Novel drug delivery nanosystems based on out-inside bifunctionalized mesoporous silica yolk–shell magnetic nanostars used as nanocarriers for curcumin

Abstract

Mesoporous silica has aroused lots of interest in biomedical fields, in which novel kinds of mesoporous silica with tunable mesoporosity and size have attracted much attention, but functionalization or multi-functionalization inside the mesopores of these structures is still rarely detected. Among all the functionalized molecules, beta cyclodextrin hydrate (β-CD), a kind of hydrophilic non-toxic carrier for hydrophobic drugs, can increase the solubility and bioavailability of drugs, acting as a pH responsive “gate” when functionalized on the surface of mesoporous silica. Herein, we functionalized β-CD inside novel kinds of mesoporous silica or magnetic mesoporous silica as the physical binding sites for the model drug curcumin through an out-inside two step bifunctionalization process including a vacuum pumping recrystallization drug loading process. According to the physical characterization with XRD, FT-IR, XPS, solid state NMR, BET, TG, DTA, TEM and DLS, the drug delivery nanosystems were successfully fabricated and contained nano-formulations of curcumin. In vitro cell testing, including Cell Counting Kit-8, Hoechst 33342 staining, hemolysis experiments and cell apoptosis, indicated that through the bifunctionalization, the biocompatibility of the mesoporous silica with cells could be improved, and the toxicity of the drug delivery nano-formulations towards two kinds of cancer cells, SK-HEP1 and HepG2, was significantly increased compared with free curcumin. In vitro release studies revealed that the nano-formulations contained more suspended curcumin molecules as described by the Higuchi models and showed enhanced pH responsive release properties, and that the magnetic nano-formulations exhibited alternating magnetically controlled release properties. The confocal microscopy analysis results revealed the obviously increasing intercellular release and uptake of both nano-formulations of curcumin by SK-HEP1 cells and the “on” and “off” stages of the alternating magnetism responsive intercellular release properties of the fabricated magnetic nano-formulations. The bifunctionalization process combined with the nanostar structure has immense potential in the drug delivery field, especially for hydrophobic drugs.

---

Introduction

Mesoporous silica has aroused lots of interest in biomaterials fields, such as drug delivery nanosystems, tissue scaffolds, tumor and cancer diagnosis, etc. There are many reports concerning novel shapes of mesoporous silica and functionalized mesoporous silica with smart controlled release. Novel kinds of mesoporous silica with tunable mesoporosity and size, first reported by Zhang et al.,¹ have aroused lots of attention among researchers in fields such as catalysis,² biomaterials and drug delivery materials,³ adsorption and environmental protection,⁴ and bioengineering⁵. However, reports of functionalizaton or multi-functionalization inside the mesopores of these materials are still rare, by which these novel structures could be functionalized with more physical or chemical drug binding sites and respond to more environmental conditions. This would promote the drug loading properties inside the mesopores,⁶ which could be widely applied in more biological fields.⁷ However, large mesopores can be functionalized with specific sites, which can not only be combined with drug molecules but can also prolong the drug release period and enhance the drug release properties. Beta cyclodextrin hydrate (β-CD), a kind of hydrophilic non-toxic carrier for hydrophobic drugs, can protect vitamins and natural colours⁸ and increase drug suspension, as well as being soluble and exhibiting stable properties in water,⁹ which could be a great help in increasing the bioavailability of drugs and delivery in vivo. What's more, cyclodextrin and beta cyclodextrin hydrate have been used to block drug molecules inside mesoporous channels, acting as a pH responsive “gate” on the surface of mesoporous silica.¹⁰ However, the surface nested β-CD are influenced by many effects and might difficultly keep nested state when applied in vivo before these drug delivery nanosystems were distributed to the targeted organs. Worse still, surface β-cyclodextrin hydrate may limit the surface functionalization of other specific binding sites, which might hinder the grafting of other specific targeted or functionalized molecules on the surface.

Curcumin, a kind of yellow hydrophobic pigment obtained from the rhizome of the herb Curcuma longa,¹¹ exhibits lots of bioactivity, which could be applied in therapy for cancer or tumors, heart failure and other diseases.¹² But curcumin molecules were shown to be unstable to a series of physical and chemical conditions, i.e., heat, ions, amino groups or alkali medium. Nano-formulations of curcumin have been previously prepared, such as silica aggregates,¹³ silica nanoparticle conjugates¹⁴ and silica xerogels,¹⁵ which aroused lots of interest and not only increased the delivery of the curcumin drugs, but also enhanced its stability and suspension properties. However, the controllable and targeted release of curcumin drugs is still neglected, leaving much to be desired in the delivery properties. In addition, curcumin molecules are sensitive to iron ions,¹⁶ as they are good chelators.¹⁷ When they are exposed to even a few Fe(III) ions or iron ions in Fe₃O₄ nanoparticles, the structure of the molecules might be affected in a manner that remains unclear. Mesoporous silica, with tunable structure and surface rich silanol groups, can connect drug molecules and inorganic nanoparticles, acting as a bridge. Hence, a novel structure of mesoporous silica, inside which the Fe₃O₄ nanoparticles can be encapsulated, is completely necessary. Using this novel structure to fabricate magnetic drug delivery nano-formulations could reduce the effect of ferric ions on curcumin molecules, and increase the controllable and targeted release properties of these drug delivery nano-formulations. Herein, we firstly functionalized β-CD inside the large mesopores of a novel nanostar structure with magnetic cores to act as the inside gate through out-inside two step functionalization technology.⁶ The surface functionalized β-CD could provide more physical binding sites for curcumin and enhance the interactions between the drug molecules and the inorganic materials,¹⁸ and was combined with a vacuum pumping recrystallization drug loading process,¹⁹ instead of using ammonia conjugated groups,²⁰ alkaline materials,²¹ covalent bonding to drug molecules²² and heat treatment during the drug loading process. This might influence the molecular structure and stability of curcumin during the drug loading process and storage.

The results indicate that the inside-grafted β-CD molecules provide more physical binding sites for drug molecules with more branched chains on the benzene rings, and increase drug loading inside the meso-channels. Through physical characterizations, such as XRD, FTIR, XPS, solid state NMR, BET, TEM, DLS, etc., the out-inside stepwise functionalization process was confirmed and the fabrication of drug delivery nanocarriers for curcumin was monitored. In vitro cell viability tests, Hoechst 33342 staining, hemolysis tests and cell apoptosis tests revealed that the bifunctionalized nanostars exhibit improved biocompatibility with cells compared with single materials, and significantly increase the toxicity of curcumin to two kinds of cancer cells, i.e., SK-HEP1 cells and HepG2 cells, compared with free curcumin. The in vitro release profile revealed that loading curcumin on this mesoporous silica can increase the amount of curcumin in suspension as described by the Higuchi models, and that the increase in the number of functionalized sites can also improve the suspension properties of curcumin molecules. Meanwhile, the pH responsive release showed that the release rate significantly increases at lower pH values. In addition, the cellular uptake experiments revealed that the intercellular release and uptake of curcumin in SK-HEP1 cells were increased due to the novel delivery properties of the nano-formulations. The “on–off” alternating magnetic field (AMF) induced release indicated that the magnetic bifunctionalized nanostar drug delivery systems are responsive to magnetism during controlled drug release, and that the intercellular drug release increases when the alternating magnetic field is applied. The bifunctionalization process combined with the nanostar structure and magnetic cores has immense potential in the drug delivery field.

