A theranostic nanocomposite system based on radial mesoporous silica hybridized with Fe3O4 nanoparticles for targeted magnetic field responsive chemotherapy of breast cancer

State Key Laboratory of New Ceramics a Science & Engineering, Tsinghua Unive lyzhao@mail.tsinghua.edu.cn; caiqiang@m Advanced Materials of Ministry of Educati and Engineering, Tsinghua University, Beijin School of Earth Sciences and Resources, C 10083, China College of Chemistry and Materials Scien 710069, China † Electronic supplementary informa 10.1039/c7ra12446e Cite this: RSC Adv., 2018, 8, 4321


Introduction
Since the rst anti-cancer drug was approved by the Food and Drug Administration (FDA) to treat hematological cancer, chemotherapy has been recognized as one of the key components of treatment for all stages of cancers, including early stage disease and patients with complete resection. 1-3 However, despite an expanding panel of chemotherapeutic agents such as paclitaxel, docetaxel, vincristine, pemetrexed, topotecan and irinotecan that show high effectiveness, the severe side effects of chemotherapy continues to be one of the signicant challenges during the treatment, due to the high dosage and indiscriminate delivery of chemotherapeutic agents. 4,5 Drug delivery systems based on nanotechnology are encouraging administration platforms for the efficient delivery of drugs to lesions and targeting sites, to avoid ineffective treatment from under-dosing or toxicity due to overdosing. [6][7][8][9] Furthermore, when these delivery systems exhibit controlled release behavior, the drug levels can be administered continuously within a desired therapeutic range for a long period of time.
Over the last decade, various drug delivery systems, such as micelles, liposomes, multifunctional dendritic polymers, nanospheres, nanocapsules, hydrogels and liquid crystals, have been successfully developed for efficient passive-or activetargeting delivery strategies. [10][11][12][13] Although each system possesses its own advantages, several signicant limitations oen affect their further application. For instance, micelles and liposomes may suffer from poor chemical stability and disintegration due to the high-density lipoproteins used. 14,15 Meanwhile, toxicity problems are the main obstacles hindering the applications of synthesized polymers, nanospheres and hydrogels, due to the problems of in vivo degradation, while dendrimers are limited by their quick elimination by the liver and kidneys and low blood stability. 16 Currently, mesoporous silica and oxidized porous silicon have attracted great attention in drug delivery research, as both materials exhibit high surface area, thermal stability, tunable pore size, chemical inertness, excellent biocompatibility, and biodegradability. [17][18][19] The large surface area and porous interior make these materials excellent reservoirs for delivering hydrophobic drugs, and their tunable pore size can be tailored to selectively store different nanoparticles and molecules of interest. In addition, the simple and adjustable synthesis process makes it easy to obtain optimized sizes and shapes for maximum cellular uptake. 20,21 Furthermore, the ease of modifying the surface of mesoporous silica makes it possible to design active-targeting and stimuliresponsive drug delivery systems.
To achieve precise chemotherapy of cancer, the drug delivery system used must be site-specic and time-release controlled. 22 Compared to passive-targeting offered by the enhanced permeability and retention effect (EPR effect), active-targeting strategies such as attaching an antibody, carrier protein or a ligand to the drug delivery system will allow it to accumulate at a target location, thus obtaining high efficacy with low dosage and low side effects and toxicity. 23,24 Meanwhile, the controlled release of drugs from delivery systems can be triggered by various internal or external stimuli such as heat, light, an electric eld, an alternative magnetic eld (AMF), pH, redox, enzyme activity, mechanical force and so on. 25,26 Among them, AMF-sensitive systems, which can be induced by integration with supermagnetic iron oxide nanoparticles, show great potential for triggering drug release due to their safety, high penetration depth and lack of radiotoxicity. 27,28 Furthermore, supermagnetic iron oxide nanoparticles loaded in a drug delivery system can be used as the T 2 -weighted MRI contrast agent to evaluate the targeting ability of the delivery system and to evaluate the therapeutic effect in real time. 29,30 Therefore, developing an effective delivery system with high drug loading efficiency, active-targeting and stimuli-responsive behavior, along with a concurrent diagnostic method with therapy (theranostic), is essential for efficient cancer chemotherapy.
Herein, in the current study, we report a novel theranostic drug delivery system based on radial mesoporous silica, which is hybridized with multiscale MNPs, for MRI-guided and AMFresponsive chemotherapy for breast cancer. Superparamagnetic iron oxide nanoparticles (IONPs) with multiscale sizes were prepared via a hydrothermal method (for larger IONPs, lIO NPs) and a thermal decomposition method (for ultra-small IONPs, uIO NPs), lIO NPs act as the mediator that is responsive to the AMF, and uIO NPs act as the T 2 contrast agent for MRI. Mesoporous silica spheres with radially oriented mesochannels were further in situ grown onto the surfaces of the lIO NPs (IOMSN NPs), and both the uIO NPs and doxorubicin, an anti-cancer drug, can be readily incorporated within the mesochannels (IOMSN@uIO(DOX) NPs). To endow the asprepared doxorubicin-loaded nano-hybrids with good targeting ability, the surfaces of the particles were modied with folic acid (FA) to obtain the nal IOMSN@uIO(DOX)-FA NPs. The results show that the as-prepared silica/magnetic nanohybrids possess uniform size and excellent drug loading, delivery efficiency and MRI contrast effect. Moreover, a series of detailed in vitro biocompatibility and H&E staining analysis of these nano-hybrids revealed excellent biocompatibility, even at a high concentration of 200 mg mL À1 . Furthermore, both in vitro cytotoxicity analysis and in vivo cytotoxicity anti-tumor effect evaluation demonstrated that IOMSN@uIO(DOX)-FA NPs are an excellent drug delivery system that possesses great potential as an MRI-guided stimuli-responsive drug delivery system theranostic platform for effective active targeting chemotherapy of cancer.

