**3. Results and Discussion**

#### *3.1. Characterization of EV Samples*

To fabricate pH-sensitive EV blends for targeting heterogeneous tumor cells, we first harvested EVs (BTEVs from BT-474 tumor cells and SKEVs from SK-N-MC tumor cells). The harvested BTEVs and SKEVs were almost spherical, as shown in the TEM image (Figure S3a). Subsequently, these EVs (BTEVs and SKEVs) were identified to exhibit specific protein expressions (Figure S3b). In particular, TSG 101, a conventional marker for EVs, was highly expressed in all EVs [46–48]. The estrogen receptor alpha (ER-α) was significantly detected in the SKEVs, whereas the tau protein (Tau) was only detected in the BTEVs (Figure S3b) [49–52]. Next, we incorporated pH-sensitive polymers (HDEA) and an antitumor model drug (DOX) to the EVs through sonication, as described in the experimental methods section. Finally, we obtained HDEA-anchored BTEVs (HDEA@BTEVs) or HDEA-anchored SKEVs (HDEA@SKEVs). HDOC@BTEVs and HDOC@SKEVs were prepared as pH-insensitive EVs to evaluate the pH-sensitive properties of HDEA@BTEVs and HDEA@SKEVs, respectively. The loading content of HDEA or HDOC in the EV samples was 29–31 wt.%, and the loading content of DOX in the EV samples was 12–16 wt.% (data not shown). Next, we prepared EV blends by physically mixing HDEA@BTEVs and HDEA@SKEVs (a weight ratio of 50/50) in PBS (pH 7.4).

As shown in Figure 1a, we expect the EV blends to home to their parent cells owing to the EVs' homing ability [4,11–17]. Here, the DEAP moieties in the EVs can be protonated at the endosomal pH 6.5, inducing EV destabilization and accelerating the release of encapsulated DOX [3,23–31].

Figure 1b,c shows that the average particle sizes of intact EVs and the EV samples ranged from 105 to 120 nm at pH 7.4. However, the particle size of the HDEA@BTEVs and HDEA@SKEVs increased from 105 nm at pH 7.4 to 200 nm at endosomal pH 6.5 (Figure 1b,c), likely owing to the destabilized membrane of EVs due to the protonated DEAP [3,30,31]. By contrast, intact BTEVs, intact SKEVs, HDOC@BTEVs, and HDOC@SKEVs indicated no significant difference in particle size when the pH of the solution was reduced to pH 6.5; this could be due to the absence of pH-sensitive polymers (HDEA). In addition, as the pH of the solution decreased from 7.4 to 6.5, the zeta potentials of the HDEA@BTEVs and HDEA@SKEVs increased from −18.3 and −18.6 mV to −8.6 mV and −9.2 mV, respectively (Figure 1d,e). It was assumed that the protonated DEAP at pH 6.5 elevated the zeta potential of the HDEA@BTEVs and HDEA@SKEVs [3,30,31]. However, the zeta potential of intact

BTEVs, intact SKEVs, HDOC@BTEVs, and HDOC@SKEVs indicated no significant difference at pH 7.4 and 6.5. Furthermore, the morphological images obtained from TEM reveal that almost spherical HDEA@BTEVs and HDEA@SKEVs at pH 7.4 were destabilized at pH 6.5, and that their structures were partially cracked (Figure 1f). However, the HDOC@BTEVs and HDOC@SKEVs did not show any noticeable changes between pH 7.4 and 6.5. These results demonstrated that pH-sensitive DEAP in the HDEA@BTEVs and HDEA@SKEVs mediated the destabilization of the EVs structure, owing to the protonation of HDEA [3,30,31] at endosomal pH 6.5. *Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 6 of 17 HDEA@BTEVs and HDEA@SKEVs at pH 7.4 were destabilized at pH 6.5, and that their structures were partially cracked (Figure 1f). However, the HDOC@BTEVs and HDOC@SKEVs did not show any noticeable changes between pH 7.4 and 6.5. These results demonstrated that pH-sensitive DEAP in the HDEA@BTEVs and HDEA@SKEVs mediated the destabilization of the EVs structure, owing to the protonation of HDEA [3,30,31] at endosomal pH 6.5.