Materials and methods

Reagents and materials

Hexadecyltrimethylammonium toluene-p-sulphonate (CTAT, 98%), (3-isocyanatopropyl)-triethoxysilane (IPTS, 95%), tris(hydroxymethyl)aminomethane (AHMPD, AR), (3-mercaptopropyl)trimethoxysilane (KH590, 95%), and β-cyclodextrin (β-CD, AR) were supplied by Sigma-Aldrich. Tetraethylorthosilicate (TEOs) and xylene were analytical reagent grade and were purchased from Guangzhou Chemical Reagent Factory, China. All the other reagents were supplied by Sigma-Aldrich unless otherwise noted.

Fabrication of nanostars and magnetic nanostars

Nanostar shaped mesoporous silica was fabricated through a soft-templating process as reported,¹ with minor adjustments. To a reactor, 1.50 g of CTAT, 0.11 g of AHMPD and 50 mL of deionized water were added. The mixed solution was mechanically stirred at 1200 rpm in a 353 K water bath for 1 h. After that, 7.8 mL of TEOs was quickly injected into the reactor and reacted at 353 K for the next 2 h. The as-synthesised nanostars were collected through a filtration process and dried at 353 K for further use. Magnetic nanostars were fabricated through an electrostatic process combined with a soft-templating process, which is shown in Fig. 1(a). To a reactor, 0.11 g of AHMPD, a certain amount of a water-based ferrofluid (0.025 g mL⁻¹ Fe₃O₄/water, as prepared in our previous work,²³ detailed steps are shown in the ESI†) and 50 mL deionized water were added. The mixed solution was vigorously stirred at 1200 rpm at 333 K for 10 min to settle the solution. After a while, 2 mL of TEOs was added dropwise to the reactor, and the mixed solution was left to react for 50 min. After the core forming reaction, 1.50 g of CTAT was added into the reactor, and the temperature was altered to 353 K to provide the soft-templating conditions for 1 h. Then, 5.8 mL of TEOs was injected quickly into the reactor, and the mixed suspension was stirred at 1200 rpm at 353 K for another 2 h. Finally, the as-synthesised magnetic nanostars were collected using a magnetic separation and filtration process, and dried at 353 K for further use. In this way, we synthesised three types of magnetic nanostars (MNS), namely M₂NS, M_1.5NS and M₁NS, by adding different amounts of water-based ferrofluid, i.e., 2 mL, 1.5 mL and 1 mL. The three samples contained different amounts of Fe₃O₄ nanoparticles: 0.05 g, 0.0375 g and 0.025 g, equal to 2.34%, 1.75% and 1.17% Fe₃O₄/SiO₂, respectively.

Fig. 1 Schemes of the fabrication of magnetic nanostars with template (a) and the bifunctionalized nanostar shaped drug delivery systems (b).

Fabrication of bifunctionalized nanostars and magnetic nanostars

Bifunctionalized nanostars and magnetic nanostars were synthesised through out-inside bifunctionalization technology. Firstly, 8 g of dried β-CD was reacted with 10 mL of (3-isocyanatopropyl)triethoxysilane (IPTS) with N=C=O bonds in 50 mL of N,N-dimethylformamide (DMF) at 323 K with magnetic stirring under an Ar atmosphere for 72 h to form Si–β-CD,²⁴ as shown in Fig. 1(b); the obtained solution was stored at 278 K under dry conditions for further use. 1.45 g of dried nanostars or magnetic nanostars with template (containing 1.0 g of pure SiO₂) were added into a reactor with 100 mL xylene and 10 mmol of KH590 (thiol groups). The mixture was refluxed for 24 h at 363 K under a N₂ atmosphere, and the powder was collected through a filtration process and washed three times with xylene and ethanol. To achieve the functionalization of the inside pores, the samples were further refluxed in 100 mL xylene with the same volume of DMF solutions containing 0.60 mmol Si–β-CD, 0.45 mmol Si–β-CD, 0.30 mmol Si–β-CD, 0.15 mmol Si–β-CD, 0.10 mmol Si–β-CD and 0.05 mmol Si–β-CD at 363 K for 24 h under a N₂ atmosphere, to form different bifunctionalized nanostars, i.e., F_0.60NS, F_0.45NS, F_0.30NS, F_0.15NS, F_0.10NS and F_0.05NS, and magnetic nanostars, i.e., F_0.60M₂NS, F_0.45M₂NS, F_0.30M₂NS, F_0.15M₂NS, F_0.10M₂NS, F_0.05M₂NS, F_0.60M_1.5NS and F_0.60M₁NS. These nanoparticles were centrifuged at 8000 rpm, washed three times with xylene, dried DMF and absolute ethanol, and then dried in a vacuum oven. The templates were removed from the nanostars, magnetic nanostars, and thiol group-modified nanostars or magnetic nanostars by stirring them three times at 400 rpm in a 10 g L⁻¹ NH₄NO₃–ethanol solution at 343 K for 3 h; the template-free samples were named NS, M₂NS, S-NS and S-M₂NS, respectively.

Fabrication of curcumin loaded bifunctionalized mesoporous silica nanostars and magnetic nanostars

Curcumin loaded bifunctionalized nanostars (Cur@FNS) and (Cur@FMNS) were fabricated through a vacuum pumping assisted recrystallization process, as shown in Fig. 2. Briefly, 100 mg of FNS or FMNS were mixed with 20 mL of a 2.0 × 10⁻³ M solution of curcumin in ethanol in a reactor after 15 min of ultrasonication, and the suspension was magnetically stirred for 1 h, followed by the dropwise addition of 20 mL of deionized water. The reactor was then vacuum pumped to a vacuum degree of 0.1 MPa for 15 min to remove the air inside the mesoporous channels of FNS or FMNS. The reactor was then shaken for 24 h in the dark in a constant temperature shaker at room temperature. The suspension in the reactor was finally centrifuged, and the upper solution was then diluted with 0.5% sodium lauryl sulfate (SLS) solution and analyzed using UV-vis spectroscopy at the 425 nm wavelength. The quantity of drug remaining could be calculated using a standard curve with a linear range of 0–10 mg L⁻¹. The drug loading capacity (LC%) and entrapment efficiency (EE%) could be calculated using the equations below. The drug delivery nanosystems were collected and dried in a vacuum oven at 303 K overnight.

