Smart drug delivery systems for precise cancer therapy

Abstract

Nano-drug delivery strategies have been highlighted in cancer treatment, and much effort has been made in the optimization of bioavailability, biocompatibility, pharmacokinetics profiles, and in vivo distributions of anticancer nano-drug delivery systems. However, problems still exist in the delicate balance between improved anticancer efficacy and reduced toxicity to normal tissues, and opportunities arise along with the development of smart stimuli-responsive delivery strategies. By on-demand responsiveness towards exogenous or endogenous stimulus, these smart delivery systems hold promise for advanced tumor-specificity as well as controllable release behavior in a spatial-temporal manner. Meanwhile, the blossom of nanotechnology, material sciences, and biomedical sciences has shed light on the diverse modern drug delivery systems with smart characteristics, versatile functions, and modification possibilities. This review summarizes the current progress in various strategies for smart drug delivery systems against malignancies and introduces the representative endogenous and exogenous stimuli-responsive smart delivery systems. It may provide references for researchers in the fields of drug delivery, biomaterials, and nanotechnology.

Graphical abstract

The development of smart drug delivery strategies and relevant drug delivery systems could provide solutions for the insufficient specificity and toxicity concerns in anticancer therapy.

1. Introduction

Medications against malignancies are usually hindered by poor specificity and subsequent concerns of toxicity, and their therapeutic effects may also be challenged by deficient concentrations at tumor sites, drug resistances, etc¹. Despite the availability of many other tactics for anticancer treatments, e.g., surgeries and radiation therapies, their limited ranges of application still urge the development of modern anticancer delivery strategies². Nano-drug delivery systems (NDDS) provide promising platforms for anticancer therapy, holding potential for versatile improvements in the in vivo distributions, behaviors, and performances of therapeutic agents³. Smart drug delivery systems have been a highlighted field in NDDS, which could provide targeting specificity, controlled release, as well as the ability to cross biological barriers, giving rise to enhanced therapeutic effects with minimized systemic side effects⁴.

The development of functionalized modules and materials, including modern biomaterials such as peptides and nucleotides, has supported the design and research of various smart delivery strategies. Stimuli-responsiveness towards endogenous or exogenous factors offers on-demand transitions of vehicle structures or properties, enabling spatial-, temporal- or dosage-specific deliveries⁵. On the other hand, the diverse nano-drug delivery vehicles also facilitate the rational design based on proper therapeutic agents and pro-drugs, as well as the applications of combinational therapies, multi-responsive platforms, etc.

This review focuses on the recent advances in smart drug delivery strategies and systems. In this review, we highlighted the main strategies and mechanisms of smart drug loading and release for anticancer therapies, and also summarized the current delivery systems supporting these strategies.

2. Strategies and mechanisms of smart drug loading and release

Smart nanoparticles have constituted an excellent platform for achieving efficient cancer therapy, which is considered an extensively explored stimuli-responsive approach to specifically release the cargoes at the tumor sites in response to endogenous (pH, enzymes, or redox gradients) or exogenous stimuli (light, temperature, ultrasound, magnetic field, and electric field). Such stimuli-responsive nanoparticles can provide on-demand drug release, thus achieving more delicate therapeutic effects and preventing drug leakage in blood circulation for avoiding off-target side effect⁶.

2.1. Endogenous stimulus-responsive DDSs

The intrinsic biological factors of tumor tissues differ significantly from healthy tissues, mainly including low pH values, over-expression of specific enzymes, increased redox-potential and hypoxia, etc⁷. Based on these differences, the pH-, enzyme-, and redox-responsive drug delivery systems (DDS) have been rationally engineered for smart drug loading and spatiotemporal release in specific targets for enhanced therapeutic efficacy.

2.1.1. pH-Responsive DDSs

Currently, there are two main strategies for the development of pH-responsive DDSs. One strategy is based on the structure or solubility change of polymers containing ionizable functional groups⁸. Various ionizable groups (e.g., carboxylic acids and amines) in the nanoparticles can be protonated upon pH variations, disrupting the hydrophilic–hydrophobic equilibrium and triggering a dramatic change of structure or solubility of the nanoparticles, thereby realizing pH-responsive drug release⁹. The other strategy is based on the cleavage or the degradation of acid-labile bonds. Chemical bonds such as hydrazone, ester, imine, oxime, and ketal bonds are stable at neutral pH but can be cleaved under acidic conditions. Therefore, constructing nanocarriers with pH-cleavable chemical bonds or using these bonds for drug conjugation can achieve prompt drug release in acidic environment¹⁰. Through the above strategies, various pH-responsive DDSs differentiating the pathophysiological pH gradients in the body have been designed for cancer therapy with high efficacy and low toxicity.

The pH gradients throughout the body can be divided into three levels: organ, tissue, and cellular levels. At the organ level, the most significant pH variation is that of the gastrointestinal tract (GIT). Different segments of the GIT have their own characteristic intraluminal pH levels, from the acidic stomach (pH 1–3) to the alkaline intestine and colon (pH 6.5–7.5)¹¹. Therefore, the pH difference of these parts can be exploited to design DDS for gastric retention or specific targeted intestinal drug delivery. Currently, colorectal cancer is a common cancer type and the third leading cause of death among the malignancies in the United States¹². However, therapeutic drug molecules such as proteins and peptides are vulnerable to acidic and enzymatic degradation in the stomach, raising challenges for their oral administration. Fortunately, oral pH-responsive colon-targeted DDSs hold great promise for colorectal cancer therapy¹³. Some pH-sensitive polymers, e.g., cellulose acetate phthalates (CAP) and methacrylate-based Eudragit® polymers, could withstand the acidic pH of the stomach and be dissolved in the alkaline milieu of the colon, which could be adopted to protect the encapsulated anticancer drugs from degradation in the harsh condition of the stomach. For instance, a colon-targeted, oral nanoparticle vaccine was constructed by encapsulating protein vaccine into poly(D,L-lactic-co-glycolic acid) (PLGA) nanoparticles, and then loading PLGA nanoparticles into Eudragit FS30D microparticles¹⁴. The mucosal uptake of microparticles was impeded due to their large diameters (>10 μm) until they disintegrated into PLGA nanoparticles in the colon (pH > 7). Therefore, the microparticle could avoid premature degradation and uptake of vaccines and precisely deliver vaccines into the colon, thus inducing effective protective immunity.

At the tissue level, the extracellular pH (pHe) of the tumor microenvironment (pH 6.7–7.1) is slightly lower than that of healthy tissue and blood (pH 7.4). Tumor cells exhibit vigorous growth and metabolism, and the irregular vasculature in tumors is insufficient to supply nutrients and oxygen for the fast-growing tumors, thus the tumors shift towards a glycolytic metabolism and produce acidic metabolites such as lactic acid and CO₂, lowering the pH of tumor interstitium¹⁵. However, there is only a subtle pH change in the tumor microenvironment as compared with the normal physiological environment. Thus, the ideal tumor pHe-responsive DDS should have a sharp response to pH variation. Intriguingly, based on the cooperative self-assembly of block copolymers with ionizable tertiary amine groups, Ma and coworkers¹⁶ pioneered the development of ultra-pH-sensitive (UPS) nanotechnology for cancer theranostics. The UPS nanoparticles micelles exhibited ultra pH sensitivity within 0.25 pH unit, tunable pH transition (pH_t) ranging from 4.4 to 7.1, ultra-fast pH-triggered dissociation (<5 ms), and exponential signal amplification (>100-fold) upon pH-driven micelle dissociation, with extensive utilization for cancer imaging and image-guided cancer surgery¹⁷. Based on the UPS nanotechnology, ONM-100, an indocyanine green (ICG)-encoded fluorescence nanoprobe is in phase II clinical trial for image-guided cancer resection and metastatic lymph node mapping. Moreover, the UPS nanoplatform has also been widely used in chemical drug, gene, and vaccine delivery¹⁸.

At the cellular level, the pH of endocytic organelles is much lower than that of cytoplasm (pH 7.2). Following endocytosis, endosomal acidification occurs rapidly due to the vacuolar-type ATPase (V-ATPase) proton pump-mediated proton influx into the lumen of the organelles. Accordingly, the pHs of early endosome, late endosome, and lysosome are dropped from about 6.3 to 5.5, and finally around 4.7¹⁹. The low pH of endocytic organelles is a widely used trigger for pH-responsive DDS which can realize organelle-specific activation and on-demand drug release. For instance, an EGFR-targeted ultra-pH-sensitive nano-photosensitizer was developed by Yan and coworkers for endocytic organelles-specific photodynamic therapy (PDT)²⁰. The photoactivity of photosensitizer Ce6 was “OFF” at physiological pH due to the FRET (Fluorescence Resonance Energy Transfer) quenching at micelle state, which could be turned “ON” upon micelle dissociation in the acidic endocytic organelles for cancer cell killing. A UPS nanoparticle with sequential targeting ability of early endosome and mitochondria was also developed for exponential amplification of PDT efficacy²¹. The nanoparticle could immediately dissociate at the acidic environment of the early endosome, enabling the photoactivity activation of the mitochondria-targeted photosensitizer TPP-PPa. Then, the activated TPP-PPa quickly escaped from the early endosome and targeted the mitochondria for effective PDT upon laser irradiation. However, the acidic environment and harsh enzymatic activity of endocytic organelles are also harmful to some therapeutic macromolecules such as proteins, small interfering RNA (siRNA), and DNA. Therefore, to achieve the maximum therapeutic effects of macromolecules, various pH-responsive DDS are also designed for endosomal escape and cytosolic delivery based on the “proton sponge” effect²², with cationic materials such as poly(ethylene imine)s (PEIs)²³.

