Biomaterials 35 (2014) 7951e7962

Contents lists available at ScienceDirect

Biomaterials journal homepage: www.elsevier.com/locate/biomaterials

Intracellular redox-activated anticancer drug delivery by functionalized hollow mesoporous silica nanoreservoirs with tumor specificity Zhong Luo a, b, Yan Hu a, **, Kaiyong Cai a, Xingwei Ding a, Quan Zhang b, Menghuan Li b, Xing Ma c, Beilu Zhang a, Yongfei Zeng b, Peizhou Li b, Jinghua Li b, Junjie Liu b, Yanli Zhao b, c, * a

Key Laboratory of Biorheological Science and Technology, Ministry of Education, College of Bioengineering, Chongqing University, Chongqing 400044, PR China b Division of Chemistry and Biological Chemistry, School of Physical and Mathematical Sciences, Nanyang Technological University, 21 Nanyang Link, Singapore 637371, Singapore c School of Materials Science and Engineering, Nanyang Technological University, Singapore 639798, Singapore

a r t i c l e i n f o

a b s t r a c t

Article history: Received 14 May 2014 Accepted 21 May 2014 Available online 13 June 2014

In this study, a type of intracellular redox-triggered hollow mesoporous silica nanoreservoirs (HMSNs) with tumor specificity was developed in order to deliver anticancer drug (i.e., doxorubicin (DOX)) to the target tumor cells with high therapeutic efficiency and reduced side effects. Firstly, adamantanamine was grafted onto the orifices of HMSNs using a redox-cleavable disulfide bond as an intermediate linker. Subsequently, a synthetic functional molecule, lactobionic acid-grafted-b-cyclodextrin (b-CD-LA), was immobilized on the surface of HMSNs through specific complexation with the adamantyl group, where b-CD served as an end-capper to keep the loaded drug within HMSNs. b-CD-LA on HMSNs could also act as a targeting agent towards tumor cells (i.e., HepG2 cells), since the lactose group in b-CD-LA is a specific ligand binding with the asialoglycoprotein receptor (ASGP-R) on HepG2 cells. In vitro studies demonstrated that DOX-loaded nanoreservoirs could be selectively endocytosed by HepG2 cells, releasing therapeutic DOX into cytoplasm and efficiently inducing the apoptosis and cell death. In vivo investigations further confirmed that DOX-loaded nanoreservoirs could permeate into the tumor sites and actively interact with tumor cells, which inhibited the tumor growth with the minimized side effect. On the whole, this drug delivery system exhibits a great potential as an efficient carrier for targeted tumor therapy in vitro and in vivo. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Drug delivery system Hollow mesoporous silica nanoreservoirs In vivo studies Redox-triggered release Targeted tumor therapy

1. Introduction Tumor illness is one of the leading causes of human death and difficult to be cured [1]. It is, therefore, imperative to develop potent treatment methods to cure tumor illness, while with less side effects to patients [2]. Recently, the approach of

* Corresponding author. Division of Chemistry and Biological Chemistry, School of Physical and Mathematical Sciences, Nanyang Technological University, Singapore 637371, Singapore. Tel.: þ65 63168792; fax: þ65 67911961. ** Corresponding author. College of Bioengineering, Chongqing University, Chongqing 400044, PR China. Tel.: þ86 23 65126127; fax: þ86 23 65106127. E-mail addresses: [email protected] (Y. Hu), [email protected] (Y. Zhao). http://dx.doi.org/10.1016/j.biomaterials.2014.05.058 0142-9612/© 2014 Elsevier Ltd. All rights reserved.

“nanomedicine” has provided an immense potential to revolutionize tumor treatments by designing nanoscale drug delivery systems for targeted administration in order to achieve the optimal treatments [2,3]. Mesoporous silica nanoparticles (MSNs), as one of representative chemotherapeutic agent delivery vehicles, have been utilized to fabricate controlled drug release systems on account of their unique features including ordered framework, tunable pore size, large specific surface area and very low cytotoxicity [4e6]. Moreover, the abundant original silanol groups (SieOH) on MSNs further facilitate them for post-functionalization [4e6]. Up to now, various types of MSN-based stimuli-responsive drug delivery systems have been developed [7e28]. For example, inorganic nanoparticles (e.g., Au [7,8], Fe3O4 [9,10], CdSe [11], and Zinc [12]), biomacromolecules (e.g., lactose [13,14], antibody

7952

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

[15,16], DNA [17,18], liposome [19,20], and collagen [21]) and (supra)molecular machines (e.g., pseudorotaxanes and rotaxanes [4,13,14,22e25]) serving as the end-cappers have already been incorporated onto the orifices of MSNs through cleavable intermediate linkages. However, a crucial issue still remains in these systems, which is how to efficiently “switch off” and “switch on” the mesopores in response to biological signals or external stimuli for controlled drug delivery in vivo. Generally, the approaches for stimuli-triggered “opening” of the end-cappers can be classified into the following categories, i.e., redox reaction [11,21,25e28], pH change [9,10,12,22,24], enzymemediated action [7,8,13e15,18], light irradiation [17], temperature change [29], magnetic field [23] and so on. For instance, Stoddart and Zink reported a series of (supra)molecular machinefunctionalized MSNs by employing switchable [2] pseudorotaxanes and [2] rotaxanes as the end-cappers to control the cargo loading and release under pH, enzyme and temperature stimuli [13,14,22e25]. These systems showed high drug encapsulation capability and high sensitivity to external stimuli, and were highly modular for specific requirements. We also developed diverse types of redox- [25,27,30], pH- [31,32], enzyme- [33], and light- [34] responsive release systems based on MSNs for controlled and targeted drug delivery. In our previous studies, lactobionic acid (LA) was employed as one of the targeting ligands to HepG2 cells, since the lactose group is a specific ligand binding to the asialoglycoprotein receptor (ASGP-R) on the membrane of HepG2 cells [32,35]. Since most of these studies did not involve the investigation and evaluation of (supra)molecular machine-functionalized MSNs for in vivo drug delivery, it is urgent to accumulate the proofs of the interactions between these systems and tumor cells, as well as animal curative effects of these systems in vivo for potential clinical applications. The disulfide bond, one of versatile and bio-cleavable linkages, has been used as an intermediate linker to connect the endcappers with MSNs for redox-triggered release of loaded cargos, since the amount of reducing agent (i.e., glutathione (GSH)) within tumor cells is nearly 103-fold higher than that of GSH in the extracellular matrix [26,28]. Herein, we reported the fabrication of intracellular redox-responsive hollow mesoporous silica nanoreservoirs (HMSNs), in which adamantanamine was grafted onto the orifices of HMSNs through a redox-cleavable disulfide bond linkage followed by end-capping with lactobionic acid-grafted-bcyclodextrin (b-CD-LA) via strong complexation between the adamantane unit and b-CD (Fig. 1). In this work, HMSNs were used as the carriers of anticancer drug doxorubicin (DOX) with enhanced loading capacity as compared with ordinary MSNs. bCD-LA could act as both an end-capper of HMSNs and a targeting ligand towards HepG2 cells [32,35]. DOX-loaded nanoreservoirs were then employed to treat HepG2 cells and tumor-bearing nude mice in order to investigate their therapeutic effects in vitro and in vivo, respectively. 2. Materials and methods 2.1. Materials All chemical reagents were purchased from SigmaeAldrich and used without further purifications. Cell culture medium and cellular imaging dyes were purchased from Invitrogen Company (USA). All solvents and inorganic reagents were commercially available. 2.2. Synthesis of 2-carboxyethyl 2-pyridyl disulfide 2-Carboxyethyl 2-pyridyl disulfide was synthesized according to our previous report [27], and its preparation was also described in the Supplementary data. 2.3. Synthesis of b-CD-LA Firstly, b-CD was reacted with p-toluenesulfonyl chloride (p-TsCl) according to a previous report with some changes [36]. Briefly, b-CD (31.7 mmol) was dispersed

