Exosome Uptake through Clathrin-mediated Endocytosis and Macropinocytosis and Mediating miR-21 Delivery*

Background: Exosomes can transfer information between cells and facilitate tumor development. Results: PC12 cell-derived exosomes enter into BMSCs through clathrin-mediated endocytosis and macropinocytosis, and decrease the expression of TGFβRII and TPM1 through miR-21. Conclusion: The results dissect the pathway of exosome internalization and demonstrate their regulation ability. Significance: These findings enhanced our knowledge of the internalization and function of tumor exosomes. Exosomes are nanoscale membrane vesicles secreted from many types of cells. Carrying functional molecules, exosomes transfer information between cells and mediate many physiological and pathological processes. In this report, utilizing selective inhibitors, molecular tools, and specific endocytosis markers, the cellular uptake of PC12 cell-derived exosomes was imaged by high-throughput microscopy and statistically analyzed. It was found that the uptake was through clathrin-mediated endocytosis and macropinocytosis. Furthermore, PC12 cell-derived exosomes can enter and deliver microRNAs (miRNAs) into bone marrow-derived mesenchymal stromal cells (BMSCs), and decrease the expression level of transforming growth factor β receptor II (TGFβRII) and tropomyosin-1 (TPM1) through miR-21. These results show the pathway of exosome internalization and demonstrate that tumor cell-derived exosomes regulate target gene expression in normal cells.

Our previous work used rat pheochromocytoma PC12 cellderived exosomes as a general model in which to study tumor exosomes (17). The cellular uptake and intracellular trafficking were visualized by real-time fluorescence microscopy and single particle tracking (SPT) (18). However, the endocytic pathway of exosomes has not been further examined. There are multiple pathways that can mediate endocytosis, including phagocytosis, macropinocytosis, clathrin-mediated endocytosis, caveolae-mediated endocytosis, and clathrin-and caveolaeindependent endocytosis (19). These pathways lead to different sorting and fate of endocytic cargo. It has been reported that exosomes derived from erythroleukemia cells or oligodendroglia cells were internalized through phagocytosis or macropinocytosis, respectively (20,21). Recently, Svensson et al. found that glioblastoma cell-derived exosomes were internalized through lipid raft-mediated endocytosis, negatively regulated by caveolin-1 (CAV1) (22). The uptake pathway of exosomes may possibly be cell type specific. Furthermore, oncomiRs, a miRNA that is associated with tumor, may be enclosed in tumor exosomes and delivered to normal cells (15,16). It is still unknown whether these reduce expression of target gene and facilitate tumor development.
In this study, exosomes were isolated from the culture medium of PC12 cells. By employing selective inhibitors, molecular tools, and endocytosis markers, it was found that the exosome uptake by PC12 cells occurred through clathrin-mediated endocytosis and macropinocytosis. Moreover, using quantitative real-time PCR (RT-PCR) and immunoblot assay, it was demonstrated that PC12 cell-derived exosomes delivered miR-21 into bone marrow-derived mesenchymal stromal cells (BMSCs) and down-regulated the expression levels of their transforming growth factor ␤ receptor II (TGF␤RII) and tropomyosin-1 (TPM1). These findings add insights into the endo-cytic pathway and the biological significance of tumor exosomes.
Exosome Isolation and Labeling-The culture medium from PC12 cells (1 ϫ 10 8 ) was collected and isolated as previously described (24). Briefly, the harvested medium was centrifuged at 300 ϫ g for 10 min, 1200 ϫ g for 20 min, and 10,000 ϫ g for 30 min to remove cells and debris. The supernatant was ultracentrifuged at 200,000 ϫ g for 2 h using a Type 70 Ti rotor in an L-80 XP ultracentrifuge (Beckman Coulter). Then the exosome pellet was resuspended in phosphate-buffered saline (PBS). For labeling, the exosome solution was incubated with 5 g/ml DiD for 30 min. The unincorporated dyes were removed using 300-kDa ultrafiltration tubes (Pall Corp.) and washed in PBS with ultracentrifugation. The concentrated solutions were diluted in PBS. The amount of exosome protein was measured using the Micro BCA Protein Assay Kit (CoWin Biotechnology).
