Multistage Self-Assembled Nanomaterials for Cancer Immunotherapy

Advances in nanotechnology have brought innovations to cancer therapy. Nanoparticle-based anticancer drugs have achieved great success from bench to bedside. However, insufficient therapy efficacy due to various physiological barriers in the body remains a key challenge. To overcome these biological barriers and improve the therapeutic efficacy of cancers, multistage self-assembled nanomaterials with advantages of stimuli-responsiveness, programmable delivery, and immune modulations provide great opportunities. In this review, we describe the typical biological barriers for nanomedicines, discuss the recent achievements of multistage self-assembled nanomaterials for stimuli-responsive drug delivery, highlighting the programmable delivery nanomaterials, in situ transformable self-assembled nanomaterials, and immune-reprogramming nanomaterials. Ultimately, we perspective the future opportunities and challenges of multistage self-assembled nanomaterials for cancer immunotherapy.


Introduction
Cancer immunotherapy is an advanced therapeutic strategy for cancers that boosts the body's own immune system to fight cancer cells and has gradually changed the paradigm of cancer therapy in clinics [1][2][3].In cancer patients, the tumor and its microenvironment usually suppress the host immune response by suppressing signaling pathways and metabolisms to escape the eliminations by immune cells such as T cells and natural killer (NK) cells [4][5][6].Multiple defensive mechanisms, including defects in immune checkpoint expression, upregulations of immune-suppressive pathways, recruitments of immunesuppressive cells, etc., are involved in the immune surveillance of cancer cells.Different approaches have been developed for cancer immunotherapy, including immune checkpoint inhibitors, adoptive cell transfer, cancer vaccines, oncolytic viruses, lymphocyte-activating cytokines, etc. [7][8][9][10].Currently, immune checkpoint inhibitors are one of the most successful therapeutics in cancer immunotherapy [11].Since Ipilimumab, a monoclonal antibody that blocks cytotoxic T lymphocyte-associated protein 4 (CTLA-4), was first approved by the U.S. Food and Drug Administration (FDA) in 2011 [12], more than 20 immune checkpoint inhibitors have been marketed in the US, Asia, and Europe.Targets of these immune checkpoint inhibitors include CTLA-4, programmed cell death 1/programmed death ligand 1 (PD-1/PD-L1), and lymphocyte activation gene 3.Despite great successes, low responsive rates and immune-related adverse effects are still the main hurdles to cancer immunotherapy in the clinic [13,14].pharmacokinetics of nanomedicines and tumor microenvironments, nanomedicines have been developed from passive targeting systems to active targeting systems.However, due to the high heterogeneity of the EPR effect in the clinic [96][97][98] and various physiological barriers in human bodies, nanocarriers still encounter low delivery efficacy and poor therapeutic effect on tumors [99][100][101].Thus, new strategies to overcome sophisticated physiological barriers are still urgent to improve the delivery efficacy of nanomedicines.
One strategy is to rational design and construct nanomaterials to deliver drugs in a multistage process to overcome multiple physiological barriers, thus achieving enhanced tumor accumulation and improved immunomodulation for cancer therapy [102][103][104][105].These multistage nanomaterials can hierarchically change features, including size, morphology, surface modification, and core formation, upon external or internal stimuli, adapting to upcoming physiological environments like regional blood flow, tumor microenvironments, and intra/intercellular environments.Compared with conventional nanocarriers, these multistage nanomaterials are more likely accommodated to tumors in sophisticated physiological environments.In the following sections, we describe the typical biological barriers for nanomedicines, summarize the recent progress of multistage self-assembled nanomaterials for stimuli-responsive drug delivery, particularly in situ self-assembled nanomaterials and immune-reprograming nanomaterials, and perspective the future opportunities and challenges for multistage self-assembled nanomaterials for cancer immunotherapy.

Physiological Barriers for Nanomedicines
The pharmacokinetics of nanomedicines are strongly affected by administration routes and their physicochemical properties, including composition, size, shape, charge, surface modification, etc. [106,107].Conventionally, an optimized nanocarrier for systemic drug delivery usually possesses several parameters [108][109][110][111], including (i) a proper size within 10-100 nm, (ii) a stealth shell such as a poly(ethylene glycol) (PEG) corona, (iii) tumortargeting ligands on the surface, and (iv) a core for loading therapeutic cargos.Compared with small molecular drugs, these nano-scaled drugs present unique pharmacokinetics after systemic administration.They can circulate long in the blood flow and are more likely to escape from excreting through the kidney and being captured by the reticuloendothelial system (RES) in the liver, lung, and spleen, resulting in enhanced tumor accumulation.Some reviews have discussed the pharmacokinetics of nanoparticles after systemic administration in detail [112][113][114][115][116][117].However, conventional nanomedicines still encounter the problems of low delivery efficacy and unsatisfied therapeutic outcomes due to the various physiological barriers in the body [118].

