Milk-exosome based pH/light sensitive drug system to enhance anticancer activity against oral squamous cell carcinoma

A multimodal drug delivery system targeting the tumor microenvironment is an inspiring method for treating cancer tissues, including oral squamous cell carcinomas (OSCC). Such approaches require an efficient and safe drug carrier. Bovine milk derived exosomes are ideal because the source is adequate and have advantages of both synthetic and cell-mediated nano carriers. In the present study, we developed a pH/light sensitive drug system based on milk-exosomes for OSCC therapy. It was called exosome–doxorubicin–anthracene endoperoxide derivative (Exo@Dox–EPT1, NPs). Milk-exosomes were conjugated to doxorubicin (Dox) by a pH-cleavable bond, which can rapture under an acidic microenvironment. Besides, endoperoxides and chlorin e6 (Ce6) were also loaded and the endoperoxides undergo thermal cycloreversion and release singlet oxygen to kill cancer cells. We have also investigated the body distribution, antitumor effects, and biocompatibility of the nanoparticles. The new milk-exosome-based drug delivery system showed controlled drug-release, biocompatibility and, proved to be effective in treating OSCC.


Introduction
Oral squamous cell carcinoma (OSCC) is the most common cancer in the head and neck region. 1 Doxorubicin (Dox), paclitaxel and cisplatin are the rst-line chemotherapy drugs for treating various cancers including OSCC. 2 However, free agents usually require high dosage and result in unexpected systemic toxicity. 3 Recently, nanomedicine advancement in cancer has provided great opportunities to solve these limitations. 4 Nanoparticle-based drug delivery systems could accumulate in tumors through the enhanced permeability and retention (EPR) effect. Moreover, such drug delivery system encapsulated multifunctional drugs could be modied with cancerassociated biomarkers to realize active targets. 5,6 Exosomes are lipid bounded natural vesicles with sizes between 40-100 nm. They are secreted by almost all kinds of cells through biological uids, including blood plasma, urine, sweat and milk. 7,8 In the intercellular cross-talking, exosomes have a vital role for transmitting signaling. 8,9 Due to the characteristics that combine synthetic nano-carriers and cellmediated vehicles, 10,11 they are identied as an ideal natural drug delivery carrier. Currently, more and more types of exosomes or exosome-mimics have been exploited. [12][13][14] For example, tumor cell-derived exosomes have been applied to treat malignant ascites and pleural effusion in a phase II study. 12 More meaningfully, dendritic cells (DCs)-derived exosomes have entered in clinical trial to treat melanoma or smallcell lung carcinoma patients based on mechanism of antigenantibody reaction. 15,16 Therefore, exosomes have great potential to realize clinical translation. However, a number of disadvantages would impede the translation of autologous exosomes, such as limited yields, longer drug preparation time, cancerstimulating risk and ethical problems. 11,17 To overcome these deciencies, bovine milk-derived exosomes were rstly isolated in 2010 (ref. 18) and gradually developed to potential drug delivery vehicle to its availability, cost and non-toxicity. 10 The carrier has been proved to reduce systemic toxicity of free Dox 19 and enhance drug accumulation in tumor tissues. It was reported that bovine milk-derived exosomes loaded with chemotherapeutic drugs showed better antitumor effects against lung and breast cancer. 10,20 More importantly, oral chemotherapy might be realized in future because bovine milk-derived exosomes could pass through the gastrointestinal barrier. 21 However due to its passive targeting, the exosomes loaded chemotherapeutic drugs would massively accumulate in various organs and cause possible damage to liver, kidney or heart. 16 Therefore, it is necessary to improve the tumor-targeting of exosomes through the active targeting strategy. Alvarez et al. suggested that DCs-derived exosomes modied by neuron-specic RVG peptide and loaded with therapeutic siRNA were applied to treat Alzheimer's disease in a mouse model. 22 Similarly, exosomes from immature DCs adapted by iRGD increased Dox delivery efficiency into breast cancer cells. 23 In 2015, Gupta et al. rstly reported that milkexosomes functionalized with folic acid improved efficacy and safety of drugs for cancer with high expression of folic acid receptors. 10 Targeting acidic tumor microenvironment (TME) is a promising approach to engineer exosomes, except biomarkertargeting strategies. As we know, majority of solid tumors exhibit pH from 6.5 to 7.4, thus it give us possibility to fabricate PH-response nanoparticles. 24 The imine bond is easily disintegrated in an acidic solution (pH ¼ 6.8). 25 Thus, if exosome membrane is conjugated chemotherapy drugs with hydrazone bond, the cleavage of the imine bond triggered by acidic TME would result in quick release of drugs in the tumor site. Next, hypoxia is another the distinct hallmarks of TME that induce irreversible tumor metastasis, as well as inict hypoxiaassociated resistance. 26 Correcting the hypoxic TME with photodynamic therapy (PDT) is an effective strategies to treat cancer. 27 The reactive oxygen (ROS) generated in PDT could destroy hypoxia in TME and kill cancer cells. 26 Endoperoxide is one of the most reliable source of singlet oxygen 28 and can generate singlet oxygen when temperature rise. Under nearinfrared (NIR) light irradiation, photosensitizer-chlorin e6 (Ce6) produces plasmonic heat which results in singlet oxygen generation of endoperoxide. In general, targeting acid and hypoxic TME would achieve controlled release of chemotherapy drugs and photochemistry therapy.
In the present study, a novel PH-response nanoparticles based on the bovine milk-derived exosomes as carrier were synthesized against oral squamous cancer cell (OSCC). First, Dox was modied on the exosomes membrane with imine bond, and then the tumor site-specic release of Dox was stimulated by PH in TME. Next, anthracene endoperoxide derivative (EPT1) and chlorin e6 (Ce6) were encapsulated in milk-exosome. When the nanoparticles were in tumor site, Ce6 produced plasmonic heat and accelerated singlet oxygen generation from EPT1 under NIR irradiation. Consequently, a reliable and tumor site-specic photochemistry therapy was realized in OSCC treatment. To our knowledge, our study might be the rst report that the bovine milk-derived exosomes as carriers were applied in photochemistry synergistic therapy triggered by PH in TME and NIR irradiation. On the other hand, hematological assays and histopathology indicated that the nanoparticles had well biocompatibility in vivo without NIR stimulation. Therefore, we believed that the nanoparticles might be a promising prospect against OSCC. The main scheme of our study is depicted in Fig. 1.