2.1.2. Enzyme-responsive DDSs

The up-regulation of specific enzymes in the tumor microenvironment and inside the tumor cells has been exploited as important triggers for smart drug delivery. Many enzymes like proteases (e.g., matrix metalloproteinase/MMP and cathepsin B), phospholipases (e.g., phospholipase A2), and peptidases (e.g., aminopeptidase), etc. have been investigated to construct enzyme-responsive DDS for tumor-tropism drug delivery²⁴. The enzyme-responsive DDS could act in the following ways: (i) Enzyme-triggered drug release either by constructing nanocarriers with a structural scaffold susceptible to specific enzymes or by using an enzyme-sensitive linker between the nanocarrier and therapeutics. Du et al.²⁵ developed a pH/cathepsin B hierarchical-responsive micelle for the programmed delivery of docetaxel (DTX). The micelle remained stable in blood circulation, while dissociated into polymer-DTX conjugates under acidic tumor microenvironment for deep tumor penetration and enhanced cellular uptake. After being endocytosed into the lysosomes, the Gly–Phe–Leu–Gly (GFLG) tetrapeptide linker of polymer-DTX conjugates was cleaved by lysosomal cathepsin B, then bioactive DTX was effectively released into the cytoplasm for enhanced antitumor efficacy. (ii) Prodrugs, ligands, and probes activation through the cleavage of enzyme-sensitive bonds. Zheng et al.²⁶ conjugated a photosensitizer (Pyro) and ¹O₂ quencher (BHQ3) with a matrix metalloproteinase-7 (MMP-7)-cleavable peptide linker, enabling effective ¹O₂ quenching of Pyro through FRET effect. Therefore, this construct was photodynamically inactive until entering the tumor with high MMP-7 expression. Upon MMP-7-induced cleavage, the photoreactivity of Pyro was recovered, thus producing cytotoxic ¹O₂ under light irradiation. (iii) Enzyme-activated detachable PEGylation layer for prolonged circulation and increased cellular internalization at the target site. Han and coworkers²⁷ prepared a dual-enzyme-responsive, gemcitabine-loaded (GEM) nanovector by conjugation of matrix metalloproteinase-9 (MMP-9)-detachable poly(ethylene glycol) (PEG) protection layer, cathepsin B-cleavable GEM, and tumor-homing motif cRGD peptide to quantum dots (QDs). The nanovectors exhibited prolonged blood circulation due to the PEG decoration. Then, the PEG shield layer could be removed by the overexpressed MMP-9 after accumulation in tumor tissue, enabling the exposure of cRGD for enhanced cellular internalization. Once endocytosed into cancer cells, the GEM could be released by elevated lysosomal cathepsin B for tumor inhibition. Enzyme-responsiveness could assist in the development of size-changeable nanoparticle systems, realizing aggregation and retention at target sites, optimized targeting, and internalization, etc28, 29, 30. Despite the huge progress made in the development of enzyme-responsive DDS for cancer therapy, several challenges remain to be solved. Firstly, some enzymes share similar active sites and catalytic mechanisms, leading to their similar substrate preferences³¹. Secondly, the enzyme expression level varies greatly not only in different cancer types, but also in tumors with a similar type yet in different individuals, and in different parts of the tumor as well³².

2.1.3. Redox responsive DDSs

Redox homeostasis is critical for cell survival with diverse cellular processes involved, and glutathione (GSH), a tripeptide that can scavenge excess reactive oxygen species (ROS) through the transformation of its heterogeneous forms (GSH/GSSG), plays a key role in maintaining the intracellular redox balance³³. Due to the highly reductive and hypoxic microenvironment of tumor tissues, the intracellular GSH level in tumor cells was four times higher than that in normal cells³⁴. In addition, the concentration of GSH varies greatly between the intracellular environments (about 2–10 mmol/L) and the extracellular environments (about 2–10 μmol/L). The remarkable concentration gradient of GSH (about 100‒1000 times) serves as an attractive stimulus for tumor-targeted intracellular drug delivery, especially for cytoplasmic delivery of proteins and genes (DNA or siRNA). Various redox responsive DDS have been developed by introducing reduction-sensitive bonds (disulfide bonds, diselenide bonds, ditelluride bonds, etc.) into the nanocarrier backbones or for drug conjugation³⁵. The disulfide bond (S–S), prone to fast cleavage by GSH, is the most commonly used linker for the fabrication of redox-responsive DDS, due to its easy introduction into polymers and drugs³⁶. For instance, Wu and coworkers³⁷ prepared a nanocomplex (siRNA/DOX@HMONs-ss-PAE) for redox-responsive gene delivery to reverse the multidrug resistance (MDR). The P-glycoprotein (P-gp) modulator siRNA and anticancer drug doxorubicin (DOX) were loaded into the hollow mesoporous organosilica nanoparticles (HMONs), which were capped with poly (β-amino esters) (PAE) through a disulfide bond. When endocytosed into tumor cells, the nanocomplex could translocate from the endo/lysosome to the cytoplasm via the “proton sponge” effect of cationic PAE, and then the encapsulated DOX and siRNA were released due to the cleavage of the disulfide bond between HMONs and capping PAE in the highly reductive intracellular microenvironment. Thus, the P-gp-mediated MDR would be reversed by siRNA, leading to enhanced intracellular DOX concentration for efficient chemotherapy of cancer cells.

In addition, tumor cells also have elevated ROS (e.g., H₂O₂, superoxide, and hydroxyl radicals) generation than normal cells due to their hypermetabolism³⁸. Hence, various types of ROS-responsive DDS have been explored for tumor-specific drug delivery on the basis of the high ROS level in tumor tissues. Many ROS-responsive linkages are frequently used in cancer therapy, including thioketal, thioether, peroxalate ester, boronic ester, etc³⁹. Xu et al.⁴⁰ developed an innovative ROS-responsive poly-prodrug nanoparticle for targeted cancer therapy. The nanoparticle was composed of a poly-prodrug inner core of ROS-responsive mitoxantrone (MTO) for ROS-triggered drug release, a PEG outer shell for prolonged blood circulation, and an iRGD targeting ligand for deep tumor penetration and specific tumor targeting. Upon exposure to high-level ROS in tumor cells, the thioketal bond in the poly-prodrug was cleaved, leading to the release of intact drug molecules for efficient tumor inhibition.

2.2. Exogenous stimulus-responsive DDSs

In some cases, endogenously stimuli-responsive nanoparticles fail to overcome the biological barriers in tumors due to insufficient and intractable responses to some subtle alterations of the above endogenous factors in the tumor microenvironment. Reasonably, exogenously stimuli-responsive DDSs are considered alternative nanoplatforms that are of great importance due to target-specific and controlled drug release at the target sites.

Exogenous stimuli-responsive DDSs are smart drug delivery systems that can actively release the cargo in response to external stimuli including light, temperature, ultrasound, magnetic field, electric field, and other stimuli, maximizing therapeutic efficacy of cargos while reducing adverse side effects. The exogenous stimuli can undergo a chemical or physical change to cause the alteration of structures and surface properties of DDSs, which ensure their tumor targeting, penetration, cellular uptake, and intracellular drug delivery to be manipulated in a controllable manner⁷. There are various advantages of exogenous stimuli-responsive DDSs in cancer therapy including (i) the location and intensity can be precisely controlled by exogenous stimuli (e.g., light, magnetic or electric field); (ii) the exogenous stimuli can be flexibly applied or removed; (iii) multiple exogenous stimuli can be integrated into a single nanoplatform for providing multifunctional properties, although such DDSs would be unsuitable for treating distal or metastatic cancer since their tumor locations are unknown.

2.2.1. Light-responsive DDSs

Light-responsive DDSs for cancer therapy have been extensively developed, since light [e.g., near-infrared (NIR) and visible light] is an attractive exogenous stimulus with a possibility to remotely control the irradiation power and exposure time for selectively treating local tumors. Lights can induce physicochemical changes, such as photoisomerization/photocleavage reactions and photothermal/photodynamic effects, thus improving the performance of light-responsive DDSs. Compared with ultraviolet (UV) and visible lights, NIR light with relatively deep tissue penetration ability is of particular interest in cancer therapy. For instance, the NIR light-responsive selenium-contained polymeric nanoparticles loading indocyanine green and doxorubicin (I/D-Se NPs) were designed for synergistic thermo-chemotherapy⁴¹. The prepared I/D-Se NPs could produce ultrafast irreversible disassembly via light-mediated selenium oxidation, thus promoting the NIR light-responsive drug release in cell cytoplasm for synergistic cancer therapy (Fig. 1)⁴¹. Zhao et al.⁴² reported the bifunctional light-responsive platinum nanocomplexes (PtNCs) that produced abundant heat via photothermal conversion from the Pt⁰ core upon irradiation and simultaneously caused a rapid release of chemotherapeutic Pt²⁺ ions surrounding the Pt⁰ core, thus leading to a light-triggered synergistic chemo-photothermal therapy. In another study, the light-responsive palladium nanocrystals-assimilated nanoscale metal-organic framework (NPMOF) nanoparticles were developed for synergistic hydrogen and photodynamic therapy, with significant light-triggered singlet oxygen (¹O₂) generation and persistent reductive hydrogen release, causing an adequate disturbance in tumor milieu for synergistic tumor therapy⁴³. Recently, Chen et al.⁴⁴ synthesized NIR-light modulated BF2-azadipyrromethene (aza-BODIPY) nanoaggregates that were actively transformed from wormlike nanofibers into spherical nanoparticles in vivo under light irradiation, with prolonged circulation, deep tumor penetration, and subsequent enhanced antitumor efficacy. More recently, aggregation-induced emission (AIE) luminogen-encapsulated lipid nanoparticles with robust NIR-I two-photon absorption were developed for spatiotemporal deep-tumor imaging, and simultaneously producing toxic hydroxyl radicals (•OH) and singlet oxygen (¹O₂) upon light irradiation for tumor ablation⁴⁵. Zwitterionic luminogen nanodots with AIE features also showed enhanced photothermal conversion efficiency (35.76%) and ROS generation performance under NIR-light irradiation, providing synergistic phototherapy against breast cancer⁴⁶. In another study, Ren and co-workers⁴⁷ successfully developed a light-responsive multifunctional Bi₂Se₃-based nanoplatform loaded with glucose oxidase and oxygenated perfluorocarbon, for improved tumor starvation and light-mediated photothermal therapy (PTT). Light irradiation not only enabled Bi₂Se₃ nanoshell to generate local hyperthermia, but also triggered the release of capsulized oxygen to recede local hypoxia.

Figure 1.

Light-responsive polymeric nanoparticles (I/D-Se-NPs) with fast drug release and an efficient cytoplasmic translocation for synergistic thermo-chemotherapy. Reprinted with the permission from Ref. 41. Copyright © 2017 American Chemical Society.

