Targeted and redox-responsive drug delivery systems based on carbonic anhydrase IX-decorated mesoporous silica nanoparticles for cancer therapy

In this work, we developed a new antibody-targeted and redox-responsive drug delivery system “MSNs-CAIX” by binding the anti-carbonic anhydrase IX antibody (A-CAIX Ab) on the surface of mesoporous silica nanoparticles (MSNs) via disulfide linkages. The design of the composite particles “MSNs-CAIX” involved the synthesis and surface functionalization with thiol groups, 2,2′-dipyridyl disulfide and CAIX antibody. In vitro, CAIX capping the doxorubicin hydrochloric (DOX)-loaded nanoparticles (DOX@MSNs-CAIX) exhibited effectively redox-responsive release in the presence of glutathione (GSH) owing to the cleavage of the disulfide bond. Compared with CAIX negative Mef cells (mouse embryo fibroblast), remarkably more DOX@MSNs-CAIX was internalized into CAIX positive 4T1 cells (mouse breast cancer cells) by receptor-mediation. Tumor targeting in vivo studies clearly demonstrated DOX@MSNs-CAIX accumulated in tumors and induced more tumor cells apoptosis in 4T1 tumor-bearing mice. With great potential, this drug delivery system is a promising candidate for targeted and redox-responsive cancer therapy.

www.nature.com/scientificreports/ Utilizing the distinctive superiority of abundant active hydroxyl groups on the surface, MSNs were incorporated with appropriate ligands to satisfy the various application requirements of cancer therapy 33,34 . It is known that passive targeting through enhanced permeability and retention (EPR) effect of MSNs faces problems of varied microvascular permeability and increased interstitial pressure 6 . To actively target the tumor sites and reduce damage to normal tissues, diverse biological recognition ligands were modified on the outer surface of MSNs to specifically recognize the receptors overexpressed on tumor cells 35 , including small molecules, monoclonal antibodies, aptamers, peptides and proteins 36,37 . Zhang et al. 38 designed MSNs nanoplatform coupled with folic acid (FA), which possessed a high targeting performance to Hela and MDA-MB-231 cells because FA specifically binded to the folate-receptor sites on the surface of cells. Er et al. 39 constructed cetuximab-targeted MSNs, significantly enhancing the therapy effect on pancreatic tumors due to cetuximab monoclonal antibody efficiently targeting the epidermal growth factor receptor (EGFR).
Additionally, to prevent drugs premature and release drugs on demand at specific sites, various stimuliresponsive DDSs based on MSNs were designed including pH, redox, enzyme, light, magnetic stimuli-responsive drug release [40][41][42] . Among these, redox stimuli-responsive caused more appealing attention considering the abundant reducing glutathione (GSH) in the cancer cells. The concentration of GSH in the normal cellular cytoplasm is as high as 10 mM, which is significantly higher than that of extracellular fluid of tissues (2 μM). Notably, the presence of GSH in tumor cytoplasm cells is fourfold higher than that of normal cells 43 . Since disulfide bonds could be cleaved by GSH in cancer cells, redox-driven capped MSNs linked with a disulfide linker will release drugs on-demand 44 .
CAIX, as a peculiar member of the membrane-associated Carbonic anhydrase (CA) family, was firstly identified in 1994 45 . In general, CAIX is poorly expressed in normal tissues and highly expressed in various solid tumors, including bladder, uterine cervix, kidneys, esophagus, lungs, head and breast carcinomas 46,47 . Antibody is one of the most widely used targeting moieties based on the affinity and specificity 44 . Anti-CAIX antibody (A-CAIX Ab), as a potential targeted agent, applied to MSNs has rarely been studied directly. Furthermore, CAIX was drafted on the surface of MSNs via redox-responsive disulfide linkages, which could efficiently trigger drug release by GSH.
Here, we designed a novel targeted and redox-responsive drug delivery system "DOX@MSNs-CAIX" as shown in Scheme 1, in which MSNs were used as the vehicle to load chemotherapy drug doxorubicin (DOX) and CAIX grafted on MSNs by disulfide bonds. The drug loading capacity and the cytotoxicity of MSNs were investigated in vitro. Also, the endocytosis ability of MSNs-CAIX was investigated on CAIX expressed negative (Mef) and positive cells (4T1), respectively. DOX@MSNs-CAIX were injected to the mouse via tail vein in 4T1 tumor-bearing mice model to evaluate the targeting and the therapeutic effect. Results indicated that DOX@ MSNs-CAIX could achieve GSH-triggered release and target to tumor sites. Thus, MSNs-CAIX are a promising drug delivery system for targeted cancer therapy.