Cell culture
Three human head and neck squamous cell carcinoma cell lines (HSC-3, SCC-9, CAL-27) and a human cardiac muscle cell line HCM were used in this study. The HSC-3 cell line was obtained from Fudan University, Shanghai, China while all other 3 cell lines were purchased from the American Type Culture Collection (ATCC, Manassas, VA). All cells were propagated in the conditioned (exosome free) Dulbecco's modied Eagle's medium (DMEM) (Gibco, Life Technologies, NY, USA) with fetal bovine serum (10%) and penicillin/streptomycin (100 units per ml). All reagents were purchased from Gibco, Life Technologies. Cells were maintained in an incubator at 37 C in the presence of 5% CO 2 . Medium was renewed every 48 hours. Cells were used when grown to logarithmic phase.

Bovine milk exosomes isolation
Fresh unpasteurized milk (2 liters) was collected from the Weizhu Dairy located in Nanjing, China and stored at 4 C for 2 hours before the isolation of exosome.
Exosomes were isolated from milk using the differential centrifugation as described previously. 10,29 Briey, milk was centrifuged at 3000 Â g for 15 min at 4 C to remove fat globules, cellular debris and somatic cells. The whey was collected by passing through a cheese cloth and transferred into polycarbonate tubes. In order to remove large particles, it was centrifuged for 60 min at 100 000g in Type 45Ti xed angle rotor using Optima LE-80K Ultracentrifuge (Beckman Coulter, Brea, California). Supernatant was nally centrifuged at 135 000 Â g for 90 min at 4 C. Then the nal supernatant was discarded and the exosome pellet was washed twice using the phosphatebuffered saline (PBS). The exosome pellet was re-suspended in PBS, followed by ltration through a 0.22 mm Steritop lter (Millipore, Darmstadt, Germany). The total protein concentration was determined using the bicinchoninic acid (BCA) kit (Beyotime Biotechnology, China). The concentration of the milk derived exosomes was adjusted to 6 mg ml À1 and stored at À80 C until further use. The isolated exosomes were characterized as recommended by the International Society of Extra Cellular Vesicles (ISEV). 30
2.3.2 Synthesis of pH sensitive Exo@DOX. Doxorubicin could be conjugated to milk-exosomes through C]N bond, as previously reported by Nie Shuming. 31 In brief, milk-exosome was lyophilized at À4 C for further use. Doxorubicin (10 mg/ 0.017 mmol) dissolved in 10 ml of ethanol was added to triethylamine (10 ml/2 mg) and stirred at room temperature for 24 hours. The solution obtained was ltered to further react with milk-exosome at 60 C for 24 h. Then it was puried for 24 hours using dialyses against cellulose membranes (M w ¼ 3.5 kDa) in PBS at 4 C to remove unreacted compound.