Synthesis and characterization of IOMSN@uIO-FA NPs
The synthetic process for the uniform radial IOMSN@uIO(DOX)-FA NPs through a simple so-template method is illustrated in Scheme 1. Firstly, lIO NPs were prepared via a one-pot hydrothermal method.
Both the scanning electron microscopy (SEM) image (Fig. 1a1) and the transmission electron microscopy (TEM) image ( Fig. 1a2) show that the as-prepared lIO NPs have an irregular spherical shape with a uniform diameter of $42 nm. Then, IO-NH 2 MNPs were used as the so-template for the synthesis of IOMSN NPs. As shown in Fig. 1b1, the NPs have a mean diameter of about 750 nm and consist of a radially oriented irregular mesostructure, which was further conrmed by the TEM image (Fig. 1b2). The continuous radial mesostructure throughout indicates that this kind of mesoporous IOMSN NPs would favor the loading or adsorption and release of guest nanoparticles or molecules, such as anti-cancer therapeutic agents. To verify the components of the IOMSN NPs, Xray powder diffraction (XRD) was conducted, and the spectrum (Fig. 2a) shows that the characteristic peaks of (220), (311), (400), (511) and (440) are compatible with the iron oxide crystal structure, while the peak at 2q ¼ 23 reveals the existence of an amorphous silica structure. Finally, IOMSN-FA NPs and IOMSN@uIO-FA NPs were synthesized via modication of the surfaces with FA and loading uIO NPs in the radial mesostructure of IOMSN NPs using a crosslinker. Aer APTES was conjugated on the surface of IOMSN NPs to obtain IOMSN-NH 2 NPs, EDC could easily react with the carboxylic acid groups of FA to form an active O-acylisourea intermediate, which was easily displaced by nucleophilic attack from the primary amino groups of the pre-prepared IOMSN-NH 2 NPs in the reaction mixture. Fig. 1c1 clearly shows that the radial mesostructure of IOMSN-FA NPs was blocked by FA, thus resulting in a rougher surface (Fig. 1c2) compared to the IOMSN NPs. Aer loading with uIO NPs in the radial mesostructures, the nal IOMSN@uIO-FA NPs were obtained. The HRTEM images clearly demonstrate the morphology of the uIO NPs (Fig. S1a †) and the radial-type amorphous silica structure of the IOMSN@uIO-FA NPs (Fig. S1b †), which was further conrmed by the corresponding element mapping ( Fig. 1d-i). Furthermore, aer being suspended in DI water homogenously and then placed beside a magnet for 1 min, all the IOMSN@uIO-FA NPs gathered close to the magnet (Fig. S2 †), conrming the successful integration of IOMSN NPs and uIO NPs. Fig. 2b shows the hysteresis loops of the IOMSN NPs, IOMSN-FA NPs and IOMSN@uIO NPs measured at room temperature. All the NPs showed supermagnetic behavior with saturation magnetization (M s ) values of 2.64, 1.68 and 3.28 emu g À1 , respectively. The supermagnetic property of the IOMSN NPs indicates that the lIO NPs had successfully acted as the sotemplate to prepare the NPs and had then been encapsulated in it. Meanwhile, the modication with FA weakened the magnetic property of the IOMSN-FA NPs, while the higher M s of the IOMSN@uIO-FA NPs compared to the IOMSN NPs and IOMSN-FA NPs directly demonstrates the successful loading of uIO NPs in the radial mesostructure of the IOMSN@uIO-FA NPs. To further evaluate the FA on the IOMSN@uIO-FA NPs, UV-vis absorbance spectroscopy and FTIR spectroscopy were performed for FA, IOMSN@uIO NPs and IOMSN@uIO-FA NPs. The IOMSN@uIO-FA NPs showed a characteristic absorbance peak at $360 nm, which corresponds to the absorption spectra of FA. In addition, the FTIR results indicate successful FA modication due to the existence of the C]O bond at 1679 cm À1 . The nitrogen sorption isotherm of the IOMSN NPs (Fig. 2e) shows a type IV curve according to the IUPAC nomenclature, with a hysteresis loop in the 0.4-1.0 p/p 0 range, demonstrating the narrow pore size distribution of the IOMSN NPs. In addition, the IOMSN@uIO-FA NPs show a type II curve with a larger p/p 0 range of 0.3-1.0 and a low surface area of 68.44 m 2 g À1 compared to 477.68 m 2 g À1 for the IOMSN@uIO-FA NPs, revealing the loading of the uIO NPs and encapsulation of FA.

