Hollow gold nanoparticles characterization
Transmission electron microscopy (TEM) images of the as-prepared nanoparticles (NPs) are shown in Fig. 1. The HGNs were hollow pseudo-spheres with a thick shell (Fig. 1a), and after polyethylene glycol (SH-PEG) functionalization, a uniform shell of almost 5 nm could be observed coating the HGNs. The mean diameter of HGNs obtained from the TEM images was 38.3 ± 8.3 nm, whereas the mean size of PEGylated NPs (PEG-HGNs) (determined by negative staining the TEM sample) was 44.9 ± 7.6 nm. Both HGNs and PEG-HGNs showed a characteristic localized surface plasmon resonance peak in the NIR region around 850 nm (Fig. 1b). A slight red shift was observed for the PEGylated NPs, possibly attributed to a different dielectric value in the interfacial double layer coating on the NPs [27]. Phase analysis light scattering measurements (not shown) showed that at pH = 7 the surface charge for the HGNs in water was − 15.35 ± 0.84 mV and compared to the one obtained for the PEG-HGNs (− 10.48 ± 0.35 mV) corroborated the efficient PEG coating on the particles surface.
The biological characterization of both HGNs and PEG-HGNs is thoroughly described in Additional file 1 section. The stability and the aggregation state of both HGNs and PEG-HGNs in cell culture media supplemented with 10% FBS was initially evaluated. Additional file 1: Figure S1A shows TEM images of both naked and PEGylated NPs after being in contact with cell culture media for 24 h (naked HGNs were significantly more aggregated than the PEGylated ones). In order to study the amount of protein adsorbed on the NPs, the bicinchoninic acid (BCA) assay was performed on both NP populations in cell culture media supplemented with 10% FBS. Additional file 1: Figure S1B evidences that in the case of HGNs the total protein amount adsorbed on their surfaces increased over time. On the contrary, the protein adsorption on PEGylated NPs was significantly lower compared to the amount adsorbed on naked HGNs in agreement with the well known steric hindrance attributed to PEG.
To corroborate this observation, the zeta potential of both NP-based dispersions in cell culture media was also measured, obtaining zeta potential values of − 3.64 ± 2.73 mV and − 8.50 ± 1.50 mV for HGNs and PEG-HGNs, respectively (Additional file 1: Figure S1C) which is indicative of a higher protein adsorption on the bare nanoparticles, shielding electric charge and giving zeta potential values closer to the isoelectrical point.
The dose-dependent viability of MSCs was evaluated under the presence of HGNs and PEG-HGNs by a metabolic assay. Additional file 1: Figure S2A reveals that 0.125 mg mL−1 was the limit for the subcytotoxic effect. Furthermore, we also confirmed that the presence of both types of NPs in the cell cultures did not produce significant changes on cell cycle phases at the subcytotoxic doses (Additional file 1: Figure S2B). Finally, the presence of the NPs inside MSCs was evaluated by confocal microscopy (Additional file 1: Figure S2C) and was indirectly quantified by MP-AES using the total amount of gold measured inside the cells (Additional file 1: Figure S2D). These results indicate that significantly more PEG-HGNs were localized inside cells compared to the amounts retrieved for HGNs, probably due to the higher stability and reduced agglomeration in the culture media provided by the PEG-surface functionalization. A better dispersion might increase the chances of nanoparticle internalization. Going deeper on HGNs uptake by MSCs, we also evaluated the preferential pathways that PEG-HGNs followed in the process of being incorporated into the target cells to verify if those strategies were compatible with the exosomal pathway. Additional file 1: Figure S3 reveals that not only clathrin-mediated endocytosis, but also other energy-dependent pathways were the main internalization mechanisms involved in the PEG-HGNs capture by MSCs, and therefore nanoparticles could be efficiently incorporated into the target cells.
Secretion of empty and HGN-loaded exosomes by MSCs
Before attempting to encapsulate HGNs within exosomes, we characterized MSCs derived exosomes (MSCs-EXOs) by physico-chemical and biological techniques. TEM images of MSCs-EXOs isolated from culture supernatants by successive ultracentrifugation steps are shown in Fig. 2. A spherical shape and a characteristic lipidic bi-layer membrane were observed in most of them. The size distribution histograms of exosomes obtained from TEM images, gave an average diameter of 113.2 ± 45.6 nm. Dynamic light scattering (DLS) and nanoparticle tracking analysis (NTA) measurements gave a mean particle diameter around 180 nm in PBS. The measured surface charged of exosomes at pH 7 was − 8.28 ± 3.42 mV likely due to the negatively charged phospholipids and proteins present on the exosomal membrane.