Specifically, drug resistance has been a vital challenge in anticancer therapies. Insufficient drug delivery to the tumor cells and sublethal chemotherapeutic concentration could lead to acquired chemo-resistance48, 49, 50, 51, and the recruitment of tumor-infiltrating cytotoxic T lymphocytes (CTLs) and intratumoral secretion of proinflammatory cytokine interferon gamma (IFN-γ), which consequently eliciting surface levels of immune regulators including programmed death ligand 1 (PD-L1) could hamper immunotherapeutic effects52, 53, 54, 55, 56. NIR-responsive nanoparticles were also reported in dealing with chemo- and immuno-resistance of the tumor through single or combination of multiple mechanisms including hyperthermia effects, photothermal ablations, immunogenic cell death (ICD) induced by laser irradiation, spatiotemporally tunable release, etc., exhibiting prospects for comprehensive anticancer therapy^50,51,57, 58, 59, 60, 61, 62 (Fig. 2, reprinted from Ref. 57).

Figure 2.

(A) MMP-2-sheddable and GSH-activatable prodrug vesicle for combined photodynamic immunotherapy. (B) The mechanism of photodynamic immunotherapy against cancer combating IDO-1-induced adaptive immune resistance. Reprinted with the permission from Ref. 57. Copyright © 2019 American Chemical Society.

Researchers have also been pursuing photodynamic therapies with further reduced dark toxicity and side effects⁶³. AIE has emerged as a solution to the unwanted phototoxicity at non-target sites, and smart delivery strategies also proved to be effective by providing more specified location, on-demand activation or release at the target sites⁶⁴. Increased selectivity could be realized by the conjugation of photosensitizers with targeting moieties, for example, and the photosensitizer antibody-drug conjugates have been extensively reviewed in previous articles, with highlights on highly specific bioconjugation techniques⁶⁵. Dual-responsiveness with multiple triggers instead of single light-responsiveness also add to the specificity of photosensitizers, and the on-demand activation strategies with various on-off mechanisms further alleviated the dark toxicity. Dong et al.⁶⁶ prepared a calcium carbonate-polydopamine composite hollow nanoparticle, and the loaded photosensitizer was quenched by polydopamine and could be released in acidic environment. Feng et al.⁶⁷ developed a nanoassembly system with pyridinium-functionalized tetraphenylethylene encapsulated in the cavity of water-soluble calixarene, which only exhibits yellow fluorescence upon light triggering due to the restriction of intramolecular motion. The photosensitizer could be displaced by 4,4′-benzidine dihydrochloride and translocate to mitochondria with restored photoactivity. Another work also concerns photosensitizer activation by displacement, the photosensitizer was quenched by macrocyclic amphiphile, and the overexpressed adenosine triphosphates on tumor sites competitively bind to the vehicle macrocyclic amphiphile, realizing release and activation of the photosensitizer at the tumor site⁶⁸.

2.2.2. Temperature-responsive DDSs

Temperature-responsive DDSs have also been extensively applied as drug delivery vehicles in cancer therapy. Generally, the DDSs are required to be stable and to retain the cargoes in normal body temperature (up to 37 °C), however sensitively release the cargoes at higher temperature (e.g., >40 °C) through significant physico-chemical changes in response to a narrow temperature increase⁷. The most popular temperature-sensitive polymeric materials include poly(amidoamine) (PAMAM), poly[2-(2-methoxyethoxy)ethyl methacrylate] (PMEO₂MA), poly(N-isopropylacrylamide) (PNIPAM) and poly(2-oxazoline)s (POxs). In response to lower critical solution temperature (LCST), the temperature-sensitive polymers often show drastic changes in their aqueous solubility⁶⁹. Above their LCST, these polymers become insoluble in water, and subsequently, the DDSs are destructed to release the cargo. For example, PNIPAM-based nanocomposites were found to actively release doxorubicin in response to increased temperature, as PNIPAM endured a reversible phase change beyond their LCST (Fig. 3)⁷⁰. In another study, PNIPAM-based thermo-sensitive polypyrrole nanoplatforms were developed for synergistic photothermal-chemotherapy⁷¹. Additionally, some cancerous tissues exhibit abnormal increases (generally 1–2 °C) in temperature compared with normal tissues. Such temperature change could be a potential endogenous stimulus in drug delivery, which raises higher requirements for the precise and tunable responsiveness of the materials. Current thermosensitive polymers and nanocarriers have been intensively reviewed in Ref. 72.

Figure 3.

(A) Synthesis and thermo-triggered drug release from DOX@PNIPAM-ppy-FA nanocomposites. (B) The mechanism of temperature-responsive release of doxorubicin and enhanced thermo-chemotherapy in breast cancer. Reprinted with the permission from Ref. 70. Copyright © 2020, Springer Nature.

Besides, temperature-responsive DDSs were also developed by incorporating thermo-sensitive cargos inside the nanovehicles. For instance, gold nanorods and iron oxide nanoparticles-incorporated liposomes could release doxorubicin upon hyperemia (50 °C) caused by light irradiation due to the collapse of the liposomes, leading to the promoted tumor destruction⁷³. Despite the extensive progress in temperature-responsive DDSs, only very few thermo-sensitive vectors are currently considered for clinical translation in addition to liposomes and gold nanoparticles. It is highly important to develop temperature-sensitive materials and vectors with high biosafety.

2.2.3. Ultrasound-responsive DDSs

Ultrasound-responsive DDSs have various applications in cancer therapy as they could specifically release payloads at the tumor site in response to externally applied ultrasound, showing promise for targeting ability, deep tumor penetration, and reduced side effects by adjusting their frequency in a specific range. For instance, high-frequency ultrasound waves (>20 kHz) could be used to trigger drug release or improve the tumor permeability of nanoparticles, while low ultrasound frequencies (<20 kHz) were usually applied for imaging^7,74. Ultrasound also induces several physical effects such as cavitation, acoustic fluid streaming, and local hyperthermia, commonly adopted as triggers for ultrasound-responsive cargoes release (e.g., anticancer agents, imaging probes) from DDSs at the desired tumor sites⁷⁵.

Many researches have supported the effectiveness of ultrasound triggering in accelerating drug release, endosomal escape, tumor accumulation, and significant tumor growth reduction75, 76, 77, and gene delivery has become a highlight in ultrasound-assisted delivery against solid tumors⁷⁸. Meng et al.⁷⁹ designed an ultrasound-responsive, self-healing hydrogel system loaded with nanovaccines, realizing remote release control and repeated vaccine release for durable anticancer therapy. Ultrasound-induced mild hyperthermia could also be utilized in DDS design, and Ma et al.⁷⁷ constructed doxorubicin and oxygen-encapsulated cerasomal perfluorocarbon nanodroplets, which remained stable during blood circulation, while ultrasound-induced mild hyperthermia upon sonication could cause the disturbance and gasification of perfluorocarbon, producing drug release channels in cerasomes that significantly accelerating the release of doxorubicin and oxygen at the tumor site (Fig. 4)⁷⁷. Besides the modulation of DDSs, ultrasound has also been widely applied in modulating the delivery barriers and assisting deep penetration into the tumor sites, as well as inducing immunological responses, as reviewed in Refs. 80 and 81.

Figure 4.

Ultrasound-triggered release of doxorubicin and oxygen from cerasomal perfluorocarbon nanodroplets. Reprinted with the permission from Ref. 77. Copyright © 2020 American Chemical Society.

2.2.4. Magnetic field-responsive DDSs

Magnetic field-responsive DDSs have gained considerable interest in the field of cancer therapy, due to their targeting potentials from the intrinsic tropism towards magnetic fields, and efficient local hyperthermia under external alternating magnetic field (AMF) for promoting on-demand drug release and effective cancer imaging and therapy. Magnetic field-responsive DDSs are generally comprised of a magnetic core with materials such as magnetite (Fe₃O₄), maghemite (Fe₂O₃), hybrid iron oxide (graphene/Au/Fe₃O₄), and other magnetic materials (ZnFe₂O₄), as well as a coating outlayer with materials including polymers, lipids, proteins, and mesoporous silica, etc. Superparamagnetic iron oxide nanoparticles (SPIONs) are predominantly adopted due to their targeting performance in external magnetic fields without retaining any residual magnetism after its withdrawal⁸². Besides magnetic materials, anticancer drugs, contrast agents, photosensitizers, plasmids, and antibodies could also be assimilated inside the magnetic field-responsive DDSs for achieving multimodal therapeutic activities. For example, doxorubicin and bioactive proteins were incorporated into magnetic Mn–Zn ferrite polymeric nanoparticles for thermo-chemotherapy under AMF⁸³. Furthermore, the AMF-induced hyperthermia could also trigger on-demand drug release from the magnetic field-responsive DDSs in the diseased regions⁸⁴. Generally, the magnetic field-responsive DDSs are often required to provide target-specific delivery, because they might affect healthy organs or tissues that are distributed with magnetic field-responsive DDSs when exposed to the AMF.

2.2.5. Electric field-responsive DDSs

Based on the electro-responsive materials such as conductive polymers (e.g., doped polypyrrole, polyaniline), polymer complexes with conductive materials (e.g., graphene, metallic nanoparticles) or hydrogel materials (e.g., alginate, chitosan), as well as installable electrical drug delivery devices, the electric field also proves to be a potential trigger for on-demand drug release⁸⁵. DDSs comprised of conductive materials could provide controlled drug release through electro-chemical oxidation/reduction and the movement of charged moieties⁸⁶. Various researches have been done on electro-responsive hydrogels, e.g., an injectable chitosan-graft-polyaniline copolymer crosslinked with oxidized dextran with dual-responsive drug release to both pH and electric field⁸⁷, and different electroactive moieties (e.g., aniline trimer⁸⁸, graphene oxide⁸⁹, polyacrylamide⁸⁹, etc.) were studied, which was reviewed by Carayon et al.⁹⁰ Besides the activation of DDSs, electric field could also exhibit various physiological effects on tissues and cells, assisting the delivery of therapeutic agents. Electroporation, for example, could transiently promote the permeability of drugs through cell membranes, with intense research in the delivery of macromolecules such as proteins and genes^91,92, while strong electric fields could further realize ablation or cytotoxicity on tumor cells and microorganisms by themselves⁹³. Additionally, distinct electric fields could be generated endogenously by injured or pathological tissues compared with normal sites, e.g., Ying et al.⁹⁴ developed an angiopep–2-modified electro-responsive hydrogel nanoparticle system, with enhanced brain accumulation and in situ drug release triggered by electroencephalograph epileptiform abnormalities.

