P–aPD-1 治疗的 rag ^−/− 小鼠（d）、CD8 耗竭小鼠（e）及 PD ^{− /−} 小鼠（f）在 aPD-1 剂量为 0.5 mg 每千克时的生存曲线（n = 8）。

Full size image

To assess the effectiveness of S–P–aPD-1 in treating another type of leukaemia, we used the WEHI-3 myelomonocytic leukaemia cell line in BALB/cJ mice. As illustrated in Supplementary Fig. 35, in this leukaemia model, mice receiving S–P–aPD-1 therapy showed better leukaemia control, and 62.5% of the treated mice were alive at day 50. In contrast, mice in the control groups succumbed by day 40 and displayed larger spleens (Supplementary Fig. 35).
为了评估 S–P–aPD-1 在治疗另一种白血病中的有效性，我们在 BALB/cJ 小鼠中使用了 WEHI-3 髓单核细胞白血病细胞系。如图补充图 35 所示，在这种白血病模型中，接受 S–P–aPD-1 治疗的小鼠显示出更好的白血病控制效果，并且在第 50 天时，62.5%的受治疗小鼠仍然存活。相比之下，对照组的小鼠在第 40 天前全部死亡，并且脾脏肿大（见补充图 35）。

In summary, our study describes a ‘cell-combination’ strategy for drug delivery in which one cell type drives homing to the leukaemia site and the other delivers the drug. This combined approach allows the achievement of effective immune responses and inhibition of leukaemia growth. We obtained these effects by taking advantage of HSCs' homing capability and in situ activation of platelets. S–P–aPD-1 promotes the delivery of aPD-1 in the bone marrow where the leukaemia is localized, and effectively unleashes leukaemia-specific T cells that control leukaemia growth and recurrence. S–P–aPD-1 seems to induce a potent immune response while mitigating toxicities. Furthermore, such a cell-assembly-mediated drug-delivery approach could be adopted to incorporate other bio-particulates to treat other diseases for which spatiotemporal drug delivery is essential^53,54,55. However, whether the platelet activation of the S–P–aPD-1 could cause potential side effects in patients due to the release of pro-inflammatory molecules or systemic T-cell activation during the circulation of S–P–aPD-1 in the blood stream remains to be studied.
总之，本研究阐述了一种“细胞组合”药物递送策略，其中一种细胞类型负责归巢至白血病部位，另一种则负责药物递送。这种联合方法实现了有效的免疫应答及白血病生长的抑制。我们通过利用造血干细胞的归巢能力及血小板的原位激活，获得了这些效果。S–P–aPD-1 促进了 aPD-1 在白血病定位的骨髓中的递送，并有效释放了控制白血病生长及复发的白血病特异性 T 细胞。S–P–aPD-1 似乎能引发强效的免疫反应，同时减轻毒性。此外，这种细胞组装介导的药物递送方法可被采纳，以结合其他生物颗粒用于治疗那些对时空药物递送至关重要的其他疾病。然而，S–P–aPD-1 引发的血小板激活是否会在其在血液中循环时，因释放促炎分子或全身性 T 细胞激活而给患者带来潜在副作用，仍有待研究。

Cell lines and cells 细胞系与细胞

The murine leukaemia cell line—C1498 tagged with luciferase and DsRed fluorescence—was kindly provided by B. Blazar (University of Minnesota). The WEHI-3 cells were purchased from the University of North Carolina tissue culture facility. The C1498 and WEHI-3 cells were maintained in Dulbecco’s modified Eagle’s medium (Gibco; Invitrogen) supplemented with 10% foetal bovine serum (Invitrogen), 100 U ml⁻¹ penicillin (Invitrogen) and 100 U ml⁻¹ streptomycin (Invitrogen). The HSCs and progenitor cells (designated as HSCs in all the studies) were isolated from the femur and tibia of C57BL/6J mice. Bone marrow was first pre-enriched with a lineage depletion kit (Miltenyi). The resulting cells were subsequently incubated and selected with anti-Sca-1 microbeads (Miltenyi) to obtain lin-Sca-1⁺ HSCs with over 90% purity. Cells were cultured in Serum-Free Expansion Medium (SFEM; STEMCELL Technologies) with the addition of human interleukin-6 (50 ng ml⁻¹; Thermo Fisher Scientific), human Flt3 ligand (100 ng ml⁻¹; Thermo Fisher Scientific), murine stem cell factors (50 ng ml⁻¹; Thermo Fisher Scientific) and low-density lipoprotein (40 µg ml⁻¹; Thermo Fisher Scientific). Cells were cultured in an incubator (Thermo Fisher Scientific) at 37 °C under an atmosphere of 5% CO₂ and 90% relative humidity. The cells were sub-cultivated approximately every 2–3 d at 80% confluence at a split ratio of 1:3.
带有荧光素酶和 DsRed 荧光标记的小鼠白血病细胞系 C1498 由 B. Blazar（明尼苏达大学）慷慨提供。WEHI-3 细胞购自北卡罗来纳大学组织培养设施。C1498 和 WEHI-3 细胞在含 10%胎牛血清（Invitrogen）、100 U/ml 青霉素（Invitrogen）和 100 U/ml 链霉素（Invitrogen）的 Dulbecco 改良 Eagle 培养基（Gibco；Invitrogen）中培养。造血干细胞（HSCs）及祖细胞（所有研究中均称为 HSCs）从 C57BL/6J 小鼠的股骨和胫骨中分离。骨髓首先通过谱系耗竭试剂盒（Miltenyi）进行预富集。随后，所得细胞与抗 Sca-1 磁珠（Miltenyi）共孵育并筛选，以获得纯度超过 90%的 lin-Sca-1+ HSCs。细胞在无血清扩展培养基（SFEM；STEMCELL Technologies）中培养，并添加人白细胞介素-6（50 ng/ml；赛默飞世尔科技）、人 Flt3 配体（100 ng/ml；赛默飞世尔科技）、小鼠干细胞因子（50 ng/ml；赛默飞世尔科技）和低密度脂蛋白（40 µg/ml；赛默飞世尔科技）。 细胞在 37°C、5% CO₂和 90%相对湿度的培养箱（赛默飞世尔科技）中培养。当细胞达到 80%汇合度时，大约每隔 2-3 天以 1:3 的分裂比例进行传代培养。