Fig. 2 Scheme of the fabrication of bifunctionalized nanostar shaped drug delivery systems.

Characterization

The crystal structures of the different samples were determined by XRD (D8 Advance, BRUKER). The FT-IR spectra of the samples were recorded in KBr pellets in the wavenumber range 500–4000 cm⁻¹ using a Fourier transform infrared spectrometer (360, Nicolet Avatar). The surface functionalized molecules were analyzed by XPS (ESCLAB 250, Thermo-VG Scientific). The molecular structures were further studied by solid state ¹³C and ²⁹Si NMR (400 MHz, AVANCE AV). The thermo-properties of the different samples were tested by TGA and DTA (DTG-60, SHIMADZU) with a maximum temperature of 800 °C in dry air. The mesoporous structures of all samples were confirmed by BET analysis (GEMINI VII 2390, Micromeritics). Microstructures of the different samples were investigated by TEM (G220, Tecnai). Dynamic light scattering (DLS) experiments were performed for the different kinds of drug delivery nanosystems in two kinds of medium, i.e., PBS (pH 7.4, 0.01 M) and simulated cell supernatant (fetal bovine serum (0.05 wt%)–PBS (pH 7.4, 0.01 M)), at constant concentration (6 wt%) in order to investigate the aggregation behavior and stability properties at 25 °C using a zetasizer (Nano-ZS90, Malvern). The drug release behavior of the different samples was studied using a UV-vis spectrophotometer (UV-2550, SHIMADZU).

In vitro release properties

Drug release in 0.5% SLS solution. In order to measure the curcumin release properties of the drug delivery nanosystems, compared with curcumin alone, Cur@F_{(0.6,0.45,0.1)}NS and Cur@F_{(0.6,0.45,0.1)}M₂NS were chosen to release in a solution of 0.5% SLS (sodium lauryl sulfate) in water through a dialysis process.²⁵ Briefly, 5 mg of each of the samples were weighed precisely and dispersed with 2 mL of 0.5% SLS solution in a dialysis bag (10 kDa cut off, Sigma-Aldrich). The dialysis bag was then put into 100 mL 0.5% SLS solution in a 200 mL dark bottle at 310 K. The bottles were then shaken in a shaker while the temperature was kept at 310 K. At a given measurement time, 1 mL of drug release solution was withdrawn and 1 mL of the original 0.5% SLS solution was immediately added, in order to keep the volume constant. The release solutions were then diluted with 0.5% SLS solution and investigated by UV-vis spectroscopy at the 425 nm wavelength.

pH responsive release. 1 mg of Cur@F_0.6NS or Cur@F_0.6M₂NS was added into a 10 mL centrifuge tube and mixed with 4 mL of either PBS (0.01 M, pH 7.4) or PBS (0.01 M, pH 5.0). These release assays were performed at 37° in a water bath shaker. At each measurement point, three of the release assays were drawn out, 4 mL of 0.5% SLS solution was added to each assay to extract the released curcumin, and then the assays were centrifuged immediately. The upper solutions were then investigated using a UV-vis spectrophotometer (UV-2550, SHIMADZU) at 425 nm.

“On–off” mode alternating magnetic field (AMF) release. In order to measure the magnetism responsive properties of Cur@FMNS, different magnetic contents of Cur@FMNS, namely Cur@F_0.6M₂NS, Cur@F_0.6M_1.5NS, and Cur@F_0.6M₁NS, were chosen to study the release in an “on–off” mode alternating magnetic field (“on–off” mode AMF) of 1500 mT at the alternation frequency of 120 rpm using a dialysis process. Briefly, 5 mg of each Cur@FMNS was weighed precisely and dispersed with 2 mL of 0.5% SLS solution in a dialysis bag (10 kD cut off, Sigma-Aldrich). The dialysis bag was then put into 100 mL 0.5% SLS solution in a 200 mL dark bottle at 310 K. The “on” or “off” mode in the AMF was applied for 10 min in each sequential process; when the process ended, the bottles were shaken slowly for 30 s to settle the solution. Then, 1 mL of the solution was withdrawn and 1 mL of the original 0.5% SLS solution was immediately added, in order to keep the volume constant. The solution was then diluted with 0.5% SLS solution and investigated by UV-vis spectroscopy at the 425 nm wavelength.

In vitro cell testing

Drug delivery materials. In order to evaluate the biocompatibility of the different materials, NS, M₂NS, F_0.6NS and F_0.6M₂NS were chosen. The tests included cell viability, hemolysis experiments and cell apoptosis. Human hepatoma SK-HEP1 cells and human embryonic kidney (HEK) 293T cells (both supplied by American type culture collection, ATCC) were incubated to evaluate the cytotoxicity of these drug delivery materials through Cell Counting Kit-8 (CCK-8), and the cell apoptotic phenomena were revealed through staining with Hoechst 33342 as observed using an upright fluorescence microscope (Olympus, BX51, Tokyo, Japan). Apoptosis was detected using an annexin V-PE apoptosis detection kit (Beyotime.com) and analyzed by flow cytometry. SK-HEP1 cell lines were seeded in 6-well plates (1.0 × 10⁵ cells per well), and 24 h after plating they were incubated with F_0.6NS or F_0.6M₂NS nanoparticles for 48 h, or were left untreated as a control. Next, the cells were washed twice with cold PBS; then they were carefully trypsinized to avoid mechanical damage of the membrane and re-suspended in annexin binding buffer in the presence of annexin V-PE stain. The sample acquisition was performed using a flow cytometer (BDFACSAria, USA). The biocompatibility of the drug delivery materials with blood cells was analyzed by a hemolysis experiment with venous blood taken from SD rats. The hemolyzed samples were centrifuged and photographed, and the supernatants were analyzed at 570 nm in a microplate reader (Bio Tek Instruments, Inc.).

Different nano-formulations of curcumim drugs. The cytotoxicity of different curcumin nano-formulations was assessed using SK-HEP1 cells and HepG2 cells (both supplied by ATCC), which were analyzed in the presence of different curcumin nano-formulations, such as free curcumin, Cur@F_0.6NS and Cur@F_0.6M₂NS. The cell apoptotic phenomena were revealed through staining with Hoechst 33342 as observed using an upright fluorescence microscope (Olympus, BX51, Tokyo, Japan). Apoptosis was detected using an annexin V-PE apoptosis detection kit (Beyotime.com) and analyzed by flow cytometry as described above, for the different nano-formulations of curcumin. The endocytosis and intercellular release properties of curcumin in the different drug nano-formulations were analyzed via confocal fluorescence imaging, while the magnetism responsive properties of Cur@F_0.6M₂NS were investigated by applying an alternating magnetic field (1500 mT at the alternation frequency of 120 rpm) for different time intervals.