Physiological Barriers for Nanomedicines
Upon systemic administration, nanomedicines encounter a series of sequential barberries before successfully arriving at the tumor sites (Figure 1) [119][120][121].The rapid clearance in blood flow and uptake by the RES are the first obstacles to nanomedicines after administration, which, on average, contributes to more than 99% loss of injected nanomedicines [122,123].The major challenges of nanomedicines during blood circulation are the degradation by enzymes, uptake by the RES and mononuclear phagocyte system, and excretion by the kidney.Of note, nanomedicines in the bloodstream face the problem of the formation of protein coronas due to the coverage by serum proteins [124][125][126][127], which inactivates the targeting ability of the ligand and facilitates the uptake by macrophages in the mononuclear phagocyte system, resulting in the non-specific accumulation and side effect to health organs such as liver, spleen, and lung.In addition, the blood flow also influences the stability of nanocarriers and usually causes burst release of the payloads.Another substantial barrier to nanocarriers is the high intratumoral pressure, which is associated with interrupted blood vessels, aggressive tumor cell proliferation, stroma cells, tumor-associated fibroblasts, and the extracellular matrix, impeding the convection of nanocarriers from tumoral vessels to tumor tissues and the deep penetration of nanocarriers within tumors [128][129][130].Upon arrival at the tumor cells, cellular internalization and endosome escape are demonstrated as essential barriers for nanomedicines to approach therapeutic effects.Unfortunately, most nanomedicines possessing long blood circulation properties encounter the problem of poor uptake by the targeted cells, known as the "PEG dilemma" [131].Meanwhile, nanomedicines installed with active-targeting ligands also face the risk of off-target effects caused by the formation of protein coronas [132,133].In addition, drug resistance due to the drug efflux pumps has also proved to be a considerable obstacle for nanomedicines [134,135].These biological barriers strongly hamper the clinical translation of nanomedicines from bench to bedside.mulation and side effect to health organs such as liver, spleen, and lung.In addition, the blood flow also influences the stability of nanocarriers and usually causes burst release of the payloads.Another substantial barrier to nanocarriers is the high intratumoral pressure, which is associated with interrupted blood vessels, aggressive tumor cell proliferation, stroma cells, tumor-associated fibroblasts, and the extracellular matrix, impeding the convection of nanocarriers from tumoral vessels to tumor tissues and the deep penetration of nanocarriers within tumors [128][129][130].Upon arrival at the tumor cells, cellular internalization and endosome escape are demonstrated as essential barriers for nanomedicines to approach therapeutic effects.Unfortunately, most nanomedicines possessing long blood circulation properties encounter the problem of poor uptake by the targeted cells, known as the "PEG dilemma" [131].Meanwhile, nanomedicines installed with active-targeting ligands also face the risk of off-target effects caused by the formation of protein coronas [132,133].In addition, drug resistance due to the drug efflux pumps has also proved to be a considerable obstacle for nanomedicines [134,135].These biological barriers strongly hamper the clinical translation of nanomedicines from bench to bedside.

Passive and Active Targeting
The proposal of the EPR effect by Maeda in the 1980s established the base of nanomedicines [92].Passive targeting nanocarriers refer to those that rely on the EPR effect to accumulate in tumors.These nanocarriers have been developed as the first generation of nano-scaled drug delivery systems.The equipment of stealth shells, such as PEGs and zwitterionic polymers [136], as well as the precise control of size in 10-100 nm, are two important features for designing this generation of nanocarriers.Some works have especially highlighted the advantages of sub-50 nm nanoparticles for deep penetration in thick fibrotic tumor models and metastatic tumor models [137][138][139][140]. Until now, passive targeting nanocarriers have achieved great success in clinical translations.Current marketed nanoparticle-based anticancer drugs all rely on passive targeting pathways to accumulate in tumors.However, evidence has pointed out that the liposomal DOX failed to show improvements in the objective response, overall survival, and progression-free survival rates via a meta-analysis in a total of 2589 patients in the clinic [141].The unsatisfied performances of nanomedicines in the clinic are possibly due to the unspecific delivery and the highly heterogeneous EPR effect in patients [142][143][144][145]. Different patients, cancer types, and even different tumoral lesions in the same patient represent different responses to the EPR effect.To improve anticancer efficacy, ligand-installed nanocarriers

Passive and Active Targeting
The proposal of the EPR effect by Maeda in the 1980s established the base of nanomedicines [92].Passive targeting nanocarriers refer to those that rely on the EPR effect to accumulate in tumors.These nanocarriers have been developed as the first generation of nano-scaled drug delivery systems.The equipment of stealth shells, such as PEGs and zwitterionic polymers [136], as well as the precise control of size in 10-100 nm, are two important features for designing this generation of nanocarriers.Some works have especially highlighted the advantages of sub-50 nm nanoparticles for deep penetration in thick fibrotic tumor models and metastatic tumor models [137][138][139][140]. Until now, passive targeting nanocarriers have achieved great success in clinical translations.Current marketed nanoparticle-based anticancer drugs all rely on passive targeting pathways to accumulate in tumors.However, evidence has pointed out that the liposomal DOX failed to show improvements in the objective response, overall survival, and progression-free survival rates via a meta-analysis in a total of 2589 patients in the clinic [141].The unsatisfied performances of nanomedicines in the clinic are possibly due to the unspecific delivery and the highly heterogeneous EPR effect in patients [142][143][144][145]. Different patients, cancer types, and even different tumoral lesions in the same patient represent different responses to the EPR effect.To improve anticancer efficacy, ligand-installed nanocarriers for the active targeting of tumors have been developed as the second generation of nano-scale drug delivery systems.These active targeting nanocarriers rely on both the EPR effect to arrive at the tumor sites and the strong bind affinity to the specific biomarkers on targeted cancer cells and tumor vascular epithelial cells [146][147][148][149][150]. Up to now, different small molecules and biomolecules have been developed as targeting ligands [151][152][153], including folic acid [154][155][156], glucose [157][158][159][160], galactose [161], transferrin [162], antibodies [163][164][165], peptides [166,167], aptamers [168][169][170][171][172][173], etc. Notably, glucosylated nanocarriers have also been developed to cross the blood-brain barrier for drug delivery to the brain.For instance, Kataoka and coworkers reported a strategy for delivering glucosylated nanocarriers to the brain using glycemic control [174][175][176].They conjugated the PD-L1 antibody with multiple PEG chains equipped with glucose via the C6 position, leaving the OH groups at positions C1, C3, and C4 to bind with the glucose transporter-1 overexpressed in brain capillaries.The PEG chains could detach in the tumor microenvironment to reinvigorate the potency of the antibody.In the orthotopic glioblastoma tumor-bearing mouse model, the glucosylated antibody achieved ~20-fold improvement in tumor accumulation compared with native antibody, resulting in potent antitumor immune response and immunological memory.Besides blood-brain barrier crossing, the same group also achieved the delivery of small interfering RNA (siRNA) to cancer stem cells (CSCs) using glucose-installed nanoparticles [177].The glucose ligands on the sub-50 nm unimer polyion complex-assembled gold nanoparticle [178,179] specifically recognized the glucose transporter-1 overexpressed on the cell membranes of CSCs, resulting in a 2-fold higher delivery efficacy of siRNA to the orthotopic breast tumor model and 2.4-fold enhanced elimination of CSCs in tumor tissues compared with non-targeted nanoparticles.Up to now, at least 15 formulations of nanomedicines based on ligand-installed nanocarriers have been enrolled in clinical trials, including nine liposomal formulations (MM-302, C225-ILSDOX, anti-EGFR-IL-dox, SGT-53, SGT-94, Lipovaxin-MM, MCC-465, 2B3-101, and MBP-426), two bacterial-derived minicells (TargomiRs and EGFR(V)-EDV-Dox), two polymeric nanoparticles (BIND-014 and CALAA-01), one retroviral vector (Rexin-G), and one nanoparticle-based vaccine for smoking cessation (SEL-068).However, compared with the great success of antibody-drug conjugates in the clinic [180][181][182][183][184], the clinical translation of ligand-installed nanomedicines still encounters the hurdle of poor therapeutic outcomes in clinical trials.For instance, the BIND-014 was terminated in the phase II study due to its unsatisfactory therapeutic outcomes [185].One reason is the heterogeneity of prostate-specific membrane antigen expression in each individual patient.To facilitate successful clinical translations of active targeting nanomedicines, it is important to establish proper and reliable models more closely to human tumoral environments and to develop non-invasion companion nanodiagnostic systems to monitor the therapeutic outcomes.