In vitro cytotoxicity and cell uptake of IOMSN@uIO-FA NPs
As is well known, the biocompatibility and cytotoxicity of a drug delivery system are of great signicance for its application in the eld of nanomedicine. 31 Therefore, the cytotoxicity of the IOMSN@uIO-FA NPs against the L929 cell line was investigated using the CCK-8 assay. It was found that the L929 cells exhibited a higher cell viability of more than 90.4%, even when incubated with a high IOMSN@uIO-FA NP concentration of 200 mg mL À1 for 24 h and 48 h (Fig. 3). Although the cell viability at a concentration of 500 mg mL À1 decreased somewhat aer 48 h  of incubation, the cell viability was still as high as 82.5%, indicating that the IOMSN@uIO-FA NPs could be used as a drug delivery system. The excellent cell uptake offered by the active targeting strategies of drug delivery systems is an effective way to enhance the delivery efficiency and chemotherapy effect. 32 The cell uptake of coumarin-6-loaded IOMSN@uIO NPs and IOMSN@uIO-FA NPs were investigated using a confocal laser scanning microscope. Fig. 4 shows the increasing amount of NPs internalized in MCF-7 cells with increasing incubation time. Obviously, the efficiency of cell uptake for the IOMSN@uIO-FA NPs was much higher than that for the IOMSN@uIO NPs, which indicates that FA could enhance the cell uptake efficiency of IOMSN@uIO NPs via the folate receptor, which was further conrmed by the 3-dimensional CLSM images (Fig. S3 †). All the NPs were located in the matrix of MCF-7 cells crossing the cell membranes.
In vitro release proles of DOX A high drug loading efficiency and smart release properties are essential characteristics to measure when evaluating a drug delivery system. Doxorubicin (DOX) was chosen as a model anticancer drug to test the IOMSN@uIO-FA NPs. The results show that the loading efficiency of DOX in the IOMSN@uIO-FA NPs was 86.21 AE 3.34%, and the DOX loading amount was 0.46 mg in 1 mg of the IOMSN@uIO(DOX)-FA NPs. The release proles of DOX from the IOMSN@uIO(DOX)-FA NPs with or without exposure to an AMF (13.33 kA m À1 , 282 kHz) in PBS at 37 C are shown in Fig. 5b. It was clearly found that the release of DOX was time-dependent and sustained-release behavior was observed.
Within 300 h, the accumulative release amount of DOX reached 83.11% without exposure to an AMF. In comparison, aer being exposed to an AMF every day, the release speed of DOX was increased and a total of 91.04% of DOX was released from the IOMSN@uIO(DOX)-FA NPs, which was 1.1 times higher than that achieved without exposure. Superparamagnetic iron oxide nanoparticles can generate heat under an AMF due to the Néel relaxation and Brown relaxation. 33 The temperature response curve for the IOMSN@uIO(DOX)-FA NP suspension in DI water with a concentration of 100 mg mL À1 under an AMF (13.33 kA m À1 , 282 kHz) is shown in Fig. 5a. Aer exposure for 10 min, a temperature increase of 4.8 C was obtained. Therefore, we speculated that the heat induced by iron oxide in the IOMSN@uIO-FA NPs under the AMF accelerated the release of DOX by accelerating the thermal motion of DOX molecules.