Results and discussion

Design of drug delivery nanosystems

Most drug molecules with –OH or –COOH groups, such as curcumin monomers, naturally carry negative charges, from which we realized that endowing the surface with negative groups, such as –SH, could stop the drug from concentrating on the surface of the delivery vehicles. Many drug molecules have benzene ring structures, and we modified the large mesoporous channels with β-CD, which could provide more physical binding sites for the drug molecules, increase drug loading inside the pores and improve the sustained drug release properties. Last but not least, since the curcumin molecular structure is unstable to many chemical or physical conditions, and –NH₂ groups might affect the molecular structure of curcumin, we provided specific binding sites for curcumin molecules in order to form stable inclusion complexes. The air inside the mesoporous channels can be removed using the vacuum treatment process, curcumin solution can be driven inside, and the curcumin molecules can be nested in the β-CD molecules. As curcumin shows precipitation properties in water, the water in the curcumin solution could be the driving force for curcumin to precipitate inside the mesopores of the silica during the 24 hour drug loading process. As shown in Fig. S2 (ESI†), the drug delivery properties were improved on increasing the binding sites inside FNS or FMNS. The EE% values were 76.51 ± 6.53% for Cur@F_0.60NS and 74.16 ± 6.72% for Cur@F_0.60M₂NS, compared with 53.82 ± 5.11% for Cur@F_0.30NS and 52.96 ± 5.23% for Cur@F_0.30M₂NS. The control samples were the template-free samples NS and M₂NS, with EE% of about 40% and LC% of about 5.5%. As the degree of functionalization with Si–β-CD declined, the drug loading inside FNS and FM₂NS depended simply on the recrystallization of the drug, as compared with NS and M₂NS. The drug loading process could be divided into two stages: the first stage was the incorporation of soluble curcumin molecules within the β-CD, and the second stage could be the precipitation of curcumin molecules, which “grow” on top of the incorporated curcumin molecules during the overloading process. With more binding sites to combine with the first layer of curcumin, the F_0.6M₂NS and F_0.6NS show better drug loading performance.

XRD

XRD patterns for the drug delivery materials are shown in Fig. 3 and Fig. S3(a) (ESI†). As shown in Fig. 3, the inset low-angle XRD plot reveals an almost non-ordered mesoporous structure for NS obtained through the soft-templating process, and after the magnetic core is encapsulated, the non-ordered mesoporous structure is turned into a low-order mesoporous structure, which might be attributed to steric effects between the core structures and the CTA⁺ micelles during the fabrication process. As shown in Fig. 3, both M₂NS and F_0.6M₂NS show a main peak at 2θ = 35.6°, corresponding to the (311) reflection of the Fe₃O₄ crystal phase, showing that the magnetic core still remains after the bifunctionalization process. Both F_0.6NS and F_0.6M₂NS did not show obvious peaks corresponding to the β-CD crystal phase, as the Si–β-CD molecules were functionalized on the two kinds of mesoporous silica in the form of mono-molecules. The drug loading properties of the drug delivery nanosystems are shown in Fig. S3(a) (ESI†); through the drug loading process and the driving force of water, curcumin was recrystallized on both F_0.6NS and F_0.6M₂NS, and the prominent peak at 2θ = 17.1°, and the peaks at 2θ = 25.9° and 2θ = 14.4° are still retained in the XRD patterns of Cur@F_0.6NS and Cur@F_0.6M₂NS. Curcumin molecules may be firstly incorporated with Si–β-CD, which may have caused the recrystallized drugs to grow onto the “first layer” of drug molecule complexes during the drug loading process. The complex of curcumin and β-CD molecules would retain the crystalline nature of curcumin.²⁶ However, a lower degree of crystallization was exhibited by Cur@F_0.6M₂NS compared with Cur@F_0.6NS, from which we concluded that the low-ordered mesoporous structure could increase drug dispersion and decrease the amount of the crystalline phase in the nano-formulations.

Fig. 3 XRD patterns of different samples; the inset image shows low-angle XRD patterns of M₂NS and NS.

FTIR analysis

The bifunctionalization process could be confirmed by the change in molecular structure, as shown by the FTIR spectra in Fig. 4. As shown in Fig. 4, all the mesoporous silica samples showed vibrations of Si–O–Si bonds at similar wavenumbers, i.e., 1095 cm⁻¹, 811 cm⁻¹ and 466 cm⁻¹. The template-free S-NS and S-MNS showed vibrations of C–H bonds at 2925 cm⁻¹ and 2856 cm⁻¹ respectively, unlike template-free NS and M₂NS, revealing that KH590 had been successfully grafted onto the outer surface of both mesoporous silica samples. The vibrations of N–H bonds at 1556 cm⁻¹ and O=C–N bonds at 1706 cm⁻¹ showed that Si–β-CD had been completely functionalized on both mesoporous silica samples. What's more, the vibration of O–H bonds in β-CD molecules at ∼1400 cm⁻¹ showed that the integrity of the β-CD molecular residues is still maintained during the formation of Si–β-CD molecules and functionalization on mesoporous silica. Both M₂NS and F_0.6M₂NS still contain Fe₃O₄ cores inside, since the vibration of Fe(III)–O bonds at 565 cm⁻¹ is still present. Drug loading properties and details of the atoms can be observed in Fig. S3(b) (ESI†). The vibration of O=C–N bonds at 1706 cm⁻¹ showed that grafted Si–β-CD is retained on both FMNS and FNS during the drug loading process. The vibrations of the C=O, C=C and C=C–H bonds in curcumin molecules at 1627 cm⁻¹, 1510 cm⁻¹ and 1429 cm⁻¹, respectively, showed that the drug had been successfully loaded into F_0.6NS and F_0.6M₂NS, from which we concluded that the curcumin delivery nano-formulations, such as Cur@F_0.6NS and Cur@F_0.6M₂NS, were fabricated successfully.

Fig. 4 FTIR patterns of different samples.

XPS analysis

The surface molecular structures of F_0.6NS and F_0.6M₂NS were confirmed by photoelectron spectroscopy, as shown in Fig. 5. The high resolution S2p spectra exhibited in Fig. 5(a) and Fig. S4(a) (ESI†) revealed the binding energies of the S2p_3/2 peak at 164.7 eV and the S2p_1/2 peak at 163.5 eV for F_0.6NS, and the binding energies of the S2p_3/2 peak at 164.5 eV and the S2p_1/2 peak at 163.4 eV for F_0.6M₂NS, which verified the presence of thiol groups and C–S bonds on the surface of both F_0.6NS and F_0.6M₂NS. The high resolution N1s spectra shown in Fig. 5(b) and Fig. S4(b) (ESI†) showed the major amide binding energy of NHC=O bonds at 400.0 eV for both F_0.6NS and F_0.6M₂NS, which was attributed to surface functionalized Si–β-CD molecules. According to the results, some of the Si–β-CD molecules were inevitably functionalized onto the surface, due to the open mesoporous structure and the unsaturated –OH sites on the surface. It is notable that no significant signals from the Fe 2p spectrum can be observed in Fig. S4(c) (ESI†) for the surface of F_0.6M₂NS, and so we believe that Fe₃O₄ particles were encapsulated inside SiO₂ and acted as magnetic cores.