2.2.6. Other exogenous stimuli-responsive DDSs

Various other exogenous stimuli-responsive DDSs have also been developed for remote-controlled cancer therapy, such as high-energy radiation (X-rays and gamma-rays). For example, the diselenide block copolymers-based nanoparticles loaded with doxorubicin were constructed to trigger drug release under a reduced dose (2 Gy) of X-ray radiation at the tumor site⁹⁵. Besides, X-ray exposure can also enhance tumor accumulation and cellular uptake of albumin nanoparticles by increasing the expression of caveolin-1 on tumor cells⁹⁶.

Moreover, exogenous stimuli-responsive DDSs also offer opportunities to overcome biological barriers such as reversing MDR and promoting lysosomal/endosomal escape. Besides, they can also integrate some novel therapeutic modalities, including photodynamic therapy, phototherapy, ultrasound, and magnetic hyperthermia, which are good alternatives to conventional cancer therapy.

2.3. Receptor-ligand-based smart DDS

Targeted delivery also contributes to modern smart DDSs. In general, tumor-targeting drug delivery systems can be constructed through two strategies: passive targeting and active targeting. In the past, numerous studies were based on the passive targeting strategy, but with decades of intensive research, there are increasingly controversial opinions on the efficiency of passive targeting based on the EPR effect. In 2017, Li et al.⁹⁷ demonstrated that the active and passive effects made different contributions to the total accumulation of nanoparticles (NPs) in tumors, and they found that the receptor-mediated targeting contributed more than the EPR effect over time. In addition, NP transportation through the inter-endothelial cell gaps in the tumor blood vessels was considered to be one of the dominant factors of the EPR effect. Recently, however, Sindhwani et al.⁹⁸ proved that the overall gap coverage was only 0.048% of the blood vessel surface area, which was 60-fold less than the required amount to explain the observed NP accumulation. Therefore, the concept that the passive effect occupies a central role in targeting drug delivery was challenged, while both passive and active targeting are still under intense research in cancer nanomedicine.

Among the frequently discussed issues of active targeting, widespread concerns have been raised about the unwanted interactions of DDSs with non-target sites including nonspecific interactions, and specific interactions whereby both target sites and non-target sites could express relevant receptors, nevertheless at different expression levels (usually referred to as on-target off-tumor effect). Various strategies have been proposed accordingly. For example, Wang et al.⁵⁰ developed tumor acidity-responsive nanoparticles for reversible shielding of the targeting ligand. The iRGD moiety was shielded from interactions with non-target sites in systemic circulation, and was responsively exposed to the acidic tumor microenvironment, assisting in tumor penetration and cellular uptake.

2.3.1. The effect of spatial distribution of ligands on drug delivery

Traditionally, ligands have been attached onto drug delivery systems (liposomes, nanoparticles, micelles, etc.) with required quantities, and the targeting effect might be increased by elevating the density of ligands within certain limits⁹⁹. However, ligands are usually present on the surface of vehicles in a random distribution pattern due to their symmetrical structure, resulting in a limited ability to recognize receptors and incomplete use of ligand materials. Moreover, it will increase the probability of forming protein coronas with the plasma binding protein during the transport process in the blood circulation if the density of the ligand is greatly increased. For instance, folic acid-modified liposomes will adsorb large amounts of natural IgM after intravenous injection, leading to a series of unexpected off-target effects, rapid clearance and enhanced immunogenicity¹⁰⁰.

The above problems can be effectively relieved by accurately controlling the presentation mode of multivalent ligands on the carrier to construct a non-uniform distribution of ligands. In this way, efficient utilization of ligands and the enhancement of the effectiveness and specificity of targeted delivery could be realized. Sempkowski et al.¹⁰¹ designed lipid-based vesicles (Fig. 5A) containing HER-2-targeting short peptides (KCCYSL), and sticky vesicles were constructed so that ligands (shown in green) were uniformly distributed over the surface of the vesicle during circulation (pH 7.4), and preferably partitioned within the lipid phase-separated domain in the acidic tumor interstitium (7.0 > pH > 6.0), resulting in low reactivity in the circulation and high reactivity even in cells with few copies of targeted receptors. This strategy increased the local ligand density of lipid vesicles and their ability to recognize target tissues in acidic environments, and decreased specific interactions with normal cells with low receptor expressions. Moreover, Poon et al.¹⁰² constructed self-assembling linear dendritic polymers (LDPs) by hydrophilic/hydrophobic interaction (Fig. 5B). The experiments demonstrated that the titer relationship did not increase linearly but reached saturation dynamics. Once passing the allowable number of ligand-receptor binding events within the binding space, the presence of excess ligands clustered in a small binding area and would result in steric binding interference, lowering the binding energy. Their findings indicated that changing the focus from ligand density to specific manipulation of ligand cluster presentation on molecularly targeted NPs may have important implications for cell targeting, leading to the development of more smart and effective targeted delivery systems.

Figure 5.

(A) pH-tunable sticky vesicles (left) and conventionally functionalized nanoparticles with pH-independent uniform distributions of targeting ligands (right). Reprinted with the permission from Ref. 101. Copyright © 2016, American Chemical Society. (B) Chemical structure and self-assembly of linear dendritic polymers (LDP). Reprinted with the permission from Ref. 102. Copyright © 2010, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

In addition, Janus nanoparticles (JPs) are composed of two or more anisotropic compartments¹⁰³. It makes them ideally suitable for modification with different functional molecules to achieve asymmetrical multivalent ligand modification. Recently, Liu et al.¹⁰⁴ developed an entirely synthetic, multivalent, Janus nanotherapeutic platform named Synthetic Nanoparticle Antibodies (SNAbs). For myeloid-derived immunosuppressor cells (MDSCs)-targeting SNAbs, they modified onto one “face” of the JPs with G3 peptides which efficiently targeted MDSCs, and cp33 peptide was modified on the opposite “face” for the binding to Fcγ receptors (FcγRs) on immune effector cells. The anisotropy in biological functions of the different faces enabled the effective pairing of target cells with effector cells.

2.3.2. The effect of ligand-receptor binding/dissociation kinetics on drug delivery

Even if ligand-modified drug delivery systems are demonstrated efficient in vitro, the cellular uptake and intracellular transport of NPs are still challenging in pursuing satisfactory in vivo performances. Various extracellular physiological barriers exist, and the competitive affinities of non-target regions for the ligands act as a specific extracellular barrier sequestrating ligand-modified drug delivery systems. In the past, most studies focused only on the affinity constant (K_D) in the assessment of this competitiveness. However, ligand-receptor interaction is a dynamic process orchestrated by several parameters including binding (K_on), dissociation (K_off), and K_D = K_off/K_on, thus similar affinities can be indicative of a completely different process of interaction. Therefore, a more comprehensive discussion of the binding/dissociation process is necessary.

A high affinity for the cell membranes or the membrane receptors is commonly believed to be beneficial to the targeted delivery of anti-tumor drugs. In the extracellular process, however, when the binding affinity with non-target regions becomes too strong, the extracellular affinity barrier occurs with as-increased strength, impeding the subsequent journey to the target sites¹⁰⁵. Many solid tumors such as pancreatic ductal adenocarcinoma (PDA), non-small cell lung cancer (NSCLC) and certain breast cancers, display tumor-associated fibroblast cells (TAFs) in their microenvironment, which could also form a major component of the binding site barrier (BSB)¹⁰⁶, intercepting in the way of vehicle delivery and reducing their final accumulation in tumor cells^107,108. Miao et al.¹⁰⁹ demonstrated that NPs were preferentially distributed in fibroblasts due to the strong binding affinity between NPs and TAFs (K_D = 10.11 nmol/L). Therefore, the number of NPs uptaken by tumor cells was reduced and the barrier effect of TAFs-induced BSB was increased. Typically, for brain tumor-targeted delivery, previous efforts have exploited the transferrin (Tf)/transferrin receptor (TfR) pathway to enhance the brain uptake of macromolecules through receptor-mediated transcytosis. However, Yu et al.^110,111 demonstrated that a reduction in Tf/TfR affinity could enhance the receptor-mediated transcytosis of anti-TfR across the mouse blood–brain barrier and achieve the drug concentration needed for treatment (Fig. 6A). In a therapeutic circumstance, the enhanced peripheral antibody concentrations could compensate for the relatively reduced affinity towards the TfR, ensuring the final interaction with the target cells and the overall internalization. Strategies such as cleavable ligand modification have been proposed according to such problems¹¹². For example, Lei et al.¹¹³ developed a nanocleaner modified with KLVFF peptide and acid-cleavable DAG peptide, which could assist in cellular uptake and detach in acidic endosomes, promoting the transcytosis of the DDS from endothelial cells into brain.

Figure 6.

(A) A model in which the affinity of anti-TfR is inversely proportional to the absorption of TfR in the brain. Reprinted with the permission from Ref. 111. Copyright © 2011, The American Association for the Advancement of Science. (B) Model for the role of nanocarrier (NC) affinity on transcytosis and lysosomal trafficking. (The thickness of each black arrow represents the efficiency of the step indicated, where increasing arrow thickness corresponds to faster rate. The orange dotted lines indicate that the effect of NC valency on the efficiency of the secretion part of transcytosis was not assessed.) Reprinted with the permission from Ref. 117. Copyright © 2020, Elsevier B.V.

