Size-Tunable Strategies for a Tumor Targeted Drug Delivery System.

Nanoparticles have been widely used in tumor targeted drug delivery, while the antitumor effects are not always satisfactory due to the limited penetration and retention. As we all know, there is a paradox that nanoparticles with large sizes tend to distribute around tumor blood vessels rather than penetrate into tumor parenchyma, while smaller sizes can penetrate deeply but with poor tumor retention. In recent days, an intelligent, size-tunable strategy provided a solution to determine the size problem of nanoparticles and exhibited good application prospects. In this review, we summarize series of stimuli-induced aggregation and shrinkage strategies for tumor targeted drug delivery, which can significantly increase the retention and penetration of nanodrugs in tumor sites at the same time, thus promoting treatment efficacy. Internal (enzymes, pH, and redox) and external (light and temperature) stimuli are introduced to change the morphology of the original nanodrugs through protonation, hydrophobization, hydrogen bond, π-π stacking and enzymolysis-resulted click reactions or dissociation, etc. Apart from applications in oncotherapy, size-tunable strategies also have a great prospect in the diagnosis and real time bioimaging fields, which are also introduced in this review. Finally, the potential challenges for application and future directions are thoroughly discussed, providing guidance for further clinical transformation.

Introduction

After years of research and development, nanoparticles have been widely used in antitumor research because of their high specific surface area, easy modification, and strong targeting properties.[1,2] To passively deliver nanoparticles to tumor sites, the enhanced permeability and penetration (EPR) effect is the strategy that is mostly used, which is specific only in tumors due to the rapid proliferation of tumor cells and the abnormal tumor vasculature system.[3−5] However, more and more studies found that only delivering nanodrugs to target sites is far from enough, and accumulation and penetration problems still influence the intratumoral delivery efficacy to a great extent.[6] Therefore, scientists have tried to design nanodrugs with both good accumulation and penetration capacity in tumor tissues to achieve in situ therapeutic concentrations and good treatment efficacy.

Among all the strategies, designing nanoparticles with tunable sizes is the most intuitive and controllable approach. Many studies have found that there is a close correlation between the antitumor effect and the size of nanodrugs.[7,8] Usually, the diameter of nanodrugs is designed according to the pore size of a leaky tumor vasculature.[9] Though differences may occur owing to the variety of tumor models, subcutaneous tumors always exhibit a characteristic pore cutoff size ranging from 200 nm to 1.2 μm, and the size is further reduced in tumors that grow in the cranium such as glioma.[10] Then, size-related accumulation and penetration abilities are taken into consideration, which is a very tricky problem to keep in balance. Because of the special structure and environment of tumor tissues, there is a contradictory effect of a nanoparticle’s size on drug delivery. That is, nanoparticles with large sizes tend to be more capable of retention in tumor tissue than those with smaller sizes.[11−13] As for the permeability, things become reversed, smaller sizes have a better penetration ability in tumor tissues.[14] To fully utilize the existing paradox, researchers have designed a series of nanoparticles with intelligent tunable sizes, including intelligent size aggregation, size shrinkage, and reversible size-changing strategies, which are systematically discussed in this review.

In this review, we will summarize intelligent size-tunable strategies including size aggregation, size shrinkage, as well as reversible size changes. Each section is divided through different stimuli such as enzyme, pH, redox, light, temperature, etc. In addition to the enhanced retention and penetration, we also focus on other potential applications in different ways. Aggregation strategies can be used in enhanced cellular uptake, antimetastasis, and tumor diagnosis (photoacoustic imaging (PA), positron emission computed tomography (PET), surface-enhanced Raman scattering, and enhanced magnetic resonance imaging (MRI)), while shrinkage strategies exhibit advantages in nuclear delivery, drug release (Scheme ), etc. In the end, we conclude with the future application of size-tunable nanoparticles and existing problems that need to be solved for better treatment.

Scheme 1

Brief Illustration of Stimuli-Induced Size-Tunable Strategies with Their Potential Applications

Size Impact on Delivery Efficacy

As one of the most important characteristics of nanoparticles, size greatly influences the efficiency of tumor targeted drug delivery in many ways, including circulation, biodistribution, tumor accumulation and penetration, as well as cellular uptake and subcellular distribution. A thorough understanding of size will be introduced first to help better elucidate the importance of size-tunable strategies.

After entering into the body, the circulation time of nanodrugs basically determines the efficacy of tumor targeting as the clearance by mononuclear phagocytic system (MPS) or filtration by the liver and spleen happens very quickly and sequesters the majority of nanodrugs. There is a correlation between the circulation and particle size. The MPS clearance exhibits a size-dependent behavior such that nanoparticles with small sizes are less likely to be taken up by macrophages than large ones.[15,16] The biodistribution is also greatly influenced by the size of nanodrugs because of the different cutoff size of organs. The renal filtration cutoff size is 5.5 nm,[17] and the vascular fenestrations in liver are 50–100 nm. Particles lower than 5.5 nm are much more easily excreted through urine, and sizes smaller than 50 nm could easily penetrate the endothelial and get trapped in the liver.[18,19]

