Charge Regulation of Self-Assembled Tubules by Protonation for Efficiently Selective and Controlled Drug Delivery.

Despite the success for targeted delivery in the body, the efficient release without side effects caused by residual drug remains a challenge. For reducing residual drug, the pH-responsive carriers were prepared by self-assembly from aromatic macrocycles, which were non-toxic and biocompatible. The inner surroundings of aromatic macrocycles could be protonated positively by acid inducing the separation of neighboring macrocycles. Thus, Dox-loaded carriers successfully inhibited the proliferation of carcinoma cells (HepG2 and 4T1) rather than normal cells (HL7702). The effects were further proved in vivo without systemic cytotoxicity. Notably, the responsive environment for drug release depended on the concentration of carriers. Particularly, drug release was promoted by carrier separation. Carrier 2 exhibited preferable anticancer efficacy than carrier 1 due to the efficient release of Dox by full separation of the carrier. Collectively, we have developed a novel strategy serving as a selective and controlled drug release platform for cancer therapeutics.

Introduction

Over the past decades, nanoparticle-based drug delivery systems (NDDs) including liposomes (Mo et al., 2012, Lee et al., 2007, Linderoth et al., 2009), polymeric nanoparticles (Wang et al., 2015, Griset et al., 2009, Tong and Cheng, 2008), and metal-organic frameworks (Horcajada et al., 2010, Horcajada et al., 2012, Zheng et al., 2016) have received a great deal of attention in the field of cancer therapies. Despite great advances in successful delivery to targeted sites against side effects over free drugs, the release of drugs from carriers at the targeted tissues still remains a challenge in clinical applications. In general, drug release is achieved through acid-responsive systems in the mild acidic environment of the lysosomes or endosomes or the tumor microenvironment (Bae et al., 2003, Du et al., 2011, Zeng et al., 2017). Among them, the labile systems generally rely on pH-responsive hydrolysis, which is unable to ensure targeting delivery. Nevertheless, the hydrolysis in carriers is mostly too stable to release the anticipated dose, thus leading to reduced therapeutic efficacy. Even worse, most of the carriers with residual drugs could be trapped in the major organs such as the liver, heart, and spleen, which may cause side effects in the course of degradation, metabolism, and excretion (Almeida et al., 2011, Arami et al., 2015, Li and Huang, 2008).

The spontaneous assembly of small molecular modules by non-covalent interactions is a key to creating stimuli-responsive systems. Up to now, various strategies for the construction of rapidly responsive supramolecular carriers have been exploited by external triggers such as pH (Duan et al., 2013), redox potential (Zhao et al., 2014), enzyme (Huang et al., 2015), light (Yesilyurt et al., 2011), and ions (Cao et al., 2014). These systems could control drug release effectively through labile chemical interactions by structural collapse. Recently, scientists have developed intelligent devices into carrier systems to create contractive nanoparticles in tumors (Tong et al., 2013). Although the shrinkage of the particles enhanced tissue penetration and retention, a small amount of residual drugs in particles would generate high toxicity or serious inflammation. Among many self-assembled systems, aromatic building blocks have proved promising scaffolds for constructing dynamic architectures (Kim et al., 2011). For example, 120°-folded aromatic segment grafted by hydrophilic oligoether dendrons can easily aggregate into aromatic macrocycles that spontaneously stack on each other to form porous tubules in response to external environments (Wu et al., 2017). As hinted from the dynamic formation of porous tubules based on macrocycle stacking, here we try to vary the charge of the inner surface of tubular carriers by reassembly in mild acidic environment to enhance the repulsion between neighboring macrocycles for facilitating the release of hydrophobic drugs. The porous tubules possess high loading capacity for doxorubicin (Dox) as well as excellent recognition of the proton. Unlike conventional responsive carriers, the tubules dissolve into positively charged macrocycles with hydrophilic cavity. The cationic macrocycles induce a strong electronic repulsion within the inner surroundings of aromatic segment to result in full separation of carriers. Notably, the enhancement of charge in macrocycle leads to preferable expelling ability for drugs (Figure 1).

Figure 1

Molecular Structure and Self-Assembly

(A) pH-responsive charge regulation of hexameric macrocycles from self-assembly of bent-shaped aromatic amphiphiles 1 and 2.

(B) Schematic dissociation of tubular carriers by protonation (multi-layered toroids from self-assembly of 1 based on cation- interaction. Fully dissociated toroids from self-assembly of 2 by strong electrostatic repulsion).