Fluorescence Microscopy-A spinning disk confocal system (Revolution XD, Andor Technology, Northern Ireland) was built on the left port of a Ti-E inverted microscope (Nikon, Japan) with a PFS (perfect focus system) to keep focus plane and a motorized stage (Ludl Electronic Products) to do montage. The confocal images were collected by an electron-multiplying charge-coupled device (EMCCD) iXon DV885 (Andor) with 1004 ϫ 1002 pixels. 405 nm, 491 nm, and 640 nm solid state lasers modulated by AOTF (Acousto-optic tunable filter) were used as the illumination sources for Hoechst 33342, FITC, and DiD, respectively. Fluorescence emission was collected by a 40 ϫ oil-immersion objective (S Fluor, NA. ϭ 1.30, Nikon), passed through EM 452/45, EM 520/15, or EM 685/40 emission filters (Semrock). All images were acquired and processed by iQ v. 2.0 software (Andor).
Uptake Inhibition Studies-To study the pathway of exosome uptake, cells were preincubated with some pharmacological/ chemical inhibitors before exosome addition. 1-50 M CPZ, 10 -200 M genistein, 10 -100 M EIPA, or 10 -100 M LY294002 was applied to pretreat cells at 37°C for 30 min and were present throughout the experiments, respectively. 5-50 M nystatin, 10 mM M␤CD, or 0.1 mg/ml CtxB was pre-incubated with cells for 60 min and washed excessively prior to exosome addition. For a given experiment, the final concentration of dimethyl sulfoxide (DMSO) was not above 0.1%. As control, cells were cultured in the presence or absence of 0.1% DMSO. To study the influence of K ϩ depletion to exosome uptake, cells were incubated in a buffer containing 20 mM HEPES, 140 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 5.5 mM D-glucose, and 0.5% BSA for 1 h at 37°C. As control, cells were incubated in the buffer supplemented with 10 mM KCl. Calcein AM was applied to test cell viability after 4 h treatment of drugs.
40 g/ml DiD-labeled exosomes were added to cells, and subsequently kept at 37°C for 3 h. After washing with PBS, fixed by 2% paraformaldehyde and incubated with 5 g/ml Hoechst 33342 for 20 min, cells were imaged by confocal microsopy. To quantify the cellular uptake of exosomes, the experiments were repeated three times, and for each individual experiment more than 800 cells were imaged at random locations. All settings of imaging and processing were kept constant, and the relative fluorescence intensities were calculated. The cell numbers were determined by counting the nuclei. The significance of the effects of various treatments versus control was evaluated by Student's t test (p Ͻ 0.05). To exclude the disturbance of DiD diffusion, the blocking degree of uptake using CFSE-labeled exosomes by CPZ was evaluated as control. Moreover, to determine whether PC12 cells can phagocytose exosomes, 10 7 /ml latex beads with 40 g/ml exosomes were incubated with cells at 37°C for 3 h, and confocal imaging was performed.
Knockdown and rescue efficiency was tested by PCR and Western blots 48 h after transfection. 48 h after transfection, cells were incubated with exosomes for 3 h, and the blocking degree of uptake was evaluated as described above.
Colocalization Studies-To colocalize exosomes with dextran, 40 g/ml DiD-labeled exosomes with 3 mg/ml FITC-dextran were added to cells. Washing, fixing, nucleus staining, and imaging were carried out after 10 min, 30 min, or 1 h incubation at 37°C. For colocalization of exosomes with CtxB, incubation of cells with FITC-CtxB (0.1 mg/ml) was performed for 1 h at 4°C to allow the toxin to bind to lipid raft without substantial internalization. After excessive washing and adding exosomes, cells were shifted to the stagetop incubator on the microscope. Live-cell imaging was performed after 10 min, 30 min, and 1 h.
RT-PCR-Total RNA of cells was extracted by Trizol reagent (Invitrogen). Exosome samples were cracked by repeated freezing and thawing. The first strand synthesis of cDNA was performed using equal amounts of RNA samples, according to M-MLV reserve transcriptase instructions (Promega). ␤-Actin or U6 was employed as the housekeeping gene for mRNAs or miRNAs analysis. PCR reactions were performed using SYBR Premix ExTaq (TaKaRa, Dalian, China) in a total volume of 20 l (1.5 l cDNA samples). RT-PCR was carried out using the 7500 system (Application Biosystems). Relative expression was calculated by the comparative 2 Ϫ⌬⌬Ct method. Each sample was assessed at least in triplicate.
Electron Microscopy-A droplet of diluted exosomes was applied onto a carbon-coated copper grid and incubated until dry. Specimens were observed with a JEM-2100 transmission electron microscope (JEOL).

