Targeted delivery of doxorubicin by nano-loaded mesenchymal stem cells for lung melanoma metastases therapy

Poor antigenic presentation of tumor tissues and a lack of specific targets currently limit the success of nanoparticle delivery system. Cellular carrier technique has been recently explored extensively as a substitutive or supplement for traditional targeting delivery system. Here, we demonstrate the usage of mesenchymal stem cells (MSCs) loaded with doxorubicin containing polymer nanoparticles in pulmonary melanoma metastases therapy, as a modified technique of targeted delivery system. The characterizations of prepared nanoparticles and MSCs sensitivity to DOX and PLGA-DOX were measured. In vitro tumor tropism, and in vivo distributions of nanoparticles loaded MSCs were also investigated. The findings have demonstrated that, the modified system not only integrates the controlled-release property of nanoparticles but also exhibits tumor tropism and penetrative characteristics of MSCs. Furthermore, the in vitro and in vivo anti-tumor study has demonstrated that drug loaded MSCs had potent efficacy in lung melanoma metastases treatment.

remain in consideration 18 . Hence, MSCs engineered with chemotherapeutics have been widely established for their tumor tropic delivery. However, MSCs overexpression of drug efflux pumps often hampers intracellular accumulation of small molecule therapeutics 19 .
In this paper, we supposed that, doxorubicin (DOX) containing nanoparticles that were formulated with poly (d, l-lactic-co-glycolic acid) (PLGA) will achieve successful drug retention in MSCs. MSCs derived from C57BL6 mice adipose tissue and loaded with PLGA-DOX nanoparticles (NP-MSCs) could target lung metastases in C57BL6 mice, suggesting a more effective methodology to treat lung metastases tumor (Fig. 1).

Results
Physicochemical properties of PLGA-DOX nanoparticles. Particle size, polydispersity index (PDI) and zeta potential of PLGA-DOX and PLGA-DiR are presented in Table 1. PLGA-DOX nanoparticles were formulated using carboxymethyl cellulose as the surfactant. Carboxymethyl cellulose can also increase the number of carboxyl groups on nanoparticles making it possible to combine PLGA and DOX more tightly. As a result, DOX was efficiently encapsulated in nanoparticles. The encapsulation efficiency (EE) and drug loading coefficient (DL) was measured 90.52 ± 0.63% and 10.77 ± 0.91%, respectively.
Transmission electron microscopy (TEM) image and the size distribution of PLGA-DOX were given in Fig. 2A,B, respectively. As the figures indicated, the nanoparticles have spherical morphology with a uniform size. In vitro release of DOX from nanoparticles is shown in Fig. 2C, illustrating an initial burst release in the first 4 hrs followed by a stable release phase.
MSCs sensitivity to DOX and PLGA-DOX. MSCs sensitivity to DOX, PLGA-DOX and PLGA were quantified. Drug-free PLGA nanoparticles were found to be rather benign toward the MSCs even at a high concentration of 1000 μ g/mL by MTT assay (Fig. 3B). As shown in Fig. 3A, MSCs survival at 24 h was also unaffected after incubation with different concentration levels of DOX and PLGA-DOX ranging from 0.78 to 25.00 μ g/mL (DOX-equiv). The MSCs viability in PLGA-DOX group remained 90.38 ± 2.52% even at a high concentration of 50 μ g/mL, suggesting that PLGA-DOX was non-toxic to MSCs in certain extent. Nevertheless, a rapid decline of cell viability was found under the concentration of 200 μ g/mL.
MSCs uptake of PLGA-DOX. To explore the optimal cellular uptake condition, qualitative assay for the effect of incubation time and concentration was conducted using flow cytometry. As shown in Fig. 4A, at the same concentration of 20 μ g/mL (DOX-equiv), intracellular uptake was found to be time-dependent. The uptake of DOX and PLGA-DOX displayed a rapid intake in the first 1 hr, followed by negligible change subsequently, suggesting the saturation of incorporation (Fig. 4A). In another experiment, DOX uptake gradually increased in a concentration dependent manner under 1 hr incubation time (Fig. 4B). A significant increase in uptake was detected in PLGA-DOX group moving from 20 μ g/mL to 50 μ g/mL (DOX-equiv). However, at 100 μ g/mL, PLGA-DOX incorporation displayed no obvious increase compared to 50 μ g/mL.