Fig. 5 The XPS S2p spectrum of F_0.6NS (a), and the N1s spectrum of F_0.6NS (b).

Solid state ¹³C and ²⁹Si NMR analysis

The solid state ¹³C and ²⁹Si NMR spectra concretely showed the out-inside functionalization process, and F_0.6NS was taken as an example. The ¹³C NMR spectra of β-CD, S-NS and F_0.6NS are shown in Fig. 6. In the spectrum of S-NS, the peaks at 12, 27 and 37 ppm were attributed to the C3, C2 and C1 carbon atoms in the surface functionalized HS–(CH₂)₃– groups, respectively. After further functionalizing S-NS with Si–β-CD, the ¹³C NMR spectrum of F_0.6NS revealed the chemical shifts of the β-CD rings, according to the spectrum of β-CD. Meanwhile, the peaks at 157, 43, 32 and 11 ppm were attributed to the C10, C11, C12 and C13 carbon atoms of the grafted Si–β-CD molecules respectively, which is evidence that β-CD rings were successfully functionalized on the F_0.6NS. What's more, the peaks at 37 and 24 ppm were assigned to the C3 and C2 carbon atoms respectively, by which we confirmed the presence of HS–(CH₂)₃– groups in F_0.6NS. In addition, the ²⁹Si NMR spectra of S-NS and F_0.6NS revealed the functionalization, as shown in Fig. S5 (ESI†). The peaks observed at −69 ppm for F_0.6NS and S-NS were attributed to the presence of T³ structures, corresponding to the modified silane coupling agent. However, the enhanced intensity of the peak at −57 ppm for F_0.6NS compared with S-NS was due to the presence of more T² structures, which reflected the fact that most of the Si–β-CD molecules were functionalized onto S-NS with the breaking of two Si–O–Me bonds, due to the steric hindrance of the β-CD rings. Both F_0.6NS and S-NS showed a Q⁴ NMR shift at −111 ppm, corresponding to Si–(OSi)₄. The sharper peak in the spectrum of F_0.6NS was associated with increasing amounts of the Q⁴ structure and decreasing amounts of the Q³ structure of Si(OSi)₃(OH) on further functionalization with Si–β-CD molecules, by which we confirmed the success of the out-inside stepwise functionalization.

Fig. 6 ¹³C NMR spectra for β-CD, S-NS and F_0.6NS.

Nitrogen adsorption–desorption isotherms

In order to confirm the alteration of the microstructure during the bifunctionalization and drug loading processes, we choose the M₂NS series of samples, the mesoporous structures of which were confirmed by nitrogen adsorption–desorption isotherms. All of the samples were described by type III isotherms for the weak adsorption on the large mesopores, and the hysteresis loops observed at high relative pressures might be attributed to slot–hole stacking by the star shaped mesoporous particles, as shown in Fig. S6 (ESI†). The BET surface areas and aperture parameters shown in Table S1 (ESI†) indicated that outer surface grafted thiol groups hardly influenced the mesoporous structure. When the inner channel was functionalized with Si–β-CD, the BET surface area and the pore volume rapidly decreased, due to the deeper inside modification. After the loading of drugs, the inside mesopores were further filled with drug molecules through the pore bifunctionalization process, the mesoporous structure of the magnetic nanostars still remained, and the drug molecules could be controllably recrystallized and loaded inside the mesopores of the nanocarriers as nano-formulations. What's more, the DTA and TG curves in Fig. S7 (ESI†) and the analysis in the ESI† also confirm the out-inside bifunctionalization process used to fabricate both drug delivery materials, F_0.6NS and F_0.6M₂NS, and to improve the drug loading properties of both drug delivery nanosystems, i.e., Cur@F_0.6NS and Cur@F_0.6M₂NS, as nano-formulations for curcumin.

TEM

Through the soft-templating conditions, we obtained mono-dispersed star shaped mesoporous silica, as shown in Fig. 7(a) and (b). Through the electrostatic and soft-templating processes, we fabricated nanostar shaped mesoporous silica, as shown in Fig. 7(c), and yolk–shell structures with Fe₃O₄ cores, as shown in Fig. 7(d) and (e). The Fe₃O₄ nanoparticles in the water-based ferrofluid had sizes of less than 20 nm, as shown in the inset in Fig. 7(e), and were suitable for encapsulation inside mesoporous silica to form magnetic cores. The particle sizes of NS and M₂NS were homogeneously distributed and concentrated in the interval of 90–150 nm. Fig. 8 reveals visually the surface functionalization of NS and M₂NS. After the bifunctionalization process, the surface and channels were covered by organic films, and the NS and M₂NS particles were wrapped inside shells of organic molecules. These results indicated that NS and M₂NS had been successfully functionalized with organic molecules to form F_0.6NS and F_0.6M₂NS. The drug loading results are shown in Fig. 8, from which we concluded that the drugs had recrystallized inside the meso-channels of F_0.6NS and F_0.6M₂NS. However, due to the overloading process, some drug crystals were inevitably grown on the surface, as shown in Fig. 8(e)–(h). The drug delivery nanosystems, i.e., Cur@F_0.6NS and Cur@F_0.6M₂NS, were fabricated as nano-formulations to act as delivery vehicles for curcumin. To further investigate the stability of these drug delivery nanosystems in different physical media, dynamic light scattering (DLS) experiments were carried out for both Cur@F_0.6NS and Cur@F_0.6M₂NS, and the results are shown in Fig. S8 and Table S2 (ESI†), in which the hydrodynamic size was revealed by the Z-average diameter and the stability was revealed by the ζ-potential. In PBS 7.4 medium, the hydrodynamic sizes were 249.09 nm and 299.07 nm for Cur@F_0.6NS and Cur@F_0.6M₂NS respectively, from which we believe that aggregates of 2–3 nanoparticles were most commonly formed in both dispersion systems, according to the particle sizes observed in Fig. 8(e–h). When scattered in cell supernatant, with electrostatic forces due to the protein molecules, the hydrodynamic sizes of both drug nanocarriers increased while the polydispersity index (PDI) values decreased, as a result of the enhanced scattering properties of the protein molecules, which might increase the bioavailability under physical conditions. The absolute ζ-potential values decreased in cell supernatant compared with pure PBS 7.4 medium, due to the increased degree of hydration and increased hydrodynamic size caused by the protein molecules present under the simulated physical conditions.