Results and Discussion

Molecular Synthesis and Self-Assembly

The responsive tubular deliverers are derived from the self-assembly of bent-shaped aromatic amphiphiles containing a proton-switchable unit of methoxyphenylpyridine-methoxyphenyl triad (ortho-Py) at the inner position, which was decorated with hydrophilic oligoether dendrons at its apex (Figure S1). The bent-shaped aromatic segment with an internal angle of 120° and hydrophilic oligoether dendrons can self-assemble into a hexameric macrocycle owing to its amphiphilic characteristics (Varghese et al., 2005). The aggregation behavior of both molecules in ethanol was studied using vapor pressure osmometry experiments. The molecular weights of 1 and 2 with formyl and methoxyl terminal groups were calculated to be 1,901 and 1,905 Da, respectively. However, in ethanol, the molecular weights of the primary aggregates were measured to be 10,749 and 10,964 Da, respectively (Figure S2), suggesting that both non-covalent macrocycles from 1 and 2 consisted of six molecules. Subsequently, the addition of water into the ethanol solution would drive the hexameric macrocycles to stack on top of each other to form hollow tubules. Transmission electron microscopy (TEM) with negatively stained samples from 0.02 wt. % aqueous solution showed that both molecules self-assembled into elongated objects with an external diameter of 8 nm (Figures 2A, 2C, S3A, and S3B). To further understand the aggregated nanostructures, scanning transmission electron microscopy (STEM) was implemented with a probe aberration corrector. STEM image showed that the 1D objects were separated by dark and white segments, indicative of the formation of hollow tubules with an inner diameter of 3 nm (Figure 2B). The internal and external diameters of the tubules were identical to the fully overlapped hexameric macrocycles, suggesting that the tubules originated from the stacking of macrocyclic hexamer. The highly ordered mesoporosity of the tubules was well suitable as scaffolds for encapsulation of hydrophobic drug in aqueous environment (Figure S4) (Peer et al., 2007, Tian et al., 2014).

Figure 2

Characterizations of Self-Assembled Tubules

(A) TEM image of negatively stained 2 from 0.02 wt. % aqueous solution; scale bar, 25 nm.

(B) STEM image with a probe aberration corrector of 2 from 0.02 wt. % aqueous solution; scale bar, 25 nm.

(C) AFM image of 2 from 0.02 wt. % aqueous solution with tapping mode; scale bar, 25 nm.

pH-Responsive Properties of Tubular Carriers

We envisioned that the external methoxyl ortho to pyridine was responsive to acid, which induced adjacent molecules to slide into a looser packing arrangement due to the stable formation of ortho-Py triad through hydrogen bonding. The proton-responsive behavior was confirmed by 1H NMR spectroscopy (Figure 3). Proton nuclear magnetic resonance (1H-NMR) measurements of 2 showed that most protons within ortho-Py triad were downfield shifted except the proton ortho to pyridine (Ho) by titration of trifluoroacetic acid (TFA), indicating the formation of hydrogen bonds. The Ho proton facing nitrogen atom in the neutral state became shielded by acidification, suggesting a rotation around the CPy-CPh bond moving away from the NH+ group (Leblond et al., 2010). The hydration process allowed the aromatic segments to slide from fully overlapped macrocycles into a looser arrangement to minimize structural crowding at the inner position of aromatic segments. This mechanical sliding of aromatic segments was further confirmed by ultraviolet-visible and fluorescence experiments. Upon addition of TFA, both molecules showed red-shifted absorption and enhanced fluorescence intensity, suggesting that the fully overlapped H-type aggregates changed into slipped J-type aggregates (Figures 4A and S3C) (Wurthner et al., 2011). Remarkably, the expanded macrocycles from J-type aggregation triggered the dissociation of tubules, which reflected in the reduction of hydrodynamic diameter from 168 to 35 nm for 1 and 141 to 26 nm for 2 as confirmed by dynamic light scattering experiments (Figures 4B and S3D). To identify the changed aggregates upon addition of TFA, TEM and atomic force microscopy (AFM) experiments were further performed. TEM images from the stock solutions with 10 equiv. TFA showed toroidal objects with a uniform external diameter of 11 nm and an internal cavity of 4 nm diameter (Figures 4C and 4D), suggesting that the macrocycles were expanded. When the acidic solutions were cast onto mica, both molecules showed discrete toroidal objects. The AFM images of 2 revealed toroidal objects of height 0.3 nm, demonstrating that the toroids are unilaminar macrocycles (Figure 4E). Nevertheless, the toroids of 1 with formyl terminal group showed height close to 3 nm (Figure 4F), suggesting these toroids were formed from multi-layered stacking of macrocycles. The layer-by-layer stacking of charged macrocycles of 1 was attributable to the cation-π interactions of formyl group with neighboring pyridinium nucleus, proved by reduced surface charge from zeta potential measurement (Figure S5) (Yamada et al., 2004). These results demonstrated that the expanded macrocycles could lead the ordered porous objects to dissolve into small aggregates. The kinetic experiment of dynamic assembly revealed that the pH-responsive property depended on the concentration of tubular carriers. In general, lysosomal or endosomal pH in cancer cells (pH 3.8–4.7) shows higher acidity than that in normal cells (pH 4.5–6.0) (Zhang et al., 2016, Kroemer and Jaattela, 2005). Figure S6 showed that when the concentration of carrier was up to 0.6 mg/mL, the structural reorganization can take place mechanically avoiding the release of drugs in the environment of normal cell. Thus the quantity of release could be monitored with 0.6 mg/mL aqueous solution in acidic buffer state (pH 4.6). The released quantity at acidic state was shown in Figure S7, demonstrating that most of the drugs were released from the carriers. Importantly, the difference appeared on the release profile of drugs. The carrier 1 dissolving into multi-layered toroids showed 82%, whereas carrier 2 separating into unilaminar toroids showed up to 95% for the enhanced charge repulsion between NH+ and cationic Dox.