The dynamic characteristics of ligand-receptor interaction can be used to design the corresponding smart drug delivery system for different tumor tissues. In the case of solid tumors above, it would be an effective solution to construct a fast-binding/fast-unbinding model to circumvent the affinity barrier and BSB. But for tumor cells at dynamic or flowing state, such as the leukemia cells and circulating tumor cells, the presence of fluidic shear stress in blood circulation might be not favorable for the binding of ligand-modified nanodrugs with their target receptor, so it would be necessary to construct a fast-binding/slow-unbinding model. Song et al.¹¹⁴ used two αvβ3 ligands (RGDm7 and DT4) with different binding rates to assemble dual-targeting nanovesicles. RGDm7 served as a “slow-binding/slow-unbinding” ligand (K_D = 6.193 μmol/L) and contributed to the stable binding with αvβ3¹¹⁵. In the meantime, DT4 showed rapid association and dissociation (K_D = 0.797 μmol/L) and made the vesicles adhere quickly to the flowing tumor cells and transfer the drug to the nucleus¹¹⁶. It was demonstrated that the potency of the dual-targeting vesicles for flowing tumor cells was superior to that for static tumor cells.

Similarly, the unsuitable affinity of receptor-ligand could also impede the intracellular transport process. The design of appropriate affinity could result in rational kinetic processes and regulations of intracellular transport, thus achieving a better therapeutic outcome. Manthe et al.¹¹⁷ used polystyrene nanocarriers targeting the intercellular adhesion molecule-1 (ICAM-1) to investigate the relationship between cellular uptake and intracellular transport of the nanocarriers with varying degrees of affinity (Fig. 6B). The results indicated that nanocarriers with high affinity (K_D = 177.2 pmol/L) had clear advantages in binding and uptake, but the nanocarrier-receptor detachment post-transport was more compromised, and this led to enhanced lysosomal transport from the basolateral side. In contrast, the low-affinity (K_D = 401.4 pmol/L) nanocarriers, which could detach more easily from the basolateral cell surface and traffic more slowly to lysosomes, would also experience lower binding and uptake on the apical side. Improving the first steps in this process could hinder the latter steps, and vice versa. Therefore, it was important to achieve a balance where nanocarriers had an intermediate affinity (K_D = 218.7 pmol/L) for apical binding and uptake without hindering the basolateral detachment or facilitating massive lysosomal trafficking. This is in accordance with the discussions in the previous section. In other words, only appropriate number of ligand modification and intermediate affinity can achieve satisfactory targeted delivery of anti-tumor drugs.

3. Carrier-based smart drug delivery

The development of varied carriers or vehicles have enabled versatile design and modification for the interface between therapeutic agents and patients. Both conventional and newly-developed carrier platforms provide fundamental supports for the various smart drug delivery strategies. In this part, several representative drug carriers and their recent developments are discussed, including lipid-based carriers, polymeric nanoparticle-based carriers, micelles, self-assembled chemical drugs-based DDSs, nucleotide-based DDSs, self-assembled peptide-based systems, and cell-derived/biomimetic delivery systems, with multiple points of view.

3.1. Lipid-based carriers for smart drug delivery

Lipid-based carriers have been under intensive investigations in anticancer delivery, demonstrating satisfactory performance in drug delivery, with relatively mature synthesis of lipid materials as well as scale-up manufacturing processes^118,119. Besides the conventional liposomes, various other lipid vehicles have emerged, e.g., lipid nanoparticles120, 121, 122. Lipid-based carriers have seen various successful clinical translations, and their capacity in cargo loading and functionalizations enabled the further development of lipid-based smart delivery systems¹¹⁸.

Liposome is the most common vehicle in lipid-based nanocarriers, and have been widely applied for chemotherapeutic delivery due to their unique ability to encapsulate hydrophilic and hydrophobic agents and targeting characteristics caused by the enhanced permeability and retention (EPR) effect or ligand modification¹²³. Since 1995, liposomal doxorubicin, Doxil® was first approved by U.S. Food and Drug Administration (FDA) to treat ovarian cancer and AIDS-related Kaposi's sarcoma¹²⁴. Subsequently, liposomal daunorubicin (DaunoXome®), mifamurtide liposomes (Mepact®), and vinCRIStine sulfate liposomes (Marqibo®), etc. were developed for various cancers¹²⁵.

Liposomes make a promising platform to improve tumor deposition and achieve on-demand drug release. Many pH-sensitive liposomes, for example, are designed to disintegrate in the acidic lysosomal environment after internalization, releasing encapsulated drugs in the desired target. Furthermore, some pH-sensitive liposomes were reported to exhibit enhanced tumor accumulation and cellular internalization, caused by changes in physicochemical properties such as size and surface charge^126,127. Additionally, other stimuli-responsive smart liposomes in response to internal triggers (e.g., enzymes, and redox potential) and external guides (e.g., magnetic field, ultrasound, and NIR excitation) have emerged as promising drug delivery systems pioneering in antitumor smart drug delivery⁵. Alpizar et al.¹²⁸ designed light-triggered liposomes for target therapy, which could successfully evade innate immune cells and then be internalized with surface potential switching from neutral to positive upon in situ irradiation. Ji et al.¹²⁹ constructed a matrix metalloproteinase-2 (MMP-2)-responsive peptide-hybrid liposome to specifically release pirfenidone at the pancreatic tumor site and down-regulate the multiple components of extracellular matrix (ECM), enhancing the antitumor efficacy of encapsulated gemcitabine. These stimuli-responsive liposomes exhibit promising application in enhancing drug delivery and achieving on-demand drug release in targets.

Liposomes could also serve as combination delivery platforms for multiple therapeutic agents, enabling smart combinational therapy strategies such as chemo-photothermal therapy. Photothermal ablation can synergistically enhance the therapeutic effect of chemotherapeutics by elevating cell membrane permeability and triggering drug release at the target site¹²³. ICG is a quintessential NIR dye for several clinical applications approved by FDA. Liposomal co-encapsulation of chemotherapeutics and ICG remarkably prolonged the blood circulation time and enhanced the tumor accumulation, proving satisfactory chemo-photothermal combinational therapeutic effect¹³⁰. Liposomes with chemo-photothermal therapy are expected to become the next clinically effective anti-tumor strategy.

Lipid nanoparticle (LNP) is another lipid-based carrier with a solid core structure, and has become a promising platform for the delivery of hydrophobic and hydrophilic small-molecule drugs as well as biological agents such as oligonucleotides, peptides, and vaccines^120,131. Various LNP vaccines have been quickly developed and played vital roles in the COVID-19 pandemic, e.g., the Moderna vaccine¹³², and the application of LNP in nucleic acid (NA) delivery has also been highlighted¹³³. Endosomal escape is a vital process for many NA drugs, and the utilization of ionizable lipids and functional helper lipids could provide essential pH-responsiveness in the endosome¹³⁴. Onpattro® (anti-transthyretin siRNA-loaded LNP) contains an ionizable cationic lipid (DLin–MC3–DMA) which could convert the potential of liposomes to positive in the endosomal environment, resulting in the release of siRNA in target cells^135,136. With the accumulating library of lipid materials with various functions and characteristics, LNPs holds great potential for versatile delivery of a universal selection of therapeutic agents, with greatly shortened time for formulation screening and optimization. To this end, however, the actual quality of relevant lipids in both laboratory research and scaled-up production requires more emphasis.

Besides liposomes and LNPs, lipids assist in various promising platforms such as nanostructured lipid nanocarriers (NLCs), lipid-drug conjugates, self-emulsifying systems, etc137, 138, 139, 140. Lipid-based smart vehicles promise a bright future for cancer treatment in the next decade or so. They might become a major arsenal for safer and more efficient treatments by ensuring proper drug localization in tumors and on-demand drug release. Although substantial advances have been made, efforts are still needed to solve the problems of scale-up and clinical verification to transfer lipid-based therapy from laboratories to clinics.

3.2. Polymeric nanoparticles-based smart drug delivery

Nanoparticles based on biocompatible and biodegradable polymers such as polylactic acid (PLA), PLGA, PEG, and N-(2-hydroxypropyl) meth-acrylamide (HPMA) have gained popularity for nanocarrier fabrication. Through chemical conjunction or physical encapsulation, polymer nanoparticles could deliver chemotherapeutics, proteins, and nucleic acids to the tumor site¹⁴¹. Several polymeric nanoparticles have been approved for clinical application, e.g., Eligard® (Tolmar) containing poly(d,l-lactide-co-glycolide) has been approved by the FDA for prostate cancer therapy¹⁴². The clinical trials of BIND-014 (docetaxel-loading polylactide core with PEG corona and ACUPA-targeting ligands), CRLX101 (nanoparticle with camptothecin-conjugated polymer backbone based on cyclodextrin and PEG), and AZD-2811 Accurin (AZD-2811-encapsulated PLA–PEG block copolymer nanoparticle) also provide valuable experiences for the translation of smart polymeric nanoparticles. In recent years, combined with specific ligands and responsive groups, smart polymeric nanoparticles are at the forefront of antitumor drug delivery with their unique advantages.

Due to high synthetic versatility and ease of conjugations, polymers were suitable for functionalization and ligand modification. For example, Qiu et al.¹⁴³ synthesized an IL12 plasmid (pIL12)-loaded polyplex constructed with esterase-responsive cationic polymer PQDEA, which was further coated with various lipids as well as DSPE–PEG conjugated with tumor-targeting ligand AEAA. Responsive to the high-level esterase in tumor cells and tumor-associated macrophages (TAMs), these smart nanoparticles turned anionic and quickly released the cargo, thus efficiently producing IL-12, activating anticancer immune responses, and remodeling the tumor microenvironment. Zou et al.¹⁴⁴ functionalized cNGQGEQc peptide onto reversibly crosslinked PEG–P(TMC–DTC)–PEI and cNGQ–PEG–P(TMC–DTC) chimeric polymersomes (cNGQ/RCCP), achieving prolonged circulation, efficient targeting, and GSH-responsive release. Yin et al.¹⁴⁵ took advantage of the affinity of hyaluronic acid (HA) for CD44 to create HA–ss–PTX, which could be specifically absorbed by tumor cells via CD44-mediated endocytosis and disrupted by reducing glutathione to release PTX.