Cold Tumors and Hot Tumors
Besides the physiological barriers for nano delivery systems, the immunosuppressive microenvironment is also a critical hurdle for cancer immunotherapy, resulting in low response rates [186].The immune checkpoint inhibitors-mediated antitumor response relies on the infiltration of T cells that recognize and kill tumor cells.However, immune checkpoint inhibitors are ineffective against cold tumors with little or no immune infiltration around cancer cells, leaving the immune system unable to attack and obliterate them effectively.This type of cancer is usually not sensitive to immunotherapy.In contrast, hot tumors have a large number of immune infiltrates around the cancer cells, and the cancer cells release signaling substances that attract immune cells and activate the immune response.This type of cancer is usually sensitive to immunotherapy.Current obstacles to treating cold tumors include the lack of effective antigens which provide targets for immunotherapy.In patients with cold tumors, there are few or no antigens on the surface of cancer cells, making it difficult for immune cells to identify and attack the cancer effectively.Therefore, it is necessary to further study the mechanism of tumor immune microenvironment in cold tumors to improve the outcomes of patients with cold tumors [187].Tuning cold tumors into hot tumors is promising to improve the therapeutic effect of immune checkpoint inhibitors [188,189].To this end, several strategies have been reported, such as promoting T cell priming and activation by increasing the antigen processing and presentation, enhancing T cell expansion by increasing the numbers of antigen-specific T cells, and augmenting T cell trafficking and infiltration by remodeling the tumor immune microenvironment, etc. [190].Nanomedicines can contribute to these processes by targeting the cancer cells, tumor immune microenvironments, and peripheral immune system [191,192], providing assessments to overcome the barrier of tumor immunosuppressive microenvironment [193,194].

Tumor Microenvironment-Responsive Nanomedicines
Tumor microenvironments refer to the surrounding environments in tumor regions, including the abnormal vasculature, acid, hypoxia, tumor-associated enzymes, redox, the extracellular matrix, stroma, intratumoral pressure, cancer stem cells, immune cells, etc. [251,252], which provide hotbeds for tumor proliferation, immune evasion, metastasis, and recurrence [253][254][255][256].It has been recognized that the normalizing and remodeling of tumor microenvironments using antiangiogenic agents and antifibrosis drugs is a promising strategy to improve the therapeutic efficacy of cancers [257][258][259][260]. Advances in this topic are reviewed in detail elsewhere [260][261][262][263]. Additionally, stimuli-responsive nanomaterials targeting tumor microenvironments have attracted increasing attention in recent decades [264][265][266][267][268]. We summarized recent examples of stimuli-responsive nanosystems for cancer therapy in Table 2.These nanomaterials not only improve the efficacy of cancer therapeutics [269][270][271][272][273] but also provide opportunities for in situ monitoring of the levels and heterogeneities of contents in the tumor microenvironment [274,275].For instance, Gao and coworkers reported a pH-responsive PEGylated polymer bearing a heptatomic ring with a tertiary amine (PC7A) that could simulate the stimulator of interferon genes (STING) pathway to enhance the cancer immunotherapy (Figure 2a) [276].They showed that innate immunity was activated via the formation of STING-PC7A biomolecular condensates [277].The polymer bound to a non-competitive pocket that differed from the natural STING ligand, resulting in a prolonged activation of the pathway.Besides the pathway activation, Liu et al. reported a pH-ultrasensitive transistor-like nanodetergent for selective cancer therapy via membranolysis [204].This membranolytic block copolymer comprised a PEG shell, a pH-responsive segment with ionizable tertiary amines, and a hydrophobic segment.It achieved a >32-fold change in cytotoxicity with a 0.1 pH change.
To monitor the tiny differences in endosome maturation pathway, Chen et al. engineered a library of pH-ultrasensitive polymeric nanophotosensitizer with a pH transition from 6.9 to 5.3.These nanophotosensitizers divided the endosome maturation into ten endocytic regions with a pH interval of 0.2, allowing the adjustment of pyroptosis-inducing activity by the targeted introduction of photodynamic oxidative stress into each region [278].Besides pH-responsive polymers, pH-low insertion peptides (pHLIP) have also garnered much interest in cancer theranostics due to their unique ability to selectively target tumor acid and transform into transmembrane α-helix within tumor cell membranes [279][280][281][282][283]. For instance, Chu et al. reported a fusion protein of pHLIP and interleukin-2 (IL-2) for antitumor immunotherapy [284].This protein was created by fusing the N-terminus of pHLIP with the C-terminus of IL-2, allowing for selective delivery to the acidic tumor microenvironment due to the low pH insertion property of pHLIP, thereby reducing the side effects.The presence of IL2 in tumor tissues promoted the proliferation of the CD8 + T and NK cells to suppress tumor growth, resulting in a 68% tumor inhibitory rate in a subcutaneous 4T1 tumor-bearing mouse model.In addition to tumor acid, Hu et al. reported a ROS-responsive delivery system for codelivery of anti-PD-L1 peptide and paclitaxel [285].The peptides were modified on the nanoparticle's surface, which could bind to the PD-L1 and induce its lysosomal degradation.The encapsulated chemodrugs were released under the overexpressed ROS in tumor cells for chemotherapy.This synergetic nanosystem promoted T cell infiltration and improved the anticancer potency for triple-negative breast cancers.Besides the single-responsive ones, multi-responsive nanosystems can further improve the specificity.For instance, Xia et al. reported a pH/enzyme-responsive nanoparticle for selective delivery of Toll-like receptor (TLR) agonists to active TLR7/8 receptor signaling at the endosomal membrane in dendritic cells (Figure 2b) [286].They synthesized a pH-sensitive PEGylated polymer in which the TLR7/8 agonist imidazoquinoline was conjugated onto the side chain via a cathepsin B-cleavable GFLG peptide linkage.This nanosystem could release imidazoquinoline under the acidic environment and cathepsin B in the endosome and activate the TLR7/8 signaling, resulting in dendritic cell maturation and antigen presentation for immunotherapy.