into deionized water (300 mL) by using a 500 mL round bottom flask. NaOH (98.4 mmol) in deionized water (8 mL) was added dropwise into the b-CD solution within 5 min. After the solution became clear, p-TsCl (31.7 mmol) in acetonitrile (18 mL) was added dropwise into the above mixture solution under an ice bath within 8 min. The solution was continuously stirred at room temperature for another 3 h. Then, the solution was adjusted to neutral condition (pH ¼ 7.0) by using HCl and then placed in a refrigerator under 4  C overnight. After the filtration, the precipitate was collected and re-dispersed into acetone and ethanol to remove unreacted p-TsCl. The precipitate was recrystallized with distilled water for several times, and dried by vacuum freeze-dryer to obtain the final product denoted as tosyl-b-CD. Secondly, tosyl-b-CD was functionalized with ethylenediamine to obtain amino b-CD according to a previous report with some modifications [36]. Briefly, tosyl-bCD (3.0 g, 2.33 mmol) was suspended in dry DMF (15 mL) in a 250 mL two-necked round bottom flask under nitrogen protection. Ethylenediamine (3 mL) was then added into the flask via a syringe and the mixture was refluxed at 60  C overnight. After cooling the solution to room temperature, the resulted product was added dropwise to acetone (150 mL) for purifications. The obtained precipitate was filtered and washed extensively with ethanol (50 mL) and acetone (50 mL  2). After that, the powder was re-dissolved in water (3 mL) followed by the precipitation with acetone (150 mL) for two times. Finally, the precipitate was collected, dialyzed with a cellulose bag filter (MW: 500) for 3 days, and dried at vacuum drying oven for overnight to afford the final product denoted as b-CD-NH2. Thirdly, b-CD-NH2 was reacted with LA to produce b-CD-LA. Briefly, a mixture of LA (0.3 g, 0.85 mmol), 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC, 0.64 mmol) and N-hydroxysuccinimide (NHS, 0.64 mmol) was dissolved into hexamethylenetetramine/HCl buffer solution (20 mL, pH ¼ 5.4) under gentle stirring for 4 h. Then, b-CD-NH2 (0.5 g, 0.43 mmol) was added into the above solution, which was stirred for overnight. The resultant mixture was added dropwise to acetone (150 mL) with stirring. After the filtration, the precipitate was washed extensively with ethyl ether (50 mL). The obtained powder was re-dissolved in water (5 mL) followed by the precipitation with cold ethyl ether (40  C). Finally, the product was dialyzed with a cellulose bag filter (Mw: 500) for 3 days, and dried under vacuum freeze-dryer for overnight.

2.4. Preparation of solid SiO2 nanoparticles € ber method acSolid SiO2 nanoparticles (~100 nm) were synthesized via Sto cording to previous reports [25,37]. Briefly, ammonium hydroxide (10 mL) and tetraethoxysilane (TEOS, 10 mL) were sequentially dissolved into an ethanol/water (428 mL/60 mL, v/v) mixture, which was stirred at 30  C for 2 h. After centrifugation (10,000 rpm), the product was rinsed with ethanol and distilled water for several times, respectively. The final nanoparticles were dispersed into distilled water by ultrasonication for further uses. 2.5. Preparation of SiO2@CTAB-SiO2 core/shell nanoparticles The core/shell silica nanoparticles were synthesized according to previous reports with some modifications [25,37]. Briefly, hexadecyltrimethylammonium bromide (CTAB, 150 mg) was dissolved into a mixture solution of water/ethanol (30 mL/30 mL, v/v) containing concentrated ammonia water (0.55 mL). After treatment with ultrasonication for 30 min, the solid SiO2 nanoparticles (100 mg) in distilled water (20 mL) were added into the above solution, which was stirred at room temperature for 30 min. TEOS (0.25 mL) was subsequently added to the mixture solution, and the reaction was continued for another 6 h. After centrifugation (10,000 rpm), the crude material was sequentially rinsed with ethanol and water for several times, and then re-dispersed into distilled water (15 mL) to afford the final product denoted as SiO2@CTAB-SiO2. 2.6. Preparation of HMSNs HMSNs were prepared by a selective etching method based on previous reports [25,37]. Briefly, the water suspension of the obtained SiO2@CTAB-SiO2 was ultrasonicated for 20 min and then vigorously stirred at room temperature for 4 h. Sodium carbonate (Na2CO3, 470 mg) in water (5 mL) was subsequently added into the mixture solution, which was remained at 50  C for 8 h to selectively etch the SiO2 core. After centrifugation and extensive washing with water (30 mL  3), the obtained HMSNs were dispersed in distilled water under stirring for overnight to remove the residues. 2.7. Sulfhydration of HMSNs The obtained HMSNs were dispersed into anhydrous toluene (50 mL) containing 3-mercaptopropyl-trimethoxysilane (MPTS, 0.5 mL), and the suspension was refluxed at 60  C with gentle stirring for 24 h [25]. After centrifugation, the products were extensively washed with acetone and ethanol to remove excess MPTS. To further extract CTAB from the hollow silica nanoparticles, the samples were dispersed into methanol/hydrochloric acid mixture solution under reflux at 80  C for 36 h, leading to MPTS-functionalized HMSNs denoted as HMSNs-HS.

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

7953

Fig. 1. (a) Fabrication of redox-triggered HMSNs by using a disulfide bond as the intermediate linker. (b) Schematic illustration of the intracellular redox-triggered HMSNs for targeted tumor therapy in vitro and in vivo.

2.8. Functionalization of nanoreservoirs with disulfide bond The obtained HMSNs-HS (0.5 g) was dispersed in ethanol solution (30 mL) containing acetic acid (1.0 mL). Then, the synthesized 2-carboxyethyl 2-pyridyl disulfide (0.3 g) was added to the solution, which was stirred at room temperature for another 48 h [27]. After centrifugation, the resulted product was washed with ethanol and distilled water for 3 times, respectively. After drying under vacuum at 30  C for 24 h, the final product was achieved, denoted as HMSNs-S-S-COOH. 2.9. Loading of model drug into hollow nanoreservoirs The loading of anticancer drug (doxorubicin hydrochloride, DOX) or fluorescein isothiocyanate (FITC) was achieved via diffusion effect [25,32]. For the DOX loading, the synthesized HMSNs-S-S-COOH (20 mg) and DOX (15 mg) were dispersed into distilled water (12 mL) by ultrasonication, and then the suspension was stirred gently for 24 h, leading to DOX-loaded nanoparticles denoted as HMSNs-S-SCOOH@DOX. For the FITC loading, HMSNs-S-S-COOH (20 mg) and FITC (5 mg) were dispersed into PBS (phosphate buffered saline) solution (12 mL) by ultrasonication, and the solution was stirred gently for 24 h, leading to FITC-loaded nanoparticles denoted as HMSNs-S-S-COOH@FITC. When HMSNs-S-S-COOH was replaced by free HMSNs (10 mg), the corresponding products were denoted as HMSNs@DOX and HMSNs@FITC. 2.10. Fabrication of redox-triggered hollow nanoreservoirs with tumor specificity The drug-loaded nanoparticles (HMSNs-S-S-COOH@DOX or HMSNs-S-SCOOH@FITC) were finally capped with b-CD or b-CD-LA. After adjusting the pH value of aqueous suspension containing drug-loaded nanoparticles to approximately 5.4, a mixture of 1-adamantanamine hydrochloride (Ada, 20 mg), 1-ethyl-3-(3dimethylamino propyl)carbodiimide hydrochloride (EDC, 15 mg) and N-

hydroxysuccinimide (NHS, 12 mg) in buffer solution (2 mL, pH ¼ 5.4) was added to the above suspension, and the obtained mixture solution was stirred at room temperature for 24 h. The resulted products were denoted as HMSNs-S-S-Ada@DOX or HMSNs-S-S-Ada@FITC. When free b-CD (0.16 g) or b-CD-LA (0.18 g) in water (3 mL) was subsequently added into the suspension of HMSNs-S-S-Ada@DOX or HMSNs-S-S-Ada@FITC, the complexes between the b-CD ring and the adamantyl group on the surface of HMSNs were formed through a hydrophobic interaction [30]. The inclusion reaction was carried out under gentle stirring for 24 h. After centrifugation, the samples were washed with distilled water for several times and dried by lyophilization. The resulted corresponding products were denoted as HMSNs-S-S-Ada/b-CD@DOX, HMSNs-S-S-Ada/b-CD-LA@DOX, HMSNs-S-S-Ada/b-CD@FITC, and HMSNs-S-S-Ada/ b-CD-LA@FITC. When the HMSNs-S-S-COOH nanoparticles without the loading of model drugs were used initially, the resulted products were donated as HMSNs-S-SAda/b-CD and HMSNs-S-S-Ada/b-CD-LA. 2.11. In vitro DOX release study In vitro DOX release of the HMSNs-S-S-Ada/b-CD-LA@DOX nanoparticles was monitored by a fluorospectrophotometer. To investigate the real-time drug release behavior, an improved crystal cuvette was firstly prepared based on our previous study [25] and filled with Tris-solution buffer (4 mL, pH ¼ 7.0). The HMSNs-S-S-Ada/ b-CD-LA@DOX (1 mg) nanoparticles were subsequently placed on the topside of the crystal cuvette with the addition of different amount of intracellular reductive GSH (10 mM and 1 mM). Finally, the real-time release behavior over 120 min was monitored at an emission wavelength of 545 nm under an excitation wavelength of 480 nm. The long-term drug release behavior was also investigated. Briefly, the HMSNsS-S-Ada/b-CD-LA@DOX (3.5 mg) nanoparticles were firstly suspended in Tris buffer

7954

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

(0.75 mL) with different amount of GSH (10 mM, 1 mM and 0 mM). Then, the incubation medium was removed for analysis at given time intervals, which was replaced with the same volume of fresh Tris buffer. Finally, the extracted incubation medium was studied by fluorescence spectrophotometer at a wavelength of 545 nm. 2.12. Characterizations NMR and mass spectrometry (MS) were operated on a Bruker BBFO-400 spectrometer and a Thermo Finnigan LCQ Fleet MS spectrometer, respectively, for the characterizations of the synthesized molecules. SEM (scanning electron microscope) and TEM (transmission electron microscope) images were performed on the JSM7100F (JEOL, Japan) and JEM-1400 (JEOL), respectively, for the observations of the morphologies and size distributions of nanoparticles. Fourier transform infrared spectroscopy (FITR) by IR Prestige-21 spectrometer (Shimadzu Corporation, Japan), BrunauereEmmetteTeller (BET) by Autosorb-IQ ASIQM 0000-3 (Quantachrome, USA), and thermogravimetric analysis (TGA) by TA-Q500 analyzer (TA Company, USA) were employed to monitor the preparation processes of the drug delivery system. The drug release behavior was investigated by a fluorospectrophotometer (RF5301PC, Shimadzu company, Japan) using a 1 cm quartz cell. The distribution of nanoparticles within cells was determined by TEM and confocal laser scanning microscopy (CLSM, LSM 510 Metanlo, Zeiss Co., Germany). Flow cytometry (Coulter Epice XL, Beckman Coulter, USA) was used to quantitatively determine the percentage of cells containing nanoparticles and the amount of nanoparticles endocytosed by cells. DNA ladder assays were performed on a wide mini-sub cell GT cell (110 V, Bio-rad Co) and observed by Gel Doc XR þ imager (Bio-rad Co., USA). After the tumor tissues were frozen and cut into thin sections, they were stained with TUNEL (terminal deoxynucleotidyl transferase dUTP nick-end labeling) kits and observed by CLSM.