Fig. 7 TEM images of (a) and (b) NS; (c), (d) and (e) M₂NS; the inset image in (e) shows the Fe₃O₄ nanoparticles in the water-based ferrofluid.

Fig. 8 TEM images of (a) and (b) F_0.6NS, and (c) and (d) F_0.6M₂NS; the blue arrows indicate the surface functionalized molecules; (e) and (f) Cur@F_0.6NS, and (g) and (h) Cur@F_0.6M₂NS; the orange arrows indicate the recrystallized drugs.

In vitro cell testing

The drug delivery materials, such as NS, M₂NS, F_0.6NS and F_0.6M₂NS, were tested for cell apoptosis with the Hoechst 33342 staining experiment, and hemolysis to evaluate the cytotoxicity and biocompatibility before and after bifunctionalization. From the hemolysis rates shown in Fig. 9, we can see that the NS and M₂NS exhibit obvious hemolysis properties at high concentrations of 1000–2000 μg mL⁻¹; the hemolysis rates were measured as over 10%. The hemolysis rate was lowered through the bifunctionalization process, since the hemolysis rates of both F_0.6NS and F_0.6M₂NS were less than 10% when the concentrations were 1000–2000 μg mL⁻¹, from which we confirmed that the bifunctionalization could increase the biocompatibility of the drug nanocarriers with blood cells. Some of the Si–β-CD molecules were inevitably functionalized on the silanol surface, reducing and modifying the silanol groups with β-CD rings and saccharide hydroxyl groups, which could reduce the toxicity to the blood cells.²⁷ The cell apoptosis results are shown in Fig. S9 and S10 (ESI†), and there was no obvious cytotoxicity of the drug delivery materials towards normal cells or cancer cells.

Fig. 9 Hemolysis assays for NS, M₂NS, F_0.6NS and F_0.6M₂NS at concentrations from 31.2 μg mL⁻¹ to 2000 μg mL⁻¹.

The cell viability testing results for 293T cells are shown in Fig. S11 (ESI†), and those for SK-HEP1 cells are shown in Fig. 10(a), in order to evaluate the cytotoxicity of the drug delivery nanocarriers. The cytotoxicity of both the NS and M₂NS samples increased when the concentrations increased to above 500 μg mL⁻¹, which might be attributed to their star-shaped structures and cluster surfaces. However, after the bifunctionalization process, both the F_0.6NS and F_0.6M₂NS samples retained high cell viability for both 293T cells and SK-HEP1 cells, and even at high concentrations of 2000 μg mL⁻¹, the cell viability was measured as over 80%, revealing improved biocompatibility and lower toxicity to cells. Additionally, the SK-HEP1 cell viability of F_0.6NS and F_0.6M₂NS was measured as over 95% through flow cytometry at concentrations of 125 and 500 μg mL⁻¹, as shown in Fig. S12 (ESI†). Meanwhile, the cytotoxicities of different drug nano-formulations of curcumin were evaluated using cancer cells, such as SK-HEP1 cells as shown in Fig. 10(b), and HepG2 cells as shown in Fig. S13 (ESI†). However, pure curcumin did not show any significant toxicity to HepG2 cells; the cell viability was measured as 100% at the given concentrations, and the IC₅₀ value was 121.3 μg mL⁻¹ for SK-HEP1 cells, which reveals the poor bioavailability and low uptake properties of free curcumin. The IC₅₀ values of Cur@F_0.6NS and Cur@F_0.6M₂NS were 13.8 μg mL⁻¹ and 8.57 μg mL⁻¹ respectively for SK-HEP1 cells, and 10.9 μg mL⁻¹ and 7.21 μg mL⁻¹ respectively for HepG2 cells. The toxicity was increased 9-fold for Cur@F_0.6NS nano-formulations, and 14-fold for Cur@F_0.6M₂NS nano-formulations, verifying the improved bioavailability of the nano-formulations, due to the novel bifunctionalized magnetic star shaped mesoporous silica. The cell apoptosis results in Fig. 11 reveal the cell apoptosis during incubation with different curcumin nano-formulations, such as free curcumin, Cur@F_0.6NS and Cur@F_0.6M₂NS; the cell apoptosis responses to Cur@F_0.6NS and Cur@F_0.6M₂NS were obviously increased compared with free curcumin, even at low concentrations of 7.8 μg mL⁻¹, showing visually the increased toxicity of the curcumin nano-formulations to SK-HEP1 cells. Last but not least, the apoptosis rates were significantly increased for both designed nanocarriers compared with free curcumin, as shown in Fig. 12. The early apoptosis rates were 10.2%, 30.5% and 34.4% for free curcumin, Cur@F_0.6NS and Cur@F_0.6M₂NS respectively, and the late apoptosis rates were 4.2%, 30.3% and 32.0% respectively, showing the enhanced ability of the designed nanocarriers to kill SK-HEP1 cells.

Fig. 10 In vitro SK-HEP1 cell viability after incubation for 48 h with NS, M₂NS, F_0.6NS and F_0.6M₂NS (a), and with drug nano-formulations of pure curcumin, Cur@F_0.6NS and Cur@F_0.6M₂NS, with curcumin concentrations ranging from 7.8 to 125 μg mL⁻¹ (b).

Fig. 11 Immunofluorescence microscopy analysis of apoptosis in SK-HEP1 cells induced by exposure for 48 h to free curcumin (a), Cur@F_0.6NS (b), and Cur@F_0.6M₂NS (c) at different curcumin concentrations (ranging from 7.8 to 31.25 μg mL⁻¹). Red arrows indicate apoptotic cells. These drug delivery nanosystems had significant effects on the nuclei (blue) stained with Hoechst 33342 compared with free curcumin. Scale bar: 100 μm.

Fig. 12 Apoptosis induction in the SK-HEP1 cell line for 48 h by different drug nano-formulations, namely curcumin, Cur@F_0.6NS and Cur@F_0.6M₂NS, with curcumin concentration of 31.25 μg mL⁻¹.