The acidic tumor microenvironment provided the basis for intelligent response at low pH. For example, dual sensitive dual drug backboned shattering polymer self-assembled nanoparticle (DD-NP) was triggered intracellularly to break down and release the dual drugs payload in a chain-shattering manner under the intracellular acidic microenvironment for optimal anticancer efficacy. Cong et al.¹⁴⁶ claimed that DD-NP could be a good example of nanomedicine tackling the major challenges together, including precise composition, direct fate monitoring of drug, drug evaluation, and screening on reliable cancer models, validating the possible use of DD-NP in the clinic. In another example, Tang et al.¹⁴⁷ synthesized charge-switchable polymeric nanoparticles, which were consisted of PEG–block–poly[(1,4-butanediol)-diacrylate-β-5-amino-1-pentanol] (PEG–PDHA), and conjugated with 2,3-dimethyl maleic anhydride (DMA). In the slightly acidic environment of tumor tissues, the anionic shell was removed, inducing the conversion of the surface charge from negative to positive, which resulted in higher intra-tumor accumulation, more efficient cellular uptake, and stronger cytotoxicity. These charge-convertible polymers represented a class of typical intelligent polymers that can afford tunable physical and structural changes that are envisioned to address critical issues in response to the stimulus at the tumor site¹⁴⁸.

Photoactivatable polymeric nanoparticles enabled remotely controlled drug and imaging agent delivery. For example, Senthilkumar reported photo-responsive poly(p-phenylene vinylene) conjugated polymer nanoparticles (CPNs) functionalized with donor-acceptor Stenhouse adduct (DASA) and folic acid units for drug delivery and imaging¹⁴⁹. Notably, drug-loaded CPNs exhibited excellent biocompatibility in the dark, indicating perfect control of the light trigger over drug release. In another work, Zeng et al.¹⁵⁰ reported the synthesis of an amphiphilic triblock copolymer, PolyPt/Ru, consisting of biocompatible PEG, reduction-responsive Pt(IV), and red-light-responsive Ru(II) moieties, which further improved the selectivity of the cancer treatment, though photoactivation at the irradiated tumor tissue (Fig. 7).

Figure 7.

Illustration of amphiphilic triblock copolymer PolyPt/Ru (A) and the self-assembly, circulation and internalization process of PolyPt/Ru (B). Reprinted with the permission from Ref. 150. Copyright © 2020, Wiley-VCH GmbH.

Functionalization always leads to a more complex structure and changes the original characteristics of the polymer. Polymeric nanoparticles without biodegradability may cause serious side effects after their accumulation at the tumor site. With the rapid emergence of novel polymeric nanoparticles, safety evaluation is important to identify polymeric candidates.

3.3. Micelle-based smart drug delivery

Many chemotherapeutics are hydrophobic small molecules, and the hydrophobicity often becomes a major barrier for their systemic delivery to solid tumors. Over the past few decades, polymeric micelles (PMs) have provided a promising platform for the delivery of poorly soluble drugs. PMs are usually self-assembled from amphiphilic block copolymers with nanoscale size (commonly 10–100 nm). Among these polymers, poly(ethylene glycol) (PEG), poly(vinyl alcohol) (PVA), poly(N-vinyl-2-pyrrolidone) (PVP), poly[N-(2-hydroxypropyl) methacrylamide] (PHPMA), and polysaccharides are commonly used as core-forming segments, enabling the encapsulation of hydrophobic molecules. Besides, polyesters and poly(amino acid) (PAAs) have drawn broad attention for shell forming¹⁵¹, and the hydrophilic outer shell could protect the drug-loaded core. The core-shell structure of micelles has enhanced the encapsulation and selective tumor targeting of many therapeutic agents. For example, PEG–b–PLA micelles for PTX have realized enhanced drug solubilization, biocompatibility, and dose escalation¹⁵².

Up to now, various micelle formulations including Paclical, Genexol-PM, and Nanoxel-PM have been approved for PTX delivery. Also, hydrophilic polypeptides have been developed to mimic the conformation of synthetic hydrophilic polymers. Banskota et al.¹⁵³ developed new unstructured polypeptides called zwitterionic polypeptides (ZIPPs), which could self-assemble into micelles after conjugation with hydrophobic paclitaxel, with a 17-fold-longer half-life compared to free paclitaxel in the HT-29 colon cancer model.

Recently, intelligent polymeric micelles have been developed to respond to different stimuli, including light-, ultrasound-, temperature-, pH-, enzyme-, redox-sensitive systems, for achieving spatiotemporal control of their therapeutic effects. For example, Li and coworkers¹⁵⁴ fabricated a micelle drug delivery system based on poly(AAm–co–AN)–g–PEG with an upper critical solution temperature (UCST) of 43 °C, with thermal-sensitive drug release combined with microwave hyperthermia. Su et al.¹⁵⁵ developed a pH and MMP-2 dual-responsive Azide–PEG–PAsp (Dip/Bz) copolymer for antitumor drug delivery, spatiotemporally controlling the release of PD-1 and PTX from MMP-2 enriched tumor tissues. Additionally, surface modification of PMs with ligand molecules such as peptides, antibodies, and small molecules also has wide applications for advanced tumor therapy. For example, iRGD-modified micelles were developed for αvβ integrin and neuropilin-1-mediated blood–brain barrier penetration and targeted delivery to glioma cells¹⁵⁶. PMs system has also been regarded as a promising platform for imaging and diagnosis, such as tumor micro-environment and tumor tissue imaging, etc. Among these materials, star polymer-based unimolecular micelles are widely investigated, which could be obtained through rational structure design, and different kinds of imaging probes can be encapsulated or labeled¹⁵⁷.

The past decade has witnessed explosive development of biodegradable micelles for targeted and controlled anticancer drug delivery. Considering their advantages such as reduced immunogenicity, biocompatibility, biodegradability, highly structural and chemical variability, amphiphilic block copolymers like PHPMA, PVA and polyesters have the potential to achieve clinical translation. Nevertheless, the clinical translation is a lengthy, costly, and complex process, successful examples of PMs from bench to bed are few. Several challenges associated with their use concern the disassembly, degradation, clearance, and metabolism of PMs in the blood stream, besides, the long-term effects of some PMs remain to be studied, especially following repeated and cumulative administration¹⁵¹. Thus, polymers for stable micellar structure, active targeting ligands, and combination delivery of multiple drugs are potentially effective approaches.

3.4. Self-assembled chemical drugs-based smart drug delivery

Conventional organic nanocarriers usually consist of lipids or polymeric materials, such as liposomes, vesicles, micelles, and polymeric nanoparticles (NPs)¹⁵⁸. Interestingly, some chemical drugs or prodrugs are also adopted to participate in the construction of these nanocarriers themselves159, 160, 161. The most remarkable feature of chemical drug-based nanoassembly is its self-delivery capacity, namely drugs or prodrugs could simultaneously serve as both cargoes and vehicle materials in the nanosystems^159,160,162. As a result, chemical drug-based nanoassembly usually shows ultrahigh drug loading capacity, sometimes even reaching 50%^159,160,162. Recently, this kind of nanocarrier has drawn more and more attention as a promising and efficient drug delivery system. In this section, we focus on the latest trends in chemical drug-based nanoassembly (Fig. 8), including pure drug-based nanoassembly and prodrug-based nanoassembly.

Figure 8.

Schematic diagram of chemical drug-based nanoassembly.

3.4.1. Pure drug-based nanoassembly

Over the years, several anticancer drugs have been formulated into nanomedicine with the help of carrier materials, such as paclitaxel, doxorubicin, and irinotecan¹⁵⁸. However, low drug-loading efficiency and carrier material-related toxicity have been widely regarded as the main obstacles to the clinical translation of nanomedicines. Therefore, the rational design of carrier-free nanomedicines has always been a high priority. Several chemotherapeutic drugs have been recently found to show self-assembly characteristics in water, such as 10-hydroxycamptothecin, doxorubicin, and curcumin^159,163, 164, 165. Moreover, photosensitizers such as 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide (DiR) and zinc meso-tetra(4-pyridyl) porphyrin (ZnTPyP) were also found to successfully self-assemble into stable NPs for imaging-guided photothermal therapy (PTT) or photodynamic therapy (PDT)^166,167. In addition to single drug nanoassembly, the co-assembly of two or more drugs was also explored for multimodal cancer therapy168, 169, 170.

The self-assembly or co-assembly of drugs was found to be driven by multiple interactions, including hydrophobic force, hydrogen bonding, π–π stacking, and electrostatic interactions, etc.¹⁵⁹. Despite the self-assembly capacity of hydrophobic drug molecules, amphipathic polymers or PEG were usually utilized in these systems to further enhance the colloidal stability and prolong the circulation time¹⁵⁹. Distinguished from polymeric micelles, PEG polymers mainly acted as PEGylation modifiers instead of carrier materials in pure drug-nanoassembly¹⁵⁹.

3.4.2. Prodrug-based nanoassembly

Despite their simple fabrication process and ultrahigh drug loading capacity, it is still challenging for pure drug-based nanoassemblies to achieve tumor stimuli-responsive drug release. In contrast, rational design of small-molecule prodrugs and prodrug-nanoassemblies could not only modify the physicochemical properties of drugs but also realize selective drug release, e.g., by inserting tumor-specific stimuli-sensitive linkers in the prodrug molecules¹⁶⁰. Moreover, many drugs are not able to form nanoassemblies themselves, but proper chemical modifications could endow them with self-assembly capability¹⁶⁰.

According to the molecular structures of prodrugs, small molecular prodrug-based nanoassembly could be summarized as three types: (i) nanoassembly of amphiphilic prodrugs; (ii) nanoassembly of hydrophobic prodrugs; and (iii) nanoassembly of dimeric prodrugs^160,171. Among them, relatively few studies have been focused on amphiphilic prodrug-based nanoassembly, due to their relatively large critical aggregation concentrations (CACs) and unsatisfactory dilute stability in blood^160,171. An increasing number of hydrophobic drugs and prodrugs have been found to have self-assembly capacity, despite the past views that they could hardly form stable NPs in water^159,160,171. Notably, the nanoassembly of hydrophobic prodrugs showed excellent stability after modification with a small amount of PEG polymers. The self-assembly ability and good stability of hydrophobic drug nanoassemblies could be attributed to multiple intermolecular interactions/forces including hydrophobic forces, hydrogen bonding, π–π stacking, and electrostatic interactions; while the assembly of amphiphilic prodrugs is mainly considered to be driven by hydrophobic forces^159,160,171.