Figure 3

pH-responsive Properties of Aromatic Segment

Representation of transformation of aromatic segment and 1H NMR spectra of 2 (8.4 mM) upon addition of trifluoroacetic acid (TFA).

Figure 4

Characterizations of Discrete Toroidal Structures

(A) Absorption (left) and emission (right) of 2 in neutral solution (black line) and CH3COOH/CH3COONa buffer solution (red line).

(B) Size distributions of 2 in neutral solution (black line) and CH3COOH/CH3COONa buffer solution (red line).

(C and D) TEM images of 1 (C) and 2 (D) from 0.02 wt. % aqueous solution with 10 equiv. TFA; scale bars, 25 nm.

(E and F) AFM images of 1 (E) and 2 (F) from 0.02 wt. % aqueous solution with 10 equiv. TFA; scale bars, 25 nm.

Evaluation of Carriers In Vitro and In Vivo

The spontaneous release of drugs in mild acidic state holds great promise for successfully delivering drugs to cancer cells and reducing side effects for normal cells simultaneously. Thus human hepatoma carcinoma cells (HepG2), 4T1 mammary carcinoma cells (4T1), and normal cells (HL7702) were employed as the model to evaluate the in vitro cytotoxicity of carriers using the Cell Counting Kit-8 assay. Cell viabilities of HL7702, HepG2, and 4T1 cells co-incubated with carriers 1 and 2 showed above 90% after 72 h (Figures 5A and S8), which meant that both carriers were biocompatible as drug carriers. After loading Dox, both loaded carriers 1 and 2 exhibited higher inhibition of HepG2 and 4T1 cell proliferation but less cytotoxicity to normal cells (HL7702) with the same treatment (Figures 5B and S9–S11). The selective cytotoxicity to HepG2 against HL7702 cells could be explained by selective pH-responsive disaggregation of tubular carriers in the acidic lysosomes of HepG2. To understand the selective behavior, intracellular trafficking of the Dox in cancer cell was observed by confocal laser scanning microscopy (CLSM). The images of CLSM showed that Dox was dominantly co-localized in lysosomes within 6 h and then successfully escaped from lysosomes and entered to the nucleus after 24 h (Figures 5C and S12). Notably, the loaded carrier 2 showed enhanced cytotoxicity than 1 against HepG2 cells (Figures 5B, S9 and S10). To understand preferable inhibition, the quantitative amount of Dox uptaken by HepG2 was observed by flow cytometry. As shown in Figure 5D, accumulations of Dox delivered by both carriers 1 and 2 were gradually increased up to 24 h, which sharply differed from the behaviors of free Dox for the controllable and sustainable drug release. Importantly, the loaded carrier 2 exhibited a preferable accumulation than 1 after 24 h. It can be explained by enhanced charge repulsion between NH+ and the protonated Dox (Figure S5), which was demonstrated by the release profile of drugs (Figure S7).

Figure 5

Evaluation of Carriers In Vitro and In Vivo

(A) Cell Counting Kit-8 assay on cell viability following incubation with carrier 1 for HL7702 (gray without twill) and HepG2 (gray with twill) as well as with carrier 2 for HL7702 (red without twill) and HepG2 (red with twill) for 72 h. Results are presented as mean ± SD (n = 3).