External Stimuli-Responsive Nanomedicines
Triggering via external stimuli is also a promising strategy for approaching the sitespecific delivery of therapeutic agents to tumors.The near-infrared (NIR) light (700-900 nm) is widely investigated in controlled drug delivery due to its relatively low scattering and tumor selectivity by local irradiation [324,325].Currently, two NIR dyes, including the indocyanine green and the methylene blue, have been approved by the FDA for tumor diagnosis and image-guided surgeries.Compared with NIR light, NIR-II light (1000-1700 nm) has recently attracted increasing attention due to its reduced photon scattering and tissue autofluorescence [326,327].For instance, Jiang et al. reported a NIR-II light activatable polymeric pronanoagonist for photothermal immunotherapy (Figure

External Stimuli-Responsive Nanomedicines
Triggering via external stimuli is also a promising strategy for approaching the sitespecific delivery of therapeutic agents to tumors.The near-infrared (NIR) light (700-900 nm) is widely investigated in controlled drug delivery due to its relatively low scattering and tumor selectivity by local irradiation [324,325].Currently, two NIR dyes, including the indocyanine green and the methylene blue, have been approved by the FDA for tumor diagnosis and image-guided surgeries.Compared with NIR light, NIR-II light (1000-1700 nm) has recently attracted increasing attention due to its reduced photon scattering and tissue autofluorescence [326,327].For instance, Jiang et al. reported a NIR-II light activatable polymeric pronanoagonist for photothermal immunotherapy (Figure 3a) [328].They construct the pronanoagonist by conjugating an immunostimulant onto an NIR-II semiconducting transducer using a thermo-responsive linker.Upon NIR-II irradiation, the photothermal effect of the pronanoagonist led to tumor ablation and immunogenic cell death, as well as the cleavage of a thermo-responsive linker to release the agonist for in situ immune activation in deep solid tumor (8 mm).Chen et al. developed a gold nano-adjuvant for the NIR-II light-triggered in situ tumor vaccine [329].The nano-adjuvant comprised a multi-branched gold nanoparticle core with a localized surface plasmon resonance peak at 1032 nm and a shell containing hyaluronidases and CpG oligodeoxynucleotides.The hyaluronidases loosened the dense extracellular matrix of tumors by degrading the hyaluronic acids to make the nano-adjuvant penetrate the tumor tissue deeply, whereas CpG oligodeoxynucleotides bound the endosomal Toll-like receptor 9 to activate the antigen presentation cells.After penetrating deeply into the tumor, the nano-adjuvant induced the immunogenic cell death (ICD) effect under the irradiation of NIR-II light, thereby inhibiting tumor growth.Besides NIR and NIR-II lights, ultrasounds are also widely used in the diagnoses and treatments of many types of diseases in the clinic due to their advantages of deep tissue penetration, thermal effects, cavitation, and acoustic radiation forces [330,331].Ultrasounds have been demonstrated to facilitate the release of drugs from liposomes, polymeric micelles, and micro/nanobubbles, improving the therapeutic efficacy [332][333][334][335].However, the clinical translation of ultrasound-assisted nanomaterials is bumpy.A phase III clinical trial of Ther-moDox, a thermosensitive liposomal DOX, combined with the high-intensity focused ultrasound for treating liver metastases tumors, did not meet the primary outcome.A post hoc analysis showed that ThermoDox was safe but invalid to increase the progression-free survival and the overall survival for the radiofrequency ablation [336].An improvement in the overall survival was observed in a subgroup of 285 patients (41% of total) who underwent ultrasound treatments for 45 min or more, suggesting an opportunity for increasing efficacy.Combining immunotherapy provides new possibilities for ultrasound-assisted nanosystems.Li et al. reported a microbubble-assisted ultrasound-guided nanoplatform for cancer immunotherapy [337].This platform was composed of a microbubbles core and a shell containing the spermine-modified dextran, 2 3 -cyclic guanosine monophosphate-adenosine monophosphate (cGAMP), and anti-CD11b antibodies.The decorated anti-CD11b antibodies enabled the nanocomplex to target antigen-presenting cells and efficiently deliver cGAMP under sonoporation, activating STING-mediated antitumor immunity.In another example, Jiang et al. reported a sono-activated semiconducting polymeric nanoagonist for immunotherapy of head and neck squamous cell carcinoma [338].The sonodynamic semiconducting polymer was conjugated with a STING agonist MSA-2 via a singlet oxygen cleavable linker.The nanoagonist could generate singlet oxygen under ultrasound, resulting in the tumor cell death for triggering the ICD effect and the release of conjugated STING agonists for in situ activation of the STING pathway in synergy.Radiofrequency ablation is a commonly used thermal therapy in clinics.Zhang et al. recently reported a bi-valent gold nanocluster with a precise Au(I) ion/Au(0) ratio for cancer immunotherapy by inducing pyroptosis via radiofrequency (Figure 3b) [339].The nanoclusters were synthesized by sequential reduction by a weak reducer lipoic acid and a strong reducer NaBH 4 with the further modification of temperature-sensitive block poly(N-isopropylacrylamideb-acrylic acid).Under radiofrequency, the nanoclusters induced pyroptosis of tumor cells and further elicited an ICD effect, resulting in an improved antitumor efficacy of αPD-1 immunotherapy to 4T1 tumor-bearing mouse model with synergy of decitabine.