Finally, the percentage of cells containing nanoparticles and the amount of endocytosed nanoparticles were analyzed by flow cytometry, respectively. 2.15. Intracellular DOX release observation The intracellular release of DOX from nanoparticles was observed by CLSM [25,30]. After the cells were incubated with TCPS, DOX (2 mg mL1), HMSNs, HMSNs@DOX, HMSNs-S-S-Ada/b-CD@DOX, and HMSNs-S-S-Ada/b-CD-LA@DOX (10 mg mL1) for 12 h, respectively, the culture medium was removed and the cells were washed with fresh PBS for 3 times. Then, HepG2 cells were fixed with 2% glutaraldehyde for 20 min, washed with PBS for 3 times, and stained with Hoechst 33258 (10 mg mL1) for 5 min. The resultant samples were mounted with 90% glycerinum and observed by CLSM. 2.16. Cytotoxicity assay The influences of free nanoparticles and DOX-loaded nanoreservoirs on the proliferations of HepG2 cells were determined quantitatively by using 3-(4,5dimethylthiazol-2-yl)-2,5 diphenyl tetrazolium bromide (MTT) assay [21,25,27,34]. When cell confluence reached around 60e70%, the culture medium in a 96-well plate was replaced with fresh medium. DOX (2 mg mL1), HMSNs, HMSNs-S-SAda/b-CD, HMSNs-S-S-Ada/b-CD-LA, HMSNs@DOX, HMSNs-S-S-Ada/b-CD@DOX and HMSNs-S-S-Ada/b-CD-LA@DOX (10 mg mL1) were then added into the cell culture medium, respectively [21,28]. After incubation at 37  C for 6 h, 12 h and 24 h, respectively, the medium was further replaced with fresh culture medium (200 mL) containing MTT solution (20 mL, 5 mg mL1) and then cultured for another 4 h. Finally, MTT containing medium was discarded, and dimethyl sulfoxide (DMSO, 0.15 mL) was used to dissolve formazen crystals. The optical density of the solution was measured by enzyme linked immunosorbent assay (ELISA) at a wavelength of 490 nm [21,25,27,32]. 2.17. Measurement of apoptosis by DNA fragmentation

2.13. Cell culture HepG2 cells and human endothelial cells were cultured with DMEM (Dulbecco's modified eagle medium) and RPMI1640 medium with 10% fetal bovine serum (FBS, Gibco), penicillin (100 U mL1) and streptomycin (100 mg mL1) at 37  C under 5% CO2 atmosphere [21,25,27]. The cell culture medium was replaced every 48 h. When the cell confluence reached around 60e70%, the medium was replaced with a fresh medium, and the samples of DOX, HMSNs, HMSNs-S-S-Ada/b-CD-LA, HMSNs@DOX, HMSNs@FITC, HMSNs-S-S-Ada/b-CD@DOX, HMSNs-S-S-Ada/b-CD@FITC, HMSNs-SS-Ada/b-CD-LA@DOX and HMSNs-S-S-Ada/b-CD-LA@FITC were added, respectively, for the following studies. 2.14. Cell uptake assay To explore the distribution of nanoparticles within tumor cells, TEM was employed to observe the intracellular specific location of nanoparticles according to previously reported procedures [25,38]. Briefly, HepG2 cells were treated with tissue culture polystyrene (TCPS) as a control group, HMSNs, HMSNs-S-S-Ada/b-CD and HMSNs-S-S-Ada/b-CD-LA (70 mg mL1), respectively. After the culture for 24 h, cells were collected by using centrifugation (2000) and washed with cacodylate buffer for 3 times. A mixture solution of glutaraldehyde (2% w/v) and paraformaldehyde (2% w/v) was further used to fix the samples at 4  C for 2 h. Then, the samples were post-fixed in osmic acid (2%) and stained in a uranyl acetate solution for every 15 min. After the dehydration in a mixture solution of dehydrated ethanol and spurr's plastic (1:1, v/v) for another 1 h, spurr's plastic was changed and placed in a vacuum oven at 60  C overnight. Finally, the samples were cut into ultrathin sections using a microtome, stained with uranyl acetate on a Cu grid for 5 min, and observed by TEM. To investigate the general distribution of nanoparticles within tumor cells, CLSM was utilized to trace the endocytosed nanoparticles according to previously reported procedures [21,25,27,34]. Briefly, HepG2 cells were incubated with HMSNs, HMSNs@FITC, HMSNs-S-S-Ada/b-CD@FITC, and HMSNs-S-S-Ada/b-CD-LA@FITC (70 mg mL1), respectively [9,19,21]. After the culture at 37  C for 12 h and 24 h, typan blue (200 mg mL1) was introduced into the medium for quenching intracellular fluorescence for 10 min. Then, cells were fixed with 2% glutaraldehyde for 20 min and permeabilized with 0.2% Triton X-100 at 4  C for 3 min. Subsequently, the samples were stained with rhodamine-phalloidin (5 U mL1) at 4  C overnight. After that, the cell nuclei were further stained with Hoechst 33258 (10 mg mL1) for 5 min. The resultant samples were finally mounted with 90% glycerinum and observed by CLSM. To further quantify the fraction (percentage (%) of FITC positive cells) of cells that endocytose nanoparticles, flow cytometry was used to determine the percentage of cells containing nanoparticles and the amount of nanoparticles endocytosed by cells [21,25]. Briefly, HepG2 cells and human endothelial cells were cultured into a sixwell plate at an initial seeding density of 2  104 cells cm2, respectively. After treated with nanoparticles (30 mg mL1) at 37  C for 2 h and 4 h respectively, the cells were collected and the extracellular fluorescence was quenched by typan blue (200 mg mL1) at room temperature for 10 min. The samples were then washed with PBS and subsequently treated with trypsin to prepare the single cell suspension.

DNA ladder analyses were employed to reveal the apoptotic mechanism of tumor cells after treatment with DOX and DOX-loaded nanoparticles [25,27,32]. As described in Section 2.16, HepG2 cells were treated with TCPS, DOX (2 mg mL1), HMSNs, HMSNs-S-S-Ada/b-CD@DOX, and HMSNs-S-S-Ada/b-CD-LA@DOX (10 mg mL1), respectively. After incubation at 37  C for 12 h and 24 h respectively, HepG2 cells were collected and DNA was extracted according to the operation manual of cell apoptosis DNA ladder isolation kit (Invitrogen Co., Ltd.), followed by the identification via gel electrophoresis with 0.8% agarose gel. 2.18. Establishment of tumor models in vivo The normal nude mice (average weight 19.2 ± 0.5 g) were kindly provided by the Xingqiao Hospital (Chongqing, China). In vivo experiments were operated in compliance with the Animal Management Rules of the Ministry of Health of the People's Republic of China (Document NO. 55, 2001). Briefly, HepG2 cells (0.2 mL, 1.8  107 cells mL1 in PBS) were injected into the subcutaneous tissue of the nude mice [25,39]. The nude mice were monitored every day for real-time recording the changes of tumor sizes by a digital caliper. When the tumor size reached to an average volume of 25 mm3 calculated by a formula: (volume ¼ (tumor length)  (tumor width)2/2), the nude mice were used for further studies [40]. Six representative mice were sacrificed for observing the initial tumor morphology. 2.19. Treatment of tumor-bearing nude mice in vivo To evaluate the curative effects of DOX-loaded nanoreservoirs in vivo, sixty mice with tumor models were employed, which were divided into five groups. They were intravenously treated with saline, HMSNs, DOX, HMSNs-S-S-Ada/b-CD@DOX, and HMSNs-S-S-Ada/b-CD-LA@DOX, respectively, for three times per week [29,30]. The mice treated with saline (100 mL) were used as the control group. The mice treated with DOX-loaded nanoparticles were injected by a dose of 0.1 mg nanoparticles (100 mL, 1 mg mL1) per mouse for each time, equivalent to a DOX dose of 0.02 mg per mouse for each time. The injection amount of free HMSNs used for the investigation of its in vivo performance was kept the same (100 mL, 1 mg mL1). The tumor sizes and mouse weights were recorded with a digital caliper and a digital calculation balance before every injection, respectively. Some mice were sacrificed at one week and three weeks after the treatment, and the tumor tissues were extracted from the hosts for the morphology and other related studies [41]. 2.20. Observation of apoptosis from tumor tissues in vivo To further investigate the apoptosis of tumor cells in vivo, tumor tissues were frozen-sliced into thin sections and stained with TUNEL kits [41]. The sections of tumor tissues were firstly placed onto glass slides via glue, fixed with 4% paraformaldehyde for 60 min, and washed with PBS (2  2 min). After that, they were incubated with 0.2% Triton X-100 at 4  C for 2 min. Then, the samples were stained with TUNEL staining kits (50 mL, rhodamine B labeled tunnel apoptosis assay kits, Invitrogen Co. USA) at 37  C for another 60 min. For nucleus visualization, cells were stained with Hoechst 33258 (10 mg mL1) for 10 min. Finally, the stained samples were mounted with 90% glycerinum and observed by CLSM.

