Glioblastoma-Derived Exosomes as Nanopharmaceutics for Improved Glioma Treatment

The use of cancer-derived exosomes has been studied in several cancer types, but the cancer-targeting efficacy of glioma-derived exosomes has not been investigated in depth for malignant glioblastoma (GBM) cells. In this study, exosomes were derived from U87MG human glioblastoma cells, and selumetinib, a new anticancer drug, was loaded into the exosomes. We observed the tropism of GBM-derived exosomes in vitro and in vivo. We found that the tropism of GBM-derived exosomes is in contrast to the behavior of non-exosome-enveloped drugs and non-GBM-specific exosomes in vitro and in vivo in an animal GBM model. We found that the tropism exhibited by GBM-derived exosomes can be utilized to shuttle selumetinib, with no specific targeting moiety, to GBM tumor sites. Therefore, our findings indicated that GBM-derived exosomes loaded with selumetinib had a specific antitumor effect on U87MG cells and were non-toxic to normal brain cells. These exosomes offer improved therapeutic prospects for glioblastoma therapy.


Introduction
Exosomes are nano-sized extracellular vesicles with a bilayer membrane [1,2]. They play a vital role in multiple normal physiological and pathological processes, acting as communication mediators between cells [3,4]. Exosomes can act as nanocarriers, releasing substances such as proteins and ribonucleic acids (RNAs) into target cells. They can deliver both hydrophilic and hydrophobic molecules and can efficiently target cancer sites [5][6][7]. Among the nanocarriers proposed as valuable for cancer therapy and diagnosis, exosomes have recently received the most attention as promising drug delivery vesicles that can overcome the shortcomings of artificial nanocarriers. In contrast to artificial nanocarriers, exosomes elicit relatively low immune recognition and have low toxicity because exosomes are membrane vesicles that are naturally released from cells and thus have high biocompatibility [8][9][10].
Nanomedicine has made many important contributions to cancer therapy and diagnosis by improving the loading capacity, biodistribution, and target accumulation of therapeutic molecules [11]. Advances in the field of image-guided cancer treatment have been applied to exosomes through the encapsulation of therapeutic molecules and the
To track the exosomes, we labeled the exosomes with fluorescent dye using Vybrant Multicolor Cell-Labeling Kits (V22889, Thermo Fisher Scientific) and an Exosome Spin Column (MW3000) (4484449, Invitrogen, Carlsbad, CA, USA) was used to remove the free fluorescent dye. To track the exosomes, we labeled the exosomes with fluorescent dye using Vybrant Multicolor Cell-Labeling Kits (V22889, Thermo Fisher Scientific) and an Exosome Spin Column (MW3000) (4484449, Invitrogen, Carlsbad, CA, USA) was used to remove the free fluorescent dye.

Size Distribution Analysis
To quantify the purified exosome, the exosome solution (5 mg/mL sterilized PBS, pH 7.4, 10010-023, Gibco) was filtered through a cellulose syringe filter with a 0.20 µM pore size (13CP020AS, ADVANTEC, Dublin, CA, USA). The diameter and concentration of the exosomes were determined with an NTA system (Nanosight NS 300, Malvern Panalytical, Malvern, UK), and imaging parameters were as follows: camera level = 14; screen gain = 3.0; detection threshold = 3; the number of frames = 1498. All size distribution analysis was performed in triplicate for independent batches.

Size Distribution Analysis
To quantify the purified exosome, the exosome solution (5 mg/mL sterilized PBS, pH 7.4, 10010-023, Gibco) was filtered through a cellulose syringe filter with a 0.20 µM pore size (13CP020AS, ADVANTEC, Dublin, CA, USA). The diameter and concentration of the exosomes were determined with an NTA system (Nanosight NS 300, Malvern Panalytical, Malvern, UK), and imaging parameters were as follows: camera level = 14; screen gain = 3.0; detection threshold = 3; the number of frames = 1498. All size distribution analysis was performed in triplicate for independent batches.

Transmission Electron Microscope (TEM)
A total of 10 µL exosome solution (5 µg/mL, in PBS) was put on a copper grid for 10 min. Then, the same grid was put on a uranyl acetate solution (2% w/v, 10 µL) to stain the sample for 10 min. The copper grid was photographed using transmission electron microscopy (TEM), (JEM-2200FS, JEOL, Tokyo, Japan) at 200 keV.

Western Blot Analysis
Protein expression was assessed using Western blot analysis. Protein was extracted in radioimmunoprecipitation assay lysis buffer (Millipore) containing a protease and phosphatase inhibitor cocktail (Roche Diagnostics, Basel, Switzerland). Protein lysates (35 µg exosome) were separated using electrophoresis in sodium dodecyl sulfate polyacrylamide gels, and the separated proteins were electrotransferred onto PVDF membranes. The membranes were blocked in a 3% bovine serum albumin (BSA) Tris-buffered saline solution containing 0.05% Tween 20 (TBS-T) for 1 h at room temperature and then incubated with the following diluted primary antibodies: CD9 recombinant rabbit m antibody (

Stability of Selumetinib-Loaded Exosomes
The stability of the U87 or A549-Selu exo samples (2.5 µg/µL in sterilized PBS) was checked by monitoring size distribution changes with a Nanoparticle Tracking Analysis (NTA) system (camera level: 14, screen gain: 3.0, detection threshold: 3) at a physiological pH of 7.4 for 7 days. The stability assessment was performed in triplicate for independent batches.

