New Optical Imaging Reporter-labeled Anaplastic Thyroid Cancer-Derived Extracellular Vesicles as a Platform for In Vivo Tumor Targeting in a Mouse Model

Extracellular vesicles (EVs), originating from multivesicular bodies by invagination of the endosomal membrane, are communication channels between distant cells. They are natural carriers of exogeneous cellular materials and have been exploited as drug delivery carriers in various diseases. Here, we found that tumor cell-derived EVs can be used as efficient targets in tumors by monitoring with an optical reporter system. Anaplastic thyroid cancer (CAL62) cell-derived EVs with Renilla luciferase (Rluc) were used to target CAL62 tumors in a mouse model. Optical imaging revealed that cancer cell-derived EVs (EV-CAL62/Rluc) targeted the original tumor (CAL62) in mice within 30 min after systemic injection. Furthermore, fluorescence imaging revealed that EV-CAL62/Rluc were internalized into CAL62 tumors in the mice. Ex vivo Optical imaging further confirmed the in vivo finding. Here, we successfully monitored the tumor targeting ability of tumor cell-derived EVs by optical imaging. Based on these results, tumor cell-derived EVs are highly effective natural carriers for drug delivery for cancer therapies.

delivery vehicles. EVs are excellent endogenous nanocarriers for exogeneous drug delivery systems. In recent years, several studies showed that drug-loaded EVs improved disease conditions 19 , including studies performed using different cancer models [20][21][22][23] . One way of reducing side effects is to target the delivery of an anti-cancer drug to the tumor.
Recent studies showed that EVs can be visualized in in vivo animal models by using a lipophilic dye 24 , radionuclides 25,26 , magnetic particles 27,28 and bioluminescence reporter system 29 . The labeling process with lipophilic dyes is simple and the labeling is suitable for real-time monitoring of EVs, but lipophilic dyes promotes clumping of EVs 30 ; significant EV damage 29 ; non-specific signals from dye remains in tissues 5,29 . Nuclear imaging could be a good candidate for tracking EVs in both preclinical and clinical studies, a limitation of this technology is the possibility of altering EV characteristics by the labeling procedure 5 . In recent studies, EVs were loaded with MRI contrasts and visualized using MRI 27,28,31 . A large amount of iron oxide-loaded EVs are needed as low sensitivity of MRI technology 5,28 . In preclinical studies bioluminescent imaging (BLI) has an extremely high signal-to-noise ratio, low auto-luminescence in mammalian tissue; low background emission compared to fluorescent-based imaging 32,33 .
Noninvasive or in vivo bioluminescent imaging is particularly advantageous for studying various live cells in small animals 32,34,35 . We recently developed a highly sensitive in vivo visualization method for EVs by employing a new BLI reporter (Renilla luciferase) system 33 . Studies involving real-time visualization of tumor-derived EVs as tumor-targeting agents are urgently needed. The present study was performed to determine whether EVs can preferentially target their parent cell, which may be useful for tumor targeting and drug delivery. Here, we tested a newly developed bioluminescent EV reporter system in an in vivo animal model to monitor the targeting ability of thyroid cancer-derived EVs to original tumors.

Results
Generation of Stable Cell Lines Expressing Luciferase Reporter Genes. Anaplastic thyroid cancer cells (CAL62) transfected with retrovirus containing the Rluc gene or effluc gene were used to generate cells expressing a reporter gene. CAL62 cells transduced with the Rluc gene were named as CAL62/Rluc and those transduced with the Effluc gene were named as CAL62/Effluc. Successful insertion of the Rluc or effluc gene into CAL62 cells was confirmed by BLI, as shown in Fig. 1A-D. As the BLI signal in CAL62/Rluc and CAL62/Effluc cells increased, there was increase in BLI signal in dose-dependent manner. (Cal62/Rluc: R 2 = 0.985; Cal62/Effluc: R 2 = 0.976) and no signals were observed in the parental CAL62 cells. In addition, successful transduction of the Rluc or Effluc gene into cells was confirmed by RT-PCR and western blotting (Figs 1E,F and S1A,B). Taken together, these results indicate that Rluc and Effluc were stably expressed in CAL62 cells. CAL62/Rluc cells were used for the isolation of EVs, and CAL62/Effluc cells were used to prepare a subcutaneous tumor mouse model. Rluc in EV-CAL62/Rluc Showed High Stability in the Serum. The Rluc stability of EV-CAL62/Rluc and recombinant Rluc (r-Rluc) in serum was tested. Rluc activity was decreased marginally EV-CAL62/Rluc, the percentage of serum stability of Rluc in EV-CAL62/Rluc was approximately 98% (1 h), 98% (2 h) and 97% (3 h) but r-Rluc serum stability was significantly decreased (P < 0.001) compare to EV-CAL62/Rluc and it was approximately 77% (1 h), 58% (2 h) and 47% (3 h) (Fig. 3D).