Synthesis of MSNs, MSNs-SH, MSNs-S-S-P, MSNs-CAIX.
MSNs were synthesized using the reported method 48 . 1.04 g, 25 wt% cetyltrimethyl ammonium chloride (CTAC) (Sigma) solution, 6.4 mL of deionized water, 0.02 g of diethanolamine (DEA), 0.9 g of ethanol were mixed and stirred in a water bath at 40 ℃ for 30 min. 0.73 mL of tetraethylorthosilicate (TEOS) (Sigma) was added dropwise into the mixture within 2 min followed by vigorously stirring for 2 h. The surfactant was removed by extraction at 80 ℃ ethanol acid solution (2 mL 37% HCl in 250 mL ethanol) for 8 h. Afterward, MSNs were washed thoroughly and dried under vacuum.
1 mL 3-mercaptopropyltrimethoxysilane (MPTMS) (Sigma) and 1 mL ethanol was mixed and stirred at room temperature for 24 h. Before the end of the reaction of MSNs, 200 μL of the mixture solution was added and stirred for another 2 h under nitrogen atmosphere. The mercaptopropyl-functionalized MSNs (MSNs-SH) were recovered by centrifugation, washed by ethanol three times. The surfactants were extracted as MSNs.
The as-prepared MSNs-S-S-P were suspended in 15 mL PBS (pH 7.4) containing 9.3 mL dimethyl sulfoxide (DMSO). Then, the CAIX-SH was added and stirred gently at room temperature for 24 h. Subsequently, the resulting particles were washed by water and collected by centrifuging. The antibody functionalized MSNs (MSNs-CAIX) were obtained and dried by freezing drier.
Characterization of the nanoparticles. The morphology, particle size and dispersion of MSNs were analyzed by HITACHI SU8220 field emission scanning electron microscope (SEM) with the working voltage at 10 kV and Tecnai G2 F20 (FEI, American) transmission electron microscopy (TEM) at an exciting voltage of 200 kV. The small angle X ray diffraction (XRD) pattern was recorded on D8 Advance (Bruker) by continuous scanning mode from 0.6° to 6°, with a scanning interval of 0.02°. The nitrogen (N 2 ) adsorption desorption isotherm was operated on ASAP 2020 by static adsorption at 77 K. The analysis of infrared spectrum was revealed by Fourier Transform Infrared Spectroscopy (FTIR) Vertex 80v&HYPERION 2000 (Bruker, Germany) using KBr pellets as background. The raman spectrum was measured on Senterra Laser Confocal Raman Microspectroscopy (Bruker, Germany) using a charge coupled device (CCD) detector with an excitation wavelength of 785 nm and a cumulative number of 20 times. 29 Si magic-angle-spinning nuclear-magnetic-resonance ( 29 Si-MAS-NMR) was observed on Advance III HD 600MHZ (Bruker, Germany).
Loading and redox-responsive release of DOX. For preparing DOX@MSNs-CAIX, doxorubicin hydrochloric (DOX) (Adams-beta) loaded MSNs-S-S-P (DOX@MSNs-S-S-P) were prior prepared. 1 mg/mL MSNs-S-S-P and 200 μg/mL DOX were suspended in PBS (pH 7.4). After ultrasonic dispersion, the mixture was shaken for 24 h at room temperature. After centrifugation, DOX@MSNs-S-S-P were collected by vacuum drying. DOX@MSNs were similarly prepared as DOX@MSNs-S-S-P. The supernatant was measured by ultraviolet-visible (UV-Vis) spectrometer (Evolution 60, Thermo) at 480 nm, the drug loading capacity and loading efficiency were calculated by the following formula: 50 mg DOX@MSNs-S-S-P were dispersed in 15 mL PBS containing 9.3 mL DMSO following by adding the as-prepared CAIX-SH. The solution was stirred at room temperature for 24 h. After being washed with deionized water, DOX@MSNs-CAIX were collected by drying vacuum.