Animals
For all experiments, a male Balb/c-nude mouse (Nara Biotech, Seoul, Korea) weighing 16-18 g was used. Animals were housed in controlled environmental conditions with a light-dark cycle of 12 h at an ambient temperature of 23 ± 1 • C and relative humidity of 50 ± 10% and had free access to food and water. All animal experiments were approved and performed according to the guidelines of Kyungpook National University Animal Research Center (IACUC) (No. KNU-2021-0225; Daegu, Korea, 24 December 2021).

U87MG Xenograft Model
For the generation of a tumor xenograft, U87 MG cells (3 × 10 6 cells/100 µL EMEM medium, serum-free) were subcutaneously injected into the upper-left flank region of the male Balb/c-nude mice. Then, the mice recovered on the 37 • C warm pad. When the tumor reached a mean size of about 100 mm 3 , mice were randomized into each experimental group: saline (n = 6), U87-Selu exo (n = 6), and A549-Selu-exo (n = 5).

In Vivo Exo and Selu-Exo Targeting to the Parental Tumor
To observe the in vivo targeting effect of exosomes on tumor tissue, DiI-labeled exosomes (DiI-U87 exo, DiI-A549 exo, DiI-U87-Selu exo and DiI-A549-Selu exo; 8-9 µg/µL exosome) were injected intravenously into U87 xenograft mice tail vein with 50 mg exosome/kg dose (n = 5 for each group) ( Figure 3). A total of 24 h after injection, the mice were sacrificed by exsanguination from the vena cava for tumor harvesting. The harvested tumors were embedded in optical cutting temperature (OCT) compound. The OCT compound sections were then cut into thicknesses of 8 µm. After the specimens were washed 3 times with TBS (100 mL), the slides were covered with VECTASHIELD mounting medium with DAPI (Vector Laboratories; Catalog No. H-1200). Images of the slide sections were captured using a Nikon fluorescence microscope with the software NIS-Elements BR 5.11 (Nikon). The relative fluorescence intensity was analyzed by ImageJ software (Version 1.50i; US National Institutes of Health, Bethesda, MD, USA).

In Vivo Anticancer Effect of U87-Selu Exo and A549-Selu Exo
U87-Selu exo (n = 6), A549-Selu exo (n = 6), or saline (n = 5) were intravenously injected into U87MG xenograft mice tail vein once every 2 days for 10 days (Figure 4). After injection, the mice recovered on the 37 •C warm pad. The injected volume was 100 µL for each of the three groups, and the amount of selumetinib loaded was 1 mg for U87-Selu exo and A549-Selu exo (8-9 µg/µL exosome). The tumor size was checked using a digital caliper every 2 days, and the tumor volume was calculated with the following formula: tumor volume = length × (width) 2 × 0.5. The anticancer effect was expressed as a percentage of tumor growth inhibition (% TGI) calculated using the following equation: 100 − (T/C × 100), where T is the mean relative tumor volume (RTV) of treated tumors and C is the mean RTV of the saline control group at the time of sacrifice. RTV = Vx/V1, where Vx is the volume in mm 3 at a given time and V1 is the start of treatment. Mean TGI (%) and standard deviation were calculated for each group.

In Vivo Anticancer Effect of U87-Selu Exo and A549-Selu Exo
U87-Selu exo (n = 6), A549-Selu exo (n = 6), or saline (n = 5) were intravenously injected into U87MG xenograft mice tail vein once every 2 days for 10 days (Figure 4). After injection, the mice recovered on the 37 ˚C warm pad. The injected volume was 100 µL for each of the three groups, and the amount of selumetinib loaded was 1 mg for U87-Selu exo and A549-Selu exo (8-9 µg/µL exosome). The tumor size was checked using a digital caliper every 2 days, and the tumor volume was calculated with the following formula: tumor volume = length × (width) 2 × 0.5. The anticancer effect was expressed as a percentage of tumor growth inhibition (% TGI) calculated using the following equation: 100 − (T/C × 100), where T is the mean relative tumor volume (RTV) of treated tumors and C is the mean RTV of the saline control group at the time of sacrifice. RTV = Vx/V1, where Vx is the volume in mm 3 at a given time and V1 is the start of treatment. Mean TGI (%) and standard deviation were calculated for each group.