To effectively accumulate in tumor sites, there is one more barrier to cross: the leaky tumor vasculature. As for the rapid growth of tumor cells, the pore cutoff size of the tumor vessel ranges from 200 nm to 1.2 μm, depending on the type of tumor.[20,21] Diameters of nanoparticles below the cutoff size are required for effective passive targeting. Generally, after entering the tumor region, nanoparticles with a large size are capable of being well retained in the tumor surroundings but it is hard for them to penetrate deeply in the dense matrix,[22,23] while the small-sized ones remain sufficiently penetrated in the tumor, whereas they can be easily pumped back into the bloodstream by the high interstitial fluid pressure of the tumor.[24−26] Although there are investigations exploring strategies to modulate the tumor microenvironment, consequently improving nanoparticles’ intratumoral distribution,[27,28] optimizing the nanoparticles’ size is deemed as a critical solution. Also, regarding the principle of the size-dependent tumor distribution, it is defined that the unique and ideal size remains exactly the equilibrium point between penetration and retention. To optimize the situation, size-tunable nanoparticles show their extraordinary talent. Compared with nanoparticles that are unique in size, size-tunable nanoparticles can reach both better penetration and retention, so long as they change their sizes at the proper time.

After nanoparticles enter into tumor sites, their size also influences the following cellular uptake process in which nanoparticles with varied sizes exhibit different internalization rates.[29−31] Sizes were found to be the determining factor for the pathway of cellular internalization. Most of the nanoparticles were internalized into cells through two pathways: phagocytosis and pinocytosis. The phagocytosis pathway was only in some special cells such as macrophages, neutrophils, or dendritic cells, or particles larger than 500 nm in diameter. Also, the pinocytosis pathway can be further divided into four different types based on the mediated proteins: caveolae-mediated endocytosis, clathrin-mediated endocytosis, noncaveolin nonclathrin mediated endocytosis, and macropinocytosis.[32] Nanoparticles between 20 and 100 nm are internalized through caveolae-mediated endocytosis,[33] 120 to 150 nm are mediated by clathrin,[34] and macropinocytosis occurs in the internalization process of nanoparticles larger than 1 μm.[35] As a reverse mechanism of endocytosis, exocytosis also plays an important role in enhanced cellular accumulation of nanoparticles. Decreased exocytosis avoids nanoparticles secretion to the extracellular space and therefore improves the therapeutic efficacy.[36] Quantities of research studies found that there was a negative correlation between the size of nanoparticles and the exocytosis process, and nanoparticles with relatively large sizes exhibit decreased exocytosis and enhanced intracellular retention.[37,38]

As with a brain for a human, the nucleus in the cell plays the most important role among all the organelles and is responsible for controlling genetics and metabolism.[39−41] Various therapies aim at the nucleus to achieve an anticancer effect, such as anthraquinone derivatives and short hairpin RNA, which restrictively exhibit activities inside the nucleus. However, the major obstacle to nuclear delivery is the nuclear membrane, in which nuclear pores are generally ∼9 nm in size.[42,43] Even though there are other strategies to promote the nuclear delivery of nanoparticles with relatively larger sizes, such as shape modulation,[44] ligand modification for active targeting,[45] or artificially opening the nuclear envelope by singlet oxygen,[46,47] the simplest and most effective way yet is to control the nanoparticles’ size below 9 nm. However, the ultrasmall size ensuring the passage across nuclear pores is comparable with the renal filtration cutoff, risking nanoparticles’ circulation and distribution at the organ level. To address the contradiction, size-tunable nanoparticles arise at the appropriate time, gathering both good blood circulation and nuclear distribution.

Aggregation Strategies

As a contradictory behavior of varied sizes described above, small-sized nanoparticles tend to have good penetration but are cleared away rapidly, while nanoparticles with large sizes have enhanced retention time but cannot penetrate deeply into tumors. Researchers proposed an aggregation strategy to enhance the accumulation and penetration of nanoparticles in the meantime. The initial small-sized nanoparticles were first used for deep penetration, and form large agglomerates for enhanced retention once deep into the tumor after specific stimulations. The stimulations can be divided into several types due to the tumor heterogeneity, such as hypoxia, slightly acidic microenvironment, and specifically upregulated enzymes. In addition, some external stimuli such as light and temperature are also utilized to design the intelligent, size-aggregatable nanoparticles (Table ). Under external or internal stimuli, initial nanoparticles with relatively small sizes interact with each other through click reaction, self-assembly, electrostatic interaction, or phase transition to form aggregations.

Table 1

Mainly Used Aggregation Strategies

stimuli	mechanism	ref
enzyme	click reaction (cyano and 1,2-thiolamino, azide and alkyne, amine and acyl)	(51−56)
	self-assembly (amphiphilic block copolymers)	(60−63)
pH	electrostatic interactions (zwitterionic compounds, hemoglobin)	(65−180)
light	C–C, C–H, O–H, X–H insertions (azobenzene, diazirine, triphenylmethane, cinnamenyl)	(72−181)
temperature	phase transition (PPCs, PNIPAm, PDEAm, PEO, PPO, polyphosphoesters)	(79−90)
redox	click reaction and host–guest interaction (disulfide bond, Fc+)	(97, 99)