Despite the specific self-assembly features, the effective hydrolysis of hydrophobic prodrugs remains challenging. To address this problem, tumor stimuli-responsive strategies by inserting chemical linkers in the prodrugs has been proved to be effective¹⁶⁰. Recently, a wide range of redox-responsive prodrug-nanoassemblies have been developed by utilizing reduction-, oxidation- or redox dual-sensitive chemical linkers, such as thioether bond, disulfide bond, trisulfide bond, thioketal bond, and diselenide bond, etc172, 173, 174, 175, 176, 177. In addition to triggering the tumor-specific activation of prodrugs, the bond angles and/or dihedral angles of these redox-responsive linkers play crucial roles in the stability and in vivo drug delivery efficiency of prodrug-nanoassemblies174, 175, 176, 177.

In addition to monomeric prodrug-based nanosystems, the nanoassembly of dimeric prodrugs represents another unique nanoplatform, including homodimer-nanoassembly and heterodimer-nanoassembly¹⁷¹. The nanoassemblies of homodimer synthesized by conjugating two identical drug molecules could exhibit higher drug loading efficiency than monomeric prodrug-nanoassemblies, sometimes even up to 80%^171,175,176. Moreover, the nanoassembly of heterodimers synthesized by coupling two different therapeutic agents provides a new platform for combination cancer treatment^171,178. Notably, most dimeric prodrugs demonstrated relatively poor assembly ability when compared with monomeric prodrugs, especially the homodimers with symmetric molecular structures¹⁷⁵. Moreover, although heterodimer-nanoassembly provides a natural platform for combination cancer therapy, the molar ratios of drugs were strictly limited to 1:1, which is detrimental to exerting the synergistic effects in different cases¹⁷¹. Still, heterodimers of two drugs with different therapeutic modes might be more flexible in producing synergistic effects, such as chemo-photodynamic heterodimers¹⁷⁸.

3.5. Nucleotide-based smart drug delivery

The nucleotide is a type of organic molecule consisting of a nucleobase, a sugar ring, and a phosphate, while such simple structure is highly important in biology. It is the basic element of endogenous nucleic acids for a broad range of physiological functions, such as DNA, mRNA, and microRNA. From a chemical standpoint, DNA has ∼10⁶-fold higher stability than RNA due to its deoxyribose structure¹⁷⁹. As such, DNA is preferably chosen by nature as the carrier to store genetic information, forming a Watson–Crick base pairing structure as discovered in 1953. Later in the 1980s, it was found that DNA was able to form much more complicated architectures than merely linear duplexes, based on which the concept “structural DNA nanotechnology” was coined to describe the higher-order DNA structures. Since then, various elegant DNA nanostructures have been designed with arbitrary sizes, shapes, and functions, such as DNA polyhedra, DNA origami, DNA dendrimer, DNA nano-train, DNA nano-flower, providing versatile vehicles for smart drug delivery^161,180. Besides the evolution of structure, the functions of DNA have also been extensively expanded by virtue of the rapid progress of molecular biology and chemical biology, resulting in the discovery of various types of functional nucleic acids¹⁸¹. One typical example is the aptamer, which is artificially obtained through a combinatorial process called SELEX. Theoretically, aptamers can be isolated to selectively bind any target of interest, and currently, a wide range of aptamers have been discovered with binding substrates of metal ions, small molecules, proteins, and even the whole cells¹⁸². Among them, the tumor cells recognition aptamers have attracted particular attention, which elicits the development of both tumor diagnostic probes and tumor-targeting drug delivery systems¹⁸³.

Compared with many other synthetic polymers, the endogenously derived DNA has various appreciable advantages as drug loading carriers for in vivo delivery. Owing to the development of the solid-phase synthesis technique, DNA is readily commercially available for large-scale synthesis via automated phosphoramidite chemistry with relatively low prices, making it accessible to most ordinary research labs. In addition, DNA with various modifications for subsequent conjugation as well as fluorophore-labeling can be easily designed, which significantly expanded the functionalities of the DNA-assembled nanostructures, such as the delivery of multiple drugs, active targeting modifications, and imaging-guided tumor therapy. The biomimetic DNA is highly biocompatible and biodegradable, which meets the stringent demand as drug carriers for in vivo applications. More importantly, DNA has a well-characterized conformation and geometry, in which the double-stranded B-DNA is ∼2 nm in diameter with a helical repeat of 3.4 nm for every 10.5 base pairs. Through the precise A/T and G/C base-pairing, DNA nanostructures with different sizes, shapes, and dimensions can be rationally designed via tuning the building block DNA sequences and incubation conditions¹⁶¹. Besides base-pairing-based assembly, DNA nanostructures can also be formed by more elegant strategies, such as polymerase chain reaction (PCR), rolling circle amplification (RCA), enabling the in situ generation of nanoparticles with a high level of reproducibility and efficiency¹⁶¹.

While the nucleotide has quite a simple chemical structure, the nucleotide-based nanoassemblies are able to deliver a wide range of therapeutic drugs, including chemotherapies, photodynamic agents, radio-therapeutics, oligonucleotides-based, and even protein-based drugs^161,180. The drugs can be loaded into DNA nano-architectures via either physisorption or chemical conjugation. Among various antitumor drugs, the most commonly used example is DOX, which can readily sandwich into two adjacent pairs of bases (especially G/C pairing), achieving ultra-high drug loading into DNA nano-assembly through a simple preparation procedure¹⁸⁴. This is inspired by the antitumor mechanism of DOX, which enters the cell nucleus and intercalates among base pairs in genetic DNA, thereby preventing DNA replication and ultimately inhibiting protein expression. A similar binding pattern was also employed to load porphyrin derivative (TMPyP4), a photodynamic therapy agent¹⁸⁵. Overall, such strategy is simple, cost-effective, yet highly efficient, while it has specific requirements for the loaded drugs, such as planar or extended planar components within their structures. More generalized methods to load the payloads are through sequence hybridization and chemical conjugations. For nucleic acids-based drugs, direct hybridization is the most straightforward way to integrate into DNA nanostructure, which has been used to deliver siRNA, microRNA, DNAzyme, and CRISPR/Cas9 system¹⁸⁶. For many other drugs, however, tedious chemical conjugations are required to incorporate them into DNA nanostructures. Although DNA itself has limited functional groups for chemical conjugations, the chemical solid-phase synthesis generates ester, amide, and disulfide bonds within DNA, allowing for subsequent conjugation of drugs with different chemical properties. However, chemical modifications are only feasible for limited lengths of DNA sequences, and the following drug conjugations need complicated chemical synthesis with low yield. In some cases, the combination of different strategies was employed to construct multiple-model therapeutic systems to conquer the extreme complexity and heterogeneity of tumor¹⁸⁷. To enable tumor targeting delivery, the aptamers and other active ligands have been equipped into DNA nanostructures for targeted tumor therapy^188,189.

Smart delivery systems should release their payload after reaching the targeting site for action. DNA nanostructures can exploit various internal and external stimuli to trigger drug release. For example, pH and GSH are the most widely employed internal triggers for anti-tumor nano-delivery systems, by virtue of acidic tumor microenvironment and up-regulated GSH levels in cancer cells. It is known that cytosine-rich sequences (i-motif domain) undergo a conformational change in response to slightly acidic pH, based on which the pH-responsive DNA nano-delivery systems were formed by the integration of i-motif¹⁹⁰. Li and coworkers¹⁹¹ developed a DNA hydrogel that was crosslinked via disulfide bonds, which was cleavable in presence of GSH, resulting in burst drug release inside cells. Likewise, a photodegradable DNA-cross-linked hydrogel was designed for drug loading, and external stimulus of light-induced gel-sol conversion to trigger the rapid release of the encapsulated molecules¹⁹². Compared with other polymers, the elegance of DNA as drug carriers is their capability to sense various biological molecules. For instance, complementary sequences can be designed to hybridize with target mRNA/microRNA¹⁹³, and aptamers can recognize ATP and nucleolin^194,195, all of which have been employed to construct DNA-based smart drug delivery system. The landmark work was reported by Li and coworkers¹⁹⁵, who developed a DNA nanorobot to deliver thrombin via incorporating an aptamer that targeted nucleolin, a protein specifically expressed on tumor-associated endothelial cells. The aptamer served both as a targeting domain and also a molecular trigger to activate the DNA nanorobot, through which thrombin could be specifically delivered to tumor-associated blood vessels for intravascular thrombosis, resulting in tumor necrosis and tumor growth inhibition.

3.6. Self-assembled peptide-based smart drug delivery

Biological and bio-inspired building blocks have attracted widespread attention in smart drug delivery technologies owing to their advanced functions for both nanotechnology and biomedicine. Among various building-block molecules, peptides with self-assembly characteristics are of particular interest, not only because of their biological origin and biodegradability, but also because of their specific bioactivity derived from rationally designed sequences¹⁹⁶. In this section, we focus on the bioactivated in vivo assembly (BIVA) nanotechnology, the modular structures of peptide materials, and corresponding conditioned self-assembly mechanism, as well as their application in the imaging and therapies against tumor (Fig. 9). BIVA technology could provide active targeting and assembly, inducing tumor-specific accumulation and prolonged retention. Based on different modules, versatile functions could be endowed onto the BIVA system, including but not limited to prolonged circulation, targeting, self-assembly, imaging, etc. Ren et al.¹⁹⁷ reported a BIVA optical nanofiber probe with five different functional modules which could self-assemble on the tumor surface, with prolonged tumor imaging and precise recognition of <2 mm orthotopic pancreatic tumor in vivo.

Figure 9.

Schematic illustration of bioactivated in vivo assembly (BIVA) nanomaterials including modular molecule designs, conditioned assembly mechanisms, and biomedical applications.

Modular design enables efficient production, maintenance, and customization across modern engineering technologies. In the terms of peptide design, modular peptide building blocks with various functions, such as targeting motif, response motif, self-assembly motif, and functional motif, are employed for the construction of delivery platforms, providing advanced scope and accuracy in their metabolic steps, associated enzymes, and biological function parts, and assisting in the programmable and controllable regulation of the assembled structure and the morphology¹⁹⁸. Combining with molecular dynamics (MD) simulation, the pre-designed compound and its assembled behavior can be further predicted. Remarkably, the molecular assemblies can be synthesized, assembled, and characterized in a rapid manner with desirable performance¹⁹⁹. Also, recent reports suggest the potential of biosynthesis for the efficient combination of diverse modules and cargos with diverse structures in anticancer immunotherapies and relevant other applications²⁰⁰.