**Figure 1.** *Cont.*  **Figure 1.** *Cont*.

*Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 7 of 17

*Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 7 of 17

**Figure 1.** (**a**) Schematic illustration of tumor-homing pH-sensitive extracellular vesicles (EVs). Particle size distribution of (**b**) BTEVs and (**c**) SKEVs at pH 7.4 or 6.5. Zeta potential changes of (**d**) BTEVs and (**e**) SKEVs at pH 7.4 or 6.5 (*n* = 3, as multiple experiments, \*\* *p* < 0.01 compared to EVs at pH 7.4). (**f**) TEM images of EVs at pH 7.4 or 6.5. **Figure 1.** (**a**) Schematic illustration of tumor-homing pH-sensitive extracellular vesicles (EVs). Particle size distribution of (**b**) BTEVs and (**c**) SKEVs at pH 7.4 or 6.5. Zeta potential changes of (**d**) BTEVs and (**e**) SKEVs at pH 7.4 or 6.5 (*n* = 3, as multiple experiments, \*\* *p* < 0.01 compared to EVs at pH 7.4). (**f**) TEM images of EVs at pH 7.4 or 6.5. **Figure 1.** (**a**) Schematic illustration of tumor-homing pH-sensitive extracellular vesicles (EVs). Particle size distribution of (**b**) BTEVs and (**c**) SKEVs at pH 7.4 or 6.5. Zeta potential changes of (**d**) BTEVs and (**e**) SKEVs at pH 7.4 or 6.5 (*n* = 3, as multiple experiments, \*\* *p* < 0.01 compared to EVs at pH 7.4). (**f**) TEM images of EVs at pH 7.4 or 6.5.

#### *3.2. pH-Triggered DOX Release 3.2. pH-Triggered DOX Release 3.2. pH-Triggered DOX Release*

The DOX release profile of the EV samples was monitored at pH 7.4 or 6.5 (Figure 2). At pH 7.4, all the EV samples released DOX gradually, and no significant differences were observed in the DOX release rates. However, at pH 6.5, the HDEA@BTEVs and HDEA@SKEVs exhibited a significant increase in the DOX release rates. Specifically, in 48 h, they released approximately 80–85 wt.% of the encapsulated DOX. In addition, regardless of pH change, the DOX@BTEVs, HDOC@BTEVs, DOX@SKEVs, and HDOC@SKEVs showed a passive DOX release of 40–45 wt.%. These results indicate that the HDEA@BTEVs and HDEA@SKEVs recognized a slight pH change, resulting in an accelerated DOX release at pH 6.5. The DOX release profile of the EV samples was monitored at pH 7.4 or 6.5 (Figure 2). At pH 7.4, all the EV samples released DOX gradually, and no significant differences were observed in the DOX release rates. However, at pH 6.5, the HDEA@BTEVs and HDEA@SKEVs exhibited a significant increase in the DOX release rates. Specifically, in 48 h, they released approximately 80–85 wt.% of the encapsulated DOX. In addition, regardless of pH change, the DOX@BTEVs, HDOC@BTEVs, DOX@SKEVs, and HDOC@SKEVs showed a passive DOX release of 40–45 wt.%. These results indicate that the HDEA@BTEVs and HDEA@SKEVs recognized a slight pH change, resulting in an accelerated DOX release at pH 6.5. The DOX release profile of the EV samples was monitored at pH 7.4 or 6.5 (Figure 2). At pH 7.4, all the EV samples released DOX gradually, and no significant differences were observed in the DOX release rates. However, at pH 6.5, the HDEA@BTEVs and HDEA@SKEVs exhibited a significant increase in the DOX release rates. Specifically, in 48 h, they released approximately 80–85 wt.% of the encapsulated DOX. In addition, regardless of pH change, the DOX@BTEVs, HDOC@BTEVs, DOX@SKEVs, and HDOC@SKEVs showed a passive DOX release of 40–45 wt.%. These results indicate that the HDEA@BTEVs and HDEA@SKEVs recognized a slight pH change, resulting in an accelerated DOX release at pH 6.5.

