Macrophage-Derived Small Exosomes: Efficient Transmission and Cytotoxicity to Cancer Cells

Tumor-associated macrophages are a key component in the tumor microenvironment, secreting extracellular vesicles (EVs) such as exosomes and other various factors for intercellular communication. However, macrophage-derived EVs heterogeneity and their cytotoxicity to cancer cells has not been well understood. Here, we aimed to separately isolate various types of macrophageEVs by size exclusion chromatography (SEC) method and investigate EV transmission and cytotoxicity to oral cancer cells. For fluorescence-labeling of cellular and EV membranes, palmitoylation signal-fused GFP and tdTomato were expressed in THP-1 monocytic cells and HSC-3 oral cancer cells, respectively. We found that fluorescence-labeled EVs secreted by macrophages were highly transmissive to oral cancer cells than those from parental monocytic cells. In a co-culture system and conditioned medium (CM), a macrophage-secreted unidentified factor was cytotoxic to oral cancer cells. We fractionated macrophage-derived EVs by the SEC method and performed western blotting to characterize various EV types. Three fractions were characterized: small exosomes (EXOS: < 50 nm) fraction containing HSP90α, HSP90β, CD63 (EV marker) and β-actin; large exosomes (EXO-L: 50-200 nm) fraction containing CD9 (EV marker) and HSP90β; large EVs (100-500 nm) fraction. Notably, the macrophage-derived small exosomes fraction was cytotoxic to oral cancer cells, while large exosomes and large EVs were not. Therefore, it was implicated that macrophage-derived small exosomes are cytotoxic with high transmission potential to cancer cells.

Many methods have been developed for EV preparation, including ultracentrifugation (UC) method [21], filter centrifugation (FC), size exclusion chromatography (SEC) [22,23], affinity purification (AP), immunoprecipitation (IP), sucrose cushion ultracentrifugation (SCU) [24], density gradient centrifugation (DGC), and polymer-based precipitation (PBP) method aka pellet-down method [25]. The SEC method is based on the differential elution profiles of particles of different sizes running through a porous polymer (gel filtration matrix). Small particles such as proteins are slowed down by entering the pores of the polymer's pores while EVs travel quicker along the column and elute first [26,27]. Compared with other methods, SEC could separate different EV subtypes and proteins with minimal effect on EV composition and structure [28]. In the present study, we use the SEC method to separately collect different EV fractions such as large and small EVs and investigate EV markers such as tetraspanins (CD9, CD63) and two homologs of HSP90-α and β derived from macrophages.
Visualization of EVs enables tracking the dissemination of the vesicles in vitro and in vivo. Palmitoylation is the covalent attachment of fatty acids to cysteine (S-palmitoylation) and less frequently to serine and threonine (O-palmitoylation) residues of membrane-associated proteins. The cellular and vesicular membrane can be labeled with palmitoylation signal (Palm)-tagged fluorescent proteins [3,29]. We have developed stable cell lines that express Palm fused with green fluorescent proteins (palmGFP) and with tandem dimer Tomato (palmTomato), which visualize the membrane of the cells and their EVs [20]. The uptake of EVs by recipient cells are also visible and quantitative using the palm fluorescent EV system [3]. However, it has been uncertain how differently monocyte or macrophage-derived EVs are transmitted to cancer cells. Therefore, in the present study, we investigated EVs' transmission efficiencies derived from monocytes and macrophages to oral cancer cells.
Tumor-derived EVs were able to modulate the tumor microenvironment to foster tumorigenesis, tumor progression, and metastasis. The tumor-derived EVs often contain HSPs [19,30]. We have shown that metastatic oral cancer and metastatic prostate cancer secrete the high levels of HSP90-positive EVs and EV-free HSP90 while targeted knockdown of HSP90α, HSP90β, and their cochaperone CDC37 powerfully diminish EV-driven malignancy events and macrophage M2 polarization in oral cancer [17,19,20,31]. At the same time, HSP90-positive TAMs accumulate in the metastatic oral cancer tissues and infiltrate into the tumors [20].
However, macrophage-derived EVs heterogeneity and their cytotoxicity to cancer cells has not been well understood. In the present study, we aimed to (i) separete small exosomes, large exosomes, and large EVs and to investigate the transmission efficiency and cytotoxic effect of macrophage-derived EVs to oral cancer cells.