Enzyme-Induced Aggregations

The enzyme-induced strategy has great potential because of its highly specific substrate selectivity and varied expression of enzyme in different organs. There are many specific upregulated enzymes in tumor tissues such as matrix metalloproteinase (MMP), legumain, hyaluronidase (HAase), gelatinase,[48] furin,[49] caspase 3/7,[50] etc. Click reactions are commonly used in this strategy because of their easy and quick procedure, which is perfectly suitable for in vivo synthesis. Many functional groups can be utilized such as azide and terminal alkynes,[51] sulfhydryl and maleimide,[52] 1,2-thiolamino and cyano groups,[53] etc. By decorating the above corresponding chemical groups on the surface of nanoparticles, aggregation will happen very quickly once the chemical groups are exposed and interact with each other. Rao et al. first used the click reaction between the cyano group of 2-cyano-benzothiazole (CBT) and the 1,2-aminothiol group of cysteine (Cys), which was the last step of the synthesis of d-luciferin and displayed biological friendliness.[54] Our group utilized the substrate of legumain (Ala-Ala-Asn↓Cys-Lys) and codelivered 2-cyano-6-amino-benzothiazole (CABT) to tumor sites where legumain was found overexpressed,[55] which generated aggregations of GNPs from 35.6 to 309.6 nm in 12 h (Figure A) and endowed GNPs not only the prolonged tumor accumulation but also bioimaging of multispectral optoacoustic tomography (MSOT). In addition to enzymes that are upregulated in tumor sites, there are circumstances when the triggered enzymes of click reactions are not or are lowly expressed in tumors. In that case, enzymes can be delivered together with nanocarriers. Gu et al. codelivered transglutaminase (TG) in a hyaluronic acid (HA)-made nanogel shell on the surface of a carrier (Figure B).[56] The protected nanogel shell collapsed after entering into the tumor and released sufficient TG to catalyze the formation of an isopeptide bond between free amine group from lysine and the acyl group from glutamine, which led to aggregation from 10 to 120 nm and prolonged tumor retention time to more than 72 h.

Figure 1

(A) Diagram depicting the legumain-triggered aggregation and composition of GNPs-DOX-A&C. Copyright 2016, American Chemical Society. (B) Schematic of tumor microenvironment-mediated construction of combination drug-delivery depots for sustained drug release using CS-NG. Copyright 2016 American Chemical Society. (C) A variety of morphologies of polymeric amphiphile aggregates depending on the design of the peptide substrate and enzymes added. Copyright 2011, American Chemical Society.

Unlike click reactions which need to modify other functional chemical groups to make aggregation happen, the enzyme-induced self-assembly makes full use of the character of the nanocarrier itself. Such nanocarriers are usually composed of amphiphilic block copolymers, which can lead to formation, destruction, or morphological transformation after changes in the chemical of physical nature.[57−59] Gianneschi et al. designed a polymer–peptide block copolymer amphiphiles system containing substrates for four different cancer-associated enzymes: protein kinase A (PKA), protein phosphatases-1 (PP1), MMP-2, and MMP-9 (Figure C).[60] Phosphorylation by PKA caused a 50-fold increase in hydrodynamic diameter together with the appearance change of amorphous structure, and the aggregation could be reversed after being successively treated with PP1 for dephosphorylation, providing feasibility for enzymatically size-switchable strategy. Alkaline phosphatases (ALPs) is another enzyme widely used to instruct the self-assembly of nanofibers in vivo,[61] which was found to be highly expressed on the cell membrane or secreted out of some cancer cells with the ability to responsively cleave the phosphate groups. Xue et al. and Liang et al. both utilized ALPs to induce nanofibers through π–π stacking.[62,63]

pH-Triggered Aggregations

Different parts of the human body exhibit different pH levels as we all know. Because of the rapid proliferation of tumor cells, the microenvironment of tumors is slightly acidic at around pH 6.5, and after being internalized into cells, some cytoplasmic organoids such as endosomes and lysosomes exhibit an even lower pH degree at around pH 5.0–5.5, while blood and normal tissues are maintained at 7.4.[64] Different from enzyme-induced aggregations, pH-triggered reactions possess the advantages of quick response and ultrasensitivity. Upon the electrostatic interactions of decorated pH-sensitive surface molecules[65,66] or natural pH-sensitive proteins,[67] the nanocarriers will aggregate and prolong retention in tumors due to the increased sedimentation-driven uptake and decreased extracellular efflux.

pH-responsive molecules usually are zwitterionic compounds, such as hydrolysis-susceptible citraconic amide,[68,69] 11-mercaptoundecanoic acid, and (10-mercaptodecyl)trimethylammonium bromide).[14] Citraconic amides are rapidly converted to positively charged primary amines by hydrolysis at a mildly acidic pH, which would in turn react with the negatively charged carboxyl groups through electrostatic attraction and lead to aggregation. However, it should be noted that mixed charged ligands with the same strong electrolytes would not aggregate and remain stable over the entire pH range.[70] Therefore, disrupting the balanced electrolytes was a feasible way to form aggregation (Figure A),[180] and the commonly used combination is a weak acid and strong base such as a weak electrolytic 11-mercaptoundecanoic acid and strong electrolytic (10-mercaptodecyl)trimethylammonium bromide.

Figure 2

(A) Schematic illustration of pH-responsive “smart” gold nanoparticles (SANs). Copyright 2013, Royal Society of Chemistry. (B) TEM images of the Hb-IR780 complex at pH 7.4 (top, the red arrows highlight the Hb-IR780 nanoparticles) and at pH 6.5 (bottom). (C) Increase in hydrodynamic size of the Hb-IR780 complex from normal tissue pH 7.4 to tumor acidic pH 6.5. Copyright 2018 American Chemical Society.