In vitro cytotoxicity of IOMSN@uIO(DOX)-FA NPs
The in vitro AMF stimuli-responsive property of DOX release inspired us to use an external AMF to control the release of DOX for enhanced chemotherapy efficacy. Herein, the cytotoxic effects of the IOMSN@uIO(DOX) NPs and IOMSN@uIO(DOX)-FA NPs, with or without AMF exposure against MCF-7 cells were tested. The results (Fig. 6) show that all the NPs exhibited time-dependent cell proliferation inhibition compared to a control group. The IOMSN@uIO(DOX)-FA NPs revealed higher   cytotoxicity to MCF-7 cancer cells than the IOMSN@uIO(DOX) NPs, as modication with FA could increase the amount of NPs taken up by cells via FA receptor-mediated endocytosis. Thus, more DOX would be released inside the cells and cause more cell death. Importantly, for the IOMSN@uIO(DOX)-FA NPs group, which were exposed to an AMF (13.33 kA m À1 , 282 kHz) for 10 min, a more efficient anti-cancer effect was obtained with a low cell viability of 38.86% aer 48 h of incubation compared to 58.26% without exposure. Therefore, the IOMSN@uIO(DOX)-FA NPs with active targeting and stimuli-responsive properties show great potential for targeting cancer tissues and enhancing chemotherapy in vivo.

In vitro MRI measurements
To evaluate the T 2 -weighted MR imaging contrast effect of the IOMSN@uIO-FA NPs, a 3.0 T whole-body MR scanner was used to obtain T 2 -weighted MR images of the IOMSN@uIO-FA NPs suspended in DI water with different Fe concentrations of 0.026, 0.029, 0.032, 0.036 and 0.039 mM (Fig. 7). The Fe concentration of the IOMSN@uIO-FA NPs was determined using inductively coupled plasma optical emission spectrometry (ICP-OES). As shown in Fig. 7a, the IOMSN@uIO-FA NPs induced a dark signal enhancement in a concentration-dependent manner. With increasing Fe concentration, a stronger signal (dark signal) was obtained, which was further conrmed by the transverse relaxivity values (r 2 ) (Fig. 7b). The r 2 value of the IOMS@uIO-FA NPs was 308 mM À1 s À1 in terms of Fe, indicating that the IOMS@uIO-FA NPs show great potential as a contrast agent for T 2 -weighted MR imaging of tumors.