Histological Analysis for Tumor Tissue
After five intravenous injections, the mice were dissected and perfused. The tumor was harvested from each mouse, and the harvested tumors were fixed with 4% PFA for 3 days. Fixed tumors were treated with an alcohol concentration gradient (50, 70, 95, and 100%), xylene (Junsei Chemical, Tokyo, Japan), and paraffin for 30 min, respectively. The paraffin-embedded tumor tissues were sectioned (5 µm) and subjected to immunofluo-

Histological Analysis for Tumor Tissue
After five intravenous injections, the mice were dissected and perfused. The tumor was harvested from each mouse, and the harvested tumors were fixed with 4% PFA for 3 days. Fixed tumors were treated with an alcohol concentration gradient (50, 70, 95, and 100%), xylene (Junsei Chemical, Tokyo, Japan), and paraffin for 30 min, respectively. The paraffin-embedded tumor tissues were sectioned (5 µm) and subjected to immunofluorescence (IF) and immnunohistochemical (IHC) staining by treatment with xylene and alcohol concentration gradients (100, 95, 70, and 50%) in an oven at 65 • C for 1 h and 10 min, respectively.
For IHC staining, endogenous peroxidase activity was inactivated by incubation in 0.3% H 2 O 2 in methanol (Sigma-Aldrich, St. Louis, MO, USA) for 10 min. The sections were then rinsed with 0.1 M TBS (pH 7.4) and boiled in citrate buffer (pH 6.0) containing 0.03% Tween-20 for 4 min. Finally, the sections were incubated with a blocking solution (5% normal goat serum (NGS) and BSA in TBS) at 25 • C for 1 h, and indirect immunization occurred with an antibody to anti-Ki-67 (diluted 1:100) for 1 h applied to the histochemistry. For negative controls, the primary antibody was omitted, and slides were incubated with a blocking solution. Sections were then incubated with horseradish peroxidase (HRP)- conjugated anti-rabbit IgG for 1 h at 25 • C, stained with VECTOR1 NovaRED (Vector Laboratories, Inc.), and counterstained with hematoxylin (BBC Biochemical, Mount Vernon, WA, USA). Sections were dehydrated, cleaned, and mounted with Permount (Fisher, Fair Lawn, NJ, USA). Images were captured with a fluorescence microscope (ECLIPSE Ti, Nikon), and the fluorescence intensity of immunostaining was quantified using ImageJ.
For IF staining, specimens were blocked with TBS supplemented with 5% NGS and 5% BSA for 2 h and incubated overnight with primary antibody (anti-cleaved-caspase-3, diluted 1:100) in blocking solution (5% NGS and BSA) at 4 • C. Sections were then washed three times in TBS and incubated for 1 h in the presence of Alexa Fluor conjugated IgG labeled secondary antibody (Invitrogen, Carlsbad, CA, USA). Sections were washed and mounted with Vectashield mounting medium containing DAPI (Vec-tashield H-1500; Vector Laboratories, Inc., Burlingame, CA, USA). Fluorescence microscopy (ECLIPSE Ti, Nikon) was used to capture images, and ImageJ was used to quantify the fluorescence intensity of immunostaining.

In Vivo Toxicity of U87-Selu Exo and A549-Selu Exo
To evaluate in vivo toxicity of U87-Selu exo and A549-Selu exo, the bodyweight of the mice (n = 5 for each group) used for the assessment of the anticancer effect was measured using an electronic scale at the end of the dosage.
Additionally, glutamate oxaloacetate transaminase (GOT) and glutamic-pyruvic transaminase (GPT) levels were measured to evaluate the liver function of the mice. For the GOT and GPT measurements, we extracted blood (500 µL) from the mice's abdominal vena cava. The extracted blood was incubated at room temperature for 2 h and then centrifuged at 2000× g for 15 min. Finally, the supernatant serum (500 µL) was used for GOT and GPT measurements.
To evaluate the histological toxicity, the liver and kidney were collected and fixed with 4% paraformaldehyde buffer over 3 days. After washing and dehydrating the tissues, sections were made by paraffin embedding and cut into 5 µm slices. Finally, the sections were stained with hematoxylin and eosin (H&E) following the protocol using Eosine Y Alcoholic (BBC Biochemical Corp., Mount Vernon, WA, USA; catalog No. 3605). Some sections were stained using Picro Sirius Red stain kits (Abcam, Cambridge, UK; Catalog No. ab150681).

In Vitro Anticancer Effect of U87-Selu Exo
U87MG cells were seeded onto a 60 mm 2 cell culture plate (8 × 10 5 cells/plate, SPL Life Sciences). After 12, 24, or 48 h, they were treated with 100 µg/µL of U87-Selu exo and selumetinib. The medium was discarded, and the cells were washed with DPBS. The extracted protein lysates (10 µg/µL) that were used to determine protein concentration by a BCA assay were separated via electrophoresis on sodium dodecyl sulfate-polyacrylamide gels, and the separated proteins were electrotransferred onto PVDF membranes. Membranes were incubated with the following diluted primary antibodies: 3 h at room temperature. The membrane was then incubated with horseradish-conjugated secondary antibody (1:2000, Cell Signaling Technologies) for 2 h at room temperature. After washing with TBS-T, immunoreactive bands were visualized using the Chemiluminescence Western Imaging System (Supernova-Q1800TM; Centronics, Daejeon, Korea). All Western blotting was performed in triplicate.