Antibody 抗体

The aPD-1 was obtained from BioLegend (catalogue number 114114, clone: RMP1-14). The antibodies used for immunostaining were specific for CD3 (BioLegend; catalogue number 100236, clone: 17A2), CD4 (BD Biosciences; catalogue number 553046, clone: RM4-5), CD8a (BioLegend; catalogue number 100708, clone: 53-6.7), IFN-γ (BioLegend; catalogue number 505806, clone: XMG1.2), CD122 (BioLegend; catalogue number 123207, clone: TM-β1), CD41 (BioLegend; catalogue number 133904, clone: MWReg30), CD9 (BioLegend; catalogue number 124807, clone: MZ3), CD61 (BioLegend; catalogue number 104307, clone: 2C9.G2 (HMβ3-1)), CD62P (BioLegend; catalogue number 148305, clone: RMP-1), CD36 (BioLegend; catalogue number 102605, clone: HM36), CD154 (BioLegend; catalogue number 106505, clone: MR1), CD62L (BioLegend; catalogue number 104405, clone: MEL-14), CD44 (BioLegend; catalogue number 103024, clone: IM7), CD34 (BioLegend; catalogue number 128609, clone: HM34), CD38 (BD Biosciences; catalogue number 558813, clone: 90/CD38), CD117 (BioLegend; catalogue number 105812, clone: 2B8), CD366 (BD Biosciences; catalogue number 566346, clone: 5D12/TIM-3), CD223 (BioLegend; catalogue number 125207, clone: C9B7W), CD274 (BioLegend; catalogue number 124311, clone: 10F.9G2), Granzyme B (BioLegend; catalogue number 372204, clone: QA16A02), CD25 (BioLegend; catalogue number 101904, clone: 3C7) and CD69 (BioLegend; catalogue number 104508, clone: H1.2F3). The cells after staining were subjected to fluorescence-activated cell-sorting analysis following the manufacturers’ instructions. Multicolour flow cytometry was used with appropriate compensation. All antibodies were used following the manufacturers’ instructions. The fluorochromes conjugated on the antibody were exactly matched to the same fluorochrome channel. After staining, cells were analysed on a fluorescence-activated cell-sorting Calibur instrument (CytoFLEX; BD Biosciences), using the FlowJo or CytExpert software packages. CD8⁺ T-cell depletion antibody (clone: YTS169.4) was purchased from Bio X Cell. The rat IgG was purchased from Invitrogen. Recombined mPD-1 was obtained from R&D Systems. The secondary antibodies used for immunostaining were goat anti-rat IgG (H+L) (Thermo Fisher Scientific; catalogue number A18866).
aPD-1 抗体购自 BioLegend（目录号 114114，克隆号：RMP1-14）。用于免疫染色的抗体特异性针对 CD3（BioLegend；目录号 100236，克隆号：17A2）、CD4（BD Biosciences；目录号 553046，克隆号：RM4-5）、CD8a（BioLegend；目录号 100708，克隆号：53-6.7）、IFN-γ（BioLegend；目录号 505806，克隆号：XMG1.2）、CD122（BioLegend；目录号 123207，克隆号：TM-β1）、CD41（BioLegend；目录号 133904，克隆号：MWReg30）、CD9（BioLegend；目录号 124807，克隆号：MZ3）、CD61（BioLegend；目录号 104307，克隆号：2C9）。G2（HMβ3-1））、CD62P（BioLegend；目录号 148305，克隆号：RMP-1）、CD36（BioLegend；目录号 102605，克隆号：HM36）、CD154（BioLegend；目录号 106505，克隆号：MR1）、CD62L（BioLegend；目录号 104405，克隆号：MEL-14）、CD44（BioLegend；目录号 103024，克隆号：IM7）、CD34（BioLegend；目录号 128609，克隆号：HM34）、CD38（BD Biosciences；目录号 558813，克隆号：90/CD38）、CD117（BioLegend；目录号 105812，克隆号：2B8）、CD366（BD Biosciences；目录号 566346，克隆号：5D12/TIM-3）、CD223（BioLegend；目录号 125207，克隆号：C9B7W）、CD274（BioLegend；目录号 124311，克隆号：10F.9G2）、Granzyme B（BioLegend；目录号 372204，克隆号：QA16A02）、CD25（BioLegend；目录号 101904，克隆号：3C7）和 CD69（BioLegend；目录号 104508，克隆号：H1.2F3）。染色后的细胞根据厂家说明进行荧光激活细胞分选分析。采用适当补偿的多色流式细胞术。所有抗体均按照厂家说明使用。 抗体上偶联的荧光染料与相同荧光通道完全匹配。染色后，细胞在荧光激活细胞分选仪 Calibur（CytoFLEX；BD Biosciences）上进行分析，使用 FlowJo 或 CytExpert 软件包。CD8 ⁺ T 细胞耗竭抗体（克隆号：YTS169.4）购自 Bio X Cell。大鼠 IgG 购自 Invitrogen。重组 mPD-1 由 R&D Systems 提供。用于免疫染色的二抗为羊抗大鼠 IgG（H+L）（Thermo Fisher Scientific；目录号 A18866）。

Mice 小鼠

C57BL/6J mice, T-cell knockout mice (B6.129S7-Rag1^tm-Mom/J), PD-1 knockout mice (B6.Cg-Pdcd1^tm1.1Shr/J) and Blab/c mice were purchased from the Jackson laboratory. Six- to eight-week-old male animals were used throughout all the experiments. All the animal studies strictly followed the animal protocol approved by the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill and North Carolina State University. The number of mice used for each experiment was determined by G*Power analysis software.
C57BL/6J 小鼠、T 细胞敲除小鼠（B6.129S7-Rag1 ^tm-Mom /J）、PD-1 敲除小鼠（B6.Cg-Pdcd1 ^tm1.1Shr /J）以及 Blab/c 小鼠均购自 Jackson 实验室。所有实验均使用 6 至 8 周龄的雄性动物。所有的动物研究均严格遵循北卡罗来纳大学教堂山分校和北卡罗来纳州立大学机构动物护理和使用委员会批准的动物实验方案。每项实验所用小鼠数量由 G*Power 分析软件确定。