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962 2.21. Statistical analysis All results were expressed as means ± standard deviation (SD) for n ¼ 6. The statistical analysis was performed using Student's T-test and one-way analysis of variance (ANOVA) at confidence levels of 95% and 99% (OriginPro version 7.5).

3. Results and discussions 3.1. Synthesis of functional molecules 2-Carboxyethyl 2-pyridyl disulfide and b-CD-LA were synthesized and characterized for integrating with HMSNs [25,32,36]. 2Carboxyethyl 2-pyridyl disulfide was prepared by the disproportionation reaction between 2,20 -dipyridyl disulfide and 3mercaptopropionic acid in methanol solution (Scheme S1 in the Supplementary data), and it was purified by a chromatographic column [27]. Tosyl-b-CD was prepared by activating b-CD with pTsCl through the reaction between the hydroxyl group (eOH) of bCD and toluenesulfonyl group (Scheme S2 in the Supplementary data). After that, Tosyl-b-CD was conjugated with ethylenediamine to afford amino b-CD (b-CD-NH2) in basic condition (Scheme S3 in the Supplementary data). In this reaction, the tosyl group in Tosylb-CD could be easily replaced by primary amino group (eNH2) of ethylenediamine in strong alkaline environment. Then, LA was grafted onto b-CD-NH2 via the reaction between the carboxylic group of LA and primary amino group of b-CD-NH2 under EDC/NHS as the coupling reagents, leading to the formation of b-CD-LA (Scheme S4 in the Supplementary data). The successful synthesis of these molecules was confirmed by NMR (Fig. S1 in the Supplementary data) and MS (Fig. S2 in the Supplementary data). As compared with the 1H NMR peaks of 2,20 dipyridyl disulfide at da (7.11e7.14 ppm), db (7.61e7.66 ppm) and dc (8.49e8.50 ppm), the peaks of synthesized 2-carboxyethyl 2pyridyl disulfide appeared at da (7.15e7.19 ppm), db (7.59e7.69 ppm), dc (8.47e8.48 ppm), dd (3.06e3.09 ppm), and de (2.78e2.81 ppm). The specific peak assignments of 1H NMR spectra for 2,20 -dipyridyl disulfide and 2-carboxyethyl 2-pyridyl disulfide are shown in Fig. S1a,b. Similar phenomena were observed for the case of b-CD derivatives. The 1H NMR characteristic peaks of b-CD mainly located at da (5.07e7.08 ppm), db (3.87e4.02 ppm) and dc (3.58e3.69 ppm). After functionalization with ethylenediamine, the 1H NMR peaks of b-CD-NH2 appeared at dd (3.39e3.47 ppm), de (2.95e3.11 ppm), and df (2.78e2.89 ppm). As compared with 1H NMR spectrum of b-CD-NH2, some new peaks of b-CD-LA further appeared at da (~4.57 ppm), db (~4.55 ppm), dc (4.21e4.22 ppm), and dd (4.11e4.12 ppm). The peak assignments of 1H NMR spectra for bCD, b-CD-NH2 and b-CD-LA are shown in Fig. S1cee. In addition, the MS peak of 2-carboxyethyl 2-pyridyl disulfide located at 221.09 (Fig. S2a), whereas that of 2,20 -dipyridyl disulfide was at 216.02 (Fig. S2b). The MS peak of b-CD was 1135.46 (Fig. S2c), and that of bCD-NH2 was at 1177.95 (Fig. S2d). Overall, the NMR and MS observations indicate that 2-carboxyethyl 2-pyridyl disulfide, b-CDNH2 and b-CD-LA were synthesized successfully. 3.2. Preparation of HMSNs HMSNs were prepared by using a hard-template method according to previous reports with slight changes [25,37]. In this case, the dense SiO2 nanoparticles were synthesized and employed as the hard template (Fig. S3a,c in the Supplementary data). The CTAB coating was achieved via the interactions between eOH groups of SiO2 nanoparticles and ammonium units in CTAB. TEOS was hydrolyzed with CTAB through hydrophilic/hydrophobic interaction, leading to the formation of the SiO2@CTAB-SiO2 core/shell nanoparticle (Fig. S3b,d in the Supplementary data). After the dense SiO2 cores were selectively etched by Na2CO3 where the CTAB molecules

7955

protected the silica shell from etching, the hollow silica nanoparticles were achieved. The SEM and TEM images indicate that HMSNs presented uniform hollow structure and well-defined mesoporous shell with average diameter of 150 ± 25 nm (Fig. 2a,c). The average thickness of the shell was 20 ± 3 nm. Previous studies reported that the nanoparticles smaller than 200 nm could be eliminated from animal models via circulation in vivo [42]. Therefore, the obtained HMSNs were exploited as promising drug delivery vehicles. 3.3. Fabrication of redox-triggered HMSNs The multifunctional redox-responsive HMSNs were finally fabricated after being loaded with anticancer drug DOX and capped with b-CD-LA (Fig. 1). The synthetic procedures include five steps: (1) HMSNs were first grafted with sulfhydryl (eSH) groups using mercapto propyltrimethoxysilane [25], denoted as HMSNs-SH; (2) HMSNs-SH was reacted with 2-carboxyethyl 2-pyridyl disulfide through the disproportionation reaction to obtain carboxylic group (eCOOH) functionalized HMSNs containing the disulfide linkage [27], denoted as HMSNs-S-S-COOH; (3) model drug of DOX or FITC was loaded into HMSNs-S-S-COOH via diffusion, denoted as HMSNs-S-S-COOH@DOX or HMSNs-S-S-COOH@FITC; (4) Ada was immobilized onto the drug-loaded nanoparticles through the reaction between the eNH2 group of Ada and the eCOOH group of nanoparticles in the presence of EDC/NHS as coupling reagents [43], denoted as HMSNs-S-S-Ada@DOX or HMSNs-S-S-Ada@FITC; and (5) b-CD-LA or free b-CD was utilized as the end-cappers of functionalized HMSNs through the inclusion complexation between the hydrophobic cavity of b-CD and 1-adamantyl group on the HMSN surface [43], denoted as HMSNs-S-S-Ada/b-CD-LA@DOX, HMSNs-S-S-Ada/b-CD-LA@FITC, HMSNs-S-S-Ada/b-CD@DOX, or HMSNs-S-S-Ada/b-CD@FITC. 3.4. Characterizations of drug delivery system To monitor the fabrication processes, the obtained materials were fully characterized and verified by SEM, TEM, BET adsorption/ desorption measurements, BarretteJoynereHalenda (BJH) pore size and volume analysis, FTIR, and TGA. The SEM and TEM images show that both HMSNs and functionalized HMSNs present uniform hollow structures with average diameter of 150 ± 25 nm and average shell thickness of 20 ± 3 nm (Fig. 2aed). It is noteworthy that the mesoporous structure of HMSNs-S-S-Ada/b-CD-LA became blurry (Fig. 2b,d) as compared with that of bare HMSNs (Fig. 2a,c) on account of the surface functionalization [21,25]. The BET surface area of HMSNs-S-S-Ada/b-CD-LA@DOX decreased to 423.6 m2 g1, whereas that of free HMSNs reached to 1284.5 m2 g1 (Fig. S4a and Table S1 in the Supplementary data). The same trend was observed for the BJH pore size measurements after the grafting processes (Fig. S4b and Table S1 in the Supplementary data). These results suggest that the mesopores (~4.0 nm in diameter) of HMSNs were end-capped by the b-CD-LA molecules. FTIR spectra (Fig. S5 in the Supplementary data) further proved that HMSNs-S-S-Ada/b-CD-LA was successfully obtained after the step-by-step reactions and the detailed analysis was provided in the Supplementary data. Furthermore, TGA confirmed that there was approximately 20 wt% of DOX in the DOX-loaded HMSNs (Fig. S5), which was much higher than that of traditional MSNs (less than 10 wt%) [25]. The contents of other functional species on the surface of HMSNs were also determined quantitatively (Fig. S6 in the Supplementary data). The amount of loaded FITC per unit weight of HMSNs was also determined using the fluorescence spectrophotometer. The results show that the FITC amount in 1 mg of HMSNs@FITC, HMSNs-S-S-Ada/bCD@FITC, and HMSNs-S-S-Ada/b-CD@FITC was 0.031 ± 0.007 mg,

7956

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

Fig. 2. Representative morphologies of free HMSNs and end-capped HMSNs (HMSNs-S-S-Ada/b-CD-LA). SEM images of (a) free HMSNs and (b) end-capped HMSNs, scale bar: 100 nm. TEM images of (c) free HMSNs and (d) end-capped HMSNs, scale bar: 50 nm.