In vivo anti-cancer efficacy
The excellent in vitro anti-cancer effect of the IOMS@uIO(DOX)-FA NPs against MCF-7 cells and prominent MR imaging behavior inspired us to further test the MRI-guided chemotherapy in vivo. Therefore, MCF-7 tumor-bearing mice were divided into ve groups randomly (n ¼ 5): control group, DOX group, IOMSN@uIO(DOX) group, IOMSN@uIO(DOX)-FA group and IOMSN@uIO(DOX)-FA + AMF group. Here, the DOX dosage of the DOX group was equivalent with that of the NP groups. All agents were intravenously (i.v.) injected through the tail vein of the mice and the therapeutic efficacy was evaluated by measuring the relevant tumor volume (Fig. 8a) and body weight    (Fig. 8b) during the treatment. The mice in the DOX group showed rapid tumor growth, indicating that the dosage of administrated free DOX was insufficient to inhibit the tumor growth. Meanwhile, for the mice treated with the IOMS-N@uIO(DOX) NPs, the tumor was inhibited in the rst 10 days, but recurrence occurred aer that, indicating that the passive targeting delivery strategy via the ERP effect had a weak delivery efficiency. As is well known, DOX is an effective chemotherapeutic drug for MCF-7 tumors, but it usually requires multiple and high dosages to achieve a satisfying anticancer effect. 34 In comparison, the IOMSN@uIO(DOX)-FA NPs exhibited effective tumor inhibition with no growth of the tumor volume during the 21 days of treatment, which was further conrmed the prominent accumulation of IOMSN@uIO(DOX)-FA NPs at the tumor site induced by active targeting of FA. Because the external AMF could trigger the release rate of DOX from the delivery system, the IOMSN@uIO(DOX)-FA NPs plus AMF led to complete remission of the MCF-7 tumors, as shown in Fig. 8c. The body weight of mice is usually used to evaluate the toxicity of chemotherapeutic agents. As shown in Fig. 8b, no obvious weight loss was observed, implying that the as-synthesized NPs were biocompatible in vivo. Haematoxylin and eosin (H&E) staining of the heart, liver, spleen, lung, kidney (Fig. S4 †) and tumor (Fig. 8d) in all treatment groups was performed aer 21 days. Compared to the control group, no observable changes were found, which further conrmed that the IOMSN@uIO(DOX)-FA NPs with excellent biocompatibility show great potential for use as an MRI-guided stimuliresponsive drug delivery system for effective chemotherapy of cancer.

Conclusions
In summary, highly uniform radial IOMSN@uIO(DOX)-FA NPs exhibiting MRI contrast effects, active targeting properties and stimuli-responsive behaviors were successfully prepared through a simple so-template process for efficient chemotherapy of cancer. The prepared IOMSN@uIO(DOX)-FA NPs showed high drug loading efficiency and AMF stimuli-responsive release properties, due to the high surface area of their radial mesostructure and induced heating property. Meanwhile, in vitro MRI measurements revealed the excellent MRI contrast effect with a high relaxation value of 308 mM À1 s À1 . Finally, both the in vitro and in vivo cytotoxicity experiments demonstrated that IOMSN@uIO(DOX)-FA shows great potential as an MRI-guided stimuli-responsive drug delivery system for effective active targeting chemotherapy of cancer. Synthesis of larger amino-terminal iron oxide nanoparticles (lIO NPs) and ultra-small iron oxide nanoparticles (uIO NPs) lIO NPs were synthesized via a hydrothermal method, in which Fe 3+ was partially reduced before the Fe 3 O 4 MNPs were formed. Briey, 1.350 g FeCl 3 $H 2 O, 2.325 g NaAc and 1.4 g SDBS were added into a mixture of 10 mL EG and 30 mL DEG. Then the mixture was vigorously stirred for 5 h and sealed in a 50 mL Teon-lined stainless-steel autoclave (with a lling ratio about 70%) at 200 C for 12 h. Aer cooling down to room temperature, the product was washed with ethanol and DI water 3 times, then collected by magnetic separation and dried in a vacuum oven at 60 C for 24 h. Finally, APTES was used as the source of amino groups to modify the lIO NPs under acidic conditions, and lIO NPs were obtained by removing the free APTES. The uIO NPs were synthesized according to a previous method using thermal decomposition. 35 Preparation of radial Fe 3 O 4 @SiO 2 mesoporous nanoparticles (IOMSN NPs) Typically, 1.0 g of CTAB was added into a solution of NH 3 $H 2 O (2 mL) and DI water (180 mL). Aer CTAB had completely dissolved, 35 mL diethyl ether and 50 mL anhydrous ethanol were added. During stirring, 0.2 g lIO NPs, which were homogeneously dispersed in 10 mL anhydrous ethanol, were added into the mixture. Aer stirring for 10 min, 5 mL of TEOS was added dropwise to the reaction mixture under vigorous stirring at room temperature for another 2 h. The nal IOMSN NPs were separated by magnetic separation and washed with DI water and ethanol two times, and eventually dried at 70 C for 12 h. Finally, the as-prepared IOMSN NPs were calcined in an argon atmosphere at 550 C for 4 h to remove CTAB.