Synthesis of Exo@DOX-EPT1 (NPs).
The prepared solution of Exo@DOX-NPs was mixed with 5 mg of EPT1 and Ce6 (1 mg) and oscillated ultrasonically oscillation for 5 hours. The unloaded EPT1 and Ce6 were removed by passing through the disposable PD-10 desalting columns. The resulting solution was directly used without any further treatment. Synthesis process and therapy mechanism was illustrated in Fig. 1.

Transmission electron microscopy (TEM)
Milk-derived exosomes (6 mg ml À1 ) were diluted to 1000 folds using deionized water and added onto the clean silica wafer and lyophilized. The extracted milk-exosome and NPs were characterized using a TEM (JEM-2010, JEOL, Japan) at an acceleration voltage of 80 kV ( Fig. 2A and B). 10 ml sample was pipetted onto the formvar/carbon-coated nickel grid, and stained by phosphotungstic acid (PTA, 1%) solution.

Dynamic light scattering (DLS)
Hydrodynamic particle size analysis was performed using dynamic light scattering by ZetaSizer Nano-ZS90 (Malvern Instrument, Worcestershire, United Kingdom). Puried milkexosomes and synthetic NPs were resuspended in 1 ml PBS and analyzed for particle size distribution.

Western blot analysis
Milk-derived exosomes for western blotting were suspended in 100 ml RIPA buffer (Solarbio, Shanghai, China). The exosomal surface proteins were analyzed by western blotting as described previously. 10 Blots were probed for CD9, CD63, Tsg101 (Abcam, Cambridge, UK) using secondary antibody anti-GM130 (Abcam, Cambridge, UK). All antibodies were used following the manufacturer's instructions.

Drug loading and encapsulation
The encapsulation efficiency and drug loading content of Dox were calculated using the following equation; Drug loading content of Dox ¼ total amount of Dox À weight of free Dox/weight of Dox-NPs

In vitro uptake of NPs
Because Dox is intrinsically uorescent, uptake of free Dox or NPs can be observed at 594 nm using confocal microscopy aer excitation at 480 nm. Briey, HSC-3, SCC-9, CAL-27 and HCM cells were seeded into a 6-well plate at a density of 1 Â 10 5 cells per well. All cells treated with free Dox or NPs (5 mg per well, Dox). Cells were labeled with uorescent probe Dil (Beyotime Biotechnology, China) aer 8 hours of treatment, and in vitro uptake was calculated detected by a confocal laser scanning microscopy (Carl Zeiss Microscope Systems, Jena, Germany).

2.9
In vitro release ability of Dox

In vitro cytotoxicity assay
The cytotoxicity of NPs on HSC-3, SCC-9, CAL-27 and HCM cells was tested using the Cell Counting Kit-8 (CCK-8) assay (Med-Chem Express, USA). Briey, cells were seeded into a 96-well plate overnight until the conuence reached 80-85%. Then, the medium was replaced by 100 ml of milk-exosomes, free Dox or NPs (with equivalent Dox 0.5 mg ml À1 ) for 6 hours. One plate of cells was treated with NPs only while the other plate of cells was irritated at 808 nm for 3 min following NPs administration. Aer 24 hours, an aliquot of 10 ml of CCK-8 solution was added in each hole and incubated for another 2 hours in 5% CO 2 at 37 C. Absorbance at 450 nm was measured and cell viability was calculated based on the absorbance data. 32

In vivo anti-tumor activity
Nude mice bearing HSC-3, SCC-9 and CAL-27 cancer cells respectively were administered several types of drugs intravenously Paper RSC Advances including milk-exosome, free Dox or NPs when tumor volume reached 100 mm 3 . According to the drug administration, mice were divided into 5 groups (n ¼ 5 for each group): a single dose of 100 ml milk-exosomes (60 mg kg À1 exosome protein), laser (aer milkexosomes injection), free Dox (0.25 mg kg À1 , amount of Dox), NP nl (0.25 mg kg À1 , amount of Dox, no laser), and NP 808 (0.25 mg kg À1 , amount of Dox, laser). Drugs were administered every 3 days for 10 times and tumor volumes were recorded simultaneously. The 808 nm laser (1 W cm À2 ) was used in additional radiation for 3 min in laser and NP 808 groups. The longest diameter (L) and shortest diameter (S) were measured every two days by the same observer. The tumor volume was calculated using the following formula: V ¼ (L Â S 2 )/2. All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of Nanjing University and approved by the Animal Ethics Committee of Nanjing.