Flow Cytometry Analysis
U87MG cells were seeded on a 60 mm 3 cell culture dish (8 × 10 5 cells/dish, SPL Life Sciences). After 24 h, the cells were treated with 200 µg/µL of native U87MG cell-derived exosome (exo), U87-Selu exo and selumetinib for 24 h. The medium was discarded, and the cells were washed with DPBS. Trypsin-2,2 ,2",2 "-(Ethane-1,2-diyldinitrilo)tetraacetic acid (EDTA) was added, and the cells were collected in DPBS + 2% FBS. The cells were centrifuged at 4000 rpm for 3 min, washed, and resuspended in DPBS + 2% FBS. The cells were fixed with 100% ethanol and incubated overnight at 4 • C. The next day, they were centrifuged, the ethanol was discarded, and the cells were washed with DPBS + 2% FBS. The pellet was resuspended in 1.12% sodium citrate buffer (pH 8.4) with 50 µg/mL RNase and incubated at 37 • C for 30 min. Finally, propidium iodide (PI) working solution (50 µg/mL) was added and incubated for 30 min at room temperature. The PI (50 µg/mL, propidium iodide, Invitrogen, Catalog No. P21493)-stained cells were analyzed using a CytoFLEX Flow Cytometer (Beckman Coulter Life Science, Brea, CA, USA) to determine the relative DNA content based on the red fluorescence. The experiment was performed in triplicate.

Statistical Analysis
Statistical analysis was done using one-way ANOVA followed by the Dunnett comparison test also using two-way ANOVA followed by Bonferroni post-tests to compare replicate means by row. Values of p < 0.05, p < 0.01 and p < 0.001 are represented by *, ** and *** vs. nontreated control and #, ##, and ### vs. treated control, respectively. GraphPad Prism (version 5.02; GraphPad Prism Software Inc., SanDiego, CA, USA) and Excel programs were used to draw figures and graphs.

Exosome Characterization
The mean size of exosomes from the NTA and TEM measurements was 170.0 ± 3.3 nm for U87 exosomes, 150.6 ± 2.5 nm for A549 exosomes, 134.0 ± 0.8 nm for U87-Selu exosomes, and 96.8 ± 4.6 nm for A549-Selu exosomes (Figure 5a,b). The average size and the distribution slightly differed after selumetinib loading into each native exosome, but the morphology was maintained as a round shape. The Western blot data also showed that the exosome-specific protein markers CD9, CD63, and CD81 (endosome-specific tetraspanins) and TSG101 (an exosome biogenesis protein) were not changed by loading selumetinib into the exosomes (Figure 5c). These data indicated that the exosomes were successfully isolated from the cell culture media. The selumetinib loading efficiency (%) was calculated using a UV-vis spectrometer to be 80-90% for the U87 exo and A549 exo. U87-Selu exo showed size stability for up to 7 days at pH 7.4 (Supplementary Materials, Figure S3). raspanins) and TSG101 (an exosome biogenesis protein) were not changed by loading lumetinib into the exosomes (Figure 5c). These data indicated that the exosomes were s cessfully isolated from the cell culture media. The selumetinib loading efficiency (%) w calculated using a UV-vis spectrometer to be 80-90% for the U87 exo and A549 exo. U Selu exo showed size stability for up to 7 days at pH 7.4 (Supplementary Materials, Figu S3). showed that all exosomes were round in shape. (c) Immunoblotting resu revealed that U87-derived exosome-specific exosomal markers (CD proteins and TSG 101) w highly expressed in U87-derived exosomes, but not in A549-derived exosomes. A non-exosom marker (Calnexin) was used as reference, n = 3.