EV fractionation (SEC and PBP)
For extraction of EVs, the size exclusion chromatography (SEC) method and the modified polymer-based precipitation (PBP) method were used as charted. THP-1 cells were seeded at 1 × 10 6 cells in a 10-cm dish and cultured for 24 h. Cells were then treated with 20 nM PMA for 24 h or untreated. The medium was changed to 4 mL serum-free fresh medium, and the cell culture supernatant was collected at 48 h after the medium changing. The cell culture supernatant was centrifuged at 2,000 × g for 30 min at 4℃ to remove detached cells, and then the supernatant was centrifuged at 10,000 × g for 30 min at 4℃ to remove cell debris.
The concentrate was incubated with Total Exosome Isolation Reagent (Thermo Fisher Scientific) at 4℃ overnight and then centrifuged at 10,000 × g for 60 min for precipitation of EVs. The pellet was diluted within 100 μL PBS.
Protein concentrations of EV fractions were analyzed using micro BCA protein assay (Thermo Fisher Scientific).

Particle size distribution
We applied 40 μL of each concentrated fraction to a disposable solvent resistant micro cuvette (ZEN0040, Malvern Panalytical, UK). Particle diameters of the fractions in a range between 0 and 10,000 nm were analyzed in Zetasizer nano ZSP (Malvern Panalytical).

Transmission electron microscopy
A 400-mesh copper grid coated with formvar/carbon films was hydrophilically treated. The concentrated fraction (10 μL) was placed on Parafilm, and the grid was floated on the fraction and left for 15 min. The sample was negatively stained with 2% uranyl acetate solution for 2 min. EVs on the grid were visualized with 10,000 times magnification with an H-7650 transmission electron microscope (TEM) (Hitachi, Tokyo, Japan) at Central Research Laboratory, Okayama

Fluorescence-labeling of cells and EVs
The fluorescent EV reporter constructs were kindly gifted from Dr. Charles P. Lai. Briefly, The lentiviral reporter constructs of CSCW-palmitoylation signal-tandem dimer Tomato (palmT) and CSCW-palmitoylation signal-EGFP (palmG) as described [3,20]. PalmG and PalmT sequences were inserted in SCSGW2 lentivector. For virus production, HEK293T cells were transfected with PalmG or PalmT constructs, psPAX2 packaging plasmid, and pMD2.G envelope plasmid using polyethyleneimine (PEI) Max transfection reagent (Polysciences, Warrington, PA). THP-1 or HSC-3 cells were infected using the spinfection method with the viral solution. Infected/transduced stable cells were selected using puromycin. Single clones were isolated by limiting dilution method. The stable cells were designated as THP1/palmG and HSC-3/palmT.

EV transmission assay
As donor cells, THP-1/palmG cells were seeded at 1 × 10 6 cells in a 10-cm dish and cultured for 24 h in RPMI-1640 with 10% FBS and then treated with 20 nM PMA for 24 h or untreated. The cells were washed twice with PBS, and then the culture media was replaced with serum-free medium and cultured for a further 2 days. The culture supernatant was collected and centrifuged at 2,000 × g for 30 min at 4℃ to remove detached cells and the supernatant were tenfold concentrated with Amicon Ultra-0.5 Centrifugal Filter Unit and used as CM.
For qualitative analysis, fluorescence images were taken using a confocal laser scanning microscopy (CLSM) system (LSM 780 META, Carl Zeiss) as described before [20,32]. Briefly, recipient cells were seeded on a type I collagen-coated coverslip in a 24-well plate at a density of 2.2 × 10 4 cells per well and cultured for 24 h in D-MEM with 10% FBS. For quantitative analysis, the fluorescence intensity of EV transmission was measured using ArrayScan High Content Screening (HCS) System (Thermo Fisher Scientific) with channel 485/549/bright field as described previously [3,32,33]. Briefly, as recipient cells, HSC-3/palmT were seeded 5,000 cells per well in a 96-well NanoCulture Plate (NCP) (Medical & Biological Laboratories, Nagoya, Japan) and cultured for 24 h in 200 μL mTeSR1. CM was applied in a ratio 1:1 with a fresh culture medium.