Preparation of aPD-1-conjugated platelets
制备抗 PD-1 抗体偶联的血小板

Murine platelets were isolated from whole mouse blood. Briefly, whole blood was collected from the C57BL/6J mice (non-terminal collection from the orbital sinus) with dipotassium ethylenediaminetetraacetic acid-treated tubes. The platelet-rich plasma (PRP) was then collected by centrifuging the whole blood at 100g for 20 min at room temperature. Thereafter, PGE1 was added to PRP at a final concentration of 1 μM, and PRP was centrifuged for another 20 min at 100 g to further get rid of red blood cells. To isolate the platelets, the PRP was centrifuged at 800g for 20 min. Next, the pellet was collected and resuspended in phosphate buffered saline (PBS) containing 1 µM PGE1. Some 20–25 ml of blood was collected from the mice and ~10 × 10⁸ platelets were isolated each time. Platelets were resuspended in 1 ml PBS with 1 µM PGE1 at the concentration of 1 × 10⁸ platelets per 100 µl PBS. For the in vitro activation of platelets, the platelet solution was centrifuged at 800g for 20 min and suspended in PBS buffer for the following activation experiments. The number of platelets was counted using a haemocytometer under a microscope.
从小鼠全血中分离出小鼠血小板。简言之，使用二钾乙二胺四乙酸处理的试管从 C57BL/6J 小鼠（通过眼眶静脉非致死性采集）收集全血。随后，在室温下以 100g 离心 20 分钟，获取富含血小板的血浆（PRP）。接着，向 PRP 中加入最终浓度为 1 μM 的 PGE1，并在 100g 下再次离心 20 分钟，以进一步去除红细胞。为分离血小板，将 PRP 在 800g 下离心 20 分钟。然后收集沉淀物，并将其重新悬浮于含有 1 µM PGE1 的磷酸盐缓冲液（PBS）中。每次从鼠中采集约 20-25 毫升血液，每次可分离出约 10 × 10^10 个血小板。血小板以 1 × 10^11 个/100 µl PBS 的浓度重新悬浮于 1 ml 含 1 µM PGE1 的 PBS 中。对于血小板的体外激活，将血小板溶液在 800g 下离心 20 分钟，并悬浮于 PBS 缓冲液中，用于后续的激活实验。通过显微镜下的血细胞计数器计数血小板数量。

To decorate aPD-1 on the surface of platelets through an SMCC linker (Sulfo-SMCC; Pierce), the free thiol groups on the surface of the platelets after incubation with Traut’s reagent were first examined using flow cytometry (1 × 10⁴ events were collected for analysis)^33,56. Briefly, 1 × 10⁶ platelets were mixed with 0.1 mg ml⁻¹ maleimeide fluorescence probe (Mal-FITC; Sigma–Aldrich) in PBS and centrifuged at 800g for 10 min after 15 min reaction. The platelets were then washed by PBS, centrifuged at 800g for 20 min and subjected to flow cytometry. Red blood cells that have been reported to have minimal free thiol groups were used as the control group here. Next, aPD-1 was reacted with Sulfo-SMCC at a molar ratio of 1:1.2 for 2 h at 4 °C. The mixture was then centrifuged in an ultrafiltration tube (molecular weight cut-off = 3 kDa) to discard the excess SMCC linker. Thereafter, SMCC-aPD-1 (0.4 pg per platelet) was added to platelets and maintained at room temperature for 1 h to obtain P–aPD-1. The excess antibodies were removed by centrifugation at 800g for 20 min. The pellet was washed and stored in PBS containing 1 μM PGE1 at room temperature before use in the experiments. To further evaluate the aPD-1 conjugation efficiency, various volumes of aPD-1 were reacted with platelets, as described above, and the resulting P–aPD-1 were centrifuged at 800g, washed with PBS and lysed using ultrasonication in 0.1% Triton buffer. The amount of aPD-1 conjugated to the platelets was measured via ELISA (Rat IgG Total ELISA Kit; eBioscience). To evaluate the effects of aPD-1 on the platelets, the stability of platelets was investigated by counting the numbers of platelets using a microscope at 24 h post-reaction. To study the aPD-1 release, 0.5 U ml⁻¹ thrombin was added to 1 × 10⁸ P–aPD-1 suspension (500 µl) to activate the platelets at 37 °C without stirring. At pre-arranged time intervals, 50 µl samples were collected and centrifuged at 800g for 20 min and the supernatant was detected by ELISA following the method described above. The non-activated P–aPD-1 was used as a control.
通过 SMCC 连接子（Sulfo-SMCC；Pierce）在血小板表面修饰 aPD-1，首先使用流式细胞术检测经 Traut 试剂处理后血小板表面的游离巯基（收集 1×10⁰事件进行分析）。简言之，将 1×10⁰血小板与 0.1 mg/ml 的马来酰亚胺荧光探针（Mal-FITC；Sigma–Aldrich）在 PBS 中混合，反应 15 分钟后以 800g 离心 10 分钟。随后，血小板用 PBS 洗涤，800g 离心 20 分钟，并进行流式细胞术分析。已知游离巯基极少的红细胞作为对照组。接下来，aPD-1 与 Sulfo-SMCC 以 1:1.2 的摩尔比在 4°C 下反应 2 小时。混合物随后在超滤管（分子量截留 3 kDa）中离心，以去除多余的 SMCC 连接子。此后，将 SMCC-aPD-1（每血小板 0.4 pg）加入血小板中，在室温下保持 1 小时，得到 P–aPD-1。过量的抗体通过 800g 离心 20 分钟去除。沉淀物洗涤后，储存在含 1 μM PGE1 的 PBS 中，于室温下备用，用于实验。 为了进一步评估 aPD-1 的偶联效率，按照上述方法，将不同体积的 aPD-1 与血小板反应，所得的 P–aPD-1 在 800g 下离心，用 PBS 洗涤并用 0.1% Triton 缓冲液超声裂解。通过 ELISA（Rat IgG Total ELISA Kit；eBioscience）测定偶联到血小板上的 aPD-1 的量。为了评估 aPD-1 对血小板的影响，通过在反应后 24 小时使用显微镜计数血小板数量来研究血小板的稳定性。为了研究 aPD-1 的释放情况，向 1 × 10^5 P–aPD-1 悬浮液（500 µl）中加入 0.5 U/ml 凝血酶，在 37°C 下不搅拌的情况下激活血小板。在预设的时间间隔内，收集 50 µl 样品并在 800g 下离心 20 分钟，随后按照上述方法通过 ELISA 检测上清液。未激活的 P–aPD-1 用作对照。