*3.3. Endosomolytic Activity Test*  To evaluate the endosomolytic activity of EV samples, we performed a hemolysis test using **Figure 2.** Cumulative doxorubicin hydrochloride (DOX) release from (**a**) BTEVs and (**b**) SKEVs at pH 7.4 or 6.5 in 48 h (*n* = 3, as multiple experiments). **Figure 2.** Cumulative doxorubicin hydrochloride (DOX) release from (**a**) BTEVs and (**b**) SKEVs at pH 7.4 or 6.5 in 48 h (*n* = 3, as multiple experiments).

#### RBCs with an endosomal-like membrane (Figure 3). At pH 7.4, intact EVs and all the EV samples exhibited negligible hemolytic activity. However, in response to endosomal pH 6.5, the hemolytic *3.3. Endosomolytic Activity Test 3.3. Endosomolytic Activity Test*

To evaluate the endosomolytic activity of EV samples, we performed a hemolysis test using RBCs with an endosomal-like membrane (Figure 3). At pH 7.4, intact EVs and all the EV samples exhibited negligible hemolytic activity. However, in response to endosomal pH 6.5, the hemolytic To evaluate the endosomolytic activity of EV samples, we performed a hemolysis test using RBCs with an endosomal-like membrane (Figure 3). At pH 7.4, intact EVs and all the EV samples exhibited negligible hemolytic activity. However, in response to endosomal pH 6.5, the hemolytic activities

of the HDEA@BTEVs and HDEA@SKEVs increased significantly, likely owing to the proton sponge effect [3,30,31] of the protonated DEAP. By contrast, intact BTEVs, HDOC@BTEVs, intact SKEVs, and HDOC@SKEVs indicated no significant difference in hemolytic activity at pH 6.5. These results indicate that the endosomolytic activity of the HDEA@BTEVs and HDEA@SKEVs facilitated the cytosolic release of drugs in the tumor cells. activities of the HDEA@BTEVs and HDEA@SKEVs increased significantly, likely owing to the proton sponge effect [3,30,31] of the protonated DEAP. By contrast, intact BTEVs, HDOC@BTEVs, intact SKEVs, and HDOC@SKEVs indicated no significant difference in hemolytic activity at pH 6.5. These results indicate that the endosomolytic activity of the HDEA@BTEVs and HDEA@SKEVs facilitated the cytosolic release of drugs in the tumor cells.

*Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 8 of 17

**Figure 3.** Hemolysis activities of (**a**) BTEVs and (**b**) SKEVs (*n* = 3, as multiple experiments, \*\* *p* < 0.01 compared with the control EVs). **Figure 3.** Hemolysis activities of (**a**) BTEVs and (**b**) SKEVs (*n* = 3, as multiple experiments, \*\* *p* < 0.01 compared with the control EVs).

#### *3.4. In Vitro Cellular Uptake and Cytotoxicity of EV Blends 3.4. In Vitro Cellular Uptake and Cytotoxicity of EV Blends*