Roles of Clathrin-mediated Endocytosis and Macropinocytosis in Exosome
Uptake-In this work, the culture medium of PC12 cells was sequentially centrifuged according to a generally accepted exosome isolation protocol. The resulting pellet was confirmed consisting of vesicles of 40 -100 nm in size by electron microscopy (Fig. 1A). The isolated exosomes were labeled by lipophilic dye DiD or amino-reactive fluorophor CFSE, and detected by fluorescence microscopy (Fig. 1B). Before applying inhibiting treatments to study the uptake pathway, several control experiments were carried out. First, controls of endocytosis inhibition were performed to test the activity of the treatments (Fig. 1C). Second, the effect of solvent was excluded. Results demonstrate that DMSO in concentrations as low as 0.1% did not affect exosome uptake (Fig. 1D). Third, to optimize drug concentrations and to avoid disturbing cells, cell viability was tested under different treatments. The results are shown in Fig.  1E, and the treatments causing a decrease in cell viability of more than 15% were not used.
Clathrin-mediated endocytosis (also known as clathrin-dependent endocytosis) is inherently active in all mammalian cells (19). When some inhibiting treatments were applied, exosome uptake of several hours was quantified following previous reports. K ϩ depletion presents a useful procedure to block clathrin-coated pits (28) . Fig. 2, A and C shows that treatment of K ϩ depletion buffer induced a partial inhibition of exosome uptake, with the percentage of internalized exosomes dropping to 55.5%. Besides, CPZ, a cationic amphipathic drug likely inducing a loss of clathrin, inhibits clathrin-mediated endocytosis (28). As shown in Fig. 2, B and D, CPZ caused a dose-de-pendent inhibition of exosome uptake with 41% block at 10 M.
To exclude the disturbance of DiD diffusion, CFSE-labeled exosomes were also used to evaluate the uptake inhibition under CPZ treatment. As shown in Fig. 2E, CFSE-labeled exosomes produced similar results as DiD-labeled exosomes. Furthermore, CHC, the basic subunit of clathrin, was knocked down by siRNA successfully (Fig. 2, F and G). Exosome uptake was blocked by about 35.3% in the CHC-negative cells (Fig. 2, H and  I). In addition, 2, the subunit of clathrin adaptor complex AP2, inhibited the assembly of clathrin-coated pits (Fig. 2, J-M). Transfection of 2-shRNA or loss-of-function mutation 2-D176A caused a 37.5% or 36.8% decrease of exosome uptake. To further validate these results, a rescue experiment was performed by co-transfection of 2-shRNA and a shRNA-resistant silent mutation-containing variant of 2. This shRNA inhibitory effect on exosome uptake could be rescued by 2-res (Fig.  2, N and O). All the results above indicated that exosome uptake involves clathrin-mediated endocytosis.
Another main endocytosis pathway, macropinocytosis, was considered. EIPA, a Na ϩ -H ϩ exchanger inhibitor, and LY294002, a phosphoinositide 3-kinase (PI3K) inhibitor, were used to block macropinocytosis (28). The effects of both inhibitors on exosome uptake by PC12 cells were quantified by fluorescence microscopy, and the uptake efficiency was reduced remarkably and depended on the dose of EIPA and LY294002 (Fig. 3, A-D). Colocalization study of exosomes with dextran at the early time of cellular internalization was performed to provide further evidence. Approximately half of the colocalization emerged 10, 30, or 60 min after internalization initiation (Fig. 3, E and F). All results above indicate that macropinocytosis played a role in exosome uptake.
Roles of Caveolae-mediated Endocytosis and Phagocytosis in Exosome Uptake-Besides clathrin-mediated endocytosis and macropinocytosis, caveolae-mediated endocytosis is one of the main routes for cellular internalization (19). Disrupting lipid rafts and inhibiting tyrosine protein kinase are two of the most widely used techniques for blocking this endocytic pathway (28). Exosome uptake was analyzed in the presence of known inhibitors, genistein and nystatin. For avoiding cholesterol sequestration and property changes on exosome membrane, nystatin was washed excessively before exosome addition. Fig.  4, A and B show fluorescence images after exosome uptake for 3 h by cells with or without drug treatments. Statistical results are shown in Fig. 4, C and D. The exosome internalization was generally not affected by genistein, even at a concentration as high as 200 M. Surprisingly, lipid raft disruption by nystatin treatment enhanced exosome uptake. The exposure time for Fig. 4B was shortened to prevent overexposure. Similar results also emerged under lipid raft disruption by another cholesterol depletion-drug M␤CD or lipid rafts blocking by CtxB (Fig. 4, E and F). It was interpreted that the plasma membrane became more fluid under these treatments leading to the enhancement of uptake through other pathways (29). Therefore, caveolaemediated endocytosis cannot play roles in exosome uptake. Moreover, clathrin-and caveolae-independent endocytosis was also not involved in exosome uptake, because this class of pathways, including Arf6-dependent, flotillin-dependent, Cdc42-dependent, and RhoA-dependent endocytosis are all inhibited by genistein and cholesterol depletion (30).
