3. Transition Metal-Labeled Exosomes
Another effective strategy for Evs application as tumor diagnostic tools is labeling with metallic compounds detectable by MRI, PET, and SPECT.
In a proof of concept study, Abello et al. [85
] developed a system for in vivo tracking of gadolinium (Gd)-labeled exosomes in osteosarcoma. Exosomes were isolated from human umbilical cord mesenchymal stem cells (HUC-MSC) by sequential centrifugations; their mean hydrodynamic diameter, measured by DLS, was 171 ± 42 nm (polydispersity index PDI 0.043 ± 0.03), while the zeta potential was −16.03 ± 0.72 mV. To label the exosomes, the authors synthesized a DSPE-DOTA (1,4,7,10-tetraaza cyclododecane-1,4,7,10-tetraacetic acid) conjugate that complexed Gd. Gd-DOTA-DSPE was included in the exosome membranes by the lipid insertion technique (1:1 ratio of exosomes:Gd-DOTA-DSPE) and the resulting Gd-labeled exosomes (Gd-Exo) showed an average diameter of 148 ± 3 nm (with an increased PDI of 0.36 ± 0.001); the successful insertion of Gd-DOTA-DSPE in the membranes was demonstrated by the zeta potential decrease (−19.70 ± 0.82 mV). As the presence of free Gd is a well-known limitation in the use of Gd-based contrast agents [104
], in order to verify the absence of Gd leakage from the complex, the authors performed a Gd release study comparing Gd-Exo with Magnevist®
(gadopentetate dimeglumine), used as reference: Magnevist®
displayed a burst release (20%) of Gd in the first 4 h, while after 72 h Gd-Exo released only 2% of the ion. By CLSM the authors confirmed exosome internalization in mouse osteosarcoma K7M2 cells using Rhodamine B (RhB), and flow cytometry revealed that exosome internalization was time-dependent, reaching the maximum (40%) after 24 h. Gd-Exo and Gd/RhB double-labeled exosomes were investigated on K7M2 mouse and 14B human osteosarcoma cells: while the single and double-labeled exosomes showed the same dose-dependent antiproliferative effect on K7M2 cells, Gd-Exo caused no dose-dependent inhibition of 14B cell growth but, conversely, Gd/RhB exosomes induced a dose-dependent proliferative effect. The mechanism of proliferative inhibition has not been clarified, although induction of apoptosis was excluded. An in vivo test was carried out on immunodeficient NU/NU nude mice bearing K7M2 osteosarcoma, and MRI images evidenced a clear accumulation of the vesicles in the tumor area 90 min after injection. A biodistribution study performed by inductively coupled plasma-mass spectrometry (ICP-MS) highlighted that liver (38%), kidney (8%) and spleen (2%) were the organs involved in vesicles elimination. Importantly, the authors evidenced higher tumor accumulation of Gd-Exo (18%) compared to Magnevist®
, explainable with the biological origin of this nanocarrier. The authors also compared the fluorescence bioimaging results of exosomes and NP labeled with DiR (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide), a near-infrared fluorescent cyanine dye, using the same animal model, and observed that the tumor fluorescence intensity from Exo-DiR increased more slowly but eventually doubled the one from NP-DiR. The authors concluded that HUC-MSC exosomes accumulated in mouse osteosarcoma tumors 24–48 h post-injection and that Gd labeling was more specific and accurate compared to fluorescence imaging.