0.029 ± 0.008 mg, and 0.028 ± 0.008 mg, respectively, which are suitable for intracellular fluorescence tracing purpose. 3.5. Drug release behavior in solution The redox-triggered release of DOX from HMSNs-S-S-Ada/b-CDLA@DOX was investigated. Firstly, the real-time release behavior was monitored by a fluorescence spectrometer [22,25,43]. HMSNsS-S-Ada/b-CD-LA@DOX was dispersed in neutral Tris buffer solutions with various amounts of reductive GSH (Fig. 3a). At physiological conditions (Tris buffers), only a negligible amount of DOX was leaked from HMSNs-S-S-Ada/b-CD-LA@DOX within 2 h, indicating a good blocking capability of b-CD-LA onto the mesopores of HMSNs [22,30,34]. The fluorescence intensity of DOX drastically increased to 300 (au) and 622 (au) after the treatments with Tris buffers containing 1 mM and 10 mM GSH, respectively. It could be explained that GSH induces the cleavage of the disulfide bonds in HMSNs-S-S-Ada/b-CD-LA@DOX, thus resulting in the removal of the b-CD-LA capping molecules from the nanoreservoirs for the DOX release [44]. The DOX release rate from HMSNs-S-S-Ada/b-CDLA@DOX was linearly correlated to the concentration of GSH. Previous reports demonstrated that GSH is abundantly distributed in the cytoplasm of tumor cells [26,28]. Thus, the HMSNs-S-S-Ada/bCD-LA@DOX delivery system was able to release the drug triggered by intracellular GSH. The long-term redox-triggered release behavior of HMSNs-S-SAda/b-CD-LA@DOX was further investigated by treating the system with Tris buffers containing 1 mM and 10 mM GSH (Fig. 3b). For the control group (pure Tris buffer), the leaked amount of DOX was still at a low level (nearly 9.2%) even after incubation for 26 h, suggesting that the disulfide bonds between the b-CD-LA capping units and HMSNs were stable at physiological conditions and the b-CDLA capping molecules could keep the loaded drug within the

hollow nanoreservoirs [44]. The released amounts of DOX increased to over 38% and 82% after treated with 1 mM and 10 mM GSH for 26 h, respectively. In this case, when the disulfide bonds were cleaved after exposure to a reductive environment (i.e., GSH), the dissociation of the b-CD-LA capping units from the HMSN surfaces enabled the release of DOX from uncapped mesopores [44]. All these results confirmed that the HMSNs-S-S-Ada/b-CDLA@DOX system is highly sensitive to the GSH stimulus. 3.6. In vitro cellular uptake assay To better understand the interactions between the nanoparticles and cancer cells in vitro, TEM was firstly used to observe the cell morphologies and the specific location of endocytosed nanoparticles within HepG2 cells [25,27,32,38]. The cells on TCPS were used as the control group. It was observed that HepG2 cells treated with the nanoparticles still displayed a well-spreading morphology as the same as the control group, having intact and distinct structures of cell membrane and nucleus (Fig. 4aed). Moreover, the nanoparticles did not interact with the cell nuclei even they already located in the cytoplasm. These observations were consistent with previous studies [25,27,32,38]. The obtained results along with the cell viability assay (Fig. S7 in the Supplementary data) indicate that all of the nanoparticles (HMSNs, HMSNs-S-S-Ada/b-CD and HMSNs-S-S-Ada/b-CD-LA) were noncytotoxic under the current conditions. Interestingly, the uptake amount of HMSNs-S-S-Ada/b-CD-LA by HepG2 cells was much higher than that of HMSNs and HMSNs-S-S-Ada/b-CD (Fig. 4d vs b,c). The high uptake could be attributed to the targeting capability of the LA units in HMSNs-S-S-Ada/b-CD-LA to specifically bind with HepG2 cells. In addition, the endosome was formed to encapsulate the nanoparticles in situ after the endocytosis by cells, especially for the case of HMSNs-S-S-Ada/b-CD-LA (red dash in Fig. 4bed and

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

Fig. 3. In vitro redox-triggered DOX release profiles from the end-capped HMSNs. (a) Real-time release profiles over 2 h under the reductive GSH stimulus. (b) Long-term release profiles over 26 h under the reductive GSH stimulus.

Fig. S8 in the Supplementary data). Our previous studies confirmed that the endocytosed nanoparticles could escape from the endosome via cellular “sponge effects”, and enter into the cytoplasm of cancer cells where they are exposed to the reductive environment provided by intracellular GSH for the drug release [25,27,32,38]. Secondly, CLSM was utilized to approximately determine the amount of endocytosed nanoparticles by HepG2 cells and the overall distribution of nanoparticles within the cells (Fig. 4eeh and e1eh1) [21,25,27,32,34]. FITC, as a fluorescence tag, was encapsulated into HMSNs, HMSNs-S-S-Ada/b-CD and HMSNs-S-S-Ada/b-CD-LA to afford HMSNs@FITC, HMSNs-S-S-Ada/b-CD@FITC and HMSNs-S-SAda/b-CD-LA@FITC, respectively. After treatment with FITC-loaded nanoparticles for 12 and 24 h, respectively, HepG2 cells were fixed and stained with dyes, and their morphologies were investigated by CLSM [21,25,27,32,34]. As compared with the control group (TCPS), the CLSM observations showed that HepG2 cells still displayed wellspreading morphologies after the incubations with free FTIC and FITC-loaded nanoparticles (HMSNs@FITC, HMSNs-S-S-Ada/bCD@FITC and HMSNs-S-S-Ada/b-CD-LA@FITC). All the cytoskeletons (red) and cell nuclei (blue) were intact and regular. The endocytosed nanoparticles with green fluorescence mainly located in the cytoplasm, without permeation into cell nucleus [27,32]. The endocytosed amount of the nanoparticles by cells increased upon the incubation time (Fig. 4eeh vs e1eh1). Remarkably, the cells treated

7957

with HMSNs-S-S-Ada/b-CD-LA@FITC exhibited better internalization efficiency as compared with HMSNs@FITC and HMSNs-S-SAda/b-CD@FITC after incubations for 12 and 24 h, respectively (Fig. 4h vs f,g; h1 vs f1,g1). To further quantitatively assess the cell uptake capability, mean fluorescence intensity (MFI) per cell was measured (Fig. S9 in the Supplementary data) [21,25,32]. The MFI per cell treated with HMSNs-S-S-Ada/b-CD-LA@FITC was 1.52-fold and 1.78-fold higher than that of HMSNs-S-S-Ada/b-CD@FITC after incubations for 12 and 24 h, respectively. The MFI per cell for HMSNs-S-S-Ada/b-CD@FITC was 1.96-fold and 2.41-fold higher than that of HMSNs@FITC after incubations for 12 and 24 h, respectively. All the results further confirmed that the LA units in HMSNs-S-SAda-b-CD-LA could improve cellular internalization amount of the hollow nanoreservoirs via specific receptor-mediated endocytosis [21,25,27,32,34]. Thirdly, flow cytometry assay was employed to quantitatively analyze the endocytosis capability of HMSNs-S-S-Ada/b-CDLA@FITC by HepG2 cells (Fig. 4i,j and Fig. S10 in the Supplementary data) [21,25,32]. In this experiment, FITC was employed as an indicator for intracellular tracing. After incubations with HMSNs-S-SAda/b-CD-LA@FITC for 2 and 4 h, the percentages of HepG2 cells with fluorescence were around 1.66 times and 1.69 times as high as that of HepG2 cells incubated with HMSNs-S-S-Ada/b-CD@FITC, and also 1.99 times and 1.85 times higher than that of the cells treated with HMSNs@FITC under the same conditions. It is noteworthy that free FITC was hardly endocytosed by HepG2 cells, since the cell membrane prevents the entry of such small molecule into cells [21,25,32]. For further characterization of the targeting capability of HMSNs-SS-Ada/b-CD-LA@FITC, cancer cells (HepG2 cells) and normal cells (endothelial cells) were treated with the nanoparticles, respectively. It was observed that the endocytosed amount of HMSNs-S-S-Ada/bCD-LA@FITC by HepG2 cells was nearly 2.41 and 2.36 times higher than that of endothelial cells after incubations for 2h and 4h, respectively (Fig. 4j and Fig. S10 in the Supplementary data). One possible reason is that the amount of lactose receptors on HepG2 cells is much higher than that of normal cells, thus facilitating the internalization of HMSNs-S-S-Ada/b-CD-LA@FITC into HepG2 cells via specific receptor-mediated endocytosis [32,35,36]. 3.7. Cytocompatibility, intracellular release and apoptosis assay To investigate the inhibitory effects of anticancer drug DOX and DOX-loaded nanoparticles for HepG2 cells in vitro, the MTT method was firstly used to quantitatively evaluate their influences on the growth of tumor cells (Fig. S11 in the Supplementary data) [21,25,27,30,32]. DOX, as one of typical clinical anticancer drugs, was employed as the model drug to be trapped by HMSNs for the studies. HepG2 cells on free TCPS were used as the control group. The growth of HepG2 cells was severely inhibited after being treated with DOX, HMSNs-S-S-Ada/b-CD@DOX and HMSNs-S-SAda/b-CD-LA@DOX as compared with that of the TCPS control for 24 h. The inhibition effect of free DOX on the growth of HepG2 cells was the strongest one among all the groups over 12 h, 24 h and 48 h, respectively. A possible reason is that the high transient concentration of free DOX in the medium could inhibit the growth of HepG2 in the initial stage and enable the living cell amount at a low level [30]. In contrast, DOX could not be released from DOXloaded nanoreservoirs (HMSNs-S-S-Ada/b-CD-LA@DOX and HMSNs-S-S-Ada/b-CD@DOX) until they were endocytosed by HepG2 cells and exposed to the reductive environment provided by intracellular GSH. The viability of HepG2 cells treated with HMSNsS-S-Ada/b-CD@DOX over 24 h was higher than that of HMSNs-S-SAda/b-CD-LA@DOX, which could be interpreted that the LA targeting units enhanced the internalization amount of HMSNs-S-SAda/b-CD-LA@DOX by HepG2 cells via the lactose receptor-