1 mg/mL DOX@MSNs-CAIX were added into PBS at different pH values (5.0, 6.0, 7.4) with or without GSH (0 mM, 2 mM, 5 mM, 10 mM). The resulting supernatant at different intervals (1 h, 3 h, 5 h, 7 h, 9 h, 12 h, 24 h and 48 h) was collected and measured by UV-Vis spectrometer. The cumulative drug release (%) was calculated as the following formula: Cell culture. 4T1-Luc (Luciferase) breast cancer cells were gifted from Jiangsu Center for the Collaboration and Innovation of Cancer Biotherapy. Mef cells (mouse embryo fibroblast cell line) were kindly gifted by Xuzhou Medical University. These two cell lines were cultured in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin in a humidified incubator at 37 °C and 5% CO 2 . In vivo therapeutic assay. BALB/C mice (2 months old) were purchased from Ji Nan Peng Yue experimental animal breeding Co. Ltd., and the protocol was approved by the Institutional Animal Care and Use Committee of Xuzhou Medical University, according to National Institutes of Health guidelines. The study was approved by the Ethical Committee for Xuzhou Medical University. Mice were housed in clean plastic cages with a temperature of 25 ± 1 °C and humidity of 55-65% and maintained under specific pathogen-free conditions. The breast cancer model of the mice was established by subcutaneous inoculating 4T1-Luc breast cancer cells (5 × 10 6 cells/mouse). When the tumor volume was approximately 300 mm 3 , the 4T1-Luc tumor-bearing mice were randomly divided into 3 groups (n = 3) and treated with PBS (control), DOX@MSNs, DOX@MSNs-CAIX, respectively. The samples in PBS at the equivalent dosage of 6 mg DOX per kg mice were injected via the tail vein at the 1st, 4th, 7th, 10th day, respectively. The body weights and the tumor volumes were recorded every two days, of which the volumes were calculated by the formula: V = W 2 × L/2, (W and L: the width and the length of the tumor). After 11 days, the Luciferase-labeled Cancer Cells were detected by LB983 Night OWL II Small animal imaging system (Berthold Technologies).

Statistical analysis.
Statistical analysis was performed with prism 5 software. The two groups were compared using Student's t test, for multiple-group comparisons, one-way ANOVA was used, followed by post hoc tests of Bonferroni or Fisher least significant difference as necessary. All data were expressed as mean ± standard deviation (SD) in triplicates. The data were considered to be significant when P < 0.05.

Results and discussion
Synthesis and characterization of CAIX targeted drug delivery system "MSNs-CAIX". The www.nature.com/scientificreports/ cles were comprehensively evaluated with transmission electron microscopy (TEM), scanning electron microscopy (SEM), small angle X ray diffraction (XRD), N 2 isothermal adsorption, 29 Si magic-angle-spinning nuclearmagnetic-resonance ( 29 Si-MAS-NMR), fourier transform infrared spectroscopy (FTIR), raman spectroscopy. The morphology and the sizes of the particles were revealed by SEM and TEM (Fig. 1A) images. The average diameter of these prepared particles was about 60 nm. The small angle powder X-ray diffraction patterns (XRD) of MSNs exhibited the diffraction peak (100) within the 2θ of 2°, the ordered mesoporous structure of MSNs-SH and MSNs-S-S-P was a little weaken after modification as shown in Fig. 1B. After capping with CAIX, the XRD pattern of MSNs-CAIX performed no peaks, indicating the pores of MSNs were capped by CAIX, which was also verified by the blurred pore structure from TEM images of MSNs-CAIX (Fig. 1A). The average hydrodynamic diameter of the prepared particles was presented in Fig. 1C, the diameter for MSNs, MSNs-SH and MSNs-S-S-P were approximately 73 nm. The size of MSNs-CAIX was 104 nm, larger than that displayed in TEM images, attributing to the slight aggregation after decorated CAIX. The zeta potential of the nanoparticles was shown in Fig. 1D, the zeta potential values of MSNs, MSNs-SH, MSNs-S-S-P and MSNs-CAIX were − 21 ± 2.78 mV, − 35 ± 3.35 mV, − 58 ± 0.12 mV, − 51 ± 0.27 mV, respectively. The changes were attributed to the different functional groups grafted on the surface of MSNs. The N 2 adsorption-desorption curves of nanoparticles were shown in Fig. 1E. The specific surface area of bare MSNs was calculated to be 995 m 2 /g. Due to the modification of chemical functional groups, the specific surface area of MSNs-SH and MSNs-S-S-P slightly reduced to 881 m 2 /g and 833 m 2 /g, respectively, which still keep large surface area to load drugs. Moreover, after being wrapped with CAIX, the specific surface area of MSNs-CAIX significantly decreased to 239 m 2 /g, suggesting that CAIX was successfully linked on the surface of MSNs and sealed mesopore channels.