Taking concentration and MSCs viability into account then, the most favorable condition would be co-incubating MSCs with nanoparticles for 1 hr at the concentration of 50 μ g/mL (DOX-equiv).  The drug content loaded in MSCs was determined by HPLC. After 1 hr incubation with 50 μ g/mL DOX or PLGA-DOX, 5 × 10 5 drug-loaded MSCs was extracted for analysis. The average DOX content was measured 24.67 ± 5.33 pg/cell and 20.98 ± 4.02 pg/cell for DOX and PLGA-DOX group, respectively.
MSCs treated with DOX and PLGA-DOX (50 μ g/mL, DOX-equiv) for 1 hr were used for drug release analysis. Drug released from NP-MSCs during the first 30 min was only 4.01 ± 1.05% followed by sustained release with  Data were presented as mean ± standard deviation (n = 6 replicates for each group). *p < 0.05, compared to DOX group at the same concentration. approximately 16.28 ± 1.26% of drug release in 48 h. DOX released from MSCs in the first 30 min was significantly higher than in NP-MSCs group (p < 0.05).
In vitro tumor tropism of MSCs. To track the in vitro migration of MSCs towards B16F10 melanoma cells, a 24-well Transwell chamber was used. The number of migration MSCs significantly increased when compared with the 0.5% Serum control group (p < 0.001). As demonstrated in Fig. 5, the unloaded and loaded MSCs responsed to the B16F10 in the bottom chamber and migrated through the membrane pore, demonstrating their tumor tropism toward tumor cells. Although the migration behavior of loaded MSCs toward B16F10 showed statistical reduction in cell number (p < 0.05), their B16F10 tropism trait was noted to be similar to unloaded MSCs and also adequate for drug delivery.
In vitro cytotoxicity studies. To determine the in vitro cytotoxic potential of NP-MSCs with B16F10 cells, different numbers of NP-MSCs were added on the top chamber of a Transwell plate, with B16F10 cells in the bottom chamber. Then, MTT analysis of B16F10 viability was conducted after 24 h treatment. Figure 6B illustrated the dose-dependent inhibition in cell survival of NP-MSCs. IC 50 of DOX and DOX in nanoparticles was 0.47 ± 0.10 μ g/mL, 1.01 ± 0.12 μ g/mL, respectively, in B16F10 cells (Fig. 6A).
The penetration of MSCs into tumor spheroid. B16F10 tumor spheroids were established to evaluate the penetration capability of DOX-MSCs or NP-MSCs. The penetration was scanned by CLSM at different depth ranging from 0-100 μ m and the mean fluorescence intensity at 40, 60, 80 μ m were analyzed with Image J software. As seen in Fig. 7B, red fluorescence of DOX-MSCs or NP-MSCs appeared inside tumor mass, demonstrating their substantial penetration into tumor nest. The attenuation in mean fluorescence intensity was recorded when the depth was increased in both groups. In comparison with MSC-DOX group, the mean fluorescence intensity of PLGA-DOX loaded MSCs was relatively higher in each scanned depth, especially at 40 μ m (p < 0.01) and 60 μ m (p < 0.05).
In vivo distribution of nanoparticles loaded MSCs. For in vivo investigation, DiR was used as the near-infrared fluorescent probe. The distribution of PLGA-DiR-MSCs and single nanoparticles were tracked in lung melanoma metastasis-bearing C57BL6 mice as shown in Fig. 8A. As the fluorescent images clearly displayed, PLGA-DiR-MSCs remained in the lung for at least 3 hrs after systemic injection implemented and fluorescence was still detectable after 24 hrs. On the contrary, bare nanoparticles could only be tracked in liver and spleen. Furthermore, in a separate experiment, frozen lung sections prepared on day 3 after NP-MSCs injection further demonstrated the tumor homing and penetrative feature of MSCs. DOX fluorescent was found to be distributed in tumor tissues (Fig. 8B). In vivo anti-tumor efficacy of loaded MSCs. The anti-tumor effect of loaded MSCs was evaluated on C57BL6 mice with an aggressive murine melanoma pulmonary metastasis. Body weights were monitored during the experiment as an indication of safety. As Fig. 9C indicated, the body weights showed a slight fluctuations during the treatment period but were not significantly different from one another.
Shown on Fig. 9D, on day 13 of tumor inoculation, metastasized colonies spread all over the surface of the excised lungs in mice treated with saline, implying the strong pulmonary metastasis potential of B16F10 melanoma cells. The suppression of tumor metastasis induced by DOX was found to be more robust than PLGA-DOX, which was consistent with the in vitro results. Because of active migration and penetration potential of MSCs, NP-MSCs could improve the drug concentration in lungs and sites of the metastasis, thereby enhancing anti-tumor efficacy sequentially. The data in Fig. 9B demonstrated that mice treated with NP-MSCs had a significant reduction of lung metastases (p < 0.001). Also, the pulmonary weight in Fig. 9D further supported the antitumor efficacy of NP-MSCs.