Cell viability
Cell viability was measured as described previously [34]. ATP content was quantified using CellTiterGlo (CTG) Luminescent Cell Viability Assay (Promega, Madison, WI). To examine the effects of CM, whole EV, and co-culture, HSC-3 cells were seeded at 2 × 10 4 cells per well in a 24-well plate. After 24 h of culturing, 1) CM: equal volume CM was added to the culture medium, 2) Co-culture: a culture insert with a 0.45-μm pore (Greiner, Kremsmunster, AU) was placed, and THP-1 cells (2 × 10 4 cells/well) were seeded in the culture insert, 3) EV protein: the whole EV fraction (1.25 μg /ml) prepared using the SEC or PBP method was added. After culturing for another 48 h, cells were detached using Trypsin/EDTA and suspended with 150 μL CTG solution and then incubated for 10 min at 37℃. Luminescence was measured using the plate reader Gemini XP SOK (Molecular Devices).
To examine the effects of EV fractions separated using the SEC method, HSC-3 cells (5,000 cells /well/ 200 μL) were seeded to a 96-wells plate. After 24 h of culturing, 10 μL of EV fractions (Fr. 1-6, 7-9, or 10-20) were added to each well (quadruplicate). After culturing for another 48 h, 150 μL medium was removed from each well and 50 μL CTG solution was added. Luminescence was measured on the Gemini XP SOK.

Flow cytometry
Phycoerythrin (PE)-conjugated mouse monoclonal anti-human antibodies were used for cell surface staining: anti-CD14, anti-CD80, anti-CD206 and anti-CD68 antibodies (all from BioLegend, San Diego, CA). Isotype-matched negative controls (CD14, mouse IgG2aκ; CD68, mouse IgG2bκ; CD80 and CD206, mouse IgG1κ) were used throughout the investigations. THP-1 cells and PMA-induced macrophages were cultured in a serum-free medium for 48 h. PMA-induced macrophages were unattached with Accutase (Innovative Cell Technologies, San Diego, CA) and centrifuged at 200 × g for 5 min at 4℃. The pellet was washed with PBS containing 0.5% bovine serum albumin (BSA, Wako). Fc receptors were blocked by incubating the cells with 25 μg/ml Fc receptor blocking solution (Human TruStain FcX™, BioLegend) for 10 min at RT before antibody staining. To detect cell surface markers, 5 μL of monoclonal mouse anti-human antibodies or the relevant isotypes were incubated with samples for 30 min at 4℃. For intracellular staining (CD68), cells were fixed with 4% paraformaldehyde (PFA) phosphate buffer solution (Wako) and permeabilized with Nonidet P-40 (NP-40; Sigma-Aldrich). The stained cells were analyzed using a MACSQunat ® X flow cytometer (Miltenyi Biotec, Bergisch Gladbach, DE), and more than 4,000 events were measured per sample. The flowcytometric analyses were carried out using MACSQunat2.6 software (Miltenyi Biotec). All experiments were performed in triplicate.

Statistical Analysis
Statistical significance was calculated using GraphPad Prism and Microsoft Excel. The difference between the two sets of data was examined by unpaired student's t-test. P < 0.05 was considered to indicate statistical significance. Data were expressed as means ± SD unless otherwise specified.

Differentiation of monocytic cells into macrophages
We first examined whether PMA stimulation on THP-1 monocytic cells could foster macrophage differentiation into M0, M1, or M2-types. For this purpose, we examined the size and shapes of non-treated THP-1 and PMA-stimulated THP-1. The cells were attached to the culture dish and enlarged after PMA stimulation (Figure 1 A, B). Flow cytometry showed that the cells with PMA stimulation showed a higher percentage in the larger size (vertical axis; FSC) and higher internal complexity such as granularity (horizontal axis; SSC) than non-treated cells ( Figure 1C). To determine the subtypes of these cells, we next examined the expression of CD14, CD68, CD80 and CD 206 in the untreated vs. PMA-stimulated THP-1 cells. The expression of CD14 was markedly increased in PMA-stimulated THP-1 cells (Figure 1D), indicating that the THP-1 cells were differentiated into the macrophage lineage. Meanwhile, intracellular and cell-surface CD68 (markers of pan-macrophages), CD80 (a marker of M1 macrophages), and CD206 (a marker of M2 macrophages) did not differ compared to their isotype controls ( Figure S1).
These data indicate that PMA-stimulated THP-1 monocytic cells were differentiated into macrophages but neither M1 nor M2 types.