Immunofluorescent (IF) Imaging Confirms the Presence of Rluc in EV-CAL62/Rluc. EV-CAL62/
Rluc or EV-CAL62 were labeled with NIR dye-DiD (a fluorescent lipophilic tracer) and then incubated with CAL62 cells. After 2-hour, IF was performed with Rluc antibody. IF results revealed the presence of Rluc signals in the EV-CAL62/Rluc/DiD incubated CAL62 cells and no Rluc signals present in EV-CAL62 /DiD incubated  In vivo Tumor Targeting Monitored by Bioluminescent Imaging. CAL62/Effluc cells were subcutaneously injected into mice and the mice were randomly divided into two groups (EV-CAL62/Rluc and PBS) at the 6 th week. The tumor was assessed by BLI after intraperitoneal D-luciferin injection 5 days prior to EV administration (Fig. 5A,B). We isolated EVs (EV-CAL62/Rluc) from conditioned medium obtained from CAL62/Rluc cells by ultracentrifugation as described previously 33 . We intravenously administered 50 µg of EV-CAL62/Rluc to mice with or without tumor xenograft, while control mice were administered PBS. Coelenterazine was intravenously injected at 5, 30, 60, and 120 min after EV-Cal62/Rluc (50 µg) or PBS injection and BLI was obtained using an IVIS Lumina III.
BLI showed that no signals were visualized at 5 min post-injection all the three groups and EV-CAL-62/Rluc was visualized in the region of lungs, liver and spleen of mice and in the tumor of CAL62/Effluc tumor bearing mice within 30 min after injection (Fig. 5C,D); quantitative analysis revealed that significant (P < 0.001) BLI signals were observed in the tumor (CAL62/Effluc) region at 30 min and No signals were visualized at the tumor site (CAL62/Effluc) in PBS-injected mice at 30 min by imaging (Fig. 5C,D). Furthermore, BLI imaging at 60 and 120 min revealed that EV-CAL62/Rluc remained in the region of lung in mice without tumor xenograft and in the tumor of CAL62/Effluc tumor bearing mice (Fig. 5C,D). Quantitative analysis revealed significantly (P < 0.001) higher BLI signals in the tumor region (CAL62/Effluc) at 60 and 120 min in EV-CAL62/Rluc-injected CAL62/Effluc tumor bearing mice compared to those in control PBS-injected CAL62/Effluc tumor bearing mice (Fig. 5D). Taken together, these results revealed that EV-CAL62/Rluc are predominantly accumulated in lung of mice without tumor xenograft and accumulated in the tumor of tumor bearing mice.

Ex Vivo Bioluminescent Imaging of EV-CAL62/Rluc Targeting Tumors and Subcellular
Visualization EV-CAL62/Rluc by Fluorescent Imaging. Ex vivo bioluminescent imaging of excised tumor showed bioluminescent signals after incubated with Coelenterazine in EV-CAL62/Rluc injected mice and no signals were observed in PBS control (Fig. 6A). These results further confirmed the in vivo imaging of EV-CAL62/Rluc targeting CAL62/Effluc Tumor. The quantification of bioluminescent imaging showed significant signals (P < 0.001) in EV-CAL62/Rluc injected mice compare to PBS control (Fig. 6B). Tumors were sectioned and subjected to immunofluorescent staining with Rluc-and Effluc-specific antibodies, which revealed that EV-CAL-62/Rluc localized to the CAL62/Effluc tumor. No Rluc signals were observed in PBS control tumors (CAL62/Effluc) (Fig. 6C).