A strategy to maximize Gd delivery to the target site, prolonging the contrast agent circulation time and minimizing the dose to avoid toxic side effects, was proposed by Rayamajhi and his group [95
]. These authors developed hybrid vesicles fusing macrophage-derived Evs with Gd-conjugated liposomes (Gd-HEVs) for MRI application. Evs were isolated from mouse macrophage J774A.1 cells by centrifugation, while liposomes composed of Gd-lipid, egg phosphatidylcholine and cholesterol (20:50:30) were prepared by the thin-film hydration method. The Gd-lipid amount was optimized by selecting the formulation with the best stability in terms of size, and the resulting Gd-liposomes were labeled with fluorescent RhB via hydrophobic insertion. The hybrid vesicles were obtained by mixing Gd-liposomes with the Evs in a 5:1 lipid:protein weight ratio and vortexing, followed by sonication and extrusion through a 200 nm polycarbonate porous filter. The Gd-HEV mean hydrodynamic diameter was 127 ± 2 nm (PDI 0.18 ± 0.01), and the zeta potential was −33 ± 4 mV; they featured good serum stability after 30 days with only a slight increment in size (up to 160 ± 10 nm) and PDI (up to 0.3 ± 0.05). Bradford assay, Fourier-transform infrared spectroscopy, sodium dodecyl sulfate–polyacrylamide gel electrophoresis and dot blot analysis confirmed that Gd-HEVs maintained the marker proteins (CD9, CD11b, CD63, CD81 and TSG101) of naive Evs; fluorescent-based energy transfer (FRET) analysis confirmed the success of the membrane fusion process. Gd3+
release measured through ICP-MS was less than 0.2% after 24 h, 1.5% at 48 h and 3.4% at 72 h, compared to Magnevist®
, which released approx. 10%, 18% and 23% Gd3+
at the same time points. The safety of this contrast agent was also evaluated in vitro on mouse osteosarcoma K7M2 and normal fibroblast NIH/3T3 cells, resulting in over 70% cell viability at the highest concentration tested (0.2 mg/mL). The magnetic properties of Gd-HEVs were studied using 3T clinical MRI, highlighting a contrast enhancement as compared to Magnevist®
used at a similar concentration. In vitro tests on K7M2 and NIH/3T3 cells transfected with early endosome green fluorescent protein revealed a co-localization of the red RhB-labeled HEVs and green early endosome, especially in K7M2 cancer cells, as quantified by a CLSM imaging analysis program. By the same study, Gd-HEVs showed higher internalization than naïve Evs and Gd-liposomes in K7M2 cells. Gd-HEVs retained the typical functionality of Evs, but the authors were not able to explain the doubled cytokine stimulation functionality, absent in Gd-liposomes and Magnevist®
. The biodistribution was studied by fluorescence bioimaging and MRI in NU/NU immunodeficient mice bearing osteosarcoma (Figure 4
): fluorescence bioimaging revealed Gd-HEVs enhanced accumulation in lungs, kidney and tumor, while a real-time MR image, taken during injection, showed Gd-HEVs retention in the blood vasculature without extravasation, while Magnevist®
exhibited immediate extravasation. In accordance with the magnetic property characterization, Gd-HEVs showed enhanced contrast intensity compared to Magnevist®
also in vivo, although Gd-HEVs did not show a clear improvement of contrast enhancement in the tumor area compared to the surrounding tissue. The low response was confirmed by ICP-MS on collected organs, revealing that only 0.63% of the Gd injected dose accumulated in the tumor, which, however, was still significantly higher than Magnevist®
(0.1%). The authors hypothesized that this low tumor homing might be due to a prolonged blood retention.
Banerjee et al. [86
] developed a platform based on radiolabeled small Evs (sEVs) for PET/MRI imaging. sEVs were isolated by sequential ultracentrifugation of human umbilical cord blood mononuclear cells cultured under hypoxic conditions, which makes the vesicles more bioactive than those secreted under normoxic conditions [105
]. sEVs were characterized by DLS, NTA and TEM (average diameter of about 100 nm, and zeta potential of −34 mV), and the surface expression of the protein markers CD9, CD45 and CD63 was confirmed by flow cytometry. The DOTA maleimidoethylacetamide derivative was then conjugated to the free thiol groups on sEVs surfaces (880 ± 150 DOTA per sEV) for complexation of the PET emitter 64
, which was assayed through ICP-MS, resulting in 30% complexation efficiency; DLS revealed no statistical changes in size and zeta potential of sEV-DOTA-Cu in comparison with the original sEVs. In vivo stability was evaluated by intravenous injection in C57BL/6J mice: after 4 h, more than 95% of sEV-DOTA-Cu was present as a complex in the blood. A biodistribution study on mice treated with sEV-DOTA-Cu evidenced that after 3 h the radioactive signal was still significant, compared to the control mice receiving only DOTA-Cu, which showed almost no radioactivity already after 1 h. The highest radioactive signal was localized in the liver (around 30%), followed by lung, kidney, bowel, stomach and brain (0.4–0.5%). The authors stated that this surface labeling procedure was superior to previously applied strategies using 68
Ga-NOTA (2,2′,2″-(1,4,7-triazacyclononane-1,4,7-triyl)triacetic acid), which induced vesicle aggregation, and to 99m
Tc encapsulation, which caused an increase in the salivary gland activity due to 99m
Tc leakage [96
]. However, the authors did not exhaustively deal with aspects relating to MRI.