Both Q and T signals of MSNs-SH were found in the 29 Si-MAS-NMR spectrum ( Fig. 2A). The Q peaks were located at − 92 ppm (Q 2 ), − 102 ppm (Q 3 ), and − 111 ppm (Q 4 ), exhibiting the typical Q 2-4 (Q n = (SiO) n (OH) 4-n ) network of the siloxane. The resonance peaks of T 2 and T 3 (T 2 = (SiO) 2 Si(OH) SH, T 3 = (SiO) 3 SiSH) at − 57 ppm and − 67 ppm revealed that the sulfhydryl group was covalently bonded in the silica network successfully. FTIR spectra of MSNs and MSNs-SH was illustrated in Fig. 2B. Compared with MSNs, the vibration of C-H bond at 2,850 cm −1 and 2,921 cm −1 in MSNs-SH, was attributed to the successfully modified propylidene group. In addition, the characteristic peak 2,580 cm −1 can be identified as the SH-stretching vibration. The A-CAIX Ab was grafted and characterized by raman spectroscopy (Fig. 2C). In Fig. 2C, for MSNs-SH, the stretching vibration of SH at 2,580 cm −1 can be observed. Meanwhile, the oscillation of CH 2 in the propyl-bridge between the silicon substrate and the sulfydryl group, namely CH 2 -Si and CH 2 -S, can be observed at 1,308 cm −1 and 1,256 cm −1 , respectively. To sum up, sulfydryl group was successfully grafted onto the silica, which was consistent with FT-IR. For MSNs-S-S-P, three peaks can be clearly observed. They were disulfide bonds located at 539 cm −1 , stretching vibration peaks of carbon-sulfur bonds at 618 cm −1 and 708 cm −1 , indicating the successful introduction of disulfide bonds into the particles. For MSNs-CAIX, four peaks were observed. The disulfide peak moved to 541 cm −1 due to the change of the chemical environment. The stretching peak of the carbon-sulfur bond is Loading and redox-responsive release of DOX. For DOX loading, the loading capacity and efficiency were 163 mg/g and 14%, respectively. GSH triggered redox-responsive DOX release behavior was investigated (Fig. 3A). In the absence of GSH, DOX cumulative release amount was 28%, 21% and 8%, respectively at pH 5.0, 6.0, 7.4 for 48 h (Fig. 3B). After adding 10 mM GSH into PBS at pH 5.0, 6.0, 7.4 (Fig. 3C), the cumulative release increased to 80%, 63% and 42%. The presence of GSH contributed to the cleavage of disulfide bonds between A-CAIX Ab and MSNs. Along with more A-CAIX Ab detached from MSNs, more DOX was released from the pores of MSNs. To furtherly verified the effect of GSH on redox-responsive release, the release properties of DOX@MSNs-CAIX was investigated in PBS at pH 7.4 with various concentrations of GSH (0 mM, 2 mM, 5 mM, 10 mM) (Fig. 3D). It was obvious that a higher concentration of GSH boosted more DOX release, indicating that DOX release behavior from DOX@MSNs-CAIX was GSH stimulation dependent. As known, the level of GSH in the cytoplasm of cancer cells is much higher than that of normal cells 44 , which would enhance the release due to the break of disulfide bonds on DOX@MSNs-CAIX. In addition, we can see that the release was pH-dependent, the lower pH, the higher cumulative release amount. The pH value of extracellular tumor tissues (6.5-6.8) tends to be more acidic than that of the normal tissues (7.4) and further decreases to 4.5-5 in lysosomes, 5.5-6.0 in endosomes 49 . The increased release in acidic environment facilitates pharmacotherapy aiming at tumor tissues.
In vitro cytotoxicity. The cytotoxicity of particles (MSNs, MSNs-SH, MSNs-S-S-P, MSNs-CAIX, DOX@ MSNs-CAIX and free DOX) was performed via CCK-8 assay after incubation with 4T1 cells for 6, 12 and 24 h. As shown in Fig. 4A, all the nanoparticles exhibited low cytotoxicity against 4T1 cells over time at concentrations up to 100 μg/mL, indicating the particles were of great biocompatibility and the decorated group on MSNs surface did not bring extra cytotoxicity. To evaluate the cytotoxicity of the drug-loaded particles, a uniform solution of drug-loaded particles (DOX@MSNs, DOX@MSNs-CAIX) and free DOX molecules with the same drug concentration were prepared and cultured with 4T1 cells for 6, 12 and 24 h. As shown in Fig. 4B

CAIX expression and cellular uptake in 4T1 and Mef cells. To assess CAIX expression in tumor cells
(4T1) and normal cells (Mef), total CAIX RNA from 4T1 and Mef cells were extracted and detected by qRT-PCR technology with CAIX specific primers. As shown in Fig. 5, the level of CAIX was significantly higher expressed in 4T1 cell line, but almost undetected in Mef cell line.