Incorporation of HGNs into MSCs-derived exosomes
We have used confocal microscopy to monitor the association of HGNs with exosomes. To this end, exosomes were labeled with a specific CD63-Alexa488 antibody (blue) the cell nuclei with Draq-5 (yellow) and the HGNs agglomerates were directly observed by reflection (red) (Fig. 3a). Orthogonal projections from confocal laser-scanning microscopy analysis revealed high purple fluorescence pixels on the MSCs cytoplasm, i.e., the merging of the blue and red fluorescence, which corresponds with aggregates of NPs and exosomes or late endosomes co-localization, indicating that multivesicular bodies within the cell already contain HGN-loaded exosomes. After these were excreted, the HGN-loaded exosomes could be easily recovered by ultracentrifugation and subjected to TEM analysis. TEM images of purified exosomes from MSCs culture supernatants after incubation with PEG-HGNs showed that most exosomes were loaded with at least one nanoparticle, confirming the high yield of nanoparticle loading into exosomes (Fig. 3b). The diameter distribution histogram of PEG-HGNs_MSCs-EXOs obtained from TEM images revealed average sizes of 126.9 ± 39.2 nm. These values confirmed that the presence of the NPs inside the exosomes did not affect their endogenous diameter. Furthermore, the size distributions obtained by NTA for control exosomes and for PEG-HGNs_MSCs-EXOs (Figs. 2c, 3e, respectively) were also similar. The measured surface charged of PEG-HGNs_MSCs-EXOs at pH 7 was − 10.26 ± 0.17 mV, confirming that the presence of the PEG-HGNs did not alter the negatively charged phospholipids and proteins present on the exosomal membrane. Western Blot analysis of isolated vesicles showed expression of the exosome-associate proteins CD9 and CD63 [signal at 21 and 53 kDa respectively, were clearly observed (Fig. 3c)]. Comparing total protein amounts obtained by the BCA (Fig. 3d) and exosomes concentration measured by NTA (Fig. 3e), higher amounts of exosomes were significantly secreted when cells were treated with NPs compared to the amounts measured for control samples (untreated cells).
Exosomes as specific vectors of PEG-HGNs between different cell lines
Given the way nanoparticle-loaded exosomes are generated, we postulated that exosomes derived from a cell line are fingerprinted with hallmarks of that cell type and therefore would be preferentially up-taken by the same cell line, even under co-culture conditions with other cell lines. To verify if exosomes from a cell line were specifically captured by the same cell line we developed a co-culture of stem cells and monocytes under optimized conditions for their simultaneous growth. To facilitate identification, both types of cells were labeled with PKH67 (green) and PKH26 (red) dyes, respectively (Additional file 1: Figure S4). Once co-cultures were optimized, the exchange of material between cells (i.e., the intercellular trafficking among MSCs and between MSCs and monocytes) could be easily followed by time-lapse microscopy thanks to the HGNs loaded inside the exosomes.
We performed time-lapse microscopy of MSCs and monocytes cultured separately (control) and co-cultured for 3 days to observe nanoparticle transference in real time. We also used MP-AES to determine the Au content of the different cell populations and supernatants in order to follow overall internalization of PEG-HGNs. Our results (Fig. 4) show that, when fresh nanoparticles (i.e. nanoparticles that had not previously been incubated with any type of cell) were added either to each cell type individually or to co-cultures containing simultaneously MSCs and monocytes, they were significantly up-taken by both cell types (between 1 and 2 μg of gold/cell population). In contrast, when PEG-HGNs were pre-incubated in MSCs and then PEG-HGNs loaded exosomes were put in contact with monocytes, we observed that nanoparticles transference occurs preferentially between MSCs and only a minor fraction of PEG-HGNs was captured by monocytes both at 24 h and 48 h (0.09 μg of gold/cell population at both times). The symmetric experiment was carried out by pre-incubating PEG-HGNs in monocytes and co-culturing them with MSCs. In this case, the amount of gold nanoparticles internalized in MSCs was remarkably small, in spite of the fact that the supernatant concentration was high (0.9 μg of gold/cell population), i.e. there was a large concentration of nanoparticles available. It is reported that when monocytic cells recognize nanoparticles in the media, as efficient scavengers, they firstly internalize NPs and trap them in the endosomal-exosomal pathway releasing those materials again to the media [28]. Figure 4 shows that the majority of gold released to the media by monocytes remained in the media instead of being uptaken by MSCs, thus confirming the preservation of the self-signature of monocytes.