To further investigate the functionality of platelets after conjugation of aPD-1, the collagen-binding and platelet aggregation studies were performed. Murine collagen type I/III (Bio-Rad) was reconstituted to a concentration of 2.0 mg ml⁻¹, added to a confocal dish and incubated overnight at 4 °C. The plate was further blocked with 2% bovine serum albumin for 2 h and washed with PBS for the collagen-binding study. The blank plate was only blocked but without the addition of collagen. 1 × 10⁷ P–aPD-1 or platelets stained with wheat germ agglutinin Alexa Fluor 594 were added to plate with or without collagen pre-coating. The plate was then washed with PBS after 1 min of incubation and subjected to confocal imaging. For the aggregation studies, P–aPD-1 or platelets labelled with wheat germ agglutinin Alexa Fluor 594 were incubated in complete medium with 0.5 U ml⁻¹ thrombin, then subjected to confocal laser scanning microscopy (LSM 710; Zeiss).
为深入探究 aPD-1 偶联后血小板的功能，进行了胶原蛋白结合实验和血小板聚集实验。使用 Bio-Rad 公司的小鼠 I/III 型胶原蛋白，配制成 2.0 mg/ml 的浓度，加入共聚焦培养皿中，于 4°C 下过夜孵育。随后，培养皿用 2%牛血清白蛋白封闭 2 小时，并用 PBS 洗涤，以备胶原蛋白结合实验。空白对照组仅进行封闭处理，未添加胶原蛋白。将 1×10^6 个 P–aPD-1 或经麦胚凝集素 Alexa Fluor 594 标记的血小板分别加入预涂或未涂胶原蛋白的培养皿中，孵育 1 分钟后用 PBS 洗涤，进行共聚焦显微成像。在聚集实验中，P–aPD-1 或麦胚凝集素 Alexa Fluor 594 标记的血小板在含 0.5 U/ml 凝血酶的完全培养基中孵育，随后采用蔡司 LSM 710 共聚焦激光扫描显微镜进行观察。

To visualize the decoration of aPD-1 on the surface of platelets, aPD-1 was conjugated with fluorescein isothiocyanate (FITC) and platelets were stained with rhodamine B. The P–aPD-1 was then observed via confocal laser scanning microscopy. To further characterize the P–aPD-1, the P–aPD-1 was stained with phycoerythrin (PE)-stained rat anti-IgG antibody and subjected to flow cytometry (1 × 10⁴ events were collected for analysis). The unstained platelet and simple mixture of platelet and isotype control antibody (anti-human CD8 antibody; BioLegend, clone: SK1) were used as controls.
为观察 aPD-1 在血小板表面的修饰情况，将 aPD-1 与异硫氰酸荧光素（FITC）结合，并用罗丹明 B 对血小板进行染色，随后通过共聚焦激光扫描显微镜观察 P–aPD-1。为进一步表征 P–aPD-1，采用藻红蛋白（PE）标记的鼠抗 IgG 抗体对 P–aPD-1 进行染色，并进行流式细胞术分析（收集 1×10⁶个事件用于分析）。未染色血小板及单纯混合血小板与同型对照抗体（抗人 CD8 抗体；BioLegend，克隆号：SK1）作为对照。

Preparation of S–P–aPD-1 assembly
S–P–aPD-1 组装体的制备

HSCs were isolated from the femur and tibia of C57BL/6J mice and cultured in 40 µM Ac₄GalNAz (Thermo Fisher Scientific) containing medium for 72 h. The HSCs were stained with APC-anti-c-kit antibody, FITC-anti-CD34 antibody and PE-anti-CD38 antibody and subjected to flow cytometry for analysis (1 × 10⁴ events were collected for analysis). To detect the presence of azide groups on the surface of HSCs, 1 × 10⁶ HSCs were incubated with PBS containing 50 µM Copper(ii)-TBTA complex, 2 mM sodium ascorbate and 25 µM FAM alkyne in the dark. After 15 min, the resulting HSCs were washed with PBS three times and subjected to flow cytometry analysis (1 × 10⁴ events were collected for analysis) and confocal observation. To functionalize the platelets with triple bonds for the click reaction, 1 × 10⁶ platelets were treated with 20 µM DBCO-PEG₄-NHS ester for 30 min at room temperature and then decorated with aPD-1, as described above. To examine the presence of triple bonds on the platelets, the resulting platelets were reacted with 20 µM azide-FITC probe for 15 min in the dark and then subjected to flow cytometry analysis (5 × 10³ events were collected for analysis).
从 C57BL/6J 小鼠的股骨和胫骨中分离出造血干细胞（HSCs），并在含有 40 µM Ac ₄ GalNAz（赛默飞世尔科技）的培养基中培养 72 小时。HSCs 用 APC-抗 c-kit 抗体、FITC-抗 CD34 抗体和 PE-抗 CD38 抗体染色，并通过流式细胞术进行分析（收集 1 × 10 ⁴ 个事件进行分析）。为检测 HSCs 表面是否存在叠氮基团，将 1 × 10 ⁶ HSCs 与含 50 µM 铜(II)-TBTA 复合物、2 mM 抗坏血酸钠和 25 µM FAM 炔的 PBS 孵育 15 分钟。随后，HSCs 用 PBS 洗涤三次，并进行流式细胞术分析（收集 1 × 10 ⁴ 个事件进行分析）和共聚焦显微镜观察。为使血小板通过点击反应功能化引入三键，将 1 × 10 ⁶ 血小板在室温下用 20 µM DBCO-PEG ₄ -NHS 酯处理 30 分钟，然后按照上述方法修饰 aPD-1。为检测血小板上是否存在三键，将所得血小板与 20 µM 叠氮-FITC 探针在暗处反应 15 分钟，随后进行流式细胞术分析（收集 5 × 10 ³ 个事件进行分析）。

For conjugation of platelets to HSCs, 1 × 10⁷ DBCO-functionalized platelets stained with rhodamine B were added to 1 × 10⁷ Ac₄GalNAz-treated HSCs and incubated for 45 min at 37 °C. Thereafter, excess azide-PEG (50 µM) was added to the HSC/platelet mixture and incubated for an additional 15 min to quench the additional DBCO on the surface of the platelets. After centrifugation at 400g for 5 min, the resulting platelet–HSC assembly was observed via confocal microscopy.
为了将血小板与造血干细胞（HSCs）结合，将 1 × 10^0 个罗丹明 B 染色的 DBCO 功能化血小板加入到 1 × 10^1 个经 Ac^2 GalNAz 处理的 HSCs 中，并在 37°C 下孵育 45 分钟。随后，向 HSC/血小板混合物中加入过量的叠氮化 PEG（50 µM），再孵育 15 分钟以淬灭血小板表面多余的 DBCO。经 400g 离心 5 分钟后，通过共聚焦显微镜观察所得的血小板-HSC 复合体。