In vitro drug release. Drug release profiles of different drug nano-formulations are revealed in Fig. S14(a) (ESI†). From the release curves, we found that the nano-formulations of curcumin, i.e., Cur@F_0.1NS, Cur@F_0.1M₂NS, Cur@F_0.45NS, Cur@F_0.45M₂NS, Cur@F_0.6NS and Cur@F_0.6M₂NS, could improve the suspension and delivery properties compared with pure curcumin, as shown by the obvious improvement in the release rate in 0.5% SLS water solution. According to the pharmacokinetics results in Fig. S14(b–d) and Table S3 (ESI†), the release of pure curcumin in 0.5% SLS was fitted by a first-order model, in which the release of curcumin depended on the dissolved concentration of curcumin. The drug releases of Cur@FNS and Cur@FM₂NS were similar, as the drug was loaded into similar structures; their release mechanisms were fitted by the Higuchi model, and the drug release processes were described by Fickian diffusion attributed to the drug loading properties inside bifunctionalized nanostar shaped mesoporous silica. However, increasing the amount of Si–β-CD might increase the drug suspension and the loading of the nano-formulations inside the mesoporous channels, decrease drug recrystallization and cause more drug molecules to combine with the Si–β-CD, further increasing the drug release rate and delivery compared with Cur@F_0.1NS, Cur@F_0.45NS and Cur@F_0.6NS, and Cur@F_0.1M₂NS, Cur@F_0.45M₂NS and Cur@F_0.6M₂NS. Moreover, when we adjusted the pH value from 7.4 to 5.0, as shown in Fig. S15 (ESI†), the release rates were increased for both drug nanocarriers. The cumulative rates were 16.45 ± 2.55% for Cur@F_0.6M₂NS and 13.23 ± 1.83% for Cur@F_0.6NS after 4 h in PBS 5.0 medium. The cumulative rates increased to 30.32 ± 2.85% and 25.02 ± 2.43% for Cur@F_0.6M₂NS and Cur@F_0.6NS respectively in PBS 5.0 medium after 12 h, while the cumulative rates were 17.54 ± 3.86% and 15.65 ± 3.22% respectively in PBS 7.4 medium after 12 h. In the lower pH medium, the hydrolysis and decomposition effects of the functionalized β-CD rings were enhanced, leading to significant destruction of the mesoporous organic framework in these nanocarriers and the dissolution of curcumin.

The cellular uptake was characterized by confocal fluorescence microscopy, as shown in Fig. S16 (ESI†). However, the free curcumin did not show any cellular uptake effect after 1 h and 4 h, revealing the low bioavailability and uptake properties of free curcumin. The nano-sized curcumin formulations, Cur@F_0.6NS and Cur@F_0.6M₂NS showed increased cellular uptake of curcumin, from which we concluded that the novel delivery nanosystems could increase the bioavailability and the toxicity of the drug towards cancer cells. When we fabricated bifunctionalized nanostars with Fe₃O₄ magnetic cores, the release of curcumin in 0.5% SLS could be adjusted by an AMF, and the payloads of Fe₃O₄ nanoparticles inside the bifunctionalized nanostars are shown in Fig. 13(a) and (b). As we increased the payload of Fe₃O₄ nanoparticles, the drug carriers became increasingly responsive to the “on–off” AMF, as the average release rate during the “on” stage increased from Cur@F_0.6M₁NS to Cur@F_0.6M₂NS. The inset in Fig. 13(b) also reveals the rapid magnetic collection of F_0.6M₂NS in water. It is worth noting that these samples retained their sensitivity to the AMF, even after 100 min, the 10th “on–off” alternation cycle. The ability to control these drug delivery nanosystems with an AMF could increase the suspension and release of curcumin in the 0.5% SLS solution, and the control of the release by the alternating magnetic field was attributed to intense local heat due to magnetic–thermal conversion,²⁸ by which the bioavailability and suspension of curcumin might be improved under magnetically controlled conditions. Furthermore, for the “on” stage of the alternating magnetic field, the fluorescence signal was obviously increased compared with the “off” stage during incubation with SK-HEP1 cells, as shown in Fig. 13(c). Significant cellular uptake properties were observed during the “on” stage after 45 min. As an effect of alternating magnetic fields, the signal increase depended on the stimulation time. After 45 min, the intercellular release was significantly different compared with incubation for 30 and 15 min, showing that the cellular uptake and bioavailability had been improved. These results indicate that the magnetism responsive nanocarriers exhibited a smart and sensitive response to alternating magnetic fields, not only in the in vitro release medium, but also between the cells. With magnetic cores, which are heat mediators suitable for localized or targeted hyperthermia therapy,²⁹ the bifunctionalized nanostars could be applied in MRI and diagnosis as well.

Fig. 13 Drug release of bifunctionalized magnetic nanostars in “on–off” AMF mode: release curves of different samples (a), average release rates of different samples (b), and the intercellular release properties of Cur@F_0.6M₂NS during the “on” and “off” stages for different time intervals revealed by the merged fluorescence images (c). The inset in (b) shows the magnetic collection of F_0.6M₂NS in water within 10 s.

Conclusions

In this work, we have explored the fabrication of two kinds of bifunctionalized nanostars for the delivery of nano-formulations of the curcumin drug, in which the drug delivery nanocarriers could effectively increase the biocompatibility and decrease the cytotoxicity to cells. Meanwhile, these drug nano-formulations could significantly increase the cytotoxicity to cancer cells, such as SK-HEP1 cells and HepG2 cells, compared with free curcumin, in order to kill them more effectively. The cellular uptake properties and bioavailability of these curcumin nano-formulations were increased in SK-HEP1 cells. With the use of magnetic cores, the drug delivery nanostars showed a sensitive response to magnetic fields, and exhibited excellent controllability of drug release both in vitro and intercellularly. Bifunctionalization combined with magnetic fabrication has immense potential in the drug delivery field and in biomedicine.

Acknowledgements

The work was supported by the National Natural Science Foundation (P. R. China, No. 31101854, 61427807 and 21207041), and Fujian Province Introduction of Major Research and Development Institution Funding Project (2012I2014).