was also employed by Shi et al. [87
] to label exosomes from breast cancer cells. The vesicles were functionalized with reactive-amine NOTA and were pegylated to enhance half-life and tumor retention. The average hydrodynamic diameter of NOTA-exosome-PEG, measured by DLS, showed a dramatic difference between the elaboration by intensity or number distributions (226.6 ± 17.9 nm and 63.6 ± 10.0, respectively). NOTA-exosome-PEG exhibited a clear difference of zeta potential (−3.3 ± 3.2 mV) compared to neat exosomes (−33.4 ± 2.2 mV) and NOTA-exosomes (−26.6 ± 2.5 mV). After 24 h of storage, 64
Cu-NOTA-exosomes-PEG featured significantly higher serum stability (95.7 ± 0.9%) than 64
Cu-NOTA-exosomes (80.4 ± 1.3%), probably because, according to the authors’ opinion, the PEG chains hindered the competitive non-specific binding of serum proteins. In vivo, PET imaging was performed on 4T1 tumor-bearing mice to evaluate the biodistribution of 64
Cu-NOTA-exosomes and 64
Cu-NOTA-exosomes-PEG after intravenous injection: while 64
Cu-NOTA-exosomes displayed robust hepatic clearance and very short blood circulation time, 64
Cu-NOTA-exosomes-PEG exhibited opposite features. Moreover, 64
Cu-NOTA-exosomes-PEG gradually accumulated in the tumor (2.7% ± 0.3 ID/g), with three-fold higher tumor uptake 24 h post-injection compared to 64
Cu-NOTA-exosomes. The authors observed that the presence of PEG also enhanced tumor contrast, as highlighted by the tumor/muscle ratios (3.5 ± 1.1 at 1 h, 6.1 ± 1.1 at 4 h, 7.4 ± 0.5 at 24 h). These results were confirmed by histological studies performed with pegylated exosomes labeled with a fluorescent dye.
HER2-targeted exosomes radiolabeled with fac-[99m
synthon for SPECT tumor imaging were studied by Molavipordanjani et al. [88
]. Targeted exosomes were obtained by transfecting parent human embryonic kidney HEK293T cells with a lentiviral vector to produce exosomes expressing on membranes DARPin G3, a ligand of the HER2 receptor [106
]. Radiolabeling was obtained by incubating fac-[99m
synthon with the exosomes (RCP 96.5%): the labeled exosomes showed a slight size increase at DLS, compared to unlabeled ones (92.8 nm, PDI 0.309 vs. 76.5 nm, PDI 0.299) and resulted stable in saline for 24 h (RCP 96%). Using a gamma-counter, the binding to different cancer cell lines was investigated, evidencing that SKOV-3 cells, which are characterized by the highest HER2 receptor expression, displayed the highest affinity; the authors confirmed that the binding was dependent on DARPin G3-HER2 interaction by saturating SKOV-3 cells with trastuzumab, an anti-HER2 antibody. The biodistribution in normal BALB/c mice and SKOV-3 xenografted nude mice revealed that the highest portion of radioactivity was accumulated in the liver and kidneys, while tumor tissue radioactivity uptake was only 2.75 and 1.47% of injected radioactivity/g, measured 1 and 4 h post-injection, respectively. Pre-injection of SKOV-3 xenografted mice with trastuzumab caused the disappearance of the signal in the tumor area, corroborating the hypothesis that 99m
Tc-exosomes acted through active targeting even in vivo.