U87MG Exosome Targeting In Vitro and In Vivo
We used the fluorescence dye (DiI)-labelled exosomes (DiI-U87 exo and DiI-A5 exo) to observe the targeting effect on bioimaging. As a negative control, non-treated c stained with DAPI was used for cell localization. In vitro exosome-targeting experime revealed that DiI-U87 exo showed higher fluorescence intensity than DiI-A549 exo, su gesting 1.4-fold more uptake of DiI-U87 exo than DiI-A549 exo by U87MG cells (Figu  6a,c). As a negative control, DAPI staining was performed without the fluorescent d DiI. DiI-U87-Selu exo, which was loaded with selumetinib, also showed higher fluor cence intensity than DiI-A549-Selu exo, suggesting 1.8-fold more uptake of DiI-U87-S exo than of DiI-A549-Selu exo by U87MG cells (Figure 6b,d). Additionally, to quanti tively compare the uptake of exosomes and selumetinib-loaded exosomes, we observ another targeting experiment with 96-well cell culture plates. As a result, the uptake U87 exo was 1.5-fold higher than A549 exo, and U87-Selu exo was 2.3-fold higher th A549-Selu exo (Figure 6e,f). These results, therefore, indicated that U87-exo is more e ciently taken up by their parent U87MG cell lines than by A549-exo. These in vitro resu also demonstrated that the loading of selumetinib into exosomes did not much change targeting efficiency of the exosomes to their parent cell lines. showed that all exosomes were round in shape. (c) Immunoblotting results revealed that U87-derived exosome-specific exosomal markers (CD proteins and TSG 101) were highly expressed in U87-derived exosomes, but not in A549-derived exosomes. A non-exosomal marker (Calnexin) was used as reference, n = 3.
In order to investigate whether exosomes were targeting their parent cells in vivo, DiI-U87 exo and DiI-A549 exo were intravenously injected into U87MG cell-implanted GBM xenograft mice. Non-treated negative controls did not represent any fluorescence without DAPI. DiI-U87 exo showed 6.9-folds much higher fluorescence intensity than DiI-A549 exo, suggesting more uptake of DiI-U87 exo than of DiI-A549 exo by GBM tumors (Figure 7a,b). In the case of selumetinib-loaded exosomes, DiI-U87-Selu exo showed 3.5-fold higher fluorescence intensity than DiI-A549-Selu exo, suggesting more uptake of DiI-U87-Selu exo than of DiI-A549-Selu exo by GBM tumors (Figure 7c,d). Compared with the in vitro results, the in vivo results revealed that the difference in uptake was larger between DiI-U87 exo and DiI-A549 exo and between DiI-U87-Selu exo and DiI-A549-Selu exo. Because the in vivo case is more realistic than the in vitro case for the evaluation of the targeting effect to parent cancer cells, the results from the in vitro and in vivo experiments strongly suggest that U87-derived exosomes have higher targeting efficacy than A549-derived exosomes to GBM tumors from parent U87MG cells. This high targeting efficacy of U87-derived exosomes was not changed by loading an anticancer drug, even in vivo. Although the in vivo ability of cancer cell-derived exosomes to return to the parent cancer cells is often called the "homing effect" [19,20], it is not appropriate to describe our in vivo result as a homing effect due to the ectopic tumor model. To solve this limitation, further study is warranted to investigate the possible homing effect using an appropriate tumor model. the targeting effect to parent cancer cells, the results from the in vitro and in vivo experiments strongly suggest that U87-derived exosomes have higher targeting efficacy than A549-derived exosomes to GBM tumors from parent U87MG cells. This high targeting efficacy of U87-derived exosomes was not changed by loading an anticancer drug, even in vivo. Although the in vivo ability of cancer cell-derived exosomes to return to the parent cancer cells is often called the "homing effect" [19,20], it is not appropriate to describe our in vivo result as a homing effect due to the ectopic tumor model. To solve this limitation, further study is warranted to investigate the possible homing effect using an appropriate tumor model.

Cytotoxicity of U87 Exo, Selumetinib and U87-Selu Exo
To investigate the cytotoxicity of the U87 exo, we conducted cell viability experiments using U87MG cells, A549 cells, and C8-D1A cells. Native U87 exo represented by the values of IC50 were none in all cell lines (Figure 8a). This result suggests that U87 exo, which were cancer-derived exosomes, did not have cytotoxic effects on either normal or cancer cells. However, selumetinib inhibited cellular proliferation in the same condition, representing values of IC50 = 69.08 ± 18.9 µg/mL for C8-D1A cells, 112.1 ± 17.75 µg/mL for U87MG cells, and 71.21 ± 21.24 µg/mL for A549 cells (Figure 8b). Even though U87-Selu exo has a relatively similar or low growth inhibition effect to native selumetinib for two cancer cells, it showed proliferation inhibition effects for cancer cells (IC50 value = 110.2 ± 19.11 µg/mL for U87MG and 130.3 ± 14.17 µg/mL for A549) without normal cell toxicity (IC50 value = none for C8-D1A) (Figure 8c). Additionally, the cytotoxic effect of U87-Selu exo was stronger in U87MG cells than in A549 cells and may result from the high tumor targeting of U87-Selu exo to its parental cell (U87MG). These results, therefore, suggest that U87-Selu exo had higher anticancer efficiency against U87MG cells than A549 cells.
To investigate the cytotoxicity of the U87 exo, we conducted cell viability experiments using U87MG cells, A549 cells, and C8-D1A cells. Native U87 exo represented by the values of IC50 were none in all cell lines (Figure 8a). This result suggests that U87 exo, which were cancer-derived exosomes, did not have cytotoxic effects on either normal or cancer cells. However, selumetinib inhibited cellular proliferation in the same condition, representing values of IC50 = 69.08 ± 18.9 µg/mL for C8-D1A cells, 112.1 ± 17.75 µg/mL for U87MG cells, and 71.21 ± 21.24 µg/mL for A549 cells (Figure 8b). Even though U87-Selu exo has a relatively similar or low growth inhibition effect to native selumetinib for two cancer cells, it showed proliferation inhibition effects for cancer cells (IC50 value = 110.2 ± 19.11 µg/mL for U87MG and 130.3 ± 14.17 µg/mL for A549) without normal cell toxicity (IC50 value = none for C8-D1A) (Figure 8c). Additionally, the cytotoxic effect of U87-Selu exo was stronger in U87MG cells than in A549 cells and may result from the high tumor targeting of U87-Selu exo to its parental cell (U87MG). These results, therefore, suggest that U87-Selu exo had higher anticancer efficiency against U87MG cells than A549 cells.