Notes and references

1. K. Zhang, L. L. Xu, J. G. Jiang, N. Calin, K. F. Lam, S. J. Zhang, H. H. Wu, G. D. Wu, B. Albela, L. Bonneviot and P. Wu, J. Am. Chem. Soc., 2013, 135, 2427–2430.
2. (a) M. Wu, L. Y. Kong, K. W. Wang, R. H. Jin, T. Y. Cheng and G. H. Liu, Catal. Sci. Technol., 2015, 5, 1750–1757; (b) X. Du and S. Z. Qiao, Small, 2015, 11, 392–413.
3. (a) Y. F. Zhu and C. L. Tao, RSC Adv., 2015, 5, 22365–22372; (b) L. Xiong, X. Du, B. Y. Shi, J. X. Bi, F. Kleitz and S. Z. Qiao, J. Mater. Chem. B, 2015, 3, 1712–1721.
4. J. P. Yang, W. Y. Chen, D. K. Shen, Y. We, X. Q. Ran, W. Teng, J. W. Fan, W. X. Zhang and D. Y. Zhao, J. Mater. Chem. A, 2014, 2, 11045–11048.
5. C. L. Tao, Y. F. Zhu, Y. Xu, M. Zhu, H. Morita and N. Hanagata, Dalton Trans., 2014, 43, 5142–5150.
6. F. Liu, J. N. Wang, P. L. Huang, Q. Zhang, J. T. Deng, Q. Y. Cao, J. T. Jia, J. H. Cheng, Y. P. Fang, D. Y. B. Deng and W. Y. Zhou, J. Mater. Chem. B, 2015, 3, 2206–2214.
7. P. L. Huang, J. N. Wang, S. T. Lai, F. Liu, N. Ni, Q. Y. Cao, W. Liu, D. Y. B. Deng and W. Y. Zhou, J. Mater. Chem. B, 2014, 2, 8616–8625.
8. G. Astray, C. Gonzalez-Barreiro, J. C. Mejuto, R. Rial-Otero and J. Simal-Gándara, Food Hydrocolloids, 2009, 23, 1631–1640.
9. (a) K. J. Waleczek, H. M. Marques, B. Hempe and P. C. Schmidt, Eur. J. Pharm. Biopharm., 2003, 55, 247–251; (b) V. A. Marcolino, G. M. Zanin, L. R. Durrant, M. D. Benassi and G. Matioli, J. Agric. Food Chem., 2011, 59, 3348–3357; (c) C. Jantarat, P. Sirathanarun, S. Ratanapongsai, P. Watcharakan, S. Sunyapong and A. Wadu, Trop. J. Pharm. Res., 2014, 13, 1215–1223.
10. (a) N. M. Khashab, A. Trabolsi, Y. A. Lau, M. W. Ambrogio, D. C. Friedman, H. A. Khatib, J. I. Zink and J. F. Stoddart, Eur. J. Org. Chem., 2009, 1669–1673; (b) D. P. Ferris, Y. L. Zhao, N. M. Khashab, H. A. Khatib, J. F. Stoddart and J. I. Zink, J. Am. Chem. Soc., 2009, 131, 1686–1688; (c) S. Angelos, Y. W. Yang, N. M. Khashab, J. F. Stoddart and J. I. Zink, J. Am. Chem. Soc., 2009, 131, 11344–11346; (d) J. Croissant, A. Chaix, O. Mongin, M. Wang, S. Clément, L. Raehm, J. O. Durand, V. Hugues, M. Blanchard-Desce, M. Maynadier, A. Gallud, M. Gary-Bobo, M. Garcia, J. Lu, F. Tamanoi, D. P. Ferris, D. Tarn and J. I. Zink, Small, 2014, 10, 1752–1755.
11. (a) M. Shoji, K. Nakagawa, A. Watanabe, T. Tsuduki, T. Yamada, S. Kuwahara, F. Kimura and T. Miyazawa, Food Chem., 2014, 151, 126–132; (b) C. S. Mangolim, C. Moriwaki, A. C. Nogueira, F. Sato, M. L. Baesso, A. M. Neto and G. Matioli, Food Chem., 2014, 153, 361–370.
12. (a) M. Salem, S. Rohani and E. R. Gillies, RSC Adv., 2014, 4, 10815–10829; (b) H. Yan, C. Teh, S. Sreejith, L. L. Zhu, A. Kwok, W. Fang, X. Ma, K. T. Nguyen, V. Korzh and Y. L. Zhao, Angew. Chem., Int. Ed., 2012, 51, 8373–8377; (c) H. Hatchera, R. Planalpb, J. Chob, F. M. Tortia and S. V. Tortic, Cell. Mol. Life Sci., 2008, 65, 1631–1652.
13. Y. Hao, Y. J. Pan, N. Nitin and R. V. Tikekar, LWT–Food Sci. Technol., 2014, 58, 667–671.
14. S. P. Singh, M. Sharma and P. K. Gupta, Int. J. Biol. Macromol., 2015, 74, 162–170.
15. C. L. Tong, U. H. Stroeher, M. H. Brownb and C. L. Raston, RSC Adv., 2015, 5, 7953–7958.
16. A. Saithongdee, N. Praphairaksit and A. Imyim, Sens. Actuators, B, 2014, 202, 935–940.
17. (a) J. R. Lou, X. X. Zhang, J. Zheng and W. Q. Ding, Anticancer Res., 2010, 30, 3249–3255; (b) L. Baum and A. Ng, J. Alzheimer's Dis., 2004, 6, 367–377.
18. M. Massaro, S. Riela, P. L. Meo, R. Noto, G. Cavallaro, S. Miliotob and G. Lazzara, J. Mater. Chem. B, 2014, 2, 7732–7738.
19. F. Liu, H. Zhang, Q. Y. Cao, X. H. Xiang, L. Q. Wang, T. He, W. Liu, Y. P. Fang, D. Y. B. Deng and W. Y. Zhou, RSC Adv., 2014, 4, 8918–8921.
20. H. H. Tønnesen and J. Karlsen, Z. Lebensm.-Unters. Forsch., 1985, 180, 402–404.
21. A. F. Wang, F. Muhammad, W. X. Qi, N. Wang, L. Chen and G. S. Zhu, ACS Appl. Mater. Interfaces, 2014, 6, 14377–14383.
22. (a) D. Jin, K. W. Park, J. H. Lee, K. Song, J. G. Kim, M. L. Seo and J. H. Jung, J. Mater. Chem., 2011, 21, 3641–3645; (b) R. K. Gangwar, G. B. Tomar, V. A. Dhumale, S. Zinjarde, R. B. Sharma and S. Datar, J. Agric. Food Chem., 2013, 61, 9632–9637.
23. F. Liu, J. N. Wang, Q. Y. Cao, H. D. Deng, G. Shao, D. YB. Deng and W. Y. Zhou, Chem. Commun., 2015, 51, 2357–2360.
24. C. Abbehausen, A. L. B. Formiga, E. Sabadini and I. V. P. Yoshida, J. Braz. Chem. Soc., 2010, 21, 1867–1876.
25. (a) S. Jambhrunkar, S. Karmakar, A. Popat, M. H. Yu and C. Z. Yu, RSC Adv., 2014, 4, 709–712; (b) A. Popat, S. Karmakar, S. Jambhrunkar, C. Xu and C. Z. Yu, Colloids Surf., B, 2014, 117, 520–527.
26. H. K. Syed and K. K. Peh, Lat. Am. J. Pharm., 2013, 32, 52–59.
27. (a) Y. S. Lin and C. L. Haynes, J. Am. Chem. Soc., 2010, 132, 4834–4842; (b) I. I. Slowing, C. W. Wu, J. L. Vivero -Escoto and V. S. Y. Lin, Small, 2009, 5, 57–62.
28. H. Wang, J. H. Yi, S. Mukherjee, P. Banerjee and S. Q. Zhou, Nanoscale, 2014, 6, 13001–13011.
29. A. Hervault and N. T. K. Thanh, Nanoscale, 2014, 6, 11553–11573.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5tb02184g

‡ Both authors contributed equally to this work.

---