We verified the cellular uptake behaviors of EV samples for tumor cells (BT-474, SK-N-MC, and Huh7 cells) using a FACSCaliburTM flow cytometer [3,31,37–39]. Figure 4 shows the quantitative results of the cellular uptake of EV samples in tumor cells (BT-474, SK-N-MC, and Huh7). The average fluorescence intensity of BTEVs samples (HDEA@BTEVs, HDOC@BTEVs, and DOX@BTEVs) in BT-474 tumor cells was ~1.4 × 103; however, those of the SKEV samples (HDEA@SKEVs, HDOC@SKEVs, and DOX@SKEVs) and free DOX were ~1.2 × 102 and ~45 (Figure 4a), respectively. Consequently, the uptake of the BTEV samples by the BT-474 tumor cells increased significantly, compared with that of the SKEV samples and free DOX. Meanwhile, as shown in Figure 4b, the uptake of the SKEV samples by the SK-N-MC tumor cells increased, compared with that of the BTEV samples and free DOX. The average fluorescence intensity of the SKEV samples in SK-N-MC tumor cells was ~1.6 × 103; however, those of the BTEV samples and free DOX were ~1.5× 102 and ~57, respectively. However, the uptake of all the EV samples in the Huh7 tumor cells was extremely low and did not exhibit a noticeable difference between all samples (Figure 4c). In addition, the average fluorescence intensity of all EV samples in the Huh7 tumor cells was ~1.1 × 102. We verified the cellular uptake behaviors of EV samples for tumor cells (BT-474, SK-N-MC, and Huh7 cells) using a FACSCaliburTM flow cytometer [3,31,37–39]. Figure 4 shows the quantitative results of the cellular uptake of EV samples in tumor cells (BT-474, SK-N-MC, and Huh7). The average fluorescence intensity of BTEVs samples (HDEA@BTEVs, HDOC@BTEVs, and DOX@BTEVs) in BT-474 tumor cells was ~1.4 <sup>×</sup> <sup>10</sup><sup>3</sup> ; however, those of the SKEV samples (HDEA@SKEVs, HDOC@SKEVs, and DOX@SKEVs) and free DOX were ~1.2 <sup>×</sup> <sup>10</sup><sup>2</sup> and ~45 (Figure 4a), respectively. Consequently, the uptake of the BTEV samples by the BT-474 tumor cells increased significantly, compared with that of the SKEV samples and free DOX. Meanwhile, as shown in Figure 4b, the uptake of the SKEV samples by the SK-N-MC tumor cells increased, compared with that of the BTEV samples and free DOX. The average fluorescence intensity of the SKEV samples in SK-N-MC tumor cells was ~1.6 <sup>×</sup> <sup>10</sup><sup>3</sup> ; however, those of the BTEV samples and free DOX were ~1.5<sup>×</sup> <sup>10</sup><sup>2</sup> and ~57, respectively. However, the uptake of all the EV samples in the Huh7 tumor cells was extremely low and did not exhibit a noticeable difference between all samples (Figure 4c). In addition, the average fluorescence intensity of all EV samples in the Huh7 tumor cells was ~1.1 <sup>×</sup> <sup>10</sup><sup>2</sup> . Overall, these results indicate that EVs with homing ability were efficiently endocytosed to their parent tumor cells [4,11–17].

Overall, these results indicate that EVs with homing ability were efficiently endocytosed to their parent tumor cells [4,11–17]. To further evaluate the internalization of EVs to their parent tumor cells, we also performed in vitro cell imaging studies using a VNIR hyperspectral camera system and a confocal laser scanning microscope [3,30,31,43–45]. The hyperspectral images shows that the DOX (present in EV samples) signal of BT-474 tumor cells treated using BTEV samples (HDEA@BTEVs and HDOC@BTEVs) was higher than that of BT-474 tumor cells treated with SKEV samples (HDEA@SKEVs and HDOC@SKEVs) (Figure 5a), whereas the DOX signal of SK-N-MC tumor cells treated using SKEV samples was higher than that of SK-N-MC tumor cells treated with BTEV samples (Figure 5b). Figure 6 shows confocal images of BT-474 and SK-N-MC tumor cells treated using EV samples. The treated cells were stained using DAPI and WGA-Alexa Fluor® 488 to visualize the cell nuclei and membranes [3,30,31]. The confocal images revealed that the BTEV samples were actively internalized to BT-474 tumor cells, although the SKEV samples were ineffective in interacting with BT-474 tumor cells (Figure 6a). Similarly, the SKEVs samples were efficiently internalized to SK-N-MC tumor cells, whereas the BTEV

samples were poorly internalized to SK-N-MC tumor cells (Figure 6b). These results support the *Pharmaceutics*  homing ability of EVs to their parent tumor cells, which is consistent with the results shown in Figure **2020**, *12*, x FOR PEER REVIEW 9 of 17 4.