Multistage Self-Assembled Nanomaterials for Cancer Immunotherapy
Compared with conventional active targeting nanocarriers that rely on the EPR effect and the ligand-receptor interaction, multistage self-assembled nanomaterials provide more chances to overcome the multiple biological barriers for cancer therapeutics.These barriers include but are not limited to the rapid clearance by the bloodstream and RES, off-target effect on tumors, intratumoral pressure, insufficient drug release, and immune-suppressive tumor microenvironment [340][341][342].The advantages of multistage self-assembled nanomaterials include (1) they are flexible for changing the formulation to improve the ability to overcome the multiple biological barriers; (2) they can enhance the targeting efficiency and retention via programmable response or in situ selfassembly, thereby reducing the side effects; and (3) they can improve the immune response by modulating or reprogramming the tumor immune environment in a synergetic manner.An early example of multi-stage nanomaterials for cancer therapy was the mesoporous Si nanovector developed by Ferrari and coworkers [343][344][345][346][347][348][349].These multistage nanocarriers were composed of nano-scaled pores and small therapeutic or diagnostic nanoparticles inside the pores, which could release out triggered by stimuli like acids.Recently, programmable nanomaterials that transform or self-assemble in situ have attracted increasing attention (Table 3) [350][351][352].These nanomaterials enhance therapeutic efficacy by increasing the targeting affinity, penetrating ability, tumor retention, cell uptake, etc.In this section, we discuss the design and construction strategies for multistage self-assembled nanomaterials reported recently (Figure 4).

Multistage Self-Assembled Nanomaterials for Cancer Immunotherapy
Compared with conventional active targeting nanocarriers that rely on the EPR effect and the ligand-receptor interaction, multistage self-assembled nanomaterials provide more chances to overcome the multiple biological barriers for cancer therapeutics.These barriers include but are not limited to the rapid clearance by the bloodstream and RES, off-target effect on tumors, intratumoral pressure, insufficient drug release, and immune-suppressive tumor microenvironment [340][341][342].The advantages of multistage self-assembled nanomaterials include (1) they are flexible for changing the formulation to improve the ability to overcome the multiple biological barriers; (2) they can enhance the targeting efficiency and retention via programmable response or in situ self-assembly, thereby reducing the side effects; and (3) they can improve the immune response by modulating or reprogramming the tumor immune environment in a synergetic manner.An early example of multi-stage nanomaterials for cancer therapy was the mesoporous Si nanovector developed by Ferrari and coworkers [343][344][345][346][347][348][349].These multistage nanocarriers were composed of nano-scaled pores and small therapeutic or diagnostic nanoparticles inside the pores, which could release out triggered by stimuli like acids.Recently, programmable nanomaterials that transform or self-assemble in situ have attracted increasing attention (Table 3) [350][351][352].These nanomaterials enhance therapeutic efficacy by increasing the targeting affinity, penetrating ability, tumor retention, cell uptake, etc.In this section, we discuss the design and construction strategies for multistage self-assembled nanomaterials reported recently (Figure 4).

Programmable Delivery Systems
To overcome the biological hurdles of nanomedicine, strategies such as PEGylation and crown detachment to increase blood circulation and tumor accumulation [364][365][366], installation of activable tumor-targeting ligands to increase tumor accumulation and cellular uptake [367][368][369], size-reduce to enhance tumor penetration [370][371][372][373][374][375][376][377], prodrug to reduce side effects [378][379][380][381][382][383], etc. have shown great potentials.One advantage of multistage programmable delivery systems is to integrate these individual strategies sequentially, corresponding to the sequential barriers after administration.For instance, Zhang et al. developed a programmable nanomedicine as an in situ cancer vaccine for cancer immunotherapy (Figure 5a) [384].The nanomedicine had a core composed of poly-[(N-2-hydroxyethyl)-aspartamide]-Pt(IV)/β-cyclodextrin and a shell composed of CpG/polyamidoamine-thioketal-adamantane (CpG/PAMAM-TK-Ad) and PEG-thioketaladamantane (PEG-TK-Ad).The CpG/PAMAM-TK-Ad and PEG-TK-Ad were attached to the core via host-guest interactions between β-cyclodextrin and adamantane.After administration, the PEG on the surface enabled long circulation in the blood, resulting in enhanced tumor accumulation.In tumor tissue, the high level of ROS triggered the detachments of PEG and CpG/PAMAM for improved cellular internalization of the core nanoparticles, which led to cell death and antigen release.The released antigen was further captured by the CpG/PAMAM, reached the tumor-draining lymph nodes, and internalized by dendritic cells.The activated dendritic cells presented antigens to T cells.The tumor antigen-specific effector T cells returned to tumor tissue to kill cancer cells.In a murine colorectal CT26 tumor model, this nanomedicine achieved a high growth-inhibitory activity of 73%.The further combination with anti-PD-L1 antibody resulted in a growth-inhibitory activity of 95%.Notably, 40% of the tumor-bearing mice were completely cured.In addition, the programmable nanosystem is also favorable for delivering combined drugs in different stages.For instance, Feng et al. reported an albumin nanoparticle for delivering PD-L1 inhibitor BMS-202 to the tumor microenvironment and cyclooxygenase-2 inhibitor celecoxib inside cells in a programmed manner [385].The nanoparticle was composed of PEGylated human serum albumin derivatives that contained pH-responsive hexamethyleneimino groups, BMS-202, and celecoxib-poly(ethyleneimine) conjugates linked by reduction-responsive disulfide linkers.Via programmed delivery, the nanosystem achieved 3.1-fold in the infiltration of CD8 + T cells to tumors and almost complete inhiation in tumor growth in subcutaneous 4T1-bearing mice.In addition, to improve the tumor penetration and synergistic effect of the nanocarriers, Wei et al. developed a bioactive selenopeptide nanomedicine for enhanced tumor chemoimmunotherapy (Figure 5b) [386].The selenopeptide was modularly designed with a tumor-targeting motif (RGD), a matrix metalloproteinase-2 (MMP-2) enzyme-cleavable linker (PLGVR), and a ROS-responsive seleno-amino acid tail.The selenopeptide was amphiphilic with a micelle structure in solution, which could encapsulate chemotherapeutics such as DOX.After systemic administration, the selenopeptide nanomedicine sequentially recognized α v β 3 integrins on the tumor cell surface for improving tumor accumulation, reduced the size induced by MMP-2 enzyme for enhancing the tumor penetration, released DOX payload quickly in tumor cells under the high level of ROS, and activated the NK cells by the oxidative metabolites of selenopeptide for immunotherapy.Due to the programmable delivery and synergistic effect, the selenopeptide nanomedicine achieved a tumor growth inhibition efficacy of 86% in an orthotopic human breast MDA-MB-231 tumor-bearing mouse model, compared with 48% for DOX solely.To make more programmable biomaterials, introducing logic gates into nanomedicine design has attracted much attention [387][388][389].For instance, Zhang et al. induced the concept of logic gates to construct a programmable polymer library [199].Different stimuli-responsive units (e.g., light-, ROS-, glutathione-, acid-, esterase-, phosphates-responses) were integrated into these polymers with logic gates and hierarchical organizations, allowing to receive disease biomarkers as inputs and site-specifically release therapeutics (e.g., kinase inhibitors, drugs, and siRNA) as outputs.