In the context of SPECT tracers, 111
In-radiolabeled engineered exosomes were developed by Rashid et al. [89
], with the aim of targeting pro-tumorigenic M2 macrophages, having a pivotal role in breast cancer cell dissemination. Exosomes were isolated by sequential centrifugations from human embryonic kidney HEK293 cells, transfected with a lentiviral vector to obtain vesicles expressing the CSPGAKVRC peptide, which specifically binds to CD206-positive M2 macrophages. Exosome hydrodynamic radius (92.2 ± 4.6 nm, measured by NTA) was not significantly different from the diameter of exosomes obtained from non-transfected cells (106 ± 14 nm). The binding ability of the engineered exosomes was tested in vitro on Raw264.7 macrophages treated with IL-3 and IL-4, stimulating CD206 expression. The engineered exosomes were labeled with DiI and binding and internalization were observed by fluorescence microscopy. To confirm the targeting ability in vivo, the same DiI-labeled exosomes were administered to 4T1 tumor-bearing Balb/c mice and tissue immunofluorescent staining confirmed the co-localization of exosomes and CD206-positive macrophages. The exosomes were labeled by incubation with In-111-oxine at room temperature for 30 min, following a previously described method [107
]: the labeling efficiency was 98%, and, after 24 h of storage in FBS, more than 92% of 111
In was still bound to the exosomes. In vivo SPECT and CT evidenced the accumulation of 111
In-exosomes in tumor, lung (metastatic site), spleen, lymph nodes and bones of mice bearing 4T1 breast cancer (Figure 5
). As a counterproof, a significant radioactivity reduction was observed in the tumors of a group of animals receiving Clophosome-A, a macrophage-depleting agent. A biodistribution study performed measuring the radioactive signal with a gamma-counter confirmed these results, revealing signals in kidneys and bladder, indicating that these organs were involved in the radiolabeled exosome excretion. Finally, the authors investigated the therapeutic potential of these targeted exosomes, further engineered to express the Fc portion of mouse IgG2, successfully inducing antibody-dependent cell-mediated cytotoxicity.
5. Exosomes Beyond Oncology
Although cancer diagnosis and therapy are the main application fields for exosomes and similar vesicles, a few different intervention areas have been explored in the latest years.
First, the exploitation of exosomes as diagnostic tools or drug carriers requires accurate preliminary investigation of their distribution pattern in vivo. Hwang et al. [96
] conceived to label macrophage-derived exosome-mimetic vesicles with 99m
Tc-hexamethylpropyleneamineoxime (HMPAO, also known as exametazime) for SPECT/CT tracing in living mice. 99m
Tc-HMPAO is a suitable radiotracer for Evs labeling because, being highly lipophilic, it easily penetrates the lipid bilayers and is trapped inside the vesicle by reacting with the sulfhydryl groups of glutathione. Exosome-mimetic vesicles were prepared from murine macrophage Raw264.7 cells extruded through polycarbonate porous membrane filters, followed by purification through density gradient ultracentrifugation. Vesicle radiolabeling was carried out by first labeling HMPAO with 99m
, followed by incubation of 185–370 MBq of 99m
Tc-HMPAO with 100 μg/100 μL exosome-mimetic vesicles. Purification from free 99m
Tc-HMPAO was performed by two different methods, either using a size exclusion column (RCP 99.6% ± 3.3) or by centrifugation (RCP 93.7% ± 5.4). No significant changes occurred in vesicles size (approx. 213 nm), measured by NTA before and after radiolabeling, and the presence of the exosome marker protein CD63 was confirmed. The serum radiochemical stability of 99m
Tc-HMPAO vesicles was approx. 90% after 5 h. In vivo SPECT/CT imaging performed on male BALB/c mice showed radioactivity accumulation in the liver and spleen 30 min after injection, and in the salivary glands after 3 h, while no brain accumulation was evidenced. On the contrary, control mice treated with 99m
Tc-HMPAO showed brain accumulation and delayed salivary gland uptake. Moreover, exosome-mimetic vesicles obtained from HB1.F3 human neural stem cells and labeled as described showed the same distribution as obtained with Raw264.7-derived vesicles. Finally, the in vivo biodistribution was investigated by ex vivo radioactivity counting at different time points, highlighting that the main difference in biodistribution between 99m
Tc-HMPAO vesicles and 99m
Tc-HMPAO pertained to brain and liver. The authors also evidenced the advantage of exosome mimetic vesicles over natural exosomes, as the amount of exosomes produced by Raw264.7 cells was half the amount of exosome-mimetic vesicles obtained by sequential extrusions of the parent cells, while the in vivo biodistribution was similar.