Though the commonly used pH-sensitive ligands are synthesized artificially, natural proteins with acidic or basic amino acid residues are deemed to be pH-sensitive nanoplatforms. It is generally known that proteins remain stable at different pH values, while aggregating around the isoelectric point (pI).[71] If the protein is properly designed with pI around the physiological pH of tumor, the natural protein could be chosen as the carrier for enhanced tumor accumulation. Li et al. employed hemoglobin as a smart pH-sensitive nanocarrier for near-infrared dye IR780 (Hb-IR780).[67] Dynamic light scattering (DLS) results showed that the Hb-IR780 was well-dispersed as a singular protein at pH 7.4 while aggregated severely after incubation at pH 6.5 (Figure B,C).

Light-Induced Aggregations

The light-induced strategy has shown great advantages in noninvasiveness, remote manipulation, and easy operation. The photothermal therapy even shows an extraordinary antitumor effect and is being used widely. In the aggregation strategies, light-responsive polymers are supplied with photoactive groups such as azobenzene, spirobenzopyran,[72] triphenylmethane,[73] or cinnamenyl that can undergo reversible structural changes under UV–Vis light (Figure A,B).[181] However, the application of this strategy is only limited in experiments on the cellular level due to the weak penetration ability of UV–vis and blue light. Upconversion nanoparticles (UCNPs) are novel materials consisting of rare-earth elements, which can convert a near-infrared (NIR) light to UV–vis radiation via two photons or a multiphoton mechanism.[74] The application of UCNP has been well developed so far and has aroused great interest of researchers as the extraordinary tissue penetration of long-wavelength NIR light exactly provides a solution to solve the long-standing problem of limited tissue penetration of visible light.[75] Liu et al. observed an enhanced photodynamic treatment efficacy when loaded with a photosensitizer in UCNP.[76] Zhao et al. reported that UCNP can trigger the photoisomerization of azobenzene,[77] which is a commonly used chemical group for light-induced aggregation as mentioned above. Therefore, the further application of light-induced aggregation in vivo can be achieved with great prospects by utilizing UCNP as the carrier.

Figure 3

(A) Schematic illustration of light-triggered assembles of dGNPs. (B) TEM images of dGNPs before and after illuminated with 405 nm laser in different periods of time. Copyright 2016, John Wiley and Sons.

Temperature-Triggered Aggregations

To construct temperature-responsive nanoparticles, the balance between segment–segment interactions and segment-solvent intermolecular interactions can be reversed by temperature changes. Some thermosensitive polymers can be designed to typically undergo the coil–helix transition upon decreasing temperature below the upper critical solution temperature (UCST), while others specifically respond to the increasing temperature above the lower critical solution temperature (LCST). Mostly used functional pairs are polymer–peptide conjugates (PPCs) (Figure A,B),[78] proteins, polylactic acid (PLA), and polysaccharides.[79−81] Polymers with the merit of thermoresponsive behaviors such as poly(N-isopropylacrylamide) (PNIPAm),[82] polyphosphoesters,[83] poly(N-diethylacrylamide) (PDEAm),[84] copolymers of poly(ethylene oxide) (PEO) and poly(propyleneoxide) (PPO),[85] and poly(N-vinylcaprolactam)[86] have been widely used. However, this strategy always needs the combination with other stimuli such as enzymes or pH to destroy the inherent interactions and expose the thermosensitive parts so as to respond to different temperatures.[87−89] In other cases, an additional local temperature increase was given through introducing photothermal molecules for specific responsiveness.[90]

Figure 4

(A) LCST profiles of F1–F3 before and after treatment with GSH, caspase-3, and Atg4B. (B) Specific responsive nanoaggregation in cells. Confocal images and bio-TEM images of MCF7 cells treated with PPCs and modulated agents. Copyright 2017, American Chemical Society.

Redox-Induced Aggregation

Attributed to the high proliferation nature of invasive tumors, tumor tissues and cells show elevated oxidative stress, resulting in a defined high level of reactive oxygen species (ROS).[91] As a consequence of handling the oxidative stress, the redox environment of tumor tissues and cells increases in a way. For example, the intracellular GSH level is elevated in tumors for increased antioxidant capacity and resistance against oxidative stress, as well as regulating cell differentiation, proliferation, and apoptosis.[92] According to research data, the concentration of GSH in tumor regions, especially the intracellular part, reaches 2–10 mM, whereas that in normal tissue remains 2–20 μM.[93−95] The disulfide bond is mostly used GSH-specific chemical groups that can be reduced to sulfhydryl, and therefore is widely used in redox environment responsive drug delivery.[96] Gao et al. utilized the dysregulation of GSH for aggregation strategies and tumor imaging.[97] After entering into tumor sites, the self-peptide was cleaved by GSH, and the exposed sulfhydryl interacted with maleimide on the end of PEG of the adjacent nanoparticles according to the click reaction (Figure A). The reaction rate between sulfhydryl and maleimide happened extremely rapidly so that the nanoparticles aggregated from 7.5 to 295 nm in 5 h (Figure B,C).

Figure 5

(A) Schematic drawing of the GSH-responsive antiphagocytosis 99mTc-labeled Fe3O4 nanoparticles for forming particle aggregates through the interparticle cross-linking reaction. (B) Temporal hydrodynamic size profiles of nonresponsive probe and (C) responsive probe in reaction with the GSH treatment. Copyright 2017, John Wiley and Sons. (D) Illustration of intracellular host–guest assembly of gold nanoparticles triggered by GSH. Copyright The Royal Society of Chemistry 2016.

