Engineered exosomes delivering specific tumor-suppressive RNAi attenuate oral cancer progression

Exosomes are involved in a wide range of biological processes in human cells. Considerable evidence suggests that engineered exosomes (eExosomes) containing therapeutic agents can attenuate the oncogenic activity of human cancer cells. Despite its biomedical relevance, no information has been available for oral squamous cell carcinoma (OSCC), and therefore the development of specific OSCC-targeting eExosomes (octExosomes) is urgently needed. We demonstrated that exosomes from normal fibroblasts transfected with Epstein–Barr Virus Induced-3 (EBI3) cDNA were electroporated with siRNA of lymphocyte cytoplasmic protein 1 (LCP1), as octExosomes, and a series of experiments were performed to evaluate the loading specificity/effectiveness and their anti-oral cancer cell activities after administration of octExosomes. These experiments revealed that octExosomes were stable, effective for transferring siLCP1 into OSCC cells and LCP1 was downregulated in OSCC cells with octExosomes as compared with their counterparts, leading to a significant tumor-suppressive effect in vitro and in vivo. Here we report the development of a new valuable tool for inhibiting tumor cells. By engineering exosomes, siLCP1 was transferred to specifically suppress oncogenic activity of OSCC cells. Inhibition of other types of human malignant cells merits further study.

Western blot analysis. Western blot analysis was conducted as described previously 13  Engineered cells. NB1RGB cells (human skin-derived fibroblasts) were transfected with an EBI3 vector (VectorBuilder, Chicago, IL, USA) designed to overexpress human EBI3 cDNA (oeEBI3 cells) or non-target vector (mock cells) #TR30012 (OriGene Technologies, Rockville, MD, USA) using a Neon electroporator (Life Technologies, Carlsbad, CA, USA). Western blot analysis was utilized to confirm the expression of these proteins in the cells as mentioned above.
Transmission electron microscopy (TEM). The TEM analysis was performed as described previously 14 .
In brief, exosomes in PBS were placed on the carbon film grid, and they were partially dried. Next, a staining solution of 2% uranyl acetate in water was added to grids for 2 min and the excess liquid was blotted off with filter paper. The grids were dried overnight at room temperature. Grids were analyzed through the use of a HITACHI H-7600 transmission electron microscope (TEM, Hitachi High-Technologies Corporation, Tokyo, Japan) at Hanaichi UltraStructure Research in Japan. www.nature.com/scientificreports/ Nanoparticle tracking analysis (NTA). The NTA was performed as described previously 15,16 . In brief, the size distribution of the exosomes was analyzed using a Nano Sight LM10 instrument (Marvern instruments, Worcestershire, UK) equipped with NTA software, version 2.3. The particle suspensions were diluted with PBS to a concentration of 10 8 to 10 9 particles/mL for analysis.

Visualization of exosomes.
We visualized exosomes as reported previously 17  Analysis of LCP1 gene expression in tumor samples, its clinical significance and preparation of siLCP1-loaded EBI3 exos (octExosomes). We recently showed that LCP1 is one of the essential components for OSCC progression 12 . To further identify the role of LCP1 in head and neck SCCs (HNSCC), including OSCCs, The Cancer Genome Atlas (TCGA) network cohort (n = 522) was assessed as we described previously 18 . First, we standardized loading conditions of exosomes to achieve a satisfactory and repeatable outcome. Thus, the loading conditions for electroporation were optimized for a Neon electroporation system. The EBI3 exos were mixed with the Neon electroporation buffer at a 1:1 ratio, and siLCP1 #sc-43208 (Santa Cruz Biotechnology) or siControl # sc37007 (Santa Cruz Biotechnology) was added to the mixture at a final ratio of siRNA : exosomes protein of 100 pmol:1 μg/mL. Electroporation was then performed at various voltages per the manufacturer's instructions with the pulse width set at 10 ms. The effect of pulse time was also assessed at the optimized voltage. After electroporation, one unit of RNase A was added to the mixture to eliminate free siLCP1 outside the exosomes. To reduce undesirable electroporation-induced siLCP1 precipitation during the loading process, EDTA was added.
Proliferation/migration/invasion assays. SAS and HSC-3 were placed in 6-well plates at 1 × 10 4 cells/ well in proliferation assays. The siLCP1-loaded EBI3 exos (octExosomes) or siControl-loaded EBI3 exos (siControl exosomes) were added at doses of 10 µg/mL for 120 h. Cells were counted every 24 h. The cell lines were assessed for viability using Luna Automated Cell Counter (Logos Biosystems, Annandale, VA, USA). All experiments were performed in triplicate. For live-cell imaging, the cells were seeded in 6-well plates with 10% FBS/DMEM until a confluent monolayer formed. Using a micropipette tip, one wound was created in the middle of each plate. We incubated plates at 37 °C at 5% carbon dioxide with free-serum medium, and live cell migration was captured after 12 and 24 h. The wound area that was free of cells was calculated using Lenaraf220b software (http://www.vecto r.co.jp/soft/ dl/win95 /art/se312 811.html).
Matrigel invasion assays were carried out using Matrigel-coated Transwell inserts (8 μm pores) (Becton-Dickinson, Franklin Lakes, NJ, USA) following the manufacturer's instructions. Two mL of DMEM with 10% FBS was placed in the lower wells. Proliferating siLCP1 cells or siControl cells (2.0 × 10 5 cells per mL) were loaded into each of the upper wells and monitored after 72 h for the migrating cells.