7958

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

Fig. 4. Representative TEM images showing a HepG2 cell onto TCPS (a, control group) and HepG2 cells treated with free HMSNs (b), HMSNs-S-S-Ada/b-CD (c), and HMSNs-S-S-Ada/ b-CD-LA (d) after 12 h. Scale bar: 2 mm. Representative CLSM images showing the distributions of free FITC (e and e1), HMSNs@FITC (f and f1), HMSNs-S-S-Ada/b-CD@FITC (g and g1), and HMSNs-S-S-Ada/b-CD-LA@FITC (h and h1) within HepG2 cells after incubations for 12 and 24 h, respectively. Scale bar: 50 mm. Red: cytoskeleton, blue: cell nuclei, green: FITC-loaded nanoparticles. (i) Quantitative flow cytometry analysis showing the percentage of HepG2 cells with fluorescence after being treated with FITC, HMSNs-S-S-Ada/bCD@FITC and HMSNs-S-S-Ada/b-CD-LA@FITC, respectively. (j) Quantitative flow cytometry analysis showing the percentages of HepG2 cells and endothelial normal cells having fluorescence after incubation with HMSNs-S-S-Ada/b-CD-LA@FITC, n ¼ 6. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

mediated endocytosis, leading to greater inhibition on the growth of HepG2 cells [32,35]. The CLSM assay was also employed to visualize the interactions between anticancer drug DOX and cell organs in vitro after being treated with DOX and DOX-loaded nanoreservoirs for 24 h, respectively (Fig. 5aee and a1ee1) [21,25,27,32,45]. H33258 dye was used for coupling with cell nuclei so as to reveal their morphology with blue staining. In this study, HepG2 cells treated with TCPS and HMSNs presented oval or round nucleus morphology with obvious boundaries (Fig. 5a,b and a1,b1), which indicate that cancer cells were in a normal state [21,25,27,32,45]. On the contrary, the cell nuclei became deformed and ruptured after treated with DOX, HMSNs-S-S-Ada/b-CD@DOX and HMSNs-S-SAda/b-CD-LA@DOX for 24 h, respectively (Fig. 5cee and c1ee1), suggesting that the cells were in an apoptotic stage [21,25,27,32,45]. Moreover, it was observed that the cell nuclei presented a mixed color of red and blue after treatment with DOX or DOX-loaded nanoparticles. A major cause of this phenomenon is that free DOX or released DOX from the nanoreservoirs could interact with the

7959

cell nuclei and destroy their double-stranded DNA structures [46]. Interestingly, the red brightness within HepG2 cells was different among these treatment groups. The highest brightness was observed from the cells treated with HMSNs-S-S-Ada/b-CDLA@DOX, and the cells with HMSNs-S-S-Ada/b-CD@DOX were in the second place (Fig. 5aee and a1ee1). It can be explained that bCD or b-CD-LA could keep the loaded DOX within hollow nanoreseviors until the systems were endocytosed by HepG2 cells, while uncapped HMSNs@DOX would have the DOX leakage problem during the delivery [25]. All the results demonstrate that HMSNs-SS-Ada/b-CD-LA@DOX could specifically interact with HepG2 cells and release DOX into the cytoplasm triggered by intracellular GSH, leading to the apoptosis and cell death by destroying the doublestranded DNA structures [21,25,27,32,45]. DNA ladder analysis was further performed to reveal the apoptotic mechanism of HepG2 cells after treated with DOX and DOX-loaded nanoparticles (Fig. 5f) [32,42,47]. DOX was reported to interact with DNA and topoisomerase II of cancer cell nuclei, resulting in the apoptosis by inducing the DNA fragmentation

Fig. 5. Representative CLSM images of HepG2 cells after treated with TCPS (a and a1, control group), HMSNs (b and b1), DOX (c and c1), HMSNs@DOX (d and d1), HMSNs-S-S-Ada/bCD@DOX (e and e1) and HMSNs-S-S-Ada/b-CD-LA@DOX (f and f1) for 24 h, respectively. Scale bar: 20 mm. Merged images from dual channels of H33258 and DOX (aef) and single DOX channel (a1-f1). Red: DOX, blue: cell nuclei. (f) DNA fragmentation (DNA ladder) observation for apoptosis of HepG2 cells after being treated with TCPS (lanes a and a1), free HMSNs (lanes b and b1), DOX (lanes c and c1), HMSNs-S-S-Ada/b-CD@DOX (lanes d and d1) and HMSNs-S-S-Ada/b-CD-LA@DOX (lanes e and e1) for 12 and 24 h, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

7960

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

Fig. 6. Representative morphologies of tumor tissues after being treated with saline (I, control group), HMSNs (II), DOX (III), HMSNs-S-S-Ada/b-CD@DOX (IV) and HMSNs-S-S-Ada/ b-CD-LA@DOX (V) for 0 day, 7 days, and 21 days, respectively. (b) Real-time measurements of tumor sizes by digital vernier caliper in vivo after the treatments (*p < 0.05, **p < 0.01). (c) Final weights of tumor tissues after the treatments for 21 days (*p < 0.05, **p < 0.01). Histological observation of in situ apoptosis in tumor tissues by using a TUNEL staining method, after the mice were treated with HMSNs (d and d1), DOX (e and e1), HMSNs-S-S-Ada/b-CD@DOX (f and f1), and HMSNs-S-S-Ada/b-CD-LA@DOX (g and g1) for 7 and 21 days, respectively. Red: apoptotic DNA, blue: cell nuclei. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