Experimental section
Preparation of FA-modied IOMSN NPs (IOMSN-FA NPs) and uIO-loaded IOMSN NPs (IOMSN@uIO-FA NPs) The modication of the surface of IOMSN NPs with FA was achieved through a crosslinker chemistry method. Briey, 0.2 g of IOMSN NPs were added into 40 mL of anhydrous ethanol containing 0.5 mL of APTES. Aer being stirred at room temperature for 24 h, the mixture was extensively washed with ethanol and dried at 80 C to obtain the IOMSN-NH 2 NPs. Then, 300 mg FA was activated using EDC (86 mg) and NHS (77 mg) dissolved in a DMF/DI water (27 mL, 3 : 1, v/v) solution with stirring for 24 h. Subsequently, the IOMSN-NH 2 NPs (250 mg) were added into the activated FA solution and stirred under anhydrous conditions overnight at room temperature. Finally, the mixture was centrifuged and washed with water and ethanol several times to obtain the IOMSN-FA NPs. To prepare the IOMSN@uIO-FA NPs, uIO NPs were added into an IOMSN-NH 2 NPs solution for another 24 h of stirring before modication with FA.
Drug loading and in vitro drug release assay DOX-loaded IOMSN@uIO(DOX)-FA NPs were synthesized as follows. 1 mL of DOX solution (1 mg mL À1 in PBS) was added into 10 mL IOMSN@uIO-FA NPs (100 mg mL À1 ) in PBS at pH ¼ 7.4. Free DOX was washed away with PBS by centrifugation several times aer being stirred overnight. The formed IOMSN@uIO(DOX)-FA NPs were suspended in 1 mL DI water and the loading ratio of DOX was calculated using the UV-vis-NIR spectra of the DOX solution before and aer drug loading. To evaluate the DOX release behavior, 1 mL IOMSN@uIO(DOX)-FA NPs were dialyzed against 49 mL PBS in a water bath shaker at 37 C. At certain time intervals, 1 mL of solution was taken out and tested using the absorbance spectrometer to determine the amount of released DOX and then returned back. The alternative magnetic eld (AMF) triggered drug release, and the IOMSN@uIO-FA NPs were exposed to an AMF with a power of 13.33 kA m À1 (282 kHz) for 10 min every 24 h.
In vitro cytotoxicity assay and cell uptake L929 cells obtained from the American Tissue Culture Collection (ATCC) were cultured with H-DMEM supplemented with 10% FBS and 1% penicillin and streptomycin in 5% CO 2 at 37 C. L929 cells were seeded in a 96-well plate at a density of 5000 cells per well and cultured for 24 h. Then, the old medium was removed and new media containing different concentrations (0, 50, 100, 200 and 500 mg mL À1 ) of IOMSN@uIO-FA NPs were added, and the cells were incubated for 24 h and 48 h. Cell viability was determined by CCK-8 assay. Before the evaluation of cell uptake efficiency, the green-emitting uorescent dye, coumarin-6, was attached to the IOMSN@uIO NPs and IOMSN@uIO-FA NPs for the measurement. Typically, MCF-7 cells were seeded in a 24-well plate at a density of 10 4 cells per well and incubated with 50 mg mL À1 coumarin-6-loaded IOMSN@uIO NPs and IOMSN@uIO-FA NPs for 1 h, 4 h, and 8 h. The cells were washed with PBS to remove the free nanoparticles and stained with rhodamine phalloidin and DAPI. A confocal laser scanning microscope (CLSM, Zeiss LSM 780, German) was used to view and image the distribution of nanoparticles.