**Figure 4.** Flow cytometry profiles of (**a**) BT-474, (**b**) SK-N-MC and (**c**) Huh7 cells treated with free DOX (5 µg/mL) or EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 °C. **Figure 4.** Flow cytometry profiles of (**a**) BT-474, (**b**) SK-N-MC and (**c**) Huh7 cells treated with free DOX (5 µg/mL) or EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 ◦C.

To further evaluate the internalization of EVs to their parent tumor cells, we also performed in vitro cell imaging studies using a VNIR hyperspectral camera system and a confocal laser scanning microscope [3,30,31,43–45]. The hyperspectral images shows that the DOX (present in EV samples) To verify the cellular uptake of the EV blends for BT-474 and SK-N-MC tumor cells, we performed in vitro cell imaging studies using a confocal laser scanning microscope [3,30,31]. As shown in Figure 7a, BT-474 and SK-N-MC tumor cells were first cultured on coverslips. Subsequently, the coverslips were

signal of BT-474 tumor cells treated using BTEV samples (HDEA@BTEVs and HDOC@BTEVs) was higher than that of BT-474 tumor cells treated with SKEV samples (HDEA@SKEVs and moved to empty cell-culture plates and then treated with each EV samples or the EV blends. As shown in Figure 7b, Ce6 dye-incorporated HDEA@BTEVs displayed strong Ce6 fluorescence in BT-474 tumor cells, although the fluorescence intensity of Ce6 was weak in SK-N-MC tumor cells. By contrast, the FITC dye-incorporated HDEA@SKEVs displayed a strong FITC fluorescence signal for SK-N-MC tumor cells, but a weak FITC fluorescence signal in BT-474 tumor cells. More importantly, the EV blends displayed a strong Ce6 fluorescence and a weak FITC fluorescence in BT-474 tumor cells, whereas a weak Ce6 fluorescence and a strong FITC fluorescence in SK-N-MC tumor cells. These results indicate that the EV blends were selectively internalized to each parent tumor cells. *Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 10 of 17 Figure 6 shows confocal images of BT-474 and SK-N-MC tumor cells treated using EV samples. The treated cells were stained using DAPI and WGA-Alexa Fluor® 488 to visualize the cell nuclei and membranes [3,30,31]. The confocal images revealed that the BTEV samples were actively internalized to BT-474 tumor cells, although the SKEV samples were ineffective in interacting with BT-474 tumor cells (Figure 6a). Similarly, the SKEVs samples were efficiently internalized to SK-N-MC tumor cells, whereas the BTEV samples were poorly internalized to SK-N-MC tumor cells (Figure 6b). These results support the homing ability of EVs to their parent tumor cells, which is consistent with the results shown in Figure 4.

**Figure 5.** Hyperspectral images of (**a**) BT-474 and (**b**) SK-N-MC cells treated with EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 °C. **Figure 5.** Hyperspectral images of (**a**) BT-474 and (**b**) SK-N-MC cells treated with EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 ◦C.

Figure 7c shows the cytotoxicity of the HDEA@BTEVs (equivalent DOX concentration of 5 µg/mL), HDEA@SKEVs (equivalent DOX concentration of 5 µg/mL), and the EV blends (equivalent DOX concentration of 5 µg/mL) to the mixed BT-474 and SK-N-MC tumor cells (50:50, cell number ratio). The EV blends exhibited highly increased tumor cell death for mixed BT-474 and SK-N-MC tumor cells. However, the HDEA@BTEVs or HDEA@SKEVs resulted in fewer tumor cell deaths. It is assumed that the efficient uptake of the EV blends in both BT-474 and SK-N-MC tumor cells resulted in the significantly increased antitumor efficacy.

In addition, to evaluate the cytotoxicity of the EV samples, BT-474, SK-N-MC, and Huh7 tumor cells were treated using free DOX and EV samples at the equivalent DOX concentration of 5 µg/mL.