Discussion
EVs are naturally occurring carriers that can target tumors and deliver a wide range of endogenous and exogeneous and small molecules, anti-cancer drugs, and nucleic acids 20,[36][37][38] . Thus, using EVs as drug delivery vehicles shows more potential compared to other nanoparticles that induce high immunogenicity in vivo and are easily eliminated by the immune system 17,39 . Tumor-derived EVs affect tumors in various manners, including effects on the epithelial to mesenchymal transition in nasopharyngeal carcinoma 40 , converting mesenchymal stem cells into cancer-associated fibroblasts 41 , transporting paracrine signaling factors and mRNA to induce angiogenesis in the metastatic microenvironment 42 , and facilitating organotropic metastatic growth 43,44 , among others. EV proteins and miRNA can be used as biomarkers for the early detection of tumors 45,46 . Tumor-derived EVs can be used for cell-free anti-tumor vaccination to induce T-cell activation and T-cell dependent immune responses against tumor cells in vivo 47 . They also have antitumor immunogenicity even without expressing MHC-I molecules, which can provide antigens against various cancer types 48 .
In vivo visualization and monitoring the EVs is an animal model can provide a foundation for developing EV-based therapies. Here, we tested the targeting ability of tumor-derived EVs to its parental tumor in vivo in a mouse model by using a newly developed reporter gene-based optical imaging system for in vivo EV monitoring, which may be useful as a monitoring platform for EV-based delivery vehicles. We used anaplastic thyroid cancer cells (CAL62) to test our hypotheses that CAL62 cancer cell-derived EVs can target CAL62 tumors in an in vivo animal model. We successfully inserted an optical reporter gene (Rluc) in CAL62 cells and isolated EVs with the optical reporter protein (Rluc). We also separately transduced CAL62 cells with the Effluc gene to prepare a subcutaneous tumor mouse model.
First, we isolated EVs from the CAL62/Rluc cell culture media using a conventional ultracentrifugation method. The round morphology, size, positive or negative and Rluc Protein in EV-CAL62/Rluc which is consistent with previous study 33 . Rluc in the EV-CAL62/Rluc showed a strong stability compare to r-Rluc under serum condition and consistent with other study 33 . Furthermore, Immunofluorescent assay confirmed the Rluc present in the EVs compartment. Immunofluorescent and flowcytometry analysis confirmed the EV-CAL62/Rluc are able to internalize into the same cancer cells more effectively regardless of Effluc gene expression in the cancer cells, This results support the hypothesis that cancer cell-derived EVs harbor a specific tropism, based on their cellular origin 17 , which can be exploited to deliver the drugs into cancer cells.
Furthermore, we used a newly developed optical reporter system to track EVs in vivo; the results suggested that most EVs accumulated in the tumor, while previous studies showed that a portion or more than half of tumor-derived EVs accumulated in the same tumor in in vivo 14,17 . Our previous work showed that labeling EVs with lipophilic dye affected their biodistribution in vivo 33 , as EV targeting occurs through integrins present in the EV membrane 49 , which can be blocked by dye labeling on the EV surface membrane and can lead to less accumulation at the tumor in vivo. Further, we visualized the subcellular distribution of EV-CAL-62/Rluc in excised tumors by immunofluorescence. Our ex vivo results further confirmed the in vivo finding and further results showed that most EVs were co-localized with CAL-62/Effluc cells, and previous studies found that cancer cells readily internalize EVs to a greater extent than normal cells 50 .
We confirmed that EVs derived from the tumor targeted the same tumor when injected intravenously into an in vivo mouse model. BLI imaging clearly showed that intravenously injected EV-CAL62/Rluc accumulated at the tumor region of mice at 30 min after injection but EV-CAL62/Rluc injected into nude mice without tumor xenograft by the intravenous route accumulated predominantly in the lung and liver, which is consistent with previous study 33 . Similar to our results (EV-CAL62 targeting CAL62 cells in vivo), a recent study evaluated exosomes derived from tumor cells (SKOV3) injected intravenously into the tail veins of SKOV3 xenograft mice; optical imaging revealed that Cy5.5-labeled SKOV3-Exo was significantly accumulated in tumor sites in vivo 17 . Other studies suggested that tumor-derived EVs are more highly associated with cancer cells than with normal cells, and an acidic pH may be one factor affecting tumor exosome trafficking to the tumor 16 . Another study suggested that tumor-derived EVs have unique protein and lipids compartments resembling those of cells from which they are derived and may interact very uniquely with the same tumor cells 50 . In addition, more SKOV3-derived exosomes were internalized than HEK293-derived exosomes in SKOV3 tumors 17 . Based on these data, tumor-derived EVs are an attractive therapeutic cargo carrier for targeted drug delivery systems. Furthermore, tumor derived EVs can be administered via intracardiac or intraperitoneal routes to enhance tumor targeting ability in the future.
Recently, EVs were considered as new biological drug delivery vehicles. EVs are small and have a negative zeta potential to ensure long circulation 51 and they can escape degradation 52 and can evade the immune system 53 . EVs can carry drugs, DNA, mRNA, miRNA, and proteins 54 . In this study, Rluc proteins inside the EVs were successfully delivered to the target tissue. A recent study of cancer-derived EVs suggested that they can promote cancer progression and form a pre-metastatic niche 54 ; advancements in the engineering of EVs may reduce these effects of cancer-derived EVs. Furthermore, drugs can be effectively delivered to the tumor area using cancer-derived EVs to destroy cancer cells. Cancer-derived EVs are produced in greater numbers than EVs in normal cells, reducing the effort needed to collect EVs.