In Vivo Biodistribution of U87-Selu Exo
To investigate the in vivo biodistribution of U87-Selu exo, nude mice bearing U87MG tumors were intravenously injected with DiD-labeled U87-Selu exo (DiD-U87-Selu exo). After 24 h, the mice were sacrificed, and the tumor, liver, kidney, lung, heart, spleen, and plasma were harvested. The organ images were taken using an IVIS fluorescence imaging device. The tumors showed the strongest fluorescence signal, suggesting that they had the highest uptake of U87-Selu exo (Figure 9). A fluorescent signal was also found in the liver but not in other organs. The liver uptake of cancer-derived exosomes has been reported in previous studies [38][39][40]. These previous biodistribution studies of intravenously injected exosomes showed the rapid clearance of exosomes by the liver using a reticuloendothelial system. A fluorescent signal was found in plasma samples, indicating a long

In Vivo Biodistribution of U87-Selu Exo
To investigate the in vivo biodistribution of U87-Selu exo, nude mice bearing U87MG tumors were intravenously injected with DiD-labeled U87-Selu exo (DiD-U87-Selu exo). After 24 h, the mice were sacrificed, and the tumor, liver, kidney, lung, heart, spleen, and plasma were harvested. The organ images were taken using an IVIS fluorescence imaging device. The tumors showed the strongest fluorescence signal, suggesting that they had the highest uptake of U87-Selu exo (Figure 9). A fluorescent signal was also found in the liver but not in other organs. The liver uptake of cancer-derived exosomes has been reported in previous studies [38][39][40]. These previous biodistribution studies of intravenously injected exosomes showed the rapid clearance of exosomes by the liver using a reticuloendothelial system. A fluorescent signal was found in plasma samples, indicating a long circulation of U87-Selu exo in the bloodstream. Therefore, the in vivo biodistribution results demonstrated targeted delivery of U87-Selu exo to tumors due to the targeting effect on parent tumor cells. circulation of U87-Selu exo in the bloodstream. Therefore, the in vivo biodistribution results demonstrated targeted delivery of U87-Selu exo to tumors due to the targeting effect on parent tumor cells. Figure 9. In vivo biodistribution of U87-Selu exo. A strong fluorescence signal was found in the tumor and liver, but not in other organs; # means independent mouse number. n = 3.

Liver and Kidney Toxicity of Selu-Exo (U87-Selu Exo and A549-Selu Exo) In Vivo
Due to the liver uptake of U87-Selu exo, the possible liver toxicity of U87-Selu exo and A549-Selu exo was examined. Hematoxylin and eosin-stained and Sirius red-stained images revealed no pathological changes in the kidney or liver of U87-Selu exo-or A549-Selu exo-treated mice (Figure 10a). Additionally, no change in body weight was observed after U87-Selu exo and A549-Selu exo injection in mice, and the GOT and GPT scores, which reflect liver function, were similar in control mice and saline-injected mice ( Figure  10b-d).

Liver and Kidney Toxicity of Selu-Exo (U87-Selu Exo and A549-Selu Exo) In Vivo
Due to the liver uptake of U87-Selu exo, the possible liver toxicity of U87-Selu exo and A549-Selu exo was examined. Hematoxylin and eosin-stained and Sirius red-stained images revealed no pathological changes in the kidney or liver of U87-Selu exo-or A549-Selu exo-treated mice (Figure 10a). Additionally, no change in body weight was observed after U87-Selu exo and A549-Selu exo injection in mice, and the GOT and GPT scores, which reflect liver function, were similar in control mice and saline-injected mice (Figure 10b-d).
Pharmaceutics 2022, 14,1002 14 of 20 circulation of U87-Selu exo in the bloodstream. Therefore, the in vivo biodistribution results demonstrated targeted delivery of U87-Selu exo to tumors due to the targeting effect on parent tumor cells. Figure 9. In vivo biodistribution of U87-Selu exo. A strong fluorescence signal was found in the tumor and liver, but not in other organs; # means independent mouse number. n = 3.

Liver and Kidney Toxicity of Selu-Exo (U87-Selu Exo and A549-Selu Exo) In Vivo
Due to the liver uptake of U87-Selu exo, the possible liver toxicity of U87-Selu exo and A549-Selu exo was examined. Hematoxylin and eosin-stained and Sirius red-stained images revealed no pathological changes in the kidney or liver of U87-Selu exo-or A549-Selu exo-treated mice (Figure 10a). Additionally, no change in body weight was observed after U87-Selu exo and A549-Selu exo injection in mice, and the GOT and GPT scores, which reflect liver function, were similar in control mice and saline-injected mice ( Figure  10b-d).