As an important antioxidant and free radical scavenger in the body,[98] GSH could not only reduce disulfide but also other species at an oxidative state. Therefore, Wang et al. used GSH as a reducing agent to specifically respond to reductive aggregation.[99] They modified GNPs with β-CD and incubated them with Fc+-PEG-Fc+, and aggregation was seen in HepG2 cells after 12 h (Figure D). The GSH triggered aggregation caused significant apoptosis of cancer cells, and because of the specific express site of GSH, this strategy was also promising in reducing unexpected side effects induced by traditional chemotherapy. However, the aggregation efficacy was detected only at the cellular level as it remained difficult to codeliver β-CD-GNPs and Fc+-PEG-Fc+ to tumor sites.

Salt-Induced Aggregation

As is known to all, the fluid in vivo can be seen as a buffer salt system, and a great majority of nanoparticles are unstable and easy to aggregate irreversibly in a high concentration of aqueous salts due to the disruption of the shielding layer on the surface. Researchers tried every possible way to prolong the circulation time of nanocarriers in vivo such as PEGylation and biomimetic cell membrane coating,[100,101] etc. However, Sun et al. first took this defect as a strategy to design salt-induced GNPs aggregation.[102] After intratumoral injection, GNPs formed irregular aggregation, while PEG-GNPs did not aggregate at all. The salt-induced aggregation got rid of complicated surface modifications and was completed instantaneously. Yet this strategy was only applied in superficial solid tumors treatment and needed precise operation because of the specific requirement of intratumoral injection.

Size-Shrinkage Strategies

With regard to the relationship between nanoparticles’ size and intratumoral behaviors mentioned before, the high fluid pressure and dense matrix inside solid tumors always impede the deep penetration and consequently homogeneous distribution of nanoparticles inside the tumor;[103] therefore, it is essential for nanoparticles to shrink their size and enhance penetration for homogeneous delivery.[104] Besides, the size (reported to be ∼10 nm and up to 39 nm when amplified)[105] of nuclear pores limits the nuclear-targeting nanoparticles to be in an engaged size. The postshrinkage nanoparticles not only retain a small size for enhanced penetration, but also contribute to many other properties, such as drug release,[106] rapid renal clearance,[107] secondary distribution,[108] etc. By now, plenty of nanoparticles have been investigated based on the endogenous acidic pH, overexpressed enzymes, redox conditions, and exogenous physicochemical stimuli, which are summarized in Table and discussed in the following.

Table 2

Mainly Used Shrinkage Strategies

stimuli	responsive chemical groups	refs
pH	amino polymers, DMA, Schiff base	(112, 113, 117−119)
enzyme	MMPs, HAase, amylase, thrombin	(124−128, 134−139)
redox	disulfide bond	(140, 141)
ROS	thioketal, thioester, polypropylene sulfide, phenylboronic ester	(146−149)

Size Shrinkage Triggered by Acidic pH

Amino polymers are a kind of polymer functionalized by amino groups, which are generally nonprotonated and exhibit hydrophobicity.[109−111] However, the amino groups in amino polymers change into hydrophilic when protonated at acidic pH, resulting in the disassembly of the hydrophobic core. The dramatic protonation of histamine during pH decreasing provides poly(histamine) with the same property.[66] Yuan et al. reported a kind of nanoassembly composed of nanomicelles and nanogel,[112] which were both self-assembled by amphiphilic copolymers (PDPA30-b-PAMA15 and P(EGMA-GMA-PDSEMA), respectively, in aqueous solution (Figure A). The micelle retained a hydrophobic core made up of PDPA, which was able to respond to intratumoral acidic pH and consequently transferred to a hydrophilic moiety (Figure B) with the size apparently shrinking from 35 nm to about 10 nm (Figure C). Another similar work recently reported by Ray et al. is much more suitable to address the penetration problem, since the size shrinkage is from 100 to 150 nm to 2–5 nm.[113]

Figure 6

(A) The polymerization of PDPA30-b-PAMA15 and P(EGMA-GMA-PDSEMA), and consequent construction of micelle and nanogel. (B) Illustration of nanosystem fabrication and responsive shrinkage. (C) TEM images of nanosystem before and after shrinkage, scale bars: 100 nm. Copyright The Royal Society of Chemistry 2014.

2,3-Dimethylmaleicanhydride (DMA) reacts with many amines to form acid amine, which can further respond to slightly acidic condition and break into amine and 2,3-dimethylmaleic acid.[114−116] Wang’s group further constructed a PCL-CDM-PAMAM/Pt system with a size shrinking from 100 to 5 nm under intratumoral acidic pH (Figure A).[117] In addition to directly decomposing contents to reduce size, DMA also showed another property, charge reversal, for the design of size-shrinkable nanoparticles (Figure B,C).[118] The dramatic size shrinkage and charge reversal were essential for the penetration and cellular internalization inside the tumor. Schiff base is another widely used pH-sensitive chemical group, characterized by a double bond formed between carbon and nitrogen atoms (-C=N-),which is unstable under acidic pH and easily broken by hydrolysis.[119] Although the critical acidic condition and pervasive proton make a pH-triggered strategy sensitive and specific, the acidic and hypoxic regions are commonly far from blood vessels,[120] which becomes a major concern for pH-triggered shrinkage.