To investigate the conjugated number of platelets on the surface of HSCs, the reacted amount of platelets stained with rhodamine B was increased from 1:1 (platelets:HSCs) to 8:1. After addition of excess azide-PEG, the resulting S–P–aPD-1 was washed with PBS three times and subjected to confocal microscopy. The percentage of conjugated platelets was quantified on the HSCs by counting 200 S–P–aPD-1 assemblies under confocal observation at different reaction ratios.
为了研究 HSCs 表面血小板的结合数量，将罗丹明 B 染色的血小板与 HSCs 的反应比例从 1:1（血小板:HSCs）增加至 8:1。加入过量叠氮 PEG 后，所得的 S–P–aPD-1 用 PBS 洗涤三次，并进行共聚焦显微镜观察。通过在不同反应比例下对 200 个 S–P–aPD-1 复合体进行共聚焦观察计数，定量了 HSCs 上结合血小板的比例。

For SEM characterization, S–P–aPD-1 was first fixed with 3.5% glutaraldehyde for 4 h, washed with PBS three times and then dehydrated with ethanol in a graded series (30, 50, 70, 85 and 90% each time for 15 min and 100% twice for 30 min), then treated with tert-butanol. After drying under vacuum, S–P–aPD-1 was coated with gold/palladium and examined by SEM (Verios 460L).
用于扫描电镜（SEM）表征时，S–P–aPD-1 首先用 3.5%戊二醛固定 4 小时，经 PBS 洗涤三次后，按梯度系列（30%、50%、70%、85%、90%各 15 分钟，100%两次各 30 分钟）用乙醇脱水，随后用叔丁醇处理。真空干燥后，S–P–aPD-1 表面镀上金/钯，并通过 SEM（Verios 460L）进行检测。

To test the bioactivity of platelets after conjugation with HSCs, S–P–aPD-1 was treated with 0.5 U ml⁻¹ thrombin for 30 min at 37 °C without stirring, then centrifuged at 400g for 5 min. The supernatant was collected and stained with 2% uranyl acetate, then observed with a transmission electron microscope (JEOL 2000FX; Hitachi). The functionality of platelets on S–P–aPD-1 was also detected by examination of key protein expression on the platelets. Briefly, S–P–aPD-1 was stained with various rat anti-mouse antibodies (CD61, CD41, CD9, CD36, CD62P and CD154; BioLegend) and analysed by flow cytometry (1 × 10⁴ events were collected for analysis). The platelets were also stained with anti-human CD8 antibody for isotype control. CD62P and CD154 detection was performed after the addition of thrombin for platelet activation.
为了检测血小板与造血干细胞结合后的生物活性，将 S–P–aPD-1 在 37°C 下不搅拌的情况下用 0.5 U/ml 的凝血酶处理 30 分钟，随后以 400g 离心 5 分钟。收集上清液并用 2%醋酸铀染色，然后通过透射电子显微镜（JEOL 2000FX；Hitachi）进行观察。通过检测血小板上的关键蛋白表达，也评估了 S–P–aPD-1 中血小板的功能。简言之，S–P–aPD-1 用多种大鼠抗小鼠抗体（CD61、CD41、CD9、CD36、CD62P 和 CD154；BioLegend）染色，并通过流式细胞术分析（收集 1 × 10^1 个事件进行分析）。血小板还用抗人 CD8 抗体进行同型对照染色。在加入凝血酶以激活血小板后，进行 CD62P 和 CD154 的检测。

To test the viability of the HSCs after incubation of Ac₄GalNAz, they were seeded in a 96-well plate at a density of 1 × 10⁴ cells well⁻¹ and incubated with 10, 20, 40 or 80 µM Ac₄GalNAz for 72 h. Then, we added 20 μl of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) solution (5 mg ml⁻¹). After 4 h incubation, the plate was centrifuged at 500g and the medium was replaced with 150 μl of dimethyl sulfoxide. The absorbance was measured at a wavelength of 570 nm using a microplate reader. To measure the cell viability of HSCs conjugated with platelets, the HSCs were incubated with different amounts of platelets and cultured for another 48 h. Then, the cell viability was investigated by MTT assay as described above.
为了检测 HSCs 在孵育 Ac ₄ GalNAz 后的存活能力，它们以 1 × 10 ⁴ 细胞/孔的密度接种于 96 孔板中，并与 10、20、40 或 80 µM 的 Ac ₄ GalNAz 共同孵育 72 小时。随后，加入 20 μl 的 3-(4,5-二甲基噻唑-2-基)-2,5-二苯基四氮唑溴盐（MTT）溶液（5 mg/ml ⁻¹ ）。孵育 4 小时后，将板以 500g 离心，并更换为 150 μl 二甲基亚砜。使用酶标仪在 570 nm 波长下测量吸光度。为测定与血小板结合的 HSCs 的细胞活性，HSCs 与不同数量的血小板共孵育，并继续培养 48 小时。随后，按照上述方法通过 MTT 实验检测细胞活性。

For conjugation of aPD-1 on the surface of HSCs, 1 × 10⁶ HSCs were reacted with SMCC-aPD-1 stained with rhodamine B at 4 °C for 1 h. After centrifugation at 400g for 5 min, S–aPD-1 was incubated at 37 °C for 2 and 6 h. After staining with endo-lysosomal tracker green, Hoechst 33258 and trypan blue, S–aPD-1 was subjected to confocal microscopy for observation.
为了将 aPD-1 共轭于造血干细胞（HSCs）表面，将 1 × 10^6 个 HSCs 与用罗丹明 B 染色的 SMCC-aPD-1 在 4°C 下反应 1 小时。反应后，在 400g 下离心 5 分钟，随后将 S–aPD-1 在 37°C 下分别孵育 2 小时和 6 小时。经过内溶酶体追踪绿色染色、Hoechst 33258 和台盼蓝染色后，S–aPD-1 通过共聚焦显微镜进行观察。