Next, CAV1, the key protein in caveolae-mediated endocytosis, was knocked down by CAV1-shRNA and rescued by cotransfection of CAV1-res (Fig. 4, G-J). The loss-of-function mutation CAV1-P132L was also expressed and tested. Exosome uptake was almost not affected in the CAV1-negative or CAV1-rescued cells (Fig. 4, K and L). Furthermore, the colocalization study of exosomes with CtxB at an early time of cellular internalization was performed to provide further evidence. As shown in Fig. 5, A and B, the colocalization percentages of exosomes and FITC-CtxB were very low after 10 min (3.0%) or 30 min (8.7%) uptake. And the percentage increased to 32.6% after 60 min of incubation, possibly due to the similar sorting target of CtxB and exosomes after cellular uptake. All the results above indicated that the internalization pathway of exosomes was quite different from caveolae-mediated endocytosis.
DYN2, another key protein necessary for both clathrin-and caveolin-mediated endocytosis, but not involved in macropinocytosis, was also tested. Applying DYN2-shRNA, the uptake efficiency was reduced to 50.7% (Fig. 5, C-F), indicating that dynamin-dependent endocytosis was partially involved. In general, phagocytosis is performed by specialized professional cells such as macrophages, neutrophils, or monocytes (19). Experiments were still carried out to exclude phagocytosis during exosome uptake by PC12 cells. Latex beads, a well-known marker of phagocytosis, were added to culture medium, allowing cellular uptake (28). Shown as Fig. 5G, no bead was observed to be internalized into cells at more than 3 h. This result indicated that PC12 cells cannot phagocytose efficiently.
Delivery of miR-21 into BMSCs through PC12 Cell-derived Exosomes-To study the entry of exosomes to normal cells, BMSCs were selected as the recipients of exosomes derived from PC12 cells. After 3 h of incubation with BMSCs, exosomes were detected in the recipient cells by fluorescence microscopy, showing that PC12 cell-derived exosomes enter into BMSCs (Fig. 6A). Our previous work showed that miR-21 has a rela- tively high copy number in PC12 cells detected by highthroughput sequencing (data not shown). miR-21 is a unique miRNA that is overexpressed in most tumor types (31). Thus, we selected miR-21 to study miRNA transfer through PC12 cell-derived exosomes. Using RT-PCR, it was confirmed that miR-21 was as abundantly expressed in exosomes as it was in cytoplasm; greater than miR-22, miR-26a, miR-214, miR-221, and miR-222. The expression level of miR-21 in BMSCs increased dramatically as early as 3 h after incubation with exosomes (Fig. 6B). Unlike tumor cells, the endogenous miR-21 level in BMSCs was quite low, leading to a rapid, more than 4-fold enhancement of miR-21 level after exosome uptake for 3-12 h. The level recovered at 24 h was possibly caused by miR-21 degradation. Shown in Fig. 6C, the down-regulation of pre-miR-21 levels occurred in BMSCs after a 6 -24 h incubation with exosomes. To demonstrate that the increased miR-21 was from exosomes but not from activated endogenous miR-21 during exosome incubation, cardiomyoblast H9C2 cell-derived exosomes were selected as negative controls to be incubated with BMSCs (because of lower miR-21 enclosed). After 3-24 h of incubation, the miR-21 level in BMSCs did not change significantly (Fig. 6D). Together, the miR-21 level in BMSCs was elevated by PC12 cell-derived exosomes, while the pre-miR-21 level was down-regulated. These results showed that PC12 cellderived exosomes delivered miR-21 into other cell types and suppressed the expression of endogenous pre-miR-21.