First, HDEA@BTEVs and HDOC@BTEVs exhibited relatively increased BT-474 tumor cell death compared with the SKEVs samples (HDEA@SKEVs and HDOC@SKEVs) (Figure 8a); this was likely owing to the homing ability of the BTEVs. However, the pH-insensitive HDOC@BTEVs resulted in a relatively reduced BT-474 tumor cell death, likely owing to the reduced DOX release (Figure 2). Similarly, the homing ability of the SKEVs resulted in the increased cell-cytotoxicity of the HDEA@SKEVs against SK-N-MC tumor cells (Figure 8b). However, Huh7 tumor cells treated with HDEA@BTEVs or HDEA@SKEVs did not exhibit noticeable levels of cell death. The low cellular uptake of the BTEV and SKEV samples in Huh7 tumor cells resulted in less cell-cytotoxicity for Huh7 tumor cells (Figure 8c). These results indicate that the homing ability of EVs to their parent tumor cells (Figures 4–6) and the endosomolytic activity/endosomal pH-triggered DOX releasing property of HDEA-anchored EVs (Figures 2 and 3) enabled a significantly improved tumor cell death. In addition, all EV samples without DOX showed negligible cytotoxicity up to 3 <sup>×</sup> <sup>10</sup><sup>8</sup> particles/mL in 24 h for incubation with BT-474, SK-N-MC, and Huh7 tumor cells (Figure 8d–f), supporting their non-toxicity. *Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 11 of 17

**Figure 6.** Confocal images of (**a**) BT-474 and (**b**) SK-N-MC cells treated with EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 °C. The cells were stained using 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) and Wheat Germ Agglutinin Alexa Fluor® 488 Conjugate (WGA-Alexa Fluor®488). **Figure 6.** Confocal images of (**a**) BT-474 and (**b**) SK-N-MC cells treated with EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 ◦C. The cells were stained using 40 ,6-diamidino-2-phenylindole dihydrochloride (DAPI) and Wheat Germ Agglutinin Alexa Fluor® 488 Conjugate (WGA-Alexa Fluor®488).

To verify the cellular uptake of the EV blends for BT-474 and SK-N-MC tumor cells, we performed in vitro cell imaging studies using a confocal laser scanning microscope [3,30,31]. As shown in Figure 7a, BT-474 and SK-N-MC tumor cells were first cultured on coverslips. Subsequently, the coverslips were moved to empty cell-culture plates and then treated with each EV samples or the EV blends. As shown in Figure 7b, Ce6 dye-incorporated HDEA@BTEVs displayed

FITC fluorescence signal for SK-N-MC tumor cells, but a weak FITC fluorescence signal in BT-474

each parent tumor cells.

in the significantly increased antitumor efficacy.

tumor cells. More importantly, the EV blends displayed a strong Ce6 fluorescence and a weak FITC fluorescence in BT-474 tumor cells, whereas a weak Ce6 fluorescence and a strong FITC fluorescence in SK-N-MC tumor cells. These results indicate that the EV blends were selectively internalized to

Figure 7c shows the cytotoxicity of the HDEA@BTEVs (equivalent DOX concentration of 5 µg/mL), HDEA@SKEVs (equivalent DOX concentration of 5 µg/mL), and the EV blends (equivalent DOX concentration of 5 µg/mL) to the mixed BT-474 and SK-N-MC tumor cells (50:50, cell number ratio). The EV blends exhibited highly increased tumor cell death for mixed BT-474 and SK-N-MC