[46,48]. As shown in Fig. 5f, the DNA fragmentation (bright band) was not observed when HepG2 cells were treated either on TCPS or with free HMSNs for 12 and 24 h, respectively. The phenomena indicate that free HMSNs had a good cyto-compatibility and did not destroy the DNA structures of cancer cells [21,25,27,32,48]. The DNA fragmentation (bright band) became visible after the cells were treated with DOX and DOX-loaded nanoparticles for 12 and 24 h, respectively. The HepG2 cells treated with HMSNs-S-S-Ada/b-CD-LA@DOX showed much higher degree of fragmentation than other groups. The results once again confirm that the capping agent of b-CD-LA onto the HMSN surface could keep DOX within the nanoreservoirs until they were endocytosed by HepG2 cells and exposed to the reductive environment of cytoplasm, thus avoiding the pre-leakage of the loaded DOX [21,25,27,32,48]. In addition, the LA targeting units in HMSNs-S-S-Ada/b-CD-LA@DOX played a critical role in improving the internalization amount through receptor-mediated endocytosis, further accelerating the apoptosis and cell death [32,35]. 3.8. In vivo evaluations To investigate the curative effects of DOX-loaded nanoreservoirs on tumor-bearing nude mice in vivo, the mice were firstly evaluated by real-time monitoring their weights after the treatments [40]. Tumor-bearing mice were treated with saline, HMSNs, free DOX, HMSNs-S-S-Ada/b-CD@DOX and HMSNs-S-S-Ada/b-CD-LA@DOX for three times per week, respectively. The mice treated with saline and HMSNs were used as a positive and negative control groups, respectively. As shown in Fig. S12 of the Supplementary data, the average weight of untreated nude mice was approximately 19.0 ± 0.5 g. After treatment with DOX and feeding for 21 days, the mice still maintained their weights at around 20 g (Fig. S12). For comparison, the average weights of the mice treated with saline, HMSNs, HMSNs-S-S-Ada/b-CD@DOX and HMSNs-S-S-Ada/b-CDLA@DOX increased to 25.0 g, 24.9 g, 24.0 g, and 23.6 g, respectively. The results indicate that HMSNs without cytotoxicity could relieve the side effects of DOX on nude mice when they were used as drug delivery vehicles [40]. Then, the tumor tissues were measured at fixed time intervals to investigate the changes in tumor size after being treated with DOX and DOX-loaded nanoparticles (Fig. 6aec) [25]. Tumor tissue observations (Fig. 6a) and volume measurements (Fig. 6b) show that in vivo tumor sizes increased upon feeding time after treatment with both saline and HMSNs [25]. On the other hand, the growth of tumor sizes was inhibited after treatment with DOX, HMSNs-S-S-Ada/bCD@DOX and HMSNs-S-S-Ada-b-CD-LA@DOX as compared with the control groups of saline and HMSNs. The differences of the curative effects on nude mice were also significant among the three groups of free DOX, HMSNs-S-S-Ada/b-CD@DOX, and HMSNs-S-S-Ada/b-CDLA@DOX. The best curative effect was achieved from the group of HMSNs-S-S-Ada/b-CD-LA@DOX, followed by the group of HMSNsS-S-Ada/b-CD@DOX. It was observed that HMSNs-S-S-Ada/b-CDLA@DOX could significantly inhibit the growth of tumors after administration for 21 days (Fig. 6a,b). This observation was also supported by weighing the final tumor tissues (Fig. 6c). All results suggest that high transient plasma drug concentration of free DOX might only be maintained in a short period after injection into mice via tail vein, which made the drug difficult to permeate into tumor sites for interacting with tumor cells in vivo [39]. In the worse case, free DOX in blood might quickly diffuse into other tissue organs, losing its bioactivity or being removed from the host via the blood circulation and metabolism [2]. Therefore, the administration of free DOX could also cause collateral damage to the normal cells and tissues in vivo. HMSNs-S-S-Ada/b-CD@DOX could maintain the blood drug concentration at a certain level through its redoxtriggered release and passive accumulation at the tumor sites via

7961

enhanced permeability and retention (EPR) effects after the injection. Although it might relieve some side effects of DOX on the host, it still could not facilitate the drug to have efficient tumor tissue permeation and specific interaction with the tumor cells [2]. For comparison, HMSNs-S-S-Ada/b-CD-LA@DOX not only actively accumulated at the tumor sites, but also permeated into tumor tissues to interact with tumor cells, achieving efficient inhibition of the tumor growth within nude mice. It could be interpreted that the capping agent b-CD-LA onto HMSNs keeps the loaded DOX within hollow nanoreservoirs after injected into the mice and located at the blood environments. The LA targeting units in HMSNs-S-S-Ada/bCD-LA@DOX further facilitate the nanoparticles to enter into tumor tissues followed by the DOX release into the cytoplasm under intracellular reductive GSH for tumor therapy [32,35]. Finally, the TUNEL apoptosis assay was employed to determine the apoptotic mechanism of tumor tissues in vivo by using a TUNELbased DNA fragmentation detection kit (Fig. 6deg and d1eg1) [25,41]. After treatment with HMSNs, DOX, HMSNs-S-S-Ada-bCD@DOX and HMSNs-S-S-Ada/b-CD-LA@DOX for 7 and 21 days, respectively, typical tumor tissues were extracted from the hosts and frozen-cut into thin sections for histochemical staining and observation. The histochemical staining kits composed of fluorometric DNA fragmentation detection kit III (F-dUTP) were utilized to specifically conjugate with free 30 -hydroxyl ends of the fragmented DNA. For the control group using free HMSNs, it was observed that in vivo tumor cells were still in good condition, since cell nuclei displayed round shape with a regular structure (Fig. 6d) [25,41]. The tumor cells even formed more compact structures in situ after being treated with HMSNs for 21 days (Fig. 6d1). The observations demonstrate that the activity of the tumor cells in vivo was increasing over time after treated with free HMSNs. In contrast, the growth of tumor cells was inhibited after treatment with DOX, HMSNs-S-S-Ada/b-CD@DOX and HMSNs-S-S-Ada/b-CD-LA@DOX for 7 days (Fig. 6eeg). Remarkably, the nude mice treated with HMSNs-S-S-Ada/b-CD-LA@DOX showed the highest curative effects among all the groups (Fig. 6g vs def), since a lot of apoptotic DNA (in red color) in tumor cells was observed and cell nuclei were also deformed and cracked [25,41]. The curative effects of HMSNsS-S-Ada/b-CD@DOX on nude mice were also better than that of free DOX (Fig. 6f vs e). A similar trend was achieved after the nude mice were treated for 21 days (Fig. 6d1eg1). All the results firmly demonstrate that HMSNs-S-S-Ada/b-CD-LA@DOX could efficiently target the tumor tissues and deliver DOX to tumor cells for inducing apoptosis and cell death in vivo. 4. Conclusions In summary, we have developed an intracellular redoxresponsive drug delivery system (HMSNs-S-S-Ada/b-CD-LA@DOX) by immobilizing the targeting unit of lactobionic acid-grafted-bcyclodextrin (b-CD-LA) onto hollow mesoporous silica nanoreservoirs (HSMNs) through disulfide bond linkage. b-CD-LA has been end-capped onto the orifices via the inclusion complexation between the hydrophobic cavity of b-CD and adamantyl groups on the surface of HMSNs. Thus, b-CD-LA could act as (1) the gatekeeper of HMSNs to maintain the loaded doxorubicin (DOX) within the mesopores, and (2) the targeting ligand to HepG2 tumor cells, since the lactose group of LA is a specific ligand for binding with the asialoglycoprotein receptor (ASGP-R) on the membrane of HepG2 cells. In vitro and in vivo studies have confirmed that HMSNs-S-S-Ada/bCD-LA@DOX is superb for specifically delivering DOX into tumor cells and releasing the drug into cytoplasm triggered by the intracellular reductive agents, effectively inhibiting the growth of tumors with the minimal side effects. Therefore, this system shows a great potential for efficient drug delivery in tumor therapy in vitro and in vivo.

7962

Z. Luo et al. / Biomaterials 35 (2014) 7951e7962

Acknowledgments This work was financially supported by the Natural Science Foundation of China (31200712 and 21274169), the Fundamental Research Funds for the Central Universities (Project No. CDJZR 10238801) and Natural Science Foundation of Chongqing Municipal Government (CSTC, JJA10056). It was also supported by the National Research Foundation (NRF), Prime Minister's Office, Singapore under its NRF Fellowship (NRF2009NRF-RF001-015) and Campus for Research Excellence and Technological Enterprise (CREATE) programme e Singapore Peking University Research Centre for a Sustainable Low-Carbon Future, the NTU-A*Star Centre of Excellence for Silicon Technologies (A*Star SERC No.: 112 351 0003). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.biomaterials.2014.05.058. References [1] Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J Clin 2011;61:69e90. [2] Ferrari M. Cancer nanotechnology: opportunities and challenges. Nat Rev Cancer 2005;5:161e71. [3] Farokhzad OC, Langer R. Impact of nanotechnology on drug delivery. ACS Nano 2009;3:16e20. [4] Ambrogio MW, Thomas CR, Zhao Y-L, Zink JI, Stoddart JF. Mechanized silica nanoparticles: a new frontier in theranostic nanomedicine. Acc Chem Res 2011;44:903e13. [5] Tang F, Li L, Chen D. Mesoporous silica nanoparticles: synthesis, biocompatibility and drug delivery. Adv Mater 2012;24:1504e34. [6] Chen Y, Chen H, Shi J. In vivo bio-safety evaluations and diagnostic/therapeutic applications of chemically designed mesoporous silica nanoparticles. Adv Mater 2013;25:3144e76. [7] Sun X, Zhao Y, Lin VS, Slowing II, Trewyn BG. Luciferase and luciferin coimmobilized mesoporous silica nanoparticle materials for intracellular biocatalysis. J Am Chem Soc 2011;133:18554e7. [8] Zhu CL, Lu CH, Song XY, Yang HH, Wang XR. Bioresponsive controlled release using mesoporous silica nanoparticles capped with aptamer-based molecular gate. J Am Chem Soc 2011;133:1278e81. [9] Gan Q, Lu X, Dong W, Yuan Y, Qian J, Li Y, et al. Endosomal pH-activatable magnetic nanoparticle-capped mesoporous silica for intracellular controlled release. J Mater Chem 2012;22:15960e8. [10] Gan Q, Lu X, Yuan Y, Qian J, Zhou H, Lu X, et al. A magnetic, reversible pHresponsive nanogated ensemble based on Fe3O4 nanoparticles-capped mesoporous silica. Biomaterials 2011;32:1932e42. [11] Lai C-Y, Trewyn BG, Jeftinija DM, Jeftinija K, Xu S, Jeftinija S, et al. A mesoporous silica nanosphere-based carrier system with chemically removable CdS nanoparticle caps for stimuli-responsive controlled release of neurotransmitters and drug molecules. J Am Chem Soc 2003;125:4451e9. [12] Muhammad F, Guo M, Qi W, Sun F, Wang A, Guo Y, et al. pH-triggered controlled drug release from mesoporous silica nanoparticles via intracelluar dissolution of ZnO nanolids. J Am Chem Soc 2011;133:8778e81. [13] Bernardos A, Aznar E, Marcos MD, Martinez-Manez R, Sancenon F, Soto J, et al. Enzyme-responsive controlled release using mesoporous silica supports capped with lactose. Angew Chem Int Ed Engl 2009;48:5884e7. [14] Bernardos A, Mondragon L, Aznar E, Marcos MD, Martinez-Manez R, Sancenon F, et al. Enzyme-responsive intracellular controlled release using nanometric silica mesoporous supports capped with "saccharides". ACS Nano 2010;4:6353e68. [15] Schlossbauer A, Kecht J, Bein T. Biotin-avidin as a protease-responsive cap system for controlled guest release from colloidal mesoporous silica. Angew Chem Int Ed Engl 2009;48:3092e5. [16] Li L-L, Xie M, Wang J, Li X, Wang C, Yuan Q, et al. A vitamin-responsive mesoporous nanocarrier with DNA aptamer-mediated cell targeting. Chem Commun 2013;49:5823e5. [17] Chen C, Zhou L, Geng J, Ren J, Qu X. Photosensitizer-incorporated quadruplex DNA-gated nanovehicles for light-triggered, targeted dual drug delivery to cancer cells. Small 2013;9:2793e800. [18] Zhang P, Cheng F, Zhou R, Cao J, Li J, Burda C, et al. DNA-hybrid-gated multifunctional mesoporous silica nanocarriers for dual-targeted and microRNA-responsive controlled drug delivery. Angew Chem Int Ed Engl 2014;53:2371e5. [19] Liu J, Jiang X, Ashley C, Brinker CJ. Electrostatically mediated liposome fusion and lipid exchange with a nanoparticle-supported bilayer for control of surface charge, drug containment, and delivery. J Am Chem Soc 2009;131: 7567e9.