Target Genes of miR-21 in BMSCs Regulated by PC12 Cellderived Exosomes-miR-21 is one of the most important oncomiRs. To determine biological functions of miR-21 delivered from PC12 cells to BMSCs through exosomes, several target genes were analyzed. TGF␤RII, a receptor on BMSCs that respond to an important regulator TGF␤ and target gene of miR-21, was selected to be analyzed. Interestingly, the expression level of TGF␤RII mRNA decreased substantially after 12 h and 24 h of exosome treatment (Fig. 7A). TGF␤RII levels also decreased after exosome treatment, which was revealed by Western blots. At the same time, the antisense oligonucleotides of miR-21 (anti-miR-21) resisted the exosome effect to reduce TGF␤RII expression (Fig. 7, B and C). Next, TPM1 as a tumor suppressor and miR-21 target was analyzed. After treatment with exosomes for 24 h, the expression level of TPM1 in BMSCs was reduced significantly. Such effect was resisted by anti-miR-21 (Fig. 7, D and E). Furthermore, PTEN, another important tumor suppressor and gene target of miR-21 was tested. However, the expression level of PTEN mRNA in BMSCs did not change after treatment with exosomes for 24 h (Fig. 7F). The suppression efficiencies for target genes of miR-21 may be cell type specific. These results indicated that PC12 cell-derived exosomes could decrease TGF␤RII and TPM1 in BMSCs  through miR-21. The above results show an example of information transfer from tumor cells to normal cells.

DISCUSSION
The interaction between tumor exosomes and cells plays important roles in tumor progression and immunity (10). Our previous works have shown that PC12 cells internalized exosomes through endocytosis (17,18). However, the endocytic pathway of exosomes was still unclear. Recently, phosphatidylserine on the exosome surface is reported to activate macropinocytosis in microglia (21). Svensson et al. found that glioblastoma cell-derived exosomes trigger lipid raft-mediated endocytosis through the activation of ERK1/2 signaling pathways that include an important role of heat shock protein 27 (HSP27) (22). Heparan sulfate proteoglycans (HSPGs) were reported to function as receptors of glioblastoma cell-derived exosomes (32). Hence, it is hypothesized that the uptake pathway of exosomes are dependent on the activation of the receptors on the recipient cells by exosomes. In this work, based on experimental data using selective inhibitors, molecular tools, and special endocytosis markers, it was concluded here that clathrin-mediated endocytosis and macropinocytosis were involved in the uptake of PC12 cell-derived exosomes. Both pathways can be mediated by receptor activation (33,34). Clathrin-mediated endocytosis involves engulfment of receptors associated with their ligands, while macropinocytosis occurs due to membrane ruffles triggered by receptor tyrosine kinases. Specific to the parental cell, many kinds of proteins and lipids on the exosome surface can attach to different receptors, inducing complex signaling. The uptake pathway dependent on the signaling is possibly cell type specific. At least two kinds of signaling activation may be involved in the uptake of PC12 cellderived exosomes. That was the reason that exosome uptake was at the same time dependent on two different pathways. The signaling pathway within it will be revealed in future studies. Furthermore, the intracellular trafficking occurring after the two endocytic pathways tends to sort toward lysosomes. Exosome proteins may be stored and utilized there. Whether exosome RNAs can escape from degradation and take effect in cells is an interesting question.
Tumor cells secrete exosome-packaging miRNAs, circulating in body fluids (serum, plasma, saliva, urine, and milk), and enter into neighboring and distant cells (15,35). miRNAs, protected in exosomes against degradation, is an important regulator of gene expression at the post-transcriptional level (36,37). In this work, exosomes acted as a natural liposome transferring miRNAs from tumor cells to normal cells. First, we confirmed that PC12 cell-derived exosomes enclosed abundant miR-21, an oncomiR overexpressed in most tumor types and associated with breast cancer, prostate cancer, and many other cancers (38,39). Second, it was found that exosomes entered and transferred miR-21 into BMSCs. Third, TGF␤RII and TPM1 expression were down-regulated by exosomes and through miR-21. TGF␤RII is often mutated in cancers, and the TGF␤ signaling pathway plays complex roles during cancer progression (40,41). TPM1, which belongs to the class II tumor suppressor genes, modulates cell transformation and tumor cell growth. So the tumor exosomes may regulate cancer progres-sion by reducing TGF␤RII and TPM1 in normal cells. Here, our data presents an example of information transfer from tumor cells to normal cells.
In summary, the endocytosis pathway involved in PC12 cellderived exosome uptake was shown to be clathrin-mediated endocytosis and macropinocytosis. It was demonstrated that exosomes entered and transferred miR-21 into BMSCs, and decreased their TGF␤RII and TPM1 expression. These results deepen the understanding of the internalization pathway of tumor exosomes, and verify an effect of tumor exosomes to normal cells.