Enzymatic triggers are commonly adopted self-assembly mechanisms in vivo, including enzymatic elimination, cyclization, and polymerization, as well as metal ion coordination. For in situ self-assembly of peptides in physiological conditions, a simple and efficient approach is to modify molecules with responsive molecular elimination to remove hydrophilic and steric hindrance parts, thus the overall balance of hydrophobicity and hydrophilicity could lead to in situ assembly^201,202. Additionally, π–π interactions could also serve as a driving force to influence the reaction pathways of cyclization, altering self-assembly states^203,204. Besides, the polymerization based on elongation of the peptide monomers-induced polymeric phase transition can provide an efficient route to form divers self-assembly. These nanoscale architectures and polymers have myriad possibilities when directly incorporated by polymerization and in situ assembly^205,206. Coordination-based self-assembly, a strategy that uses metal coordination as an important force to direct the association of peptide molecules, has recently developed into an elegant method for the precise design of theranostic systems by combining the advantages of both peptide self-assembly and metal coordination interactions²⁰⁷. Assembly is an equilibrium process between the individual building units and their aggregated state, the practical self-assembly of molecules should be precisely controlled by concentration in solvent²⁰⁸.

The BIVA technology-based peptide candidates exhibit promising drug delivery capability and advanced drug substitute ability for cancer therapy^209,210 and unique advantages, including reduced side effects, optimized pharmacokinetics and pharmacodynamics211, 212, 213.

3.7. Cell-derived/biomimetic smart drug delivery

Endogenous materials have been widely used in the research of drug delivery systems because of the advantages of good biocompatibility, low immunogenicity, and retention of endogenous biological functions. In recent years, endogenous materials such as biomembrane-coated vectors and extracellular vesicles (EVs) have received more and more attention. Compared with traditional nanomaterials, biomembrane-coated NDDSs can mimic many natural characteristics of their cells of origin214, 215, 216. The current biomembrane-coated nanomaterials in research mainly include red blood cell (RBC) membrane²¹⁷, platelet membrane²¹⁸, immune cell membrane²¹⁹, tumor cell membrane²²⁰, and hybrid cell membrane²²¹. RBCs and platelets have been intensively studied due to their non-nucleus characteristics. Yang et al.²²² synthesized a thermal-sensitive nitric oxide (NO) donor S-nitrosothiols (SNO)-pendant copolymer [poly(acrylamide-co-acrylonitrile-co-vinylimidazole)-SNO copolymer, PAAV-SNO] with upper critical solution temperature (UCST), and RBCs were together adopted to fabricate an erythrocyte membrane-camouflaged nanobullet for codelivery of NIR-II photothermal agent IR1061 and indoleamine 2,3-dioxygenase 1 (IDO-1) inhibitor 1-methyl-tryptophan (1-MT), realizing stealth in the circulation as well as light- and heat-regulated release at the tumor site. Platelet membranes consist of CD47 and P-selectin, which could help escape from macrophages and provide high affinity with tumor cells, respectively²²³. Wang et al.²²⁴ utilized platelet membranes to coat the nanoparticles formed by chitosan oligosaccharide (CS)-PLGA copolymer (PCS-PLGA NPs) for delivering anticancer drug bufalin, which exhibited prolonged half-life and significant accumulation in the tumor site due to P-selectin. Immune cell membranes, e.g., macrophage, T cell, and natural killer cell membranes, etc., can effectively reduce opsonization and non-specific clearance, and provide cellular recognitions or targeting abilities²²⁵. Also, the application of hybrid cell membranes might retain the inherent functions of both cell membranes. Jiang et al.²²⁶ designed erythrocyte-cancer hybrid membrane-camouflaged melanin nanoparticles using RBC membrane and MCF-7 membrane, the bionic nanoparticle exhibited prolonged blood circulation time and more accurate targeting to MCF-7 cells.

Exosomes are a type of EVs, containing components from the parent cell such as DNA, RNA, lipids, proteins, etc.²²⁷ Besides, exosomes possess natural composition, low immunogenicity, reduced cytotoxicity, favorable penetration through physiological barriers, and tumor cell-derived exosomes exhibit specific targeting towards parent tumor cells²²⁸. Also, extra characteristics and smart functions could be endowed on exosomes with ligand modification, cargo loading, etc.²²⁹ These exosomes were used as ideal carriers for delivering nucleic acids and proteins to tumor sites²³⁰. Towards the large-scale production of targeted and therapeutic exosomes, cell nanoporation technology was adopted by Yang et al.²³¹, achieving increased exosome production and transcription of therapeutic mRNA, as well as the expressing of targeting peptide on the N terminus of exosomal CD47. The produced exosomes realized efficient penetration through the blood–brain barrier and targeted delivery to the glioma sites with enhanced therapeutic effect. The exosomes could also provide various biological properties for the modification of NPs²³². In recent years, biomimetic drug delivery systems have been extensively studied, but for clinical applications, more in-depth researches are needed.

4. Future perspective

At present, with the emphasis on precision and comprehensiveness in clinical cancer treatments, new requirements are put forward for tumor-targeted NDDS, urging the development of flexible and adjustable formulation, controllable interval and dosage of administration, spatiotemporally-precise delivery, and easy clinical translations. The construction of NDDS by smart self-assembly has become an important trend in modern smart/stimuli-responsive drug delivery. Based on supramolecular in vivo self-assembly, a tumor-selective cascade activatable self-detained system (TCASS) could significantly enhance the smart accumulation of self-assembly NDDS on desired targets, and promote the penetration and retention at tumor sites via aggregation/assembly-induced retention (AIR) effect²⁰². Additionally, on the combination of host-guest interactions²³³ and biomarker displacement activation (BDA)⁶⁸ strategy, some overexpressed tumor markers in the tumor microenvironment may competitively bind with the macrocyclic host, and enable controllable release of the anti-tumor agent from its disguised form and restoring its anti-tumor activity. This host-guest complex exhibits high efficacy and low cytotoxicity, hence this kind of supramolecular system is expected to become an important development direction of multifunctional nanomedicine.

In recent years, the global trend of pharmaceutical products has shifted from chemicals to biopharmaceuticals. However, the performance of conventional NDDS proves insufficient to solve the problems of poor stability and low encapsulation in the delivery of biological macromolecules such as peptides, proteins, vaccines, and gene drugs. A strategy of supramolecular peptide therapeutics (SPT) was proposed for the improvement of the stability and anticancer efficacy of the peptides in vivo. The phenylalanine was introduced into the N terminal of anti-tumor peptide (KLAK), and the encapsulation efficiency of supramolecular peptide complex reached 97% through host-guest interactions between the phenylalanine and cucurbit uril²³⁴. Considering the huge demand for peptide-based pharmaceuticals for clinical purposes, SPT might provide a versatile and potential bridge for clinical translations of bioactive peptides and proteins. Nanotechnologies also have widespread applications in vaccinology to such an extent that there is a very active research area known as “nanovaccinology”, especially in tumor-targeting. However, conventional formulation strategies still have to face the difficulty of complicated preparation processes, low antigen loading, and weak immune response. A Nano-B5 platform was built through an in vivo production process combining the self-assembly capacities of pentamer domains from the bacterial AB5 toxin and the loading capacity of unnatural trimer peptides to encapsulate diverse antigens including peptides and polysaccharides. The full-biosynthesis process of protein skeleton could avoid the introduction of exogenous materials and ensure the safety of the vaccine, and the utilization of fusion expression or protein glycosylation modification strategies could easily achieve high-efficacy loading of different antigens without chemical coupling. Compared with pure antigen, nanovaccines could avoid the rapid clearance of antigen and achieve effective enrichment in lymph nodes, and rapidly activated antigen-presenting cells (APCs) for T cell activation without additional adjuvant, thus enhancing the level of subsequent immune response, and offering a platform technology for combining diverse modular chassis components and antigen cargos to generate various high-performance nanovaccines²⁰⁰.

In addition, various individualized gene therapy drugs have also been gradually developed. Traditionally, most gene therapy delivery systems focus only on one therapeutic target, such as DNA in the nucleus or mRNA in the cytoplasm. However, spatiotemporal features of transcription (nucleus) and translation (cytoplasm) of target genes differ widely. Thus, the utilization of branched DNA structure with circular supramolecule as the core can efficiently load components through base complementary pairing-based self-assembly of nucleic acids and supramolecular host-guest recognition mechanism, then the gene-editing (sgRNA/Cas9) and gene-silencing (antisense) components can achieve level-by-level responsive release. This strategy takes advantage of the responsiveness of the assembly structure to tumor markers, realizing the synergistic dual-drug codelivery for gene editing system and gene silencing system, and offering new strategies for the treatment of malignancies²³⁵.

Despite the great progress in smart NDDS, several challenges remain in the design of biologically functional controllable components, e.g., the prediction of in vivo stimuli-responsive behavior, the standardized evaluation criteria, clinical translation and industrial production. Although novel functional materials have been under continuous development, in-depth research of their in vivo degradation and safety issues remains in urgent need. Artificial intelligence (AI), big data and multi-scale simulation technology can be introduced into the construction of smart NDDS to change the paradigm of pharmaceutical research into data-driven mode236, 237, 238. Specifically, a structurally diverse hydrogel library with more than 2000 motifs²³⁹ can link the chemical features of carriers with their self-assembly properties and accurately predict the gel formation ability by deep learning. Hydrogels offer scalable and tunable delivery platforms which can load a variety of drugs by hydrophobic domains, affinity-mediated binding, or covalent integration^240,241. These kinds of design and prediction tools represent a future direction of the smart anticancer NDDS. Carrier module libraries and drug libraries could be established in various combinations in terms of different patients and tumor classification. Accordingly, the tumor-targeting smart NDDSs are expected to be realized with simple formulation design, easy preparation, good biocompatibility, and benefit for advancing the precise treatment for cancer patients.

Author contributions

All of the authors drafted the manuscript together. All of the authors have read and approved the final manuscript.

Conflicts of interest

The authors have no conflicts of interest to declare.