In Vivo Anticancer Effects of U87-Selu Exo and A549-Selu Exo
To evaluate the anticancer effect of two selumetinib-loaded exosomes (U87-Selu exo and A549-Selu exo), tumor volume and TGI were measured (Figure 11a,b). The tumor volume increased continuously up to 10 days in the saline group. In the A549-Selu exotreated group, the tumor volume was increased, but the volume increase leveled off after 8 days. In the U87-Selu exo-treated group, the tumor volume was slightly increased, then decreased after 4 days. The biggest anticancer effect was produced by U87-Selu exo. However, it should be noted that the small number of replicates (n = 5) is a possible study limitation to confirm the measured tumor volumes. The TGI results were in good agreement with tumor volume measurements. TGI was 31.2% for saline, 55.9% for A549-Selu exo, and 99.8% for U87-Selu exo. Cleaved caspase-3, a tumor apoptosis marker, indicated that the U87-Selu exo-treated group exhibited the highest anticancer effect (Figure 11c). ki-67, a tumor cell growth marker, also showed that tumor growth was minimal in the U87-Selu exo-treated group (Figure 11c). Taken together, these results indicate that the U87-Selu exo-treated group experienced the strongest anticancer effects in a GBM xenograft model, suggesting that the high tumor-targeting effect of U87-derived exosomes plays an important role in cancer therapy.

In Vivo Anticancer Effects of U87-Selu Exo and A549-Selu Exo
To evaluate the anticancer effect of two selumetinib-loaded exosomes (U87-Selu exo and A549-Selu exo), tumor volume and TGI were measured (Figure 11a,b). The tumor volume increased continuously up to 10 days in the saline group. In the A549-Selu exotreated group, the tumor volume was increased, but the volume increase leveled off after 8 days. In the U87-Selu exo-treated group, the tumor volume was slightly increased, then decreased after 4 days. The biggest anticancer effect was produced by U87-Selu exo. However, it should be noted that the small number of replicates (n = 5) is a possible study limitation to confirm the measured tumor volumes. The TGI results were in good agreement with tumor volume measurements. TGI was 31.2% for saline, 55.9% for A549-Selu exo, and 99.8% for U87-Selu exo. Cleaved caspase-3, a tumor apoptosis marker, indicated that the U87-Selu exo-treated group exhibited the highest anticancer effect (Figure 11c). ki-67, a tumor cell growth marker, also showed that tumor growth was minimal in the U87-Selu exo-treated group (Figure 11c). Taken together, these results indicate that the U87-Selu exo-treated group experienced the strongest anticancer effects in a GBM xenograft model, suggesting that the high tumor-targeting effect of U87-derived exosomes plays an important role in cancer therapy.  Several studies have reported quantitative analyses to get better insights into the tumor microenvironment, including proteomics [41,42]. In this study, to evaluate its anticancer effect, we performed a Western blot analysis for apoptosis factors and observed the effect of U87-Selu exo on apoptosis (Supplemental Information, Figure S4). The levels of the cleaved form of Poly (ADP-ribose) polymerase (PARP) were increased by about 5.1-fold and 7.5-fold, respectively, compared with control, when U87MG cells were treated with U87-Selu exo and selu for 48 h (Figure 12a). The level of Bcl2 was decreased by about 0.8-fold compared with control when U87MG cells were treated with U87-Selu exo and selu for 48 h (Figure 12a). Because PARP and Bcl2 are useful markers of apoptosis [43,44], our results indicated that apoptosis occurred when the cells were treated with U87-Selu exo and selumetinib. By measuring the levels of p27, PCNA, cyclin D1, and p-p38, we confirmed that the proliferation of cancer was inhibited (Figure 12b-e). In the control group, the expression of cyclin D1 was downregulated at 48 h due to serum-free media starvation.

Apoptosis of U87-Selu Exo to U87MG Cells
Several studies have reported quantitative analyses to get better insights into the tumor microenvironment, including proteomics [41,42]. In this study, to evaluate its anticancer effect, we performed a Western blot analysis for apoptosis factors and observed the effect of U87-Selu exo on apoptosis (Supplemental Information, Figure S4). The levels of the cleaved form of Poly (ADP-ribose) polymerase (PARP) were increased by about 5.1fold and 7.5-fold, respectively, compared with control, when U87MG cells were treated with U87-Selu exo and selu for 48 h (Figure 12a). The level of Bcl2 was decreased by about 0.8-fold compared with control when U87MG cells were treated with U87-Selu exo and selu for 48 h (Figure 12a). Because PARP and Bcl2 are useful markers of apoptosis [43,44], our results indicated that apoptosis occurred when the cells were treated with U87-Selu exo and selumetinib. By measuring the levels of p27, PCNA, cyclin D1, and p-p38, we confirmed that the proliferation of cancer was inhibited (Figure 12b-e). In the control group, the expression of cyclin D1 was downregulated at 48 h due to serum-free media starvation. When cells were treated with U87-Selu exo and selumetinib for 24 h, the levels of p27 and p-p38 were increased by about 4.2, 5.1, 10.8, and 7.8 times in each group, respectively. In the case of PCNA and cyclin D1, the levels were decreased by about 0.6, 0.7, 0.1, and 0.1 times for each group (Table 1). Taken together, these results indicated that selumetinib and U87-Selu exo had anticancer effects on U87MG cells.  When cells were treated with U87-Selu exo and selumetinib for 24 h, the levels of p27 and p-p38 were increased by about 4.2, 5.1, 10.8, and 7.8 times in each group, respectively. In the case of PCNA and cyclin D1, the levels were decreased by about 0.6, 0.7, 0.1, and 0.1 times for each group (Table 1). Taken together, these results indicated that selumetinib and U87-Selu exo had anticancer effects on U87MG cells.