Figure 7

(A) Illustration of construction and pH-responsive size shrinkage of DMA-based nanomicelle. Copyright 2016 National Academy of Sciences. (B) Schematic illustration of DMA-based PNV, with charge reversal for dissociation. (C) The dissociation of PNV and release of polymer-Dox inside the tumor. Copyright 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

Size Shrinkage Triggered by Overexpressed Enzyme

MMPs and HAase are the most common tumor-specific enzymes used as recognizing markers in responsive drug delivery. MMPs include a large family of proteinases, in which the MMP-2 and MMP-9 are the most known, playing an important role in tissue remodeling associated with various physiological or pathological processes.[121−123] The MMP-2 and MMP-9 are generally overexpressed and secreted by tumors for matrix digestion, recognizing plenty of substrates involving gelatin and collagens, which retain MMPs’ responsiveness when designed as a component of a drug carrier. The very first size-shrinkable nanovehicle based on MMPs was designed by Wong and co-workers.[124] They covered the large gelatin nanoparticles (GelNPs) with small quantum dots (QDs) to form QDGelNPs, which rapidly shrank from 100 nm to 9.7 nm when responding to intratumoral MMP-2 and MMP-9, as a consequence of GelNPs dissociation (Figure A). And taking advantage of the gelatin nanoparticles, our group functionalized small-sized dendrigraft polylysine (DGL) on the surface of gelatin nanoparticles (GelNP),[125−127] with DOX loaded and angiopep-2 decorated on DGL,[128] resulting in deep penetration and homogeneous tumor therapy (74.1% growth inhibition). Unlike MMPs with a range of substrates, HAase exclusively hydrolyzes HA to short chains or modified glucose units.[129] The natural negative charge on HA makes itself easy to conjugate positively charged materials, and the specific combination between HA and CD44 also provides HA with active-targeting ability.[130−133] For application of superior HA, our group cross-linked HA with positively charged DGL to form size-shrinkable HAase-responsive nanoparticles (Figure B) that exhibited extraordinary penetration ability with the further assistance of an NO donor on an HA shell.[134,135]

Figure 8

(A) Schematic of 100 nm QDGelNPs shrinking size to 10 nm QDs by cleaving away the gelatin scaffold with MMP-2. Copyright 2011 National Academy of Sciences. (B) Schematic illustration of the design and synergistic effects for deep tumor penetration and therapy effects of IDDHN. Copyright 2018 Elsevier Ltd.

As instanced, the enzyme-triggered size shrinkage is commonly achieved by shell dissociation of core–shell nanoparticles and detachment of small-sized decorations from large nanoparticles.[136] The rule is always applicable, and a recent-report nanosystem is also representative, which can release small-sized cargo nanoparticles when lactate oxidase digests its shell.[137] Other enzymes-dependent nanoparticles also obeyed this designing principle, such as α-amylase-digested hydroxyethyl starch[138] and thrombin-induced depegylation.[139] The enzyme responsiveness is of superselectivity owing to exclusive enzyme–substrate recognition, which is essential for improving the therapeutic efficacy and reducing the adverse effects of cancer therapy. However, uneven levels of enzyme expression in different tumors restrict the application scope of certain responsive nanoparticles.

Size Shrinkage Triggered by Redox Condition

Since the GSH level is critical in tumor tissue and cytoplasm, its responsiveness is also evident in size shrinkage strategies. Guo et al. constructed a nanomicelle composed of pegylated polylactide (PEG–PLA) and DMA-modified polythylenimine (PEI–DMA),[140] which was linked by a disulfide bond, forming PEG–PLA–S–S-PEI–DMA (PELEss-DA, see structures in Figure A). The high level of intracellular GSH reacted with the disulfide bond and deshielded PEI shell. Also, the dramatic size shrinkage allowed the remaining nanoparticles’ entry into nucleus (Figure B) and consequent DOX release for DNA interruption. The nucleus delivery ensured the drug’s activity toward the target, avoiding drug resistant transporters’ function, which is essential for nucleus-targeting chemotherapeutic agents. A similar investigation was found in Wang’s work, in which a PSPD/P123-Dex nanomicelle was designed (see structures in Figure C).[141] The two materials formed 120 nm PSPD/P123-Dex and dramatically shrank into approximately 30 nm P123-Dex under an intracellular GSH-abundant condition. Regarding the suitable small size and assistance of dexamethasone, P123-Dex pithily delivered encapsulated DOX into the nucleus for pronounced cytotoxicity (Figure D).

Figure 9

(A) Schematic design of the nucleus entry of size-shrinkable polymer micelles (PELEss-DA) to overcome MDR. (B) Cellular uptake and intracellular distribution analysis. Copyright 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. (C) Schematic illustration of the cooperative dimensional strategy for anticancer drug delivery mediated by hybrid micelle PSPD/P123-Dex. (D) Evaluation of HeLa cells internalizing different micelles, the number represents: 1. Free DOX; 2. PPD/DOX; 3. PSPD/DOX; 4. PPD/P123-Dex/DOX; 5. PSPD/P123-Dex/DOX. Copyright 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