In Vivo Self-Assembled Nanomaterials
In vivo self-assembled nanomaterials can transform or self-assemble in tumor tissues in situ triggered by internal or external stimuli.The self-assembly process is governed by both thermodynamics and kinetics, which provide different assembled structures and new biological functions.Via the rational design of building blocks and control of thermodynamics and kinetics, in vivo self-assembled nanomaterials have advantages such as enhanced tumor accumulation and retention, improved tumor penetration, increased cellular internalization, etc., enabling improved immunotherapy outcomes.For instance, by tuning the self-assembly properties and kinetics of peptide building blocks, either retention in the cell membrane or rapid cell entry was achieved, resulting in different biological activities [390,391].To overcome the vaccine's hurdles on poor lymph node delivery and dendritic cell uptake, Wang et al. reported an in situ phase transitional polymeric vaccine [392].The vaccine was composed of a thermoresponsive poly(N-isopropylacrylamide) backbone modified with photothermal conversion cyanine and antigen peptide OVA 257-264 (peptide sequence: SIINFEKL) on the side chains.The low critical solution temperature of the polymer was tuned to be 40 • C.During lymph node draining, the polymers retained a small size of 24 nm.Upon arrival at the lymph node, they transformed into 483 nm-sized particles triggered by laser, resulting in improved endocytosis by lymph node-resident dendritic cells.This laser-induced dynamic size modulation strategy induced a rapid and robust immune response in the subcutaneous B16-F10-OVA melanoma tumor-bearing mouse model.In addition to cancer vaccines, in vivo self-assembled systems have also shown promising potential to improve therapeutic efficacy.For instance, Wang et al. reported an enzyme-instructed self-assembly (EISA) peptide to selectively degrade PD-L1 in cancer cells for improved cancer therapy (Figure 6a) [393].The peptide included a PD-L1targeting motif, an alkaline phosphatase (ALP)-responsive phosphorylated tyrosine, and a self-assembly module containing phenylalanines and an adamantine (peptide sequence: Ada-GG D F D F D N D Y D S D K D P D T D D D R D Q D Y D H D F).After dephosphorylation by ALP and binding to PD-L1 on tumor cell membranes, the peptide self-assembled into nanofibers around PD-L1 in situ.The resulting peptide self-assemblies and PD-L1 were further internalized and degraded by the proteasome pathway in the cytoplasm.Interestingly, the in situ self-assembly happened in ALP-overexpressed murine breast cancer 4T1 cells instead of in normal human liver LO2 cells expressing low levels of ALP.Due to the selective degradation of PD-L1, the self-assembled peptide resulted in a tumor volume decrease of 23.7% in subcutaneous 4T1-bearing mice.In addition, the in situ self-assembled peptide can also induce the aggregation of receptors to activate anticancer signaling pathways.For instance, Li et al. reported an in situ self-assembled peptide system to facilitate the aggregation of tumor-specific transmembrane Eph receptor A2 (EphA2) for converting cold tumors to hot ones [394] (Figure 6b).The peptide concluded a central fluorophore 4,7-di(thiophen-2-yl)-2,1,3-benzothiadiazole and two peripheral EphA2-targeted self-assembled peptides (peptide sequence: FFGYSAYPDSVPMMS).The peptide bond specifically to EphA2 promoted cancer malignancy and induced the aggregation of the receptors, resulting in the activation of the antitumor pathway and visualization of EphA2 receptors in a fluorescence turn-on manner.By inducing immunogenic death and recruiting massive tumor-infiltrating T cells, the peptide also efficiently converted immunologically cold tumors to hot ones.To further overcome the hurdles of poor infiltration of T cells and tumor penetration of antibodies induced by ECM, Hu et al. reported an in situ self-assembled bispecific peptide that targeted both C-X-C chemokine receptor type 4 (CXCR4) and PD-L1 (peptide sequence: AMD070-DPGLGYLKLVFFGCVRARTR) [395] (Figure 6c).The rapid formation of CXCR4/PD-L1-targeted nanoclusters on tumor cell surfaces in situ could enhance the blockages of both CXCR4 and PD-L1, resulting in reductions in ECM component accumulation and solid tumor stress (to 44%).By improving T cell activation and infiltration, this in situ bispecific self-assembled system achieved a tumor growth inhibition efficacy of 74% compared with 24% for PD-L1 in the subcutaneous mouse urothelial carcinoma MB49 tumor-bearing mouse model.Of note, compared with antibodies, this nanosystem had the advantages of rapid blood clearance (elimination half-life (t 1/2β ) = 1.4 h) and prolonged tumor retention (t 1/2β = 69.3h), providing possibilities to overcome the potential systemic side effect.