In vivo pharmacokinetics and biodistribution
体内药代动力学与生物分布

All animals were treated in accordance with the Guide for the Care and Use of Laboratory Animals, approved by the Institutional Animal Care and Use Committee of the University of North Carolina at Chapel Hill and North Carolina State University. Six mice were randomly divided into two groups and intravenously injected with Cy5.5-aPD-1 (aPD-1: 1 mg per kg), S–P–aPD-1-Cy5.5 (aPD-1: 1 mg per kg; HSCs/platelets: 1 × 10⁸; in 200 μl PBS for each mouse). At predetermined time points (1, 2, 4, 8, 12, 24 and 48 h), a 10 μl blood sample was collected from the tail, diluted in 100 μl water and subjected to sonication. The aPD-1 was measured by fluorescence of Cy5.5. The concentration of aPD-1 in 0.5 min was set as 100%. Furthermore, to investigate the in vivo clearance of free aPD-1 and Cy5.5-aPD-1, six mice were randomly divided into two groups and intravenously injected with aPD-1 and Cy5.5-aPD-1 at an aPD-1 dose of 1 mg per kg. At predetermined time points (0.083, 0.5, 1, 3, 6, 12, 24 and 48 h), a 20 μl blood sample was collected and centrifuged for further ELISA assay. For ELISA assay, the 96-well plate was pre-incubated with 2 µg ml⁻¹ mPD-1 protein overnight and blocked with 2% bovine serum albumin for 2 h. Then, the blood sample was diluted and added to the platelets for incubation for 1 h. For the following procedures, we followed the manufacturer’s procedure relating to the rat IgG ELISA detection kit. For the in vivo biodistribution study, the mice were intravenously injected with PBS, Cy5.5-labelled free aPD-1, S–aPD-1, P–aPD-1, a mixture of HSCs and aPD-1, and S–P–aPD-1 at a Cy5.5 dose of 30 nmol per kg. After 6 h, the bones and other major organs were taken out and fluorescence images were recorded using an IVIS Spectrum system (PerkinElmer). The fluorescence intensities of regions of interest were analysed by Living Image Software. For bone marrow accumulation of HSC–platelet assemblies, the leukaemia-bearing mice were intravenously injected with HSC–platelet assemblies (HSCs were stained with FITC, while platelets were stained with rhodamine B; HSC–platelets: 5 × 10⁷). At 12 h post-injection, the mice were euthanized. The bones were collected and bone tissues were fixed in the 10% formalin. After 48 h, the bones were decalcified in ethylenediaminetetraacetic acid solution for 2 d, followed by incubation in 30% sucrose. Thereafter, the bone tissues were frozen in O.C.T. medium for sectioning. After staining with Hoechst, the bone slides were subjected to confocal microscopy for observation. To observe the potential generation of PMP in the bone marrow, the rhodamine B-labelled platelets, PMPs and HSC–platelets were injected intravenously into the leukaemia-bearing mice. After 24 h, the bone tissues were collected, sectioned as described above and subjected to confocal microscopy observation. Quantification of the fluorescence signal was performed using ImageJ software.
所有动物均按照《实验动物护理和使用指南》进行处理，该指南已获得北卡罗来纳大学教堂山分校和北卡罗来纳州立大学机构动物护理和使用委员会的批准。六只小鼠被随机分为两组，分别静脉注射 Cy5.5-aPD-1（aPD-1：每公斤 1 毫克），S–P–aPD-1-Cy5.5（aPD-1：每公斤 1 毫克；造血干细胞/血小板：1 × 10^6；每只小鼠注射 200 微升 PBS）。在预定的时间点（1、2、4、8、12、24 和 48 小时），从尾部采集 10 微升血液样本，稀释于 100 微升水中并进行超声处理。通过 Cy5.5 的荧光强度测定 aPD-1 的浓度，将 0.5 分钟的 aPD-1 浓度设定为 100%。此外，为了研究游离 aPD-1 和 Cy5.5-aPD-1 在体内的清除情况，六只小鼠被随机分为两组，分别静脉注射 aPD-1 和 Cy5.5-aPD-1，aPD-1 剂量为每公斤 1 毫克。在预定的时间点（0.083、0.5、1、3、6、12、24 和 48 小时），采集 20 微升血液样本并离心，以进行后续的 ELISA 检测。 在 ELISA 检测中，96 孔板预先与 2 µg ml⁻¹ mPD-1 蛋白孵育过夜，然后用 2%牛血清白蛋白封闭 2 小时。随后，血液样本稀释后加入血小板中，孵育 1 小时。后续步骤按照大鼠 IgG ELISA 检测试剂盒的说明书进行。在体内生物分布研究中，小鼠静脉注射了 PBS、Cy5.5 标记的游离 aPD-1、S–aPD-1、P–aPD-1、HSC 与 aPD-1 的混合物以及 S–P–aPD-1，Cy5.5 剂量为每公斤 30 nmol。6 小时后，取出骨骼及其他主要器官，使用 IVIS Spectrum 系统（PerkinElmer）记录荧光图像。感兴趣区域的荧光强度通过 Living Image 软件进行分析。为研究 HSC–血小板复合物在骨髓中的积累，白血病小鼠静脉注射了 HSC–血小板复合物（HSC 用 FITC 染色，血小板用罗丹明 B 染色；HSC–血小板比例为 5 × 10²）。注射后 12 小时，处死小鼠。收集骨骼，骨组织固定于 10%甲醛中。 48 小时后，骨组织在乙二胺四乙酸溶液中脱钙处理 2 天，随后在 30%蔗糖中孵育。之后，骨组织在 O.C.T.介质中冷冻用于切片。经 Hoechst 染色后，骨切片通过共聚焦显微镜进行观察。为观察骨髓中可能产生的 PMP，将罗丹明 B 标记的血小板、PMP 及 HSC-血小板静脉注射入白血病小鼠体内。24 小时后，收集骨组织，按上述方法切片，并进行共聚焦显微镜观察。荧光信号的定量分析采用 ImageJ 软件完成。