(B) Cell viability of HL7702 (without twill) and HepG2 (with twill) incubated with loaded carrier 1 (gray) as well as loaded carrier 2 (red) by indicated time points. Relative Dox concentration is 0.64 μg/mL; results are presented as mean ± SD (n = 3). *p < 0.05.

(C) Intracellular trafficking of Dox-loaded carrier 2 in HepG2 cells. Cells were harvested at the indicated time points (blue, nuclei were stained with Hoechst; red, distribution of Dox; green, lysosomes were labeled by LysoTracker). Scale bars, 20 μm.

(D) Quantitative flow cytometry analysis of Dox-positive HepG2 cancer cells after incubation with indicated groups. Free Dox (black), loaded carrier 1 (red), and loaded carrier 2 (blue). Results are presented as mean ± SD (n = 3). *p < 0.05.

(E and F) Mean tumor volumes (E) and body weights (F) of the mice in indicated groups after treatment at different time intervals. Tail intravenous injection into the 4T1 tumor-bearing mice was at a single Dox dose of 3 mg/kg every 2 days. Results are presented as mean ± SD (n = 4). *p < 0.05.

(G) In vivo real-time imaging of carrier after intravenous injection of DiR-labeled carrier 2 (at a DiR dose of 1 mg/kg).

(H) H&E analyses of the major organs after the last treatment of saline and Dox-loaded carrier 2; scale bars, 100 μm.

The advances of drug release were also observed in vivo. Owing to the enhanced permeability and retention effect, the NDDs preferentially leak into tumor tissues via a permeable tumor vasculature (Davis et al., 2008). Moreover, the carriers surrounded by ethylene oxide chains can hardly be adsorbed by proteins and cleared from the body by the reticuloendothelial system (Knop et al., 2010). To systematically investigate the drug delivery efficiency and the blood retention time of the pH-responsive carriers in vivo, the near-infrared (NIR) dye 1,1-dioctadecyl-3,3,3,3-tetramethylindotricarbocyanine iodide (DiR) was encapsulated in carriers 1 and 2. Then, the carriers labeled by DiR were intravenously injected into 4T1 tumor-bearing nude mice. Subsequently, the real-time imaging of carriers 1 and 2 in the tumor-bearing mice were monitored using in vivo imaging system (Figures 5G and S13). The NIR fluorescent signals of DiR-loaded carriers in the tumor tissue gradually increased up to 48 h and maintained the fluorescence until 96 h post-injection. The results indicated that a long retention promoted the tumor accumulation of the tubular carriers. Correspondingly, the antitumor efficiency was evaluated through tumor size changes. As shown in Figures 5E and S14, after 14 days treatment, the groups of Dox-loaded carrier showed higher antitumor efficiency than free Dox and the saline-treated groups. As expected, Dox-loaded carrier 2 exhibited the highest antitumor efficiency for the enhanced drug release, which was confirmed by the intensity of DiR in the tumor site (Figure S15). It was also worth mentioning that no significant difference in body weight was observed during the treatment (Figure 5F). Furthermore, hematoxylin and eosin staining of major organ slices of the treated mice revealed a low systemic toxicity of the tubular carriers (Figures 5H and S16).

Conclusion

We have successfully prepared a kind of pH-responsive drug carrier from the self-assembly of aromatic macrocycles. The aromatic carriers surrounded by ethylene oxide segment exhibited non-toxicity and biocompatibility. The protonation led the macrocycles to be positively charged, which resulted in the separation of porous carriers. Notably, the responsive carriers can recognize external acidic environment fully dependent on the concentration of macrocycles. We found that the carriers dissolved into discrete macrocycles in mild acidic state with the pH close to 4.6. Thus the systems provided efficient anticancer effect with a low systemic cytotoxicity in vitro and in vivo. In particular, 2 with enhanced positive charge showed preferable inhibition than 1 for the efficient release of drug from the complete separation of carriers. Given these results, the work presented herein provides a practicable strategy for enhanced intracellular drug accumulation by regulating the charge of carriers to prompt efficient release of drug for cancer therapeutics.

Limitations of the Study

Although pH-responsive tubular assembly was successfully used to deliver drugs into cancer cell and reduce the side effect of residual drugs through full dissociation by charge control, responsive carriers lack the specific selectivity for cancer cell. Thus, we will graft a ligand with active targeting into tubular surrounding to develop their special recognition of cancer cell in further study.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.