[20] Liu J, Stace-Naughton A, Jiang X, Brinker CJ. Porous nanoparticle supported lipid bilayers (protocells) as delivery vehicles. J Am Chem Soc 2009;131:1354e5. [21] Luo Z, Cai K, Hu Y, Zhao L, Liu P, Duan L, et al. Mesoporous silica nanoparticles end-capped with collagen: redox-responsive nanoreservoirs for targeted drug delivery. Angew Chem Int Ed Engl 2011;50:640e3. [22] Du L, Liao S, Khatib HA, Stoddart JF, Zink JI. Controlled-access hollow mechanized silica nanocontainers. J Am Chem Soc 2009;131:15136e42. [23] Thomas CR, Ferris DP, Lee JH, Choi E, Cho MH, Kim ES, et al. Noninvasive remote-controlled release of drug molecules in vitro using magnetic actuation of mechanized nanoparticles. J Am Chem Soc 2010;132:10623e5. [24] He Q, Gao Y, Zhang L, Zhang Z, Gao F, Ji X, et al. A pH-responsive mesoporous silica nanoparticles-based multi-drug delivery system for overcoming multidrug resistance. Biomaterials 2011;32:7711e20. [25] Luo Z, Ding X, Hu Y, Wu S, Xiang Y, Zeng Y, et al. Engineering a hollow nanocontainer platform with multifunctional molecular machines for tumortargeted therapy in vitro and in vivo. ACS Nano 2013;7:10271e84. [26] Kim H, Kim S, Park C, Lee H, Park HJ, Kim C. Glutathione-induced intracellular release of guests from mesoporous silica nanocontainers with cyclodextrin gatekeepers. Adv Mater 2010;22:4280e3. [27] Luo Z, Cai K, Hu Y, Li J, Ding X, Zhang B, et al. Redox-responsive molecular nanoreservoirs for controlled intracellular anticancer drug delivery based on magnetic nanoparticles. Adv Mater 2011;24:431e5. [28] Verma A, Simard JM, Worrall JW, Rotello VM. Tunable reactivation of nanoparticle-inhibited beta-galactosidase by glutathione at intracellular concentrations. J Am Chem Soc 2004;126:13987e91. [29] Zhang X, Yang P, Dai Y, Pa Ma, Li X, Cheng Z, et al. Multifunctional upconverting nanocomposites with smart polymer brushes gated mesopores for cell imaging and thermo/pH dual-responsive drug controlled release. Adv Funct Mater 2013;23:4067e78. [30] Zhang Q, Liu F, Nguyen KT, Ma X, Wang X, Xing B, et al. Multifunctional mesoporous silica nanoparticles for cancer-targeted and controlled drug delivery. Adv Funct Mater 2012;22:5144e56. [31] Ma X, Zhao Y, Ng KW, Zhao Y. Integrated hollow mesoporous silica nanoparticles for target drug/siRNA co-delivery. Chem e Eur J 2013;19: 15593e155603. [32] Luo Z, Cai K, Hu Y, Zhang B, Xu D. Cell-specific intracellular anticancer drug delivery from mesoporous silica nanoparticles with pH sensitivity. Adv Healthcare Mater 2012;1:321e5. [33] Hu Y, Cai K, Luo Z, Jandt KD. Layer-by-layer assembly of b-estradiol loaded mesoporous silica nanoparticles on titanium substrates and its implication for bone homeostasis. Adv Mater 2010;22:4146e50. [34] Yan H, Teh C, Sreejith S, Zhu L, Kwok A, Fang W, et al. Functional mesoporous silica nanoparticles for photothermal-controlled drug delivery in vivo. Angew Chem Int Ed Engl 2012;51:8373e7. [35] Kim E-M, Jeong H-J, Kim S-L, Sohn M-H, Nah J-W, Bom H-S, et al. Asialoglycoprotein-receptor-targeted hepatocyte imaging using 99mTc galactosylated chitosan. Nucl Med Biol 2006;33:529e34. [36] Hu S, Li J, Xiang J, Pan J, Luo S, Cheng J-P. Asymmetric supramolecular primary amine catalysis in aqueous buffer: connections of selective recognition and asymmetric catalysis. J Am Chem Soc 2010;132:7216e28. [37] Fang X, Chen C, Liu Z, Liu P, Zheng N. A cationic surfactant assisted selective etching strategy to hollow mesoporous silica spheres. Nanoscale 2011;3: 1632e9. [38] Schrand AM, Schlager JJ, Dai L, Hussain SM. Preparation of cells for assessing ultrastructural localization of nanoparticles with transmission electron microscopy. Nat Protoc 2010;5:744e57. [39] Dhar S, Kolishetti N, Lippard SJ, Farokhzad OC. Targeted delivery of a cisplatin prodrug for safer and more effective prostate cancer therapy in vivo. Proc Natl Acad Sci USA 2011;108:1850e5. [40] Yang K, Xu H, Cheng L, Sun C, Wang J, Liu Z. In vitro and in vivo near-infrared photothermal therapy of cancer using polypyrrole organic nanoparticles. Adv Mater 2012;24:5586e92. [41] Idris NM, Gnanasammandhan MK, Zhang J, Ho PC, Mahendran R, Zhang Y. In vivo photodynamic therapy using upconversion nanoparticles as remotecontrolled nanotransducers. Nat Med 2012;18:1580e5. [42] Lu J, Liong M, Zink JI, Tamanoi F. Mesoporous silica nanoparticles as a delivery system for hydrophobic anticancer drugs. Small 2007;3:1341e6. [43] Patel K, Angelos S, Dichtel WR, Coskun A, Yang Y-W, Zink JI, et al. Enzymeresponsive snap-top covered silica nanocontainers. J Am Chem Soc 2008;130: 2382e3. [44] Cui Y, Dong H, Cai X, Wang D, Li Y. Mesoporous silica nanoparticles capped with disulfide-linked PEG gatekeepers for glutathione-mediated controlled release. ACS Appl Mater Interfaces 2012;4:3177e83. [45] Wang F, Wang Y-C, Dou S, Xiong M-H, Sun T-M, Wang J. Doxorubicin-tethered responsive gold nanoparticles facilitate intracellular drug delivery for overcoming multidrug resistance in cancer cells. ACS Nano 2011;5:3679e92. [46] Tewey K, Rowe T, Yang L, Halligan B, Liu L. Adriamycin-induced DNA damage mediated by mammalian DNA topoisomerase II. Science 1984;226:466e8. [47] Huang IP, Sun S-P, Cheng S-H, Lee C-H, Wu C-Y, Yang C-S, et al. Enhanced chemotherapy of cancer using pH-sensitive mesoporous silica nanoparticles to antagonize P-glycoprotein-mediated drug resistance. Mol Cancer Ther 2011;10:761e9. [48] Wang Y, Arriaga EA. Monitoring incorporation, transformation and subcellular distribution of N-L-leucyl-doxorubicin in uterine sarcoma cells using capillary electrophoretic techniques. Cancer Lett 2008;262:123e32.