In vivo leukaemia treatment
体内白血病治疗

To build a leukaemia model, 1 × 10⁶ luciferase-tagged C1498 cells were intravenously injected into mice. The expression of PD-L1 on C1498 cells was analysed by flow cytometry after staining with anti-PD-L1 antibody (5 × 10⁴ events were collected for analysis). After 7 d, the 64 mice were randomly divided into 8 groups and intravenously administered with saline, HSCs, platelets, free aPD-1, S–aPD-1, P–aPD-1, S + P–aPD-1 or S–P–aPD-1 (5 × 10⁷ cells) at aPD-1 dose of 0.5 mg per kg via the tail vein. Non-leukaemia-bearing mice were injected with S–P–aPD-1 (5 × 10⁷ cells) at aPD-1 dose of 0.5 mg per kg, also via the tail vein. We further added an S–P–rat-IgG group and treatment groups receiving daily administration of aPD-1 at an IgG dose of 0.5 mg per kg and an aPD-1 dose of 0.25 mg per kg. The treatment was repeated every other day three times. The growth of leukaemia was monitored by detection of bioluminescence signals from C1498 cells. We used D-Luciferin (Xenogen) as a substrate for luciferase and each mouse was injected intraperitoneally with D-Luciferin at a dose of 150 mg per kg in 100 µl PBS. Bioluminescence images were collected after 5 min injection of D-Luciferin with an IVIS Spectrum Imaging System (PerkinElmer) and the acquisition time of the bioluminescence signal was 5 min. The bioluminescence signals were recorded at 1, 2 and 3 weeks after injection of the C1498 cells. Living Image Software version 4.3.1 (PerkinElmer) was used to quantitate the bioluminescence signal. To correct the background bioluminescence, we subtracted signals acquired from leukaemia-free mice (injected with D-Luciferin). After 3 weeks, a 100 μl blood sample was collected and lysed with red blood cell lysis buffer. The remaining cells were subjected to flow cytometry for investigation of the numbers of C1498 cells in the peripheral blood (2 × 10⁴ events were collected for the flow cytometry assay). The survival time of each mouse was recorded and the spleen was weighed and imaged. The following euthanasia criteria were used in leukaemia-bearing mice: (1) diminished activity and hunched, with a rough hair coat (lack of grooming); (2) moribund animals; (3) muscle atrophy or emaciation; (4) abnormal distension; (5) incontinence or prolonged diarrhoea (longer than 48 h); and (6) neurological disorders (seizures, circling). The main tissues (heart, liver, spleen, lung, liver and bone) were taken out for the haematoxylin and eosin staining. The slides were observed using an optical microscope (DM5500B; Leica).
为了构建白血病模型，将 1 × 10^6 个携带荧光素酶标签的 C1498 细胞通过静脉注射入小鼠体内。C1498 细胞上 PD-L1 的表达通过流式细胞术分析，使用抗 PD-L1 抗体进行染色（收集 5 × 10^5 个事件进行分析）。7 天后，将 64 只小鼠随机分为 8 组，并通过尾静脉分别注射生理盐水、造血干细胞（HSCs）、血小板、游离的抗 PD-1 抗体（aPD-1）、血小板修饰的 aPD-1（S–aPD-1）、HSCs 修饰的 aPD-1（P–aPD-1）、S + P–aPD-1 或 S–P–aPD-1（5 × 10^6 个细胞），aPD-1 剂量为每公斤体重 0.5 毫克。未患白血病的小鼠则通过尾静脉注射 S–P–aPD-1（5 × 10^7 个细胞），aPD-1 剂量同样为每公斤体重 0.5 毫克。此外，我们还增加了一个 S–P–rat-IgG 组，以及每日接受 0.5 毫克/公斤 IgG 剂量和 0.25 毫克/公斤 aPD-1 剂量的治疗组。治疗每两天重复三次。通过检测 C1498 细胞发出的生物荧光信号来监测白血病的生长情况。我们使用 D-荧光素（Xenogen）作为荧光素酶的底物，每只小鼠腹腔注射 150 毫克/公斤的 D-荧光素，溶于 100 微升 PBS 中。 生物发光图像在注射 D-Luciferin 后 5 分钟使用 IVIS Spectrum 成像系统（PerkinElmer）采集，生物发光信号的采集时间为 5 分钟。生物发光信号在注射 C1498 细胞后 1、2 和 3 周进行记录。Living Image 软件版本 4.3.1（PerkinElmer）用于定量生物发光信号。为校正背景生物发光，我们从无白血病小鼠（注射了 D-Luciferin）中获取的信号中减去背景信号。3 周后，采集 100 μl 血液样本并用红细胞裂解缓冲液处理。剩余细胞通过流式细胞术检测外周血中 C1498 细胞的数量（流式细胞术检测采集了 2 × 10^4 个事件）。记录每只小鼠的生存时间，并称量和拍摄脾脏图像。 在携带白血病的实验小鼠中，采用以下安乐死标准：(1) 活动减少、弓背、毛发粗糙（缺乏梳理）；(2) 濒死状态；(3) 肌肉萎缩或消瘦；(4) 异常肿胀；(5) 失禁或持续腹泻（超过 48 小时）；(6) 神经功能障碍（抽搐、转圈）。主要组织（心脏、肝脏、脾脏、肺、肝脏和骨骼）被取出进行苏木精和伊红染色。使用光学显微镜（DM5500B；徕卡）观察载玻片。

To investigate the cellular mechanism underlying the treatment efficacy of S–P–aPD-1, T-cell knockout mice (rag^−/−), CD8⁺ T-cell depleted mice (CD8^−/−) and PD-1 knockout mice (PD^−/−) were injected with PBS, free aPD-1 and S–P–aPD-1 at an aPD-1 concentration of 0.5 mg per kg. The survival time of each mouse was recorded. To deplete CD8⁺ T cells, C57BL/6J mice were injected with 500 µg of anti-CD8 monoclonal antibodies (clone YTS169.4; Bio X Cell) intraperitoneally every 72 h, beginning 2 d before S–P–aPD-1 treatment, for the duration of the treatment. Depletion was confirmed by flow cytometry analysis of T cells isolated from the peripheral blood (CytoFLEX; Beckman Coulter) (5 × 10⁴ events were collected for the flow cytometry assay). All flow cytometry data were analysed using CytExpert and Flowjo software.
为了探究 S–P–aPD-1 治疗效果的细胞机制，将 T 细胞缺失小鼠（rag ^−/− ）、CD8 ⁺ T 细胞耗竭小鼠（CD8 ^{− /−} ）以及 PD-1 缺失小鼠（PD ^−/− ）分别注射 PBS、游离 aPD-1 和 S–P–aPD-1，aPD-1 浓度为每公斤 0.5 毫克。记录每只小鼠的存活时间。为耗竭 CD8 ⁺ T 细胞，C57BL/6J 小鼠在 S–P–aPD-1 治疗前 2 天开始，每隔 72 小时腹腔注射 500 微克抗 CD8 单克隆抗体（克隆号 YTS169.4；Bio X Cell），持续至治疗结束。通过流式细胞术分析外周血中分离的 T 细胞（使用 CytoFLEX；Beckman Coulter）确认耗竭情况（流式细胞术检测收集 5×10 ⁴ 个事件）。所有流式细胞术数据均使用 CytExpert 和 Flowjo 软件进行分析。

For the leukaemia re-challenge study, C57BL/6J mice were injected with 1 × 10⁶ C1498 cells 80 d after the first injection of C1498 cells. Bioluminescence signals were then monitored weekly and the survival time was recorded.
为了进行白血病再挑战研究，C57BL/6J 小鼠在首次注射 C1498 细胞后的第 80 天被注射了 1 × 10^6 个 C1498 细胞。随后每周监测生物发光信号，并记录生存时间。