High transmission efficiency of macrophage-derived EVs to oral cancer cells
To monitor macrophage-derived EVs' uptake by oral cancer cells, we labeled the membrane of THP-1 cells and their EVs with palmGFP and of HSC-3 cells with Palm-Tomato. Then, we confirmed that the palm-fluorescent proteins were successfully expressed in these cells (Figure 2A). Besides, THP-1/palmG cells were differentiated by stimulating with PMA to macrophage/palmG, which were more attaching to the dishes and larger shapes (Figure 2A, arrows). Next, we confirmed the transmission of macrophage-derived EVs to macropahges themselves ( Figure S2A) and to oral cancer cells in 2D culture ( Figure S2B). Then, we examined the transmission efficiency of the fluorescent EVs into oral cancer cells in a 3D culture environment. The macrophage-EVs' transmission efficiency to the oral cancer cells was significantly higher than that of monocytes (Figure 2 B, C, D). The monocyte-EVs were transmitted to 15.8 % of the recipient HSC-3 cells while macrophage EVs to 36.7% of the recipient cells ( Figure 2C).
These data indicate that macrophage-derived EVs were highly transmissive to oral cancer cells compared with monocyte-derived EVs.

The cytotoxicity of unidentified macrophage-secreted factor to oral cancer cells
To examine whether macrophage-secreted factors were cytotoxic to cancer cells, we next established intercellular communication experiments using conditioned medium (CM) and transwell-based co-culture system ( Figure 3A). Then, we examined the cell viability of the HSC-3 cells after receiving macrophage-derived CM or in the co-culture system. The CM of macrophages reduced ATP activity of HSC-3 cells (Figure 3B), whereas the CM of monocytes increased the ATP activity. ATP activity of HSC-3 cells after coculturing with macrophages was reduced to 65% ( Figure 3C), whereas co-culturing with monocytes did not alter the ATP activity.
These data suggest that macrophages released cytotoxic factors to oral cancer cells.
The cup-like shapes were found in macrophage-derived and monocyte-derived EVs by TEM, indicating both cell types secreted exosomes ( Figure 5 A-E). Macrophage-EVs' size appeared to be larger and more various than monocyte-EVs (Figure 5 A, B). In the Fr. 1-6, a few "large EVs" with the size of approximately 200 nm were observed by TEM (Figure 5C). In the Fr. 7-9, small EVs (50-100 nm) were observed under TEM ( Figure 5D). These small EVs were suggested as large exosomes (EXO-L) from the size. In the fraction 10-20, few EVs (approximately 30 nm) and a high density of proteins were visible in the field of TEM ( Figure 5E), suggesting that the Fr. 10-20 could contain small exosomes (EXO-S) and vesicle-free proteins.
Next, the particle diameter distribution analysis revealed that the size of monocyte-EVs was peaked at 165.8 nm in the range between 50-500 nm, while that of macrophage-EVs was peaked at 205.1 nm in the wider range between 50-1000 nm (Figure 5 F, G), suggesting that macrophage-EVs' size was larger and more various than monocyte-EVs.
In the Fr. 1-6 of the SEC method, the particle size was between 100-500 nm peaked at 208.9 nm (Figure 5H), suggesting that the Fr. 1-6 contained large EVs (larger than exosomes). In the Fr. 7-9, particle size was ranged between 50-300 nm with a peaked size of approximately 150 nm ( Figure 5I), suggested as large exosomes (EXO-L) from the size. In the fraction 10-20, the particles' size was smaller than 100 nm and peaked at approximately 40 nm (Figure 5J), consist with the TEM. Thus, the Fr. 10-20 could contain small exosomes (EXO-S) and vesicle-free proteins.
To characterize the small and large exosomes or larger EVs by protein markers, we next performed western blotting of tetraspanins (CD9, CD63: established EV markers), HSP90α, HSP90β, and β-actin. CD9 was markedly detected in the EXO-L fraction (Fr. 7-9) (Figure 5K, L; Figure S4, S5). On the other hands, CD63, another tetraspanin family member often found in EVs, was markedly detected in the Fr. 10-20 (EXO-S and free proteins) while degradation of CD63 was seen in the Fr. 7-9 ( Figure S5), suggesting that CD63 might be selectively cleaved out from the EXO-S by metalloproteinases [3,32]. HSP90α, HSP90β and β-actin were markedly found in the Fr. 10-20 ( Figure 5 K, L), suggesting that these HSP90 homologs and β-actin mainly existed in EXO-S and/or as vesicle-free forms. Additionally HSP90β was detectable in the EXO-L fraction (Fr. 7-9) as well, while not contained in the large EV fraction.
These data indicate that different types of EVs can be separately isolated using the SEC method as EXO-L, EXO-S and large EVs. These data also suggest that macrophages released CD9-positive large exosomes (EXO-L), CD63/HSP90α/HSP90β-positive small exosomes.