Conclusion
We successfully visualized EVs derived from CAL-62 targets CAL-62 tumors in vivo using an optical imaging system. These cancer-derived EVs show potential as candidates for drug delivery systems.
In the presence of oxygen, Rluc catalyzes the non-ATP-dependent oxidation of coelenterazine to generate a luminescent signal 57 . Rluc or Effluc activity was determined by BLI using the IVIS Lumina III instrument (In Vivo Imaging System, IVIS Lumina III, PerkinElmer).

Isolation of EVs. Cal62 and Cal62/Rluc cells were cultured with DMEM supplemented with 10%
EV-depleted fetal bovine serum (filtered through a 0.22-µm syringe filter and then centrifuged for 18 h at 100,000 × g at 4 °C) for isolation of EVs. EVs were isolated as described previously 33 . Briefly, the supernatant was centrifuged at 300 × g for 10 min, 1500 × g for 20 min, and finally 2500 × g for 20 min (to remove cells and debris). The supernatant was filtered through a 0.45-µm syringe filter and centrifuged at 100,000 × g for 60 min. The pellet was resuspended and centrifuged at 100,000 × g for 60 min. The final pellet was resuspended in 50-100 µL of PBS and stored at −80 °C. All EVs were used within one week. All ultracentrifugation steps were performed (SW-28 rotor; Ultra-Clear tube) using the Optima TM L-100 XP ultracentrifuge (Beckman Coulter, Brea, CA, USA). All centrifugation steps were conducted at 4 °C.