Flow Cytometry
To further evaluate the effects of U87-Selu exo on the apoptosis pathway by examining cell cycle arrest, fluorescence-activated cell sorting (FACS) analysis was performed. Cells were treated with 200 µg/µL of exo, Selu, or U87-Selu exo for 24 h, and then stained with PI. The control and exo-treated groups did not undergo cell cycle arrest (Figure 13a). Treatment with U87-Selu exo or Selu led to an increase in the G1 phase; the extent of arrest was about 11% in cells treated with U87-Selu exo. This increase was higher than what was observed in cells treated with Selu ( Figure 13b). Because G1 phase arrest is associated with apoptosis of mesenchymal or epithelial cells [45,46], the FACS results showed that both U87-Selu exo and Selu induce G1 phase arrest, indicating apoptosis.

Flow Cytometry
To further evaluate the effects of U87-Selu exo on the apoptosis pathway by examining cell cycle arrest, fluorescence-activated cell sorting (FACS) analysis was performed. Cells were treated with 200 µg/µL of exo, Selu, or U87-Selu exo for 24 h, and then stained with PI. The control and exo-treated groups did not undergo cell cycle arrest (Figure 13a). Treatment with U87-Selu exo or Selu led to an increase in the G1 phase; the extent of arrest was about 11% in cells treated with U87-Selu exo. This increase was higher than what was observed in cells treated with Selu ( Figure 13b). Because G1 phase arrest is associated with apoptosis of mesenchymal or epithelial cells [45,46], the FACS results showed that both U87-Selu exo and Selu induce G1 phase arrest, indicating apoptosis.

Anticancer Mechanism of U87-Selu Exo
Because U87MG cells, which were treated with U87-Selu exo, were stopped in the G1 phase and went through apoptosis, we further studied the mechanism of the U87-Selu exo (Supplementary Materials, Figure S5). Because selumetinib is known to be a pMEK inhibitor [47,48], we hypothesized that U87-Selu exo shows a similar anticancer mechanism as selumetinib. To confirm our hypothesis, we conducted Western blotting to evaluate the activation of intracellular signaling pathways. Compared with the control group, cells treated with U87-Selu exo or selumetanib showed downregulated expression of pMEK and pERK (Figure 14a-c). pMEK was lower by about 0.4-and 0.3-fold than the control group, and pERK was not detected at all in the U87-Selu exo-treated and selumetinibtreated groups. Through the expression of c-Raf, Raf, and Ras, we confirmed that U87-Selu exo and selumetinib are specific pMEK inhibitors. Taken together, these results confirmed our hypothesis that the anticancer mechanism of U87-Selu exo is like that of selumetinib, even after loading selumetinib into exosomes.

Anticancer Mechanism of U87-Selu Exo
Because U87MG cells, which were treated with U87-Selu exo, were stopped in the G1 phase and went through apoptosis, we further studied the mechanism of the U87-Selu exo (Supplementary Materials, Figure S5). Because selumetinib is known to be a pMEK inhibitor [47,48], we hypothesized that U87-Selu exo shows a similar anticancer mechanism as selumetinib. To confirm our hypothesis, we conducted Western blotting to evaluate the activation of intracellular signaling pathways. Compared with the control group, cells treated with U87-Selu exo or selumetanib showed downregulated expression of pMEK and pERK (Figure 14a-c). pMEK was lower by about 0.4-and 0.3-fold than the control group, and pERK was not detected at all in the U87-Selu exo-treated and selumetinib-treated groups. Through the expression of c-Raf, Raf, and Ras, we confirmed that U87-Selu exo and selumetinib are specific pMEK inhibitors. Taken together, these results confirmed our hypothesis that the anticancer mechanism of U87-Selu exo is like that of selumetinib, even after loading selumetinib into exosomes.

Conclusions
We demonstrated that selumetinib-loaded U87MG-derived exosomes (U87-Selu exo) can be used for targeted GBM (U87MG cell) therapy using their ability to target their parent cells. U87-Selu exo did not show any cytotoxicity to normal brain cells, even at high

Conclusions
We demonstrated that selumetinib-loaded U87MG-derived exosomes (U87-Selu exo) can be used for targeted GBM (U87MG cell) therapy using their ability to target their parent cells. U87-Selu exo did not show any cytotoxicity to normal brain cells, even at high doses, and did not show any toxicity to the liver and kidney in vivo. Therefore, our findings indicated that U87-Selu exo has a specific antitumor effect on GBM with a U87MG origin. The non-toxicity of U87-Selu exo to normal brain cells and liver indicates that there are promising therapeutic options for the treatment of GBM. Furthermore, it is still challenging to demonstrate whether glioma-derived exosomes show an in vivo homing effect because the central nervous system has the blood-brain barrier. Therefore, future work is warranted to investigate the in vivo homing effect of glioma-derived exosomes using an appropriate brain tumor model, such as an orthotopic model. This future work is especially important in the clinical translation of glioma-derived exosomes for GBM treatment.