Moreover, the same trend was also observed in Additional file 1: Figure S5. Herein, we employed MP-AES to determine the specific capturing of previously purified PEG-HGNs loaded exosomes of different cellular origin. When MSCs derived exosomes loaded with PEG-HGNs (PEG-HGNs_MSCs-EXOs) were added to a co-culture of MSCs and monocytes, they were preferentially taken by MSCs (0.12 and 0.17 μg of gold/cell population at 24 h and 48 h, respectively) and only a minor fraction of PEG-HGNs was internalized into monocytes (0.015 μg of gold/cell population). In the symmetric experiment, (i.e. PEG-HGNs were pre-incubated with monocytes and then the exosomes produced were released into the co-culture of MSCs and monocytes) the amount of gold nanoparticles internalized in MSCs was remarkably small, in spite of the fact that the supernatant concentration was high (0.14 μg of gold/cell population). While the Au analysis results agree with the selective uptake of PEG-HGNs-loaded exosomes according to their cellular origin, Fig. 4 and Additional file 1: Figure S5 also seem to indicate that, to a lesser extent, some type of non-selective internalization could also be taking place, since a small fraction of Au was observed to be internalized by non-related cells. However, this can be explained as a result of the uptake of free Au nanoparticles that either were present originally in the medium or were excreted through a non-exosomal pathway and therefore lack the cell-specific exosomal coating. These free HGNs were clearly present along with exosome-enclosed HGNs in the culture medium (see Fig. 3b).
To validate the conclusions reached with other cell types, the same study was carried out using monocytes and tumoral cells (B16-F1 cells) instead of MSCs. The results (Fig. 4) show that for separate cultures of each cell, the uptake of PEG-HGNs by monocytes was significantly (ca. 5 times) higher than that obtained for the tumoral cells. When cultured together, a similar uptake ratio is observed for both types of cells. However, when the gold nanoparticles were pre-incubated in B16-F1 cells and transferred to the co-culture with monocytes, only a small amount of PEG-HGNs was captured by monocytes (0.089 μg of gold/cell population). On the contrary, when PEG-HGNs were pre-incubated with monocytes and the resulting HGN-containing exosomes were put in contact with the co-culture of monocytes and B16-F1 cells, only a small fraction was captured by B16-F1 cells (0.042 μg of gold/cell population).
Time-lapse microscopy was used to obtain a direct confirmation of the conclusions obtained from MP-AES analysis regarding the selective uptake of PEG-HGNs according to their origin. To this end, MSCs were incubated with PEG-HGNs and then the MSCs containing internalized nanoparticles were co-cultured with monocytes. As can be seen in the frames presented in Fig. 4c and d, and in the movies included as Additional files 2 and 3, a high amount of nanoparticles were present inside the MSCs, and those were very active in transferring nanoparticles between cells. However, tracking of the excreted material shows that the nanoparticles released by a donor MSC were mostly captured by another MSC and not by monocytes present in the proximity, even though monocytes were close by and often in the path of the released material. Therefore, a specific signature of the exosomes released from a cell type is conserved and used as recognition moiety for the same cell type.
Selective death by hyperthermia mediated using PEG-HGNs-loaded exosomes
As a proof of concept of the possibility of inducing selective cell death using the high intrinsic selectivity of exosome-mediated transport we analyzed the in vitro photothermal effect of HGN-loaded exosomes originated from MSCs on separate cultures containing MSCs, melanoma cells (B16-F1 and B16-F10) and monocytes. To this end, MSCs were incubated with PEG-HGNs (0.1 mg mL−1) and the resulting HGN-containing exosomes were harvested and purified. Then cultures with the above cell lines were exposed to the loaded exosomes for 24 h (the PEG-HGNs_MSCs-EXOs concentration added to each well, was estimated as the ratio between the amount of donor cells from which those exosomes were derived (MSCs) and the number of treated cells). After washing, cell cultures were irradiated with a NIR laser, and viability/toxicity was evaluated by flow cytometry as well as by direct live/dead staining.
In Fig. 5a, it is possible to observe that NIR laser irradiation did not significantly reduce cell viability on B16-F1 cells, B16-F10 cells and monocytes that had been treated with PEG-HGNs containing exosomes derived from MSCs, i.e., they present the same viability as the control samples, (cells subjected to NIR irradiation in the absence of HGNs loaded exosomes). On the contrary, a reduction in viability of almost 50% was observed for irradiated MSCs after treatment with HGN-loaded exosomes derived from MSCs. Very similar results were obtained using a fluorescence inverted microscope to quantify live/dead cells after exposure to MSC-derived exosomes followed by NIR irradiation (Additional file 1: Figure S6). These results again confirm the high specificity of exosome uptake depending on the exosomes origin. HGNs_MSCs-EXOs produced by seeded cells to treated cells, a decrease in the viability of approximately 70% was measured for irradiated MSCs demonstrating a dose-dependent process. In contrast, the viability of B16-F1 cells, B16-F10 cells and monocytes again was unaffected.
Finally, a further study was carried out to confirm selective exchange in the dynamic environment of a co-culture system. To this end, MSCs previously loaded with PEG-HGNs were co-cultured with B16-F1 cells during 48 h, a period sufficiently long to have multiple excretions/uptakes of nanoparticles by MSCs. Figure 5b shows that, after laser irradiation, only MSCs were dead while B16-F1 were still alive. This selective photothermal effect confirmed the lack of exosomal exchange between B16-F1 and MSCs also under co-culture conditions and this fact is consistent with our hypothesis of exosome fingerprinting according to the cellular origin.