In vivo targeting and biodistribution in a tumor xenograft mouse model by IVIS imaging.
Animal handling and all animal experiments followed the ARRIVE guidelines. BALB/C-nu mice were used to study the in vivo targeting and biodistribution of the octExosomes based on the previous report 19 . In brief, female BALB/cAnNCrj-nu/nu mice (Oriental Yeast Co., Ltd., Andover, MA) were engrafted subcutaneously with 1 × 10 7 SAS or HSC-3 cells on the back, and the tumor volume was calculated as (width) 2 × (length)/2. After the injection, the mice with a palpable tumor > 100 mm 3 in size were chosen for this study. For imaging of fluorescently labeled exosomes, a stock solution of the lipophilic near-infrared dye XenoLight DiR (Caliper Life Sciences, Hopkinton, MA, USA) was prepared in ethanol. The octExosomes were incubated with 2 µmol/L DiR for 30 min, washed with 10 mL of PBS, and then injected intravenously through the tail vein at a dosage of 1.5 mg/kg. Six h after injection and every 3 days, Dir fluorescence in the tumor-bearing mice was captured with a Xenogen IVIS-200 optical in vivo Imaging system (Caliper Life Sciences). The tumors were measured every 3 days. The weight of mice was measured every 3 days during the experiment. No obvious decrease in the body weight of any mouse was detected after drug treatment. Upon completion of treatment (18 days), tumor grafts were harvested. All experimental procedures were approved by The Institutional Animal Care and Use Committee of the Chiba University (approval number, 1-126).
Immunohistochemistry. For histological analysis, paraffin-embedded samples were first deparaffinized and rehydrated based on standard protocols as we described previously 13 . In brief, the blocking of non-specific antigens and endogenous peroxidase activity was carried out with a peroxidase blocking reagent (serum and 3% hydrogen peroxide), followed by overnight incubation with anti-LCP1 #HPA019493 (Atlas Antibodies), 1:200 and anti-PCNA #GTX100539 (Gene Tex), 1:500 at 4 °C. A ChemMate DAKO EnVision Detection Kit (Peroxidase/DAB, Rabbit/Mouse; DakoCytomation, Glostrup, Denmark) was used to immunostain the slides in accordance to the manufacturer's protocols. www.nature.com/scientificreports/ Statistical analyses. All data are expressed as the mean (±) standard deviation. Statistical analysis was conducted using Student's t test. P < 0.05 was considered to indicate statistical significance.