Lungs in each group were also investigated by H&E staining. As seen in Fig. 9E, many melanomas appeared in the lung sections of melanoma-bearing mice, especially in the saline group. In contrast, the lungs of mice treated with NP-MSCs showed considerable reduction in melanoma number, and many normal pores also existed.

Discussion
Virtually all, most nanomedicine including those decorated with specific binding ligands could accumulate passively in tumor tissues via EPR effect 20 . However, the accumulation of nanomedicine under ERP effect was gradually recognized to be restricted to tumors that are highly vascularized. As a result, under-perfused or hypoxic locations within tumors rarely benefit from EPR effect 21 . MSCs have shown potent tumor-homing potential in response to pro-inflammatory cytokines in tumor microenvironment, which enable the possibility of true, active carriers for tumor-targeting therapy 22 .
As it reported, MSCs have been engineered for delivery of chemotherapeutic drugs such as paclitaxel, gemcitabine and doxorubicin. Commonly-used materials in the nanoparticles engineered MSCs include polymeric micelles 23 , mesoporous silica 24 , dendrimers 25 . However, most nano-loaded MSCs were administrated intratumorally due to the limited drug loading or targeting efficiency, so that this system is not suitable for orthotropic tumor or metastases models. Some efforts have been made in lung melanoma metastases by incorporating PTX with MSCs directly. However, the anti-tumor effect was still hindered by low drug loading efficiency of 2.5 ± 0.05 pg/ cell 26 , which is potentially attributed to the efflux outside the cells by the P-glycoprotein overexpressed on MSCs. MSCs modified genetically were also investigated in pulmonary melanoma metastases model. As it documented, bone marrow-derived MSCs were transfected with adenovirus vector to express CX3CL1 and inhibit the development of B16F10 lung metastases, the number of metastatic nodules treated with MSCs-CX3CL1 was significantly reduced by 84% compared with PBS group 27 . To avoid safety problems caused by virus transfection, non-viral vectors were designed for MSCs gene-modification and endowing MSCs the potent anti-tumor efficacy in syngeneic lung melanoma metastases model [28][29][30][31] . Nevertheless, most current researches on these non-viral vectors are still restricted to the in vitro evaluation of their transfection efficiency 32 . Here, we reported for the first time, the application of mice adipose-derived MSCs were loaded with doxorubicin-containing PLGA nanoparticles (PLGA-DOX) for the treatment of pulmonary B16F10 melanoma metastases. The average DOX content was measured 20.98 ± 4.02 pg/cell in NP-MSCs extremely higher than that reported in the previous work. Due to the active migration and penetration potential of NP-MSCs, drug concentration was improved in the lungs and sites of the metastasis, which resulted in enhanced anti-tumor efficacy. In our research, controlled-release function of PLGA nanoparticles and the migratory potential of MSCs were utilized to construct a carrier-in-carrier delivery system. In the preparation of PLGA-DOX, emulsion evaporation method was employed with sodium carboxymethyl cellulose (SCMC) as the stabilizer. SCMC could adsorb on the surface of PLGA to provide ample carboxylic acid, which facilitates the loading of amide-containing drugs, resulting in high encapsulation efficiency.
MSCs uptake of DOX solution and PLGA-DOX nanoparticles are governed by passive diffusion and endocytosis, respectively 33 . Different from passive diffusion, latter manner was energy consuming, as a result, drug loaded in MSCs could get higher in DOX group when the concentration was increased, while PLGA-DOX group reached the equilibrium between endocytosis and exocytosis.
On the other side, because of the limiting number and density of MSCs in intravenous injection, enough nanoparticles incorporation in MSCs is essential for achieving high concentration in tumor tissues. Initial study showed that loading efficiency of PLGA-DOX nanoparticles reached 90.52 ± 0.63% and the drug loaded in each cell was 20.98 ± 4.02 pg/cell approximately, comparable to DOX group under the optimal incubation condition. With the addition of lung targeting, therapeutic dose could be obtained. Further conduction of nanoparticles optimization allowing more MSCs take-up and reducing the number of NP-MSCs during treatment period still needs to be investigated to improve the clinical safety.