Busato et al. [82
] developed an exosome-based diagnostic platform for neurodegenerative diseases by loading stem-cell-derived exosomes with commercial magnetite USPION. Murine adipose stem cells (ASCs), isolated from male C57BL/6 mice, were labeled with USPION following two different procedures: in the first one, a fixed number of ASCs were incubated with increasing concentrations of USPION for 24 and 72 h; in the second procedure, a fixed amount of USPION was incubated for 72 h with increasing cell number. To evaluate the internalization of USPION, the cells were stained by the Prussian blue method, and counterstained with nuclear fast red. Using a light microscope, the authors highlighted a dose- and time-dependent internalization of USPION in ASCs, and TEM analysis confirmed the success of the labeling procedure, with the USPION localized in the cytoplasmic compartment. After optimization of the MRI parameters, exosomes were isolated from ASCs incubated with USPION for 24 h, by using an exosome isolation kit based on targeted filtration. Through a bicinchoninic protein assay, exosome concentration was measured in terms of protein content. The quantification of internalized iron oxide NP in the exosomes was performed with a previously described procedure [112
] based on the use of a potassium ferrocyanide solution and determination of the absorbance at 700 nm by UV spectrophotometry. After the in vitro MRI visualization of USPION-labeled exosomes immobilized in an agarose matrix, the authors performed an in vivo test on male C57BL/6 mice, which were intramuscularly injected with USPION-loaded exosomes and with plain USPION containing similar amounts of iron; the USPION-labeled exosomes were clearly detected by MRI and by histological analysis of gastrocnemius.
Betzer et al. [84
] developed a non-invasive CT imaging platform based on exosomes labeled with gold NP (GNP) as a powerful diagnostic tool for various brain disorders. The exosomes, isolated from mesenchymal stem cells by sequential centrifugations, were analyzed by NTA for diameter (127.7 nm) and zeta potential (−29.2 mV). To optimize the size for exosome labeling, the authors prepared both 20 and 5-nm-sized GNP (TEM measured), following two different synthetic pathways. To prevent aggregation, PEG7 was used in both procedures and D-(β)-glucosamine hydrochloride was coupled to GNP to promote exosome internalization. The loading, carried out by incubation at 37 °C and confirmed by dark field microscope images, did not cause any significant change in exosome diameter, but induced a clear zeta potential decrease. Atomic absorption spectroscopy allowed the authors to observe a decrease in GNP uptake when the incubation was carried out at 4 °C. Thus, the authors hypothesized the involvement of an active energy-dependent process; in particular, they verified that the uptake was mediated by the GLUT-1 transporter by interaction with glucose moieties of the GNP coating, especially in the case of the 5 nm NP. Brain accumulation and whole-body distribution of the GNP-exosomes were studied after intravenous (IV) and intranasal administration (IN) in C57bl/6 male mice, by quantifying the gold amount in explanted organs: 1 h after administration, the amount of exosomes in the brain was significantly higher in the case of IN administration compared to the IV route, and 24 h after IN administration a substantial presence of exosomes was still detectable. Regarding whole-body distribution, the main difference between the two administration routes was found in the liver accumulation, which was higher for IV injection. In vivo CT images of a murine model of stroke showed that GNP-exosomes were detectable in the stroke area after 24 h, and this result was confirmed by region of interest and inductively coupled plasma (ICP) analysis. Finally, to establish whether the proposed exosome labeling protocol was a reliable and consistent imaging technique, the authors double-labeled the exosomes with GNP and red PKH26 dye and, comparing brain CT images with spectral unmixing fluorescence images, they were able to identify both exosomes and GNP in the same striatal slices, confirming that the nanoparticles were retained in the exosomes up to 24 h after labeling.
The unique properties of exosomes and the observation of the exosome-mediated cell-cell signaling imply that exosomes might be employed as nanocarriers for drug delivery, for both diagnostic and therapeutic purposes in oncology. Sequential ultracentrifugation is considered the gold standard procedure in exosome isolation, even if many research groups prefer the use of proprietary isolation kits based on selective exosome precipitation. Noteworthy is the modification of the parent cell culturing conditions applied by Banerjee et al. [86
] in order to maximize Ev bioactivity.
Regarding the diagnostic application, MRI and CT-based methods seem to be the most promising techniques for Ev detection, and most research groups favor exogenous loading methods for the inclusion of the detectable agent. As for the therapeutic application, hyperthermia can easily be obtained by using