In vivo distribution
Athymic nude mice (n ¼ 4 per group) bearing HSC-3 tumor were employed. Indocyanine green (ICG) encapsulated NPs were prepared and dispersed in 100 ml PBS. ICG-NPs were administrated into HSC-3 tumor bearing mice via an intravenous injection. In vivo uorescence imaging was obtained at various time intervals (5 minutes, 1 hour, 2 hours, 6 hours, 12 hours, 24 hours and 48 hours). Then mice were executed to harvest major organs and tumor tissue for uoresce imaging. All uorescence images were obtained by in vivo uorescence imaging system (Cri Inc., MA, USA).

Statistic analysis
Triplicate data were analyzed in GraphPad Prism 6 and shown as mean AE standard deviation (SD). Outcome variables were compared among treatment groups using the unpaired t-test and one-way analysis of variance (ANOVA). P-Values less than 0.05 were considered to be statistically signicant.

Characterization of milk-exosome and Exo@DOX-EPT1 NPs
The present study isolated and analyzed the milk-exosome and Exo@DOX-EPT1 NPs. Our data showed a typical exosomal characterization of milk-exosomes and NPs (Fig. 2). The diameter and size measurements of isolated vesicles and NPs using DLS indicated that the isolated milk-exosome and NPs aer synthesis have an average diameter of 105.7 nm, and 122.4 nm respectively ( Fig. 2A and B). The size and structural morphology of milk-exosome and NPs were conrmed by TEM and showing a typical bilayer membrane ( Fig. 2A and B). In addition, the western blotting analysis conrmed the existence of typical membrane proteins of exosome, including CD63 and Tsg101 (Fig. 2C). EPT1 is a reliable source of singlet oxygen that can release ROS under remote-controlled heating conditions independent from tissues' oxygen concentration. The anthracene endoperoxide can be easily monitored by UV-vis spectroscopy, as the endoperoxide does not have any absorption bands at wavelengths longer than 350 nm. In contrast, the corresponding anthracene has several diagnostic bands in the range of 350 nm to 425 nm. 28 The synthetic route of EPT1 was illustrated in Fig. 2D and the structure of loaded EPT1 was conrmed through HNMR, CNMR and UV-vis spectroscopy analysis (Fig. 2E-G), and consistent with the literature report. 28 The EPT1 drug loading content was 5.8% in this study.

PH-response Dox release
Dox concentration was calculated based on standard absorption curve as shown in Fig. 2H. The initial drug loading content of Dox was 13.4%. In order to verify Dox loading and pH-controlled release ability of NPs, they were dispersed in the PBS (pH ¼ 7.4) and acetate buffer (ABS; pH ¼ 5). Dox released from NPs in two different pH buffers was compared at variable time points (Fig. 2I). The results indicated that almost 25% Dox was released from NPs aer 9 h of incubation and maintained a stable suspension in PBS. When pH was decreased to 5 (ABS), Dox release rate aer 9 h was increased to 42.5% and almost 73% during 50 hours. These nding suggested that Dox can have sustained release from NPs under acidic conditions.

Release of ROS under light excitation
As the ROS generation diminished corresponding to decreased cellular oxygen concentration, many PDT methods based on organic photosensitizer are self-limiting (e.g. porphyrin, 33 phthalocyanine, 34 chlorine 35 ). In the present study, we built up a novel PDT system based on milk-exosome loaded anthracene endoperoxide derivative and Ce6. HSC-3, SCC-9, CAL-27 and HCM cells were treated with PBS, H 2 O 2 and NPs with or without irritation at 808 nm. The ROS production of different samples was analyzed using DCFD kit. NPs released no apparent ROS without light irritation, but produced ROS effectively aer irritation with an 808 nm NIR laser (1 W cm À2 ) (Fig. 3). These results proved the PDT ability of the fabricated NPs.

Cell uptake
Drug delivery system based on exosomes has been demonstrated to deliver various therapeutic cargos that can easily be internalized by recipient cells. 36 This study examined the uptake of NPs compared to free Dox using the confocal microscopy. Briey, HSC-3, SCC-9, CAL-27 and HCM cells were stained with dye Cy5 and treated with free Dox or NPs for 8 hours (with equivalent Dox of 5 mg per well). The cell internalization was observed by detecting emission of Dox using the uoresce microscopy. NPs-treated cells displayed more intense green uorescence in the cytoplasm compared to free Dox-treated cells (Fig. 4). It indicated that more NPs were taken up via milk-exosomes.