Furthermore, the sensitivity of MSCs to DOX and PLGA-DOX were evaluated. Due to P-glycoprotein drug efflux transporter overexpression, MSCs were notably resistant to some chemotherapeutic agents 26 . Our result showed that DOX and PLGA-DOX were both non-toxic to MSCs at 25 μ g/mL. At 50 μ g/mL of chemotherapeutic agents, MSCs viability was revealed higher than 90% for PLGA-DOX group after 24 h drug incubation, suggesting the majority of MSCs remained viable following loading and hold the inherent traits 24 .
The crucial property of MSCs as drug carriers is their migratory and permeability potential 34 . As documented, MSCs are drawn to TNF-α , TGF-β , and other cytokines secreted in tumor microenvironment and then home to tumor tissue 35 . In this project, B16F10 cancer cells were used as chemoattractant and the tropism of MSCs was demonstrated in comparison with 0.5% FBS control. While the migratory capacities of NP-MSCs were slightly affected but still retain the innate tumor homing nature. This discrepancy can be attributed to nanoparticles  attached on the membrane of MSCs or inside MSCs hampering their deformation to migrate through transwell membrane. In vivo distribution suggests that NP-MSCs could localize in lungs and more than 95.99% drug was released here according to the release profile.
Additionally, the permeability of loaded MSCs was investigated with an in vitro 3D multicellular tumor spheroid model, which mimicked physiological tumor tissue microenvironments by organized extracellular matrix and acidic pH. In the spheroid depth of 40 and 60 μ m, the mean fluorescence intensity of NP-MSCs group was measured relatively stronger than DOX-MSCs, potentially due to the powerful penetration ability of NP-MSCs. Early release of DOX in medium prior to MSCs permeation into tumor spheroid may also contribute to the result. Our studies demonstrated that PLGA-DOX loaded MSCs not only retain their tumor homing nature, but also could permeate into tumor aggregates.
In vivo anti-tumor efficacy of NP-MSCs was carried out in C57BL6 mice bearing B16F10 melanoma that metastasized to the lungs. NP-MSCs exhibited potent anti-tumor efficacy, with the main conclusions as follows. Drug loaded MSCs retained the tumor tropic and permeability trait, which facilitated an accurate targeting of metastases and permeation of tumor nest. Moreover, DOX loaded in MSCs was mainly bumped out of cells by P-gp system and then easily eliminated from tumor mass, while nanoparticles were found to be retained in MSCs for extended periods and act as drug depots which could release drug slowly.

Conclusions
In our experiment, MSCs isolated from adipose tissue in C57BL6 mice were loaded with PLGA-DOX nanoparticles to treat pulmonary metastases. The key properties of MSCs such as their tumor homing and permeable properties were found unaffected by drug loading, which facilitated an accurate targeting of metastases and permeation of tumor nest. Further, in vivo distribution studies showed that, NP-MSCs were noted to preferably reside in lungs while bare nanoparticles deposited mainly in liver and spleen. According to the photographs of lung sections and tumor spheroid scanning, PLGA-DOX carried by MSCs could penetrate into tumor nest. MSCs carrying PLGA-DOX nanoparticles were found to be effective in killing B16F10 melanoma cells in vitro with a dose dependent manner. In vivo antitumor investigations further supported the efficacy of NP-MSCs in the treatment of pulmonary metastases. Thus in this novel work, we demonstrate PLGA-DOX loaded MSCs can be promising agent for the treatment of pulmonary tumors through intravenous injection.
Preparation of drug-loaded PLGA nanoparticles. Doxorubicin and DiR were loaded in PLGA nanoparticles by an emulsion evaporation method 36,37 . In brief, 4 mg of DOX-HCL was first converted into free base with 0.01 mol/L NaOH solution and then added directly into 2 mL of acetone containing 30 mg PLGA. Next, 2.5 mL mixture of PLGA and the drug was added slowly into 0.01% sodium carboxymethyl cellulose. Followed by the evaporation of the acetone with gentle stirring overnight, the resulting DOX-PLGA nanoparticles were washed by ultrafiltration technique (500 μ L, 30k MWCO, Millipore, USA) at 3000 rpm for 25 minutes to remove unloaded doxorubicin. After the final wash, nanoparticles in the upper chamber of the ultrafiltration centrifuge filter were harvested. Unloaded doxorubicin was in the lower chamber. Nanoparticles containing DiR was also prepared similarly.
Characterization of nanoparticles. The morphology of nanoparticles was acquired using transmission electron microscope (TEM, Hitachi H-7650 and JEOL JEM-2100F, Japan) and the sample was diluted 5 times by loading sample solution on nickel grids without staining.