Size Shrinkage Triggered by ROS

Although it is widely known that there is a relatively high level of ROS in cancer cells (∼10 μM) compared with normal tissues (∼10 nM),[142] it is still not sufficient enough to trigger hydrolysis of ROS-responsive chemical groups instantly. NIR light retains better tissue penetration compared with ultraviolet and visible light.[143] Importantly, plenty of photosensitizers are able to convert NIR light into ROS, which has been widely explored as photodynamic therapy for surficial cancer treatment.[144,145] A thioketal linker is easy to synthesize and respond to ROS, resulting in two broken parts. Cao et al. designed a PCL–PEG nanomicelle with poly(thioketal phosphoester) (TK-PPE) mixed core.[146] The chlorin e6 (Ce6) loaded in the micelle could transfer NIR irradiation into ROS, further cleaving the TK-PPE and dissociating the core of micelle (Figure A). As a consequence, the nanomicelle shrank its size from 154 to 72 nm (Figure B), with DOX released. By comparison to free drug, the preshrinkage large micelle ensured its prolonged circulation (Figure C), while the postshrinkage small one promoted its distribution and deep penetration inside the tumor (Figure D). Together with the transferred ROS and released DOX facilitated by NIR irradiation, the shrinkable nanomicelle performed chemo- photodynamic therapy in deep tumor region, resulting in remarkable antitumor effect (Figure E). Other ROS-responsive structure or materials, such as polypropylene sulfide,[147] phenylboronic ester[148] and thioester[149] are also of talent in designing NIR/ROS-triggered shrinkage. Unlike other endogenous stimuli, ROS triggered by external NIR irradiation is desirable and controllable, as well as tumor-specificity. Meanwhile, no activity remains in the nonirradiation or nondrug region, ensuring the safety and avoiding adverse-effect aroused by cancer therapy. Besides, dual stimuli seem like one case with two locks, ensuring the specificity of unfolding.

Figure 10

(A) Preparation and function of the light-activated shrinkable nanoparticle TK-PPE@NPCe6/DOX. (B) TEM images of TK-PPE@NPCe6 after 660 nm laser irradiation at different times. (C) Plasma DOX concentration versus time after intravenous injection of free DOX, TKPPE@NPCe6/DOX, and TK-PPE@NPCe6/TK-PPE@NPDOX. (D) Quantification of DOX content in tumors by HPLC. Tumors were excised 12 and 24 h after intravenous injection. (E) MDA-MB-231 tumor growth curves of various groups after intravenous administration, *p < 0.05; **p < 0.01. Copyright 2017 American Chemical Society.

Major Challenges of Size-Tunable Strategies for Application

Though great progress has been made in size-tunable nanoparticles, there are still many challenges that hinder the efficacy, which need to be further optimized and solved.

Protein Corona May Shield the Responsive Ligand

Protein corona acts as the main issue to affect the targeting efficiency. The corona is quickly formed once nanoparticles are introduced into biological fluid. As we have discussed previously, the composition, size, shape, and surface chemistry of nanoparticles, and even the different fluid environment in vivo, influence the formation of protein corona.[150] In addition to acting as opsonin to mark a nanodrug and alter its biodistribution, the adsorbed protein corona on the surface may also shield the responsive ligand from stimuli and affect the sensitivity of size changing. Among which, the mechanism that need direct contact such as enzymatic reaction, pH-induced protonation and GSH-dependent redox may suffer from the protein corona a lot, while the external stimuli such as light and temperature would not be bothered. Though PEGylation has been confirmed to facilitate nanoparticles with the “stealth” property, which can reduce the protein adsorption and decrease the effect of corona, there comes a new question whether the enzymatic reaction would also be affected by PEGylation, as the nature of enzyme is protein as well. The response sensitivity may be affected in the enzyme-induced size change after PEGylation.

Accuracy of Measurement

As we are doing research studies about size changing, the detection of nanoparticles’ size undoubtably requires very high accuracy. However, the size measured by DLS always displays a relatively wide peak.[151] Nanoparticles with an average DLS size of 100 nm probably have a peak ranging from 50 to 400 nm. It is not that convincing to confirm the success of size shrinkage (from 100 to 50 nm) or size aggregation (from 100 to 400 nm) as there is an overlap of size during the size change process. The corresponding distribution intervals can be compared separately instead of just comparing the average number, which is meaningless. On the other hand, nanoparticles with a good polydispersity index are highly recommended in the study of size change, such as silica nanoparticles, gold nanoparticles, and polystyrene nanoparticles, as these nanoparticles have already been well industrialized and have quite uniform size, and the narrow size range would partly solve the problem in a way.

Off-Target Effect

The location accuracy of the responsive reaction needs to be precisely designed. Nanoparticles are designed to specifically target tumor sites and enhance the accumulation. However, the off-target effects always attenuate the antitumor efficacy and cause damage to normal organs. For example, P-aminophenyl-α-d-mannopyranoside (MAN) is a widely used glucose analogue modified on nanoparticles to help cross the blood–brain barrier and target brain tumor cells,[152] while its receptor, the mannose receptor, is also expressed on Kupffer cells of liver.[153] The majority of nanoparticles are trapped in the liver, limiting the effective delivery to tumor sites. On the other hand, since the pH and enzyme environments are somehow different on both sides of the biological barrier, it remains to be explored whether the responsiveness still exists after crossing barriers, or the responsiveness is crippled by prereacting with enzyme on the barriers. Furthermore, the most commonly used targeting strategy is ligand–receptor binding; ligands are modified on nanoparticles to help actively target the tumor sites. However, the chemical modification would destroy the innate structure of ligands and attenuate the binding efficiency to some extent. The recently rising bio-orthogonal reactions improve the targeting specificity very well.[154,155] By first giving a primary exogenous tag to label target cells, the secondary tag could specifically recognize the labeled cells and avoid the off-target effect.[156−158] However, the existing primary tag delivery still depends on the interaction of ligand and receptor, and relevant studies are only at the cellular level; further work needs to be done to make it better applied in vivo.