Development of octExsomes.
To direct the targeting of exosomes, they were engineered to express peptides that could recognize sequences abundantly present on OSCC cell membranes. We performed gene expression analyses of microarray data collected from OSCC cell lines and HNOKs. Volcano plots show that 1536 upregulated genes and 3381 downregulated genes were significantly changed more than two-fold in their z-scores (Fig. 1A). Among the significantly upregulated genes, unsupervised hierarchical clustering of all the differentially expressed genes (n = 344) showed clear OSCC cell membrane gene expression patterns. Of them, 6 were confirmed to be commonly upregulated in the expression profiles of 4 of the OSCC cell lines. The EBI3 gene was the only one to increase its mRNA expression level in all OSCC-derived cell lines that we examined (Fig. 1B). Thus, it was selected for further studies in vitro and in vivo. Expression of EBI3 protein was confirmed www.nature.com/scientificreports/ by Western blot analysis of transfected NB1RGB cells (Fig. 1C). These modifications did not appear to affect the physical properties (size and shape) of the engineered exosomes (eExosomes) and exosomes released from untransfected NB1RGB cells (NB1RGB exosomes) based on electron microscopy (Fig. 1D) and NTA (Fig. 1E). EBI3 protein was significantly upregulated in EBI3-transfected NB1RGB cells and was incorporated into the NB1RGB cell-derived exosomes according to Western blots (Fig. 1F). Stained exosomes (NB1RGB exos and EBI3 exos) (5 ng/µL, 10 ng/µL and 20 ng/µL) were added to the culture media of SAS, HaCaT cells and NB1RGB cells. Uptake was examined by confocal laser scanning microscopy. Green fluorescence was not detected in HaCaT cells and NB1RGB cells 6 h after addition of SYTO RNA Selectstained NB1RGB exos ( Fig. 2A). In addition, 5 ng/µL, 10 ng/µL and 20 ng/µL stained NB1RGB exos or EBI3 exos were added to the culture media of SAS and uptake was examined as above. Green fluorescence was detected in SAS 1 h, 3 h and 6 h after addition of SYTO RNA Select-stained EBI3 exos. We found that the optimal conditions were 10 ng/µL and 3 h (Fig. 2B, Supplementary Fig. 4). However, the detailed mechanisms of their cellular uptake are still unknown. Some recent reports suggested that endocytosis has been reported to be a major pathway for the cellular uptake of exosomes 20,21 . Therefore, these results suggested that exosomes released from oeEBI3 NB1RGB cells were more taken up in SAS, and it may be speculated via endocytosis.
Regarding the bioinformatics analysis of LCP1, although LCP1 mRNA expression status was not associated with lower overall survival (OS), its overexpression was linked to lower OS among the patients with metastatic regions (Supplementary Fig. 1A,B). Those findings suggested that siLCP1 was a good candidate for an eExosome cargo for targeting OSCC cells.

The octExosomes suppress the progression of oral cancer cells in vitro.
To assess whether the siLCP1-loaded EBI3 exos (octExosomes) would be able to specifically deliver siLCP1 in vitro, we performed www.nature.com/scientificreports/ qPCR and Western blot analyses. As expected, we detected significant downregulation of LCP1 in octExosome-treated cell lines (Fig. 3A,B). Adding siControl exosomes to the cells did not result in LCP1-inhibition (Fig. 3A,B), suggesting high efficacy of exosome-mediated delivery of siLCP1 by octExosomes. Furthermore, these modifications did not appear to affect the physical properties (size and shape) of the octExosomes and siControl exosomes based on electron microscopy (Fig. 3C) and NTA (Fig. 3D). We then examined whether octExsosomes interrupted cell growth/migration/invasion in vitro. Significant cell growth inhibition was apparent in OSCC cells treated with octExosomes, while no growth inhibition of cells treated with siControl exosomes was observed (Fig. 4A). In addition, the silencing of LCP1 expression by octExosomes drastically weakened the cells' migratory and invasive abilities, but not in the siControl exosome group. (Fig. 4B,C).

The octExosomes suppress the progression of oral cancer cells in vivo.
We next evaluated the tumor targeting effect of octExosomes in vivo. As shown in Fig. 5A, a strong accumulation of Dir fluorescence was observed by whole-mouse imaging in the tumor area for the siControl exosome and octExosome group. It is noteworthy that the color intensity for exosome accumulation was much higher in the tumor area (Fig. 5A). Furthermore, while siControl exosomes did not suppress tumor growth, adding the octExosomes significantly inhibited tumor growth. The sizes of tumors at day 18 were ranked as follows: octExosomes < siControl exosomes ≈ control, Fig. 5B, C. No significant body weight loss was observed across different groups during the experimental period (Supplementary Fig. 2). Among the xenografted tumors, both LCP1 mRNA levels and LCP1 protein levels were significantly reduced in octExosomes-tumors compared to the siControl exosome group (Fig. 6A,  B). Immunohistochemical analyses showed clear differences in LCP1 and PCNA expression status between the two groups (octExosomes versus siControl exosomes; Fig. 6C). PCNA, a marker of cell proliferation, level was considerably suppressed in the octExosome group. Therefore, we suggest that octExosomes treatment led to markedly suppression of tumor growth through downregulation of LCP1.