Particle size (diameter, nm), polydispersity index (PDI) and zeta potential (surface charge) were determined by dynamic light scattering (DLS) technique (Zeta Plus Brookehaven Instrument, UK). The PLGA NPs were diluted by deionized water before measurement, and then the analysis was performed at 25 °C with a scattering angle of 90°.
Encapsulation efficiency (EE) and drug loading coefficient (DL) were determined by ultrafiltration technique and calculated by equation (1)  6.0 and 7.0, respectively) was transferred into dialysis bags (MWCO: 14 kDa) which was placed in 40 mL of the same buffer. The sample was then incubated at 37 °C, 100 rpm. Sink conditions were maintained throughout the experiment. At predetermined time points, 1 mL dialysate was collected and analyzed using spectrophotometric method at 480 nm. The dialysate was then replenished with the same volume of citrate buffer. In vitro cytotoxicity assessment. The anti-tumor effect of NP-MSCs on B16F10 was conducted by using a Transwell assay. B16F10 cells (2 × 10 4 ) were seeded at the bottom chamber of the Transwell plates. Then, NP-MSCs of different concentration (0, 1 × 10 4 , 2.5 × 10 4 , 5 × 10 4 ) were added into the top chamber of the Transwell plates. After 24 h incubation, the viability of B16F10 was determined by MTT assay. In addition, the cytotoxicity of DOX solution and DOX loaded nanoparticles to B16F10 was also conducted. The cell viability data was analyzed using GraphPad Prism software to calculate the IC 50 of different groups.

Cytotoxicity of drug on
Melanoma tumor spheroid penetration. The in vitro three-dimensional spheroids model of B16F10 cells were employed to evaluate the penetrability of loaded MSCs. 2% (w/v) agarose solution were prepared in serum-free DMEM by heating at 80 °C for 20 min and then sterilized. 5 × 10 3 tumor cells with complete medium were seeded into a 96-well plate which was initially coated with agarose to avoid cell adhesion. The tumor spheroids were grown at 37 °C in 5% CO 2 for 5 days until the diameter of the spheroid reached 600 μ m. These compact and uniform spheroids were then used for subsequent permeation studies.
MSC-DOX and MSC-PLGA-DOX were co-cultured with tumor spheroids for 48 h. After incubation, tumor spheroids were washed twice with PBS and fixed with 4% paraformaldehyde for 20 min. The permeability of drug was then investigated with confocal laser-scanning microscopy (FV1000, Olympus, Japan).
In vivo distribution studies of nanoparticles loaded MSCs. The distribution of drug loaded MSCs in mice bearing lung melanoma metastasis was investigated with DiR as a fluorescent probe. PLGA-DiR-MSCs were injected intravenously at 1 × 10 6 cells/200 μ L/mice. In vivo distribution of PLGA-DiR-MSCs was detected at determined time points (0.5, 1, 3, 24, 72 h) in a dark room with NIR imaging system. Tumor bearing mice injected with PLGA-DiR nanoparticles solution were imaged as a control.
Additionally, PLGA-DOX loaded MSCs (NP-MSCs) were also injected into tumor bearing mice for lung frozen specimen observation on day 3 under a fluorescence microscope (IX53, Olympus, Japan).
Scientific RepoRts | 7:44758 | DOI: 10.1038/srep44758 In vivo anti-metastasis effect. All care and handling of animals were carried out according to the international laws and policies (EEC Council Directive 86/609, 1987) and approved by the animal ethics committee of China Pharmaceutical University (Nanjing, China). The lung metastases models were established in six-week-old female C57BL6 mice by intravenously inoculation of 1 × 10 6 B16F10 cells through a tail vein (0 day). The mice were then randomly divided into five groups (n = 6 animals for each group): Blank group (Saline), DOX group (DOX), PLGA-DOX group (PLGA-DOX), MSC-DOX group (MSC-DOX), and MSC-PLGA-DOX group (MSC-PLGA-DOX). All the doxorubicin preparations were injected via the tail veins at the dose of 2 mg/kg on the 3rd, 7th, 11th day. On 13th day of tumor inoculation, lung specimens in all groups were harvested and photographed. Lung weight and the number of metastatic nodules were also examined. The lung specimens were finally treated with formalin for hematoxylin and eosin (HE) staining.
Statistical analysis. Statistical analysis was performed using GraphPad Prism 6.0 and at least three independent experiments were carried out. Data are presented as mean ± standard deviation, and are analyzed using paired student's t-test and one-way analysis of variance (ANOVA).The significant difference level was set at p < 0.05.