In order to validate the targeted capacity of A-CAIX Ab, the CAIX expressed positive and negative cells (4T1 and Mef) grown on the microscope slides were transferred into one dish (Fig. 6A) and co-cultured with DOX@MSNs, DOX@MSNs-CAIX and free DOX, respectively with equivalent drug dose. As shown in Fig. 6B, for DOX@MSNs, MSNs-FITC (green) and DOX (red) were distributed in cytoplasm of 4T1 and Mef cells. On the contrary, for DOX@MSNs-CAIX, increasing fluorescent signal of MSNs-CAIX-FITC and DOX in 4T1 cells was found compared with Mef cells (Fig. 6C). And the released DOX was evenly distributed in cytoplasm and nucleus of 4T1, which was similar as the endocytosis behavior of free DOX by 4T1 (Fig. 6D). For CAIX positive 4T1 cells, A-CAIX Ab specifically recognized the CAIX antigen overexpressed on 4T1, effectively enhancing the internalization of drug-loaded particles. 4T1 and Mef cells were cultured in the same environment, with the In vivo therapeutic efficacy. To validate the in vivo targeted anti-cancer efficacy of "MSNs-CAIX", 4T1-Luc tumor-bearing mouse model was established. The mice were treated with PBS, DOX@MSNs and DOX@MSNs-CAIX, respectively via tail intravenous injection for 11 days (Fig. 7A,B). Before receiving treatment, the tumor size of tumor-bearing mice in the three groups was little difference (Fig. 7A). After intervention for 11 days the tumors of the three groups all grew with time (Fig. 7B), however, the tumors in DOX@MSNs-CAIX group grew significantly slower than the other two groups due to its remarkably efficient tumor inhibition, and the corresponding bioluminescence intensity was quantitatively demonstrated in Fig. 7C. Then, the solid tumor tissues were removed, as shown in Fig. 7D,E, the size and the weight of tumors in DOX@MSNs-CAIX group were the smallest among these three groups. Also, the tumor growth rate of DOX@MSNs-CAIX group was significantly lower than the other two groups (Fig. 7F). These results demonstrated that DOX@MSNs-CAIX exhibited the most effective cancer therapeutic efficacy, which was mainly due to the precise delivery of the drug at the targeted tumor site.
The results were furtherly confirmed by the in vivo distribution of particles and DOX in the tumor tissues (Fig. 8A). It can be seen that the fluorescence signals of DOX (red) and MSNs-CAIX (green) in DOX@MSNs-CAIX group were stronger than those in DOX@MSNs group and PBS group. Furthermore, the quantitative analysis of the particles fluorescence intensity (Fig. 8B) showed that there were more DOX@MSNs-CAIX in the tumor due to the targeted capacity. Also as shown in Fig. 8C, negligible change in mouse body weight was www.nature.com/scientificreports/ measured in the experimental process, indicating that the MSNs and MSNs-CAIX have excellent biocompatibility, which could act as the promising drug delivery vehicle in vivo. All the above results indicated that the A-CAIX Ab conjugated MSNs could significantly improve the antitumor therapy efficacy as the result of the fact that the targeted A-CAIX Ab could "navigate" the DOX loaded MSNs to the destination by specifical receptor mediated endocytosis, and the enriched drug cause more tumor cells apoptosis.

Conclusions.
In this study, we found that MSNs capped with A-CAIX Ab "MSNs-CAIX" could be used as a redox-responsive controlled-release and targeted delivery carrier. The model drug DOX could be released from the delivery system with GSH as a trigger. In vitro, the MSNs-CAIX could significantly facilitate cell internalization in CAIX-positive cancer cells owing to the targeted capacity of A-CAIX Ab. Moreover, in vivo, the DOX@ MSNs-CAIX could suppress tumor growth more efficiently, which could act as the specific drug delivery vehicle for cancer therapy. Selecting MSNs as the vehicle with the versatile features, A-CAIX Ab as the targeted agent with specific recognition of the tumor cell surface, this exciting active targeted system MSNs-CAIX was worth being further developed for tumor targeting therapy.