The cytotoxicity of macrophage-derived fraction 10-20 (small exosomes and free proteins) to oral cancer cells
To examine EVs' cytotoxicity derived from macrophages on oral cancer cells, we stimulated the HSC-3 cells with fractions (Fr. 1-6, 7-9, and 10-20) and then measured ATP activity. The morphology of HSC-3 cells was changed to spindle-like shapes after applying the Fr. 10-20 (EXO-S and free proteins) but not by the Fr.1-6 nor the Fr.7-9 ( Figure 6A). The Fr. 10-20 tended to lower the viability of HSC-3 cells (P=0.0863), which was more effective than the other two fractions ( Figure 6B). We next examined the effects of three different concentrations (100 µg/mL, 200 µg/mL, and 400 µg/mL) of the Fr. 10-20 on the oral cancer cells. The spindle shapes were found in the HSC-3 cells stimulated with these fractions, whereas not found in the unstimulated cells ( Figure 6C). Simultaneously, the cell viability of HSC-3 cells was reduced to approximately 77% after applying the higher concentration (200 µg/mL and 400 µg/mL) of the Fr. 10-20 ( Figure 6D).
These data indicate that macrophages-derived EXO-S could be cytotoxic to oral cancer cells.

Discussion
The heterogeneity of EVs and extracellular particles is recently a crucial issue in biology and medicine. Many methodologies have been developed to separate various EVs according to the surface marker specificity, EV size, and density. The SEC method enables the separation of vesicle and protein fractions according to their size. Indeed, we first collected twenty fractions from macrophage CM and then finally analyzed three fractions, including the large EV fraction (Fr. 1-6, with CD9, HSP90α, and β-actin), EXO-L fraction (Fr. 7-9, with CD9, HSP90, β-actin), and EXO-S with vesicle-free proteins (Fr. 10-20, containing CD63, HSP90, and β-actin). Our data suggest that macrophage-derived anti-cancer cytotoxic factors could be contained in EXO-S or exist in vesicle-free form (Figure 7). Protein markers of EVs had been well investigated by many researchers and organized in MISEV2018 [16]. However, some proteins such as HSP90, β-actin, and even tetraspanins have been detected in many fractions, including large and small exosomes, large and small EVs, and vesicle-free protein fractions in the context-dependent manner [17,19,20]. We previously found that EVs prepared by a modified PBP method contained marked concentrations of CD9, β-actin, and HSP90 than those from the ultracentrifugation method [36]. Reproducibly, CD9, and HSP90α were markedly detected in macrophage-derived EVs isolated by the PBP methods in the present study. However, it has been noticed that the precipitating agents may interfere with the structures in EVs, whereas SEC was proved to minimally affect EV composition [28]. In the SEC method, isolation of vesicles from plasma or serum often caused co-isolation of high-density lipoprotein (HDL) [37]. Besides, low-density lipoprotein (LDL) can bind onto the isolated EVs. These characteristics often resulted in lipoproteins contamination in EVs [38].
Consequently, it was difficult to separate lipoproteins with EVs by SEC [23,39]. The key members of the tetraspanin family, CD9 and CD63, were usually used as exosomes markers, whereas they were also detectable in other types of EVs [40]. EVs consisted of heterogeneous subtypes, among which CD9 was mainly contained in Exo-L (90-120 nm), while CD63 was in Exo-S (60-80 nm) [7]. The distinction of EV markers can be explained by their origin because CD9-positive EVs are formed at the plasma membrane and early endocytic locations, while CD63-positive EVs were specifically abundant in proteins