Immune-Reprogramming Nanomaterials
To boost the therapeutic effects of immunotherapy, reprogramming immune cells demonstrated attractive potential.One of the most successful examples is the chimeric antigen receptor T cell (CAR T) therapy [396].In this therapeutic, one's own immune cells, mainly T cells and NK cells, are genetically engineered to express chimeric antigen receptors, allowing the immune cells to recognize and kill tumor cells specifically.Currently, six CAR T therapies have been approved for cancer therapy by the FDA since the first approval in 2017.However, CAR T therapy still encounters limitations such as patient dependence, clinical toxicities (e.g., cytokine release syndrome and neurotoxicity), and resistance.Besides cell-based therapy, immune-reprogramming nanomaterials have recently demonstrated genius in cancer immunotherapy.Unlike CAR T therapy, these nanomaterials can reprogram immune cells in the body without the need to extract immune cells.In addition, compared with direct delivery of drugs to tumors that encounter a high risk of clearance by the immune system, the immune-reprogramming nanocarriers are designed to target and re-activate immune cells.For instance, Nahmad et al. directly engineered B cells in vivo to secrete neutralizing anti-HIV antibodies in mice [397].They prepared two adeno-associated viral vectors to encode Staphylococcus aureus Cas9 and broadly neutralizing antibody 3BNC117, respectively.Via the intravenous administrations of two vectors, B cells in mice were engineered to express high levels of broadly neutralizing antibodies at neutralizing titers of up to 6.8 µg ml −1 .In addition to genetic engineering, self-assembled nanomaterials that directly modify the surface of immune cells have recently attracted much attention.For instance, Jiang et al. reported a type of immunomodulating nano-adaptors to promote antibody-based cancer immunotherapy (Figure 7a) [398].These nano-adaptors comprised an anti-IgG (Fc specific)-modified polystyrene nanoparticle core and a shell consisting of two monoclonal antibodies.These antibodies were conjugated on the nanoparticle surface via Fc-specific noncovalent interactions.When conjugated with anti-PD-1 antibodies and anti-PD-L1 antibodies, the resulting nano-adapters effectively promoted T cell-tumor cell interactions and augmented the T cell-mediated immunotherapy in subcutaneous B16-F10 melanoma tumor-bearing mice.The average tumor volume in the nano-adapter group was 4.3-fold and 3.2-fold smaller than those receiving the mixture of two antibodies and the mixture of two single antibody-conjugated nanoparticles, respectively.Via conjugations of anti-killer-cell lectin-like receptor G1 antibodies and anti-PDL1 antibodies, the resulting nano-adapters enhanced the NK-cell mediated immunotherapy in B16-F10 pulmonary metastatic tumor-bearing mice.The metastatic nodules in the lungs of nano-adapter-treated mice (median, ~7) were much less than those of the mixture of two antibodies-treated mice (median, ~34), the mixture of two single antibody-conjugated nanoparticles-treated mice (median, ~27), and IgG control-treated mice (median, ~62), respectively.In addition, the nano-adapters with anti-factor 1-receptor antibodies and anti-CD47 antibodies could also improve macrophage-mediated immunotherapy by converting tumor-supportive M2 macrophages to tumoricidal M1 macrophages and physically connecting macrophages and tumor cells.Besides pre-assembled nanoparticles, Zhao et al. also reported an in vivo self-assembled glycopeptide for reprogramming tumor-associated macrophages (TAMs) to boost cancer immunotherapy (Figure 7b) [399].The glycopeptide consisted of a tumor-targeting motif, an MMP-2 cleavable linker, and a mannose moiety for targeting mannose receptors on M2-like TAMs (peptide sequence: Mannose-alkyl-PLGVRGRGD).After systemic administration, the precursor glycopeptide entered tumor tissues via active targeting and was cleaved by MMP-2 enzymes.The resulting mannose segment further self-assembled into nanoparticles with improved binding affinity to the mannose receptors (411-fold decrease in the dissociation constant), leading to the switch of M2-like microphases to M1-like ones and enhancement in tumor penetration.Owing to the advantages of deep tumor penetration and enhanced hypoxic TAMs repolarization, this glycopeptide with anti-PD-1 antibody achieved a 90.2% tumor inhibitory rate in the TAMsabundant 4T1-breast cancer model.Furthermore, to further improve the spatiotemporal specificity for immunotherapy, An et al. reported a bispecific glycopeptide that targeted CD206 on M2-like TAMs and CXCR4 receptors on tumor cells for inhibiting bladder cancer recurrence (Figure 7c) [400].The peptide consisted of 4 modules, including (i) a CD206targeting motif with (ii) an MMP-2 cleavable linker, (iii) a CXCR4-targeting motif, and (iv) a self-assembly motif (peptide sequence: LGASWHRPDKK(PLGYLG-(man) 3 )LVFFAECG).In the tumor microenvironment, the peptide repolarized M2-like TAMs to the M1 phenotype, promoting the recruitment of CD8 + T cells.Meanwhile, the peptide was cleaved by MMP-2 enzymes and formed CXCR4-binding nanofibers in situ for the long-term arrest of CXCR4 signaling, promoting T cell infiltration.Owing to the spatiotemporal regulation of tumor microenvironment, this bispecific glycopeptide reduced the recurrent rate of orthotopic bladder MB49-Luc tumor-bearing mice to 22% compared with 100% for saline and plerixafor groups and 89% for the DOX group.
for DOX solely.To make more programmable biomaterials, introducing logic gates into nanomedicine design has attracted much attention [387][388][389].For instance, Zhang et al. induced the concept of logic gates to construct a programmable polymer library [199].Different stimuli-responsive units (e.g., light-, ROS-, glutathione-, acid-, esterase-, phosphates-responses) were integrated into these polymers with logic gates and hierarchical organizations, allowing to receive disease biomarkers as inputs and site-specifically release therapeutics (e.g., kinase inhibitors, drugs, and siRNA) as outputs.