**Figure 7.** (**a**) Schematic illustration of in vitro experiments using Ce6 dye-incorporated HDEA@BTEVs, FITC dye-incorporated HDEA@SKEVs, and EV blends [Ce6 dye-incorporated HDEA@BTEVs/FITC dye-incorporated HDEA@SKEVs (50:50 wt.%)]. (**b**) Confocal images of BT-474 cells (left) and SK-N-MC cells (right) treated with Ce6 dye-incorporated HDEA@BTEVs, FITC dye-incorporated HDEA@SKEVs, or EV blends [Ce6 dye-incorporated HDEA@BTEVs/FITC dye-incorporated HDEA@SKEVs (50/50 wt.%)] (equivalent to EVs of 30 µg/mL) for 4 h incubation at 37 °C. (**c**) Cell viability determined by CCK-8 assay of tumor cells treated with EVs or EV blends [HDEA@BTEVs/HDEA@SKEVs (50/50 wt.%)] (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 °C (*n* = 7, as multiple experiments, \*\* *p* < 0.01 compared with each EV sample). In addition, to evaluate the cytotoxicity of the EV samples, BT-474, SK-N-MC, and Huh7 tumor cells were treated using free DOX and EV samples at the equivalent DOX concentration of 5 µg/mL. **Figure 7.** (**a**) Schematic illustration of in vitro experiments using Ce6 dye-incorporated HDEA@BTEVs, FITC dye-incorporated HDEA@SKEVs, and EV blends [Ce6 dye-incorporated HDEA@BTEVs/FITC dye-incorporated HDEA@SKEVs (50:50 wt.%)]. (**b**) Confocal images of BT-474 cells (left) and SK-N-MC cells (right) treated with Ce6 dye-incorporated HDEA@BTEVs, FITC dye-incorporated HDEA@SKEVs, or EV blends [Ce6 dye-incorporated HDEA@BTEVs/FITC dye-incorporated HDEA@SKEVs (50/50 wt.%)] (equivalent to EVs of 30 µg/mL) for 4 h incubation at 37 ◦C. (**c**) Cell viability determined by CCK-8 assay of tumor cells treated with EVs or EV blends [HDEA@BTEVs/HDEA@SKEVs (50/50 wt.%)] (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 ◦C (*n* = 7, as multiple experiments, \*\* *p* < 0.01 compared with each EV sample).

Huh7 tumor cells (Figure 8c). These results indicate that the homing ability of EVs to their parent tumor cells (Figures 4–6) and the endosomolytic activity/endosomal pH-triggered DOX releasing property of HDEA-anchored EVs (Figures 2 and 3) enabled a significantly improved tumor cell death. In addition, all EV samples without DOX showed negligible cytotoxicity up to 3 × 108 particles/mL in 24 h for incubation with BT-474, SK-N-MC, and Huh7 tumor cells (Figure 8d–f),

supporting their non-toxicity.

First, HDEA@BTEVs and HDOC@BTEVs exhibited relatively increased BT-474 tumor cell death compared with the SKEVs samples (HDEA@SKEVs and HDOC@SKEVs) (Figure 8a); this was likely owing to the homing ability of the BTEVs. However, the pH-insensitive HDOC@BTEVs resulted in a relatively reduced BT-474 tumor cell death, likely owing to the reduced DOX release (Figure 2). Similarly, the homing ability of the SKEVs resulted in the increased cell-cytotoxicity of the

*Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 14 of 17

**Figure 8.** Cell viability determined by Cell Counting Kit-8 (CCK-8) assay of (**a**) BT-474, (**b**) SK-N-MC and (**c**) Huh7 cells treated with free DOX (5 µg/mL) or EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 °C (*n* = 7, as multiple experiments), (\*\* *p* < 0.01 compared with the free DOX). Cell viability determined by CCK-8 assay of (**d**) BT-474, (**e**) SK-N-MC, and (**f**) Huh7 cells treated with blank EVs (without DOX) for 24 h incubation at 37 °C (*n* = 7, as multiple experiments). **Figure 8.** Cell viability determined by Cell Counting Kit-8 (CCK-8) assay of (**a**) BT-474, (**b**) SK-N-MC and (**c**) Huh7 cells treated with free DOX (5 µg/mL) or EVs (equivalent to DOX of 5 µg/mL) for 4 h incubation at 37 ◦C (*n* = 7, as multiple experiments), (\*\* *p* < 0.01 compared with the free DOX). Cell viability determined by CCK-8 assay of (**d**) BT-474, (**e**) SK-N-MC, and (**f**) Huh7 cells treated with blank EVs (without DOX) for 24 h incubation at 37 ◦C (*n* = 7, as multiple experiments).