associated with endosomes, multivesicular bodies (MVB), vacuoles, and phagocytic vesicles [7,40]. In the present study, Exo-L with a strong CD9 signal was mainly presented in faction 7-9, while Exo-S appeared to be presented only in the fraction 10-20. Moreover, HSP90 showed a stronger signal in the fraction 10-20 (containing EXO-S and free proteins) than in fraction 7-9 (EXO-L). HSP90 plays key roles in phagocytosis and antigen crosspresentation of antigen-presenting cells (APC) such as dendritic cells and macrophages [30,41,42]. Also, cancer cells often release HSP90-rich EVs and vesicle-free HSP90, which are targetable for cancer therapy [19,20,43]. Indeed, HSP90-rich tumor-infiltrating macrophages were detected in oral cancer specimens of patients [20]. The targeted knockdown of HSP90 in oral cancer cells significantly reduced exosome release and transmission efficiency to macrophages [20]. Thus, both cancer cells and TAMs express HSP90 that involves vesicle uptake, phagocytosis, and antigen presentation. Our findings of HSP90α and HSP90β in the macrophage-derived exosomes and as vesicle-free proteins are vital for further studies on the roles of HSP90 in macrophage polarization and the effect of macrophage-derived EVs on cancer cells.
Cancer cells are often sensitive to extracellular microenvironment and signals. We previously showed that cell stress could induce cell morphology changes such as increasing round and spindle-shaped cells, and these morphological changes involve epithelialto-mesenchymal transition (EMT) [19,42]. A similar morphological variation was observed in the present study after treating the oral cancer cells with the fraction 10-20 (containing EXO-S and free proteins) ( Figure 5). It has been shown that Exo-S, exomeres [7], and other types of proteins generally were contained in the Fr. 10-20 obtained by the SEC method. Comparing to the other EV fractions, the cytotoxicity in the CM and EXO-S might have been caused by cytokines such as interferon-γ (IFN-γ) [44], tumor necrosis factor α (TNFα), and effector molecules such as nitric oxide (NO) secreted by macrophages. HSPs are molecular chaperones that assist functional protein folding, presumably of cytotoxic factors [42,45,46]. Although our current study identified the transmission of macrophage-EVs and the characteristic of different EV fractions, it is still unclear what kind of factors in the CM was cytotoxic to the cancer cells.
Nonetheless, apart from cytokines, it has been shown that macrophage-derived EVs involved anti-tumor activity. For instance, macrophages exposed to environmental stresses, such as nutrient deprivation, hypoxia, and the signaling factors released from cancer cells could recruit diverse intracellular factors and package them into secretory vesicles to mediate the primary response against cancer cells until the immune surveillance system was activated [47]. It was also found that macrophages cultured in glucosedepleted medium secreted human glycyl-tRNA synthetase 1 (GARS1)-EVs, which involved immunological defense response against tumorigenesis and could promote cancer cell death [47,48]. Relevantly, we emphasize the powerful transmission ability of macrophage-derived EVs. The EV isolation method shown in the present study indicates macrophage-derived EVs' potential roles in targeted therapeutics against tumors.

Conclusions
In conclusion, using a fluorescent EV reporter system, we monitored that macrophage-derived EVs were highly transmissive to cancer cells. Our data indicate that macrophages release cytotoxic factors on oral cancer cells. The cytotoxic vesicle-free fraction