In Vivo Self-Assembled Nanomaterials
In vivo self-assembled nanomaterials can transform or self-assemble in tumor tissues in situ triggered by internal or external stimuli.The self-assembly process is governed by both thermodynamics and kinetics, which provide different assembled structures and new biological functions.Via the rational design of building blocks and control of thermodynamics and kinetics, in vivo self-assembled nanomaterials have advantages such as enhanced tumor accumulation and retention, improved tumor penetration, increased cellular internalization, etc., enabling improved immunotherapy out-

Conclusions and Perspective
In this review, we briefly presented recent progress on multistage self-assembled nanomaterials for cancer immunotherapy.The traditional "one size fits all" approach can hardly confront the sophisticated physiological and tumoral environments.Therefore, nanomaterials that can overcome multiple biological barriers are highly desirable.Among them, in situ self-assembled nanomaterials have attracted much attention, allowing precise and on-demand self-assembly in disease sites triggered by internal and external stimuli.These nanomaterials have displayed advantages such as prolonged blood circulation, enhanced tumor accumulation and penetration, increased tumor cell internalization, controlled drug release in specific sites, and improved immune responses.These features enable them to achieve improved cancer therapeutics with reduced side effects.Despite significant advancements in multistage self-assembled nano-delivery systems, key challenges associated with nanomedicines persist.These include inadequate specific targeting, limited delivery efficacy, and potential side effects.Several critical issues should be considered for the clinical success of these smart nanomaterials.
First, the multistage nanocarriers involve multiple and programmed delivery procedures in the body.Reliable and non-invasive monitoring techniques or multimode tracking probes for evaluating the process and efficacy for each stage are highly desired.Second, for many multistage nanocarriers, their architectures and constructions are highly complicated.Concerns regarding reproducibility and high quality control in scale-up manufacturing must be addressed.Third, the safety and side effects of the multistage nanocarriers should be highly considered, particularly the immune-related side effects and long-term toxicity.Fourth, human immune systems are quite different from experimental animal models.Developing reliable humanized animal models and organs-on-chips with immune systems is urgent.Fifth, a fundamental and deep understanding of the thermodynamics and kinetics of in vivo self-assembly is critical to spatiotemporally control of the self-assembly process, assembled structures, distribution, retention, and excretion in the body.Last but not least, discovering novel biomaterials using artificial intelligence is the new research paradigm in biomedical fields.High-throughput material library and screening are urgent to build up the structure-activity relationships in biomaterialomics to guide the rational material design of nanomaterials for immunotherapy driven by machine learning and data science.Bridging together the cutting edges of nanotechnology, biotechnology, and data science, multistage self-assembled nanomaterials are expected to prompt the advent of precise and efficient cancer therapy in the near future.

Figure 1 .
Figure 1.The biological barriers for nanocarriers for delivering drugs to tumors.

Figure 1 .
Figure 1.The biological barriers for nanocarriers for delivering drugs to tumors.

Figure 2 .
Figure 2. Tumor microenvironment-responsive nanomaterials for cancer immunotherapy.(a) PHsensitive block copolymers formed biomolecular condensates with STING for prolonged cancer immunotherapy.Figure adapted with permission from Ref. [277] under the Creative Commons CC BY license.Copyright 2021, Springer Nature.(b) PH/enzyme-responsive nanoparticle for selective delivery of TLR agonists to active TLR7/8 receptor signaling at the endosomal membrane in dendritic cells.Data are shown as mean ± s.d., n.s.: not significant, **** p < 0.0001.Figure adapted with permission from Ref. [286].Copyright 2022, American Chemical Society.

Figure 2 .
Figure 2. Tumor microenvironment-responsive nanomaterials for cancer immunotherapy.(a) PHsensitive block copolymers formed biomolecular condensates with STING for prolonged cancer immunotherapy.Figure adapted with permission from Ref. [277] under the Creative Commons CC BY license.Copyright 2021, Springer Nature.(b) PH/enzyme-responsive nanoparticle for selective delivery of TLR agonists to active TLR7/8 receptor signaling at the endosomal membrane in dendritic cells.Data are shown as mean ± s.d., n.s.: not significant, **** p < 0.0001.Figure adapted with permission from Ref. [286].Copyright 2022, American Chemical Society.

Figure 4 .
Figure 4. Strategies for construction of multistage self-assembled nanomaterials.

Figure 4 .
Figure 4. Strategies for construction of multistage self-assembled nanomaterials.

Figure 7 .
Figure 7. Immune-reprogramming nanomaterials for immunotherapy.(a) Immunomodulating nanoadaptors to promote antibody-based cancer immunotherapy.Figure adapted with permission from Ref. [398] under the Creative Commons CC BY license.Copyright 2021, Springer Nature.(b) In vivo self-assembled glycopeptide for reprogramming tumor-associated macrophages to boost cancer immunotherapy.Figures adapted with permission from Ref. [399].Copyright 2023, Wiley-VCH GmbH.(c) Bispecific glycopeptide for spatiotemporal regulation of tumor microenvironment to inhibit bladder cancer recurrence.Figure adapted with permission from Ref. [400] under the Creative Commons CC BY license.Copyright 2023, American Association for the Advancement of Science.

Table 2 .
Recent examples of stimuli-responsive nanosystems for cancer immunotherapy.

Table 3 .
Recent examples of multistage self-assembled nanocarriers for cancer immunotherapy.

Table 3 .
Recent examples of multistage self-assembled nanocarriers for cancer immunotherapy.