In vitro chemotherapy efficiency
To evaluate the cytotoxicity of the IOMSN@uIO(DOX)-FA NPs, 5000 MCF-7 cells (from the American Tissue Culture Collection (ATCC)) were seeded in 96-well plates for 24 h and then incubated with 200 mL IOMSN@uIO(DOX) NPs and IOMSN@uIO(DOX)-FA NPs suspensions in media at a concentration of 100 mg mL À1 for 24 h and 48 h. For the AMF-triggered chemotherapy group, the cells that were incubated with IOMSN@uIO(DOX)-FA NPs were exposed to an AMF with a power of 13.33 kA m À1 (282 kHz) for 10 min. At certain time intervals, 100 mL CCK-8 solution was added for another 2 h incubation and then the absorbance was measured with a microplate reader (Varioskan LUX, Thermo Fisher Scientic Inc. Waltham, Massachusetts, USA) at a wavelength of 450 nm. The cytotoxicity was expressed as the percentage of cell viability compared with that of control group.

In vitro MRI measurement
To evaluate the MR imaging property, IOMSN@uIO-FA NPs were suspended in DI water with different Fe concentrations of 0.026, 0.029, 0.032, 0.036 and 0.039 mM. In vitro MRI relaxation measurements were conducted on a 3.0 T whole-body MR scanner (Achieva TX, Philips Medical System, Best, The Netherlands) and a turbo spin echo sequence was performed to measure the T 2 relaxation time in terms of Fe with the concentration varying from 0.026 to 0.039 mM.

In vivo anti-tumor efficacy
Balb/c female mice weighing 15-20 g were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. and used obeying the protocol no. ZLY16, approved by Tsinghua University Laboratory Animal Research Center (Beijing, China). To determine the anti-tumor efficacy of the IOMSN@uIO(DOX)-FA NPs, the MCF-7 tumor-bearing mice were randomly divided into ve groups (n ¼ 5): control group i.v. injected with saline (200 mL), DOX group i.v. injected with free DOX (1.5 mg kg À1 , 200 mL), IOMSN@uIO(DOX) group i.v. injected with IOMS-N@uIO(DOX) NPs (20 mg kg À1 , 200 mL), IOMSN@uIO(DOX)-FA group i.v. injected with IOMSN@uIO(DOX)-FA NPs (20 mg kg À1 , 200 mL) and IOMSN@uIO(DOX)-FA + AMF group i.v. injected with IOMSN@uIO(DOX)-FA NPs (20 mg kg À1 , 200 mL) and exposed to AMF (13.33 kA m À1 , 282 kHz) for 10 min every two days. The tumor of each treated mouse was measured with a caliper every other day to calculate the tumor volume using the formula: volume ¼ (tumor length) Â (tumor width) 2 /2. Relative tumor volume and relative body weight of mice were calculated as V/V 0 (V 0 is the tumor volume when treatment was initiated) and W/W 0 (W 0 is the body weight when treatment was initiated), respectively. Aer 21 days of treatment, the mice were sacriced and the main organs (heart, liver, spleen, lung, kidney and tumor) were removed for H&E histological processing and analysis.

Conflicts of interest
There are no conicts to declare.