Discussion
Accumulating evidence has suggested that exosomes released from human cells could be ideal nanocarriers of therapeutic agents for clinical use [22][23][24] . However, it is not clear how best to endow exosomes with the efficiency to specifically deliver therapeutic molecules to target cells. In addition, cancer-derived exosomes will regulate/ facilitate organ-specific metastases 25 , indicating that the development of specific cancer-targeting exosomes from non-cancerous cells is required. Here, we report that the progression of oral cancer cells was inhibited by treatment with specific exosomes (octExosomes). They were constructed by modifying normal fibroblast-derived exosomes such that they expressed transmembrane EBI3 on the surface and enclosed siLCP1. We identified gene expression signatures common to OSCC transmembrane proteins and identified a commonly-overexpressed gene in OSCC cell lines (Fig. 1B), EBI3 (Fig. 1A, B). EBI3 regulates the immune system and multiple cytokines 26 . In addition, EBI3-deficient mice show prominent abnormalities in the glomerular basement membrane 27 . Moreover, EBI3-deficiency leads to downregulated expression of the adhesion molecule VCAM1 28 . A recent report suggested that IL-35, a subunit of which is EBI3 protein, promotes metastasis of human pancreatic cancer 29 by inducing the endothelial adhesion molecule, ICAM1. Further, EBI3 is frequently expressed in a subset of human malignancies [30][31][32] , suggesting that EBI3 may play a functional role in cell-cell adhesion during the progression of human cancers.
The interaction between transmembrane EBI3 and octExosomes in the cancer microenvironment is still unknown. Since EBI3 also constitutes homodimeric structures 33,34 , we speculated here that our octExosomes directly recognized EBI3 protein produced from the OSCC cells. Thus, further studies on cellular uptake and membrane fusion mechanisms in exosomes are necessary. Nonetheless, this molecule could be a candidate surface component of OSCC-specific exosomes.
Most human cells release exosomes 35 . However, due to the differences in their contents and their surface compositions, it is crucial to consider their characteristics with regard to therapy. From a clinical standpoint, the above-mentioned characteristics probably require modification. The contents of exosomes derived from cancer cells may promote cancer progression 36 . Thus, therapeutic exosomes must be derived from normal cells. We chose to modify normal human fibroblast (NB1RGB)-derived exosomes. After transfecting NB1RGB cells with EBI3, we observed a robust accumulation of modified exosomes. There was no significant influence on exosomal size (Fig. 1D, E). Thus, we suggest that the modified exosomes from NB1RGB cells could be useful for treating OSCC cells, possessing high efficacy and low oncogenicity on targets.
We recently found that LCP1 overexpression was an essential aspect of oral cancer progression 10 . The silencing of LCP1 by siRNA suppressed both cancer cell growth and metastatic phenotypes, and LCP1-positive OSCC cases were closely associated with the tumor size and regional metastasis 10 . TCGA cohort analysis indicated a significant relationship between LCP1 overexpression and lymph node status ( Supplementary Fig. 1C, D). Considering the impact of LCP1 oncogenic activation, we speculated that more concerted efforts were needed to reduce LCP1 expression. Towards that end, we examined the anti-tumor effects achieved by loading siLCP1 into the modified exosomes. Electroporation techniques allowed us to load large amounts of siLCP1 into highly stable exosomes (Fig. 3D). Our production method relies on ultracentrifugation, which could include non-exosomal contaminants. However, the octExosome preparations did not yield any measurable side effects and showed consistent in vitro and in vivo efficacy (Figs. 4 and 5 and Supplementary Fig. 2). In the present study, we found that octExosomes accumulated around OSCC cells and suppressed LCP1 expression in OSCC cells/tumors in vitro and in vivo (Figs. 4, 5, 6). Lower expression of LCP1 protein was observed in xenografted OSCC cells compared to the siControl group, suggesting that siLCP1 was effectively loaded into cancer cells via octExosomes (Fig. 6B, C).
The field of exosome-based drug delivery has been pursued for a number of years 37 . Here, we demonstrated that specific targeting and anti-tumor effects can be achieved with fibroblast cell exosomes. Our preclinical     Based on qPCR and Western blot analyses. Note that LCP1 was considerably suppressed in the octExosome group compared to siControl exosome and control groups at both the mRNA level and protein level. (C) Immunohistochemical analysis of tumor LCP1 and PCNA changes. LCP1 and PCNA expressions were considerably suppressed in the octExosome group. Scale bars indicate 50 μm.

Data availability
These microarray data were deposited in the NCBI Gene Expression Omnibus database under GEO accession number GSE146483. Other datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.