Antitumor effect in vivo
To further evaluate the therapeutic effects of NPs in vivo against OSCC, mice bearing HSC-3 xenogra tumors were randomly divided into ve groups (ve mice for each group) when tumor sizes reached approximately 100 mm 3 . The anti-tumor effect of different samples clearly showed that the mice treated with milk-exosomes (control) had the fastest tumor growth and reached the largest volume about 2.19 (SD ¼ 0.30) cm 3 at the end (Fig. 6). Mice treated with laser revealed no obvious changes in the tumor growth compared in the control group mice (1.85 cm 3 , SD ¼ 0.36 of tumor volume at the end, P > 0.05). In case of mice treated with free Dox and NP nl , tumor growth was significantly slower and reached 1.26 cm 3 (SD ¼ 0.12) and 0.86 cm 3 (SD ¼ 0.18) respectively. Moreover, in group NP 808 , the tumor growth was restrained effectively and almost disappeared aer treatment (average tumor volume 0.05 cm 3 , SD ¼ 0.07). Therefore, the NPs produce synergistic effects of photochemistry which can be triggered by acid TME and NIR and is a novel promising therapeutic approach against OSCC.

Biodistribution in vivo
In order to examine the in vivo distribution of the NPs, we created ICG-labeled NPs for in vivo uorescence imaging experiments. An aliquot of 100 ml of free ICG (control) and NPs was injected intravenously into HSC-3 tumor-bearing nude mice with an equivalent dose of ICG (0.3 mg ml À1 ). The uorescent imaging was performed at 1, 4, 8, 24, 48, and 72 hours aer injecting the medicaments (Fig. 7). At 24 hours aer injection, uorescent signals of tumor tissues in NP@ICG group were stronger than that of free ICG group. Moreover, NP@ICG still had remarkable accumulation in tumor tissues at 72 hours while uorescent signals of free ICG group were invisible in tumor at the same time. Aer 72 hours, the mice were euthanised and their organs and tumor tissues were harvested for uorescent imaging analysis. Strong signals were observed in the liver and kidney tissues in both groups. Furthermore, the NP@ICG group had much more uorescent intensity in tumor while that of free ICG group was invisible. In summary, NPs enhanced the drug accumulation in tumor tissues and had longer retention time in vivo.

Biocompatibility
Non-specic toxicity is one of the major limitation of traditional chemotherapeutic agents, including Dox. 4 The cardiotoxicity of Dox is a major complication for clinical treatment. Exosomebased drug delivery system was proven to reduce normal tissues toxicity from Dox. 19 In cytotoxicity evaluation, this study indicated that more numbers of HCM cells were survived aer treated with NPs alone than free Dox (Fig. 8A). To test the biocompatibility of NPs in vivo, mice received intravenous injection of Dox or NPs every 3 days at equivalent quantities. Mice treated with PBS were set as control group. The hematology assay for mice treated with NPs or PBS were administrated every 7 days. The histopathology of major organs (heart, liver, spleen, lung and kidney) were further performed. In mice treated with Dox, the pathological examination clearly showed myocardial damage including cardiomyocyte hypertrophy and myocardial tissue degeneration (Fig. 8B). However, no apparent hematological abnormality (Fig. 8C) and obvious pathological change were observed in mice treated with NPs. The results suggested that NPs had a good biocompatibility in vivo.

Conclusions
The study might be the rst study that a new sensitive drug delivery system based on milk-exosomes for photochemistry therapy against oral squamous cell carcinoma. Dox was modi-ed on the exosomes membrane with imine bond, and then the tumor site-specic release of Dox was stimulated by acid environment. Next, EPT1 and Ce6 were encapsulated in milkexosome. When the NPs were accumulated in tumor site, Ce6 produced plasmonic heat and accelerated ROS generation from EPT1 under NIR irradiation. Therefore, the photochemistry therapy against OSCC was realized based on the pH/light sensitive Exo@Dox-EPT1. In vitro and in vivo testing supported well the biocompatibility of this Exo@Dox-EPT1. Consequently, the Exo@Dox-EPT1 might be a promising prospect to treat OSCC.

Conflicts of interest
The authors declare no competing nancial interest.