Complicated Tumor Microenvironment

Because of the heterogeneous nature of tumor sites, the tumor microenvironment should be taken into consideration. The dense ECM and high interstitial pressure of the tumor tissue attenuate nanoparticle delivery efficiency. However, the intelligent size change of nanoparticles together with remodeling of the tumor microenvironment could synergistically improve the accumulation and penetration of nanoparticles at the tumor sites, such as degradation of ECM and normalization of tumor vasculature system, which could enhance the penetration ability of nanoparticles and significantly reduce the reflux of nanoparticles leading to promoted accumulation at tumor sites. However, there also exists unpredicted adverse effects. Disruption of ECM may also lead to tumor migration and metastasis, and untimely normalization of blood vessels may make it difficult for nanoparticles to passively target tumor tissues through an EPR effect. The extent and timing of remodeling of microenvironment need to be carefully balanced.

Sensitivity of Transformation

Because of the high interstitial pressure of tumor sites, nanoparticles that enter tumor sites are easily pumped back to vessels, and the retention time of the original nanoparticles will not be too long. Therefore, the sensitivity and response time with stimuli are a key factor in designing size-tunable nanoparticles. As demonstrated above, pH-triggered reactions possess the advantages of quick response and ultrasensitivity, while the reaction time between enzyme and substate is not that fast. The enzyme-induced aggregations are usually observed at the 12 or 24 h time point, and a majority of the original nanoparticles may be eliminated before aggregations happen.

Possibility of Clinical Translation

In the last few decades, an EPR effect was considered to be the basic principle for passive drug delivery to tumor sites.[159] However, there are more and more controversies and doubts about the EPR effect because of the unsatisfactory therapeutic efficacy in clinical trials. Also, it has been reported that only 0.7% nanoparticles can be finally delivered to tumor sites.[160] One of the most important reasons is the difference between human and model animals,[161] and there is no single animal model that can fully reproduce human cancers. In addition, the EPR effect only exhibits good results in animal models, which is because of the requirement of tumor volume. In tumor-bearing mice, it is until the tumor grows to 200 mm3 that it exhibit good EPR effect, which is nearly impossible for a human to have a such big tumor. Recently, scientists are more prone to intervene in tumors at an early stage after inoculation in mice or choose active-targeted strategies to treat cancers. Moreover, there are still other problems that limit the clinical translation such as the biocompatibility of nanodrugs and scalable manufacturing for industry, which is a long way to go.

Outlook and Future Directions

In addition to the above strategies, there are many other novel approaches to make nanocarriers aggregate or shrink in tumor tissues, such as using enzyme-instructed self-assembly to generate intracellular supramolecular nanofibers to achieve the aggregation goal,[162−165] or using the change of shape to achieve the purpose of aggregation in a round-about way.[166,167] Intracellular aggregation will decrease the efflux of nanodrugs in a way, which has potential in overcoming multidrug resistance.[49] Except prolonging the retention of nanocarriers after aggregation, properties of aggregated nanodrugs tend to be changed, and most of them exhibit a good photothermal, photoacoustic effect and enhanced contract of MRI, which provide a possibility for real time imaging and photothermal therapy leading to better antitumor treatment prospects.[168] Aggregation-induced emission (AIE) fluorophores are the application of aggregation strategies to solve the aggregation-caused quenching (ACQ) problems and give a new long-term tracer for in vitro or in vivo imaging.[169,170] Also, because of the good results achieved in antitumor therapy, aggregation strategies are increasingly being used in other areas, such as brain tumor surgery guidance[171] and myocardial infarction.[172]

Moreover, the shrinkage strategy also shows capabilities more than penetration promotion and nucleus delivery. The shrinking procedure always accompanies shell dissociation or volume squeezing of nanoparticles, which have potential for specific drug release. Besides, nanoparticles’ size influences their distribution not only in tumor tissue, but also in normal organs. Small-sized nanoparticles tend to accumulate in organs such as kidney, spleen, and lung, which generate different potential functions and strategies for nanoparticle-based therapy, such as secondary distribution in spleen for immune activation, rapid renal clearance for elimination of toxic materials, and lung-targeting function for alveolus remodeling.

With more and more applications of size aggregation and size-shrinkage strategies, there comes a question of which strategy is better. In fact, though the two strategies are totally opposite from the methodology, the ultimate goal and therapeutic efficacy are the same, which are to prolong the penetration and retention time in tumor sites at the same time. Besides, the emergence of the size-reversible strategies further optimizes the strategies of size change.[173,174] And recently, nanoparticles were produced that could simultaneously change the tumor microenvironment along with the size change,[175,176] such as hypoxia relief,[177] normalization of vasculature,[178] and extracellular matrix modification,[179] endowing a better therapeutic effect.

In short, it is trendy to carefully design responsive, size-changeable nanoparticles based on a certain tumor, even a certain part of the tumor tissue. We look forward the time when multiresponsive nanoparticles with multistage size changes are developed and optimize current treatment, which seems hard to achieve today but will be of great value.