To build another leukaemia model, 1 × 10⁶ WEHI-3 cells were intravenously injected into mice. After 7 d, the 64 mice were randomly divided into 8 groups and intravenously administered with saline, HSCs, platelets, free aPD-1, S–aPD-1, P–aPD-1, S + P–aPD-1 or S–P–aPD-1 (5 × 10⁷ cells) at an aPD-1 concentration of 0.5 mg per kg via the tail vein. The survival time of each mouse was recorded and the spleen was weighted.
为了建立另一个白血病模型，将 1 × 10^6 个 WEHI-3 细胞通过静脉注射到小鼠体内。7 天后，将 64 只小鼠随机分为 8 组，并通过尾静脉分别静脉注射生理盐水、造血干细胞（HSCs）、血小板、游离的 aPD-1、S–aPD-1、P–aPD-1、S + P–aPD-1 或 S–P–aPD-1（5 × 10^7 个细胞），aPD-1 浓度为每公斤体重 0.5 毫克。记录每只小鼠的生存时间并称量脾脏重量。

Cytokine and chemokine detection and T-cell analysis
细胞因子和趋化因子检测与 T 细胞分析

The plasma levels of multiple cytokines and chemokines were measured using Luminex-based detection. Peripheral blood was harvested at day 12 and centrifuged at 500g for 10 min. The supernatant was aliquoted and stored at −80 °C until analysis. Samples were diluted with Luminex assay buffer following manufacturer’s instructions. Cytokine values were z-score normalized per sample before clustering. Samples were divided into four groups arbitrarily, based on their cytokine profile by k-means clustering, using the pheatmap function. To investigate the PD-1 expression in T cells in the bone marrow, the bone marrow was collected and stained with anti-CD8 and anti-PD-1 antibodies (5 × 10⁴ events were collected for analysis). The anti-human CD8 antibody was used as an isotype control antibody. For T-cell analysis, peripheral blood and bone marrow were harvested at day 12. The blood sample was lysed first with red blood cell lysis buffer and then stained with anti-CD3, anti-CD8, anti-CD4, anti-Granzyme B, anti-CD44, anti-CD25, anti-CD69 and anti-IFNγ antibodies for 30 min. The Granzyme B staining was performed according to BioLegend’s intracellular staining protocol. To count CD3⁺ T-cell numbers, 150 µl blood was lysed first using red blood cell lysis buffer and subjected to flow cytometry analysis after staining with APC-anti-CD3 antibody with the addition of 50 µl counting beads. We collected 1 × 10⁵ events for CD8 and CD4 analysis, and 5 × 10⁴ events for IFNγ analysis. For memory T-cell analysis, the splenocytes were taken out of the saline- and S–P–aPD-1-treated mice after 30 d and stained with anti-CD8, anti-CD44, anti-CD62L and anti-CD122 antibody. Thereafter, the stained cells were subjected to flow cytometry for analysis (5 × 10⁴ events were collected for analysis). To detect T-cell exhaustion in PD^−/− mice, the lymphocytes of lymph nodes from normal mice and PD^−/− mice were collected, stained with anti-CD8, anti-TIM-3 and anti-LAG-3, and subjected to flow cytometry analysis (5 × 10⁴ events were collected for analysis). To analyse T cells from the spleen and lymph nodes, tissues were first mechanically disrupted and the cells were filtered through a 40 µm strainer for further analysis.
使用基于 Luminex 的检测方法测定了多种细胞因子和趋化因子的血浆水平。在第 12 天采集外周血，并以 500g 离心 10 分钟。将上清液分装并储存于−80°C，直至分析。按照制造商的说明，用 Luminex 检测缓冲液稀释样品。在进行聚类分析前，细胞因子值按样品进行了 z-score 标准化。根据细胞因子谱，通过 k-means 聚类使用 pheatmap 函数，将样品任意分为四组。为了研究骨髓中 T 细胞的 PD-1 表达情况，收集骨髓并用抗 CD8 和抗 PD-1 抗体染色（收集 5×10 ⁴ 个事件进行分析）。抗人 CD8 抗体作为同型对照抗体使用。对于 T 细胞分析，在第 12 天采集外周血和骨髓。首先用红细胞裂解缓冲液裂解血液样本，然后与抗 CD3、抗 CD8、抗 CD4、抗 Granzyme B、抗 CD44、抗 CD25、抗 CD69 和抗 IFNγ抗体共同孵育 30 分钟。Granzyme B 染色按照 BioLegend 的细胞内染色方案进行。 为计数 CD3 ⁺ T 细胞数量，首先使用红细胞裂解缓冲液处理 150 µl 血液，随后加入 50 µl 计数珠并使用 APC-抗 CD3 抗体染色，进行流式细胞术分析。我们采集了 1 × 10 ⁵ 个事件用于 CD8 和 CD4 分析，5 × 10 ⁴ 个事件用于 IFNγ分析。对于记忆 T 细胞分析，在盐水和 S–P–aPD-1 处理 30 天后取出脾细胞，并用抗 CD8、抗 CD44、抗 CD62L 和抗 CD122 抗体染色。随后，对染色细胞进行流式细胞术分析（收集 5 × 10 ⁴ 个事件进行分析）。为检测 PD ^−/− 小鼠中 T 细胞耗竭情况，采集正常小鼠和 PD ^−/− 小鼠的淋巴结淋巴细胞，用抗 CD8、抗 TIM-3 和抗 LAG-3 抗体染色，并进行流式细胞术分析（收集 5 × 10 ⁴ 个事件进行分析）。为分析脾脏和淋巴结中的 T 细胞，首先对组织进行机械破碎，细胞通过 40 µm 筛网过滤以供进一步分析。

Statistics 统计学

All results are presented as means ± s.d. Statistical analysis was evaluated using GraphPad Prism (7.0). The log-rank test was performed for statistical analysis of survival time, and one-way analysis of variance (ANOVA) followed by Tukey’s honest significant difference (HSD) post hoc test for multiple comparisons was performed for other statistical analyses. The differences between experimental groups and control groups were considered statistically significant at P < 0.05. *P < 0.05; **P < 0.01; and ***P < 0.001.
所有结果以平均值±标准差表示。统计分析采用 GraphPad Prism（7.0）软件进行。生存时间的统计分析采用 log-rank 检验，其他统计分析则采用单因素方差分析（ANOVA），随后进行 Tukey’s 诚实显著性差异（HSD）事后检验以进行多重比较。实验组与对照组之间的差异在 P < 0.05 时被认为具有统计学显著性。*P < 0.05；**P < 0.01；***P < 0.001。

Reporting Summary 报告摘要

Further information on research design is available in the Nature Research Reporting Summary linked to this article.
有关研究设计的更多信息，请参阅与本文关联的《自然研究报告》摘要。