Transmission Electron Microscopy (TEM).
EVs were imaged by TEM. Briefly, after isolation, EVs were fixed at 4 °C overnight. The fixative contained 2.5% glutaraldehyde in 0.01 M phosphate buffer at pH 7.4 (filtered through 0.22-µm filters) and was washed with PBS. EVs were post-fixed in 1% O s O 4 (Taab Laboratories Equipment Ltd., Reading, UK) for 30 min. EV pellets were washed with distilled water and dehydrated with graded ethanol. EV pellets were negative-stained with 1% uranyl-acetate in 50% ethanol for 30 min and then embedded in Taab 812 (Taab), followed by polymerization at 60 °C overnight and ultra-sectioning for TEM. Ultra-thin sections were examined, and images captured with a HT-7700 transmission electron microscope (Hitachi, Tokyo, Japan) operated at 100 kV. Rluc Stability Assay. EV-CAL-62/Rluc or Free Rluc protein was incubated in 20% FBS in a PBS solution at 37 °C 33 . Stability of Rluc was evaluated by measuring Rluc activity at 0, 6, 12, 18, 24 hours for EV-CAL-62/Rluc. Stability of Free Rluc protein (Creative BioMart) was evaluated by measuring Rluc activity at 0, 1, 2, 3, 4 and 5 hours. The appropriate substrate coelenterazine was added to each well. Rluc activity was determined by BLI as described above.

Rluc Stability Assay Recombinant Rluc Protein (r-Rluc). Stability of Free Rluc protein (Creative
BioMart) was evaluated by measuring Rluc activity at 0, 1, 2 and 3 hours. The appropriate substrate coelenterazine was added to each well. Rluc activity was determined by BLI as described above.
Internalization Assay. The DiD-labeled EVs were incubated together with CAL62 parental cell lines for 1 h at 37 °C under 5% CO 2 and were then fixed in methanol. Coverslips were then mounted using the VECTASHIELD anti-fade mounting medium (Vector Laboratories, Burlingame, CA, USA) and sealed. The cellular internalization of EVs was observed using an LSM 780 laser scanning microscope (Carl Zeiss). Total number of DiD (EV-CAL62/ Rluc/DiD) positive cells were counted by two independent observers from CAL62 and CAL62/Effluc cells.
Flow cytometry. Flow Cytometry Analysis was performed as described 59 , Briefly, The DiD-labeled EVs were incubated together with CAL62 or CAL62/Effluc cells for 1-hour at 37 °C under 5% CO 2 . Cells were washed with PBS (trice) and centrifuged for 3 min at 3500 g to remove free EVs. The supernatant was discarded, and the cells were resuspended in 1 mL of PBS for flow cytometry analysis on a BD Biosciences Aria III (BD Biosciences, San Jose, CA, USA).

Animals.
Five-week-old female BALB/c nude mice were purchased from Japan SLC, Inc. (Shizuoka, Japan).
All procedures were reviewed and approved by Kyungpook National University (KNU-2012-43) Animal Care and Use Committee and performed in accordance with the Guiding Principles for the Care and Use of Laboratory Animals.
Establishment of a CAL62/effluc Subcutaneous Animal Model. Nude mice were subcutaneously implanted with CAL62/effluc (5 × 10 6 cells) on the right lower flank regions and allowed to grow for 6 weeks. The mice were anesthetized with 2.5% isoflurane (Merial, Lyon, France) and bioluminescent images were captured using the IVIS Lumina III imaging system after intraperitoneal injection of 100 µL of D-luciferin (3 mg/mouse; Caliper). BLI was subsequently assessed up to the 6 th week and mice were dived into two groups (EV-Cal62/Rluc or PBS).
In vivo Imaging of EV-Cal62/Rluc Targeting Tumors. The mice were anesthetized with 2.5% isoflurane and bioluminescent images were captured using the IVIS III imaging system with an appropriate amount of coelenterazine administered intravenously after 5, 30, 60 and 120 min after EV-Cal62/Rluc (50 µg) (n = 3) injection to naïve mice, EV-Cal62/Rluc (50 µg) (n = 3) or PBS (n = 3) injection to tumor bearing mice through same route. Animals were sacrificed after 120 min and tumors were immediately excised and incubated with coelenterazine. Bioluminescent signals were captured and quantified as described above.
Statistical analysis. All data are expressed as the means ± standard deviation (SD). Two groups of data were statistically analyzed by t-test using GraphPad Prism7 software version 7.04 (GraphPad Software, Inc., La Jolla, CA, USA). A P value less than 0.05 was considered statistically significant.

Data Availability
The authors declare that all the relevant data supporting the findings of this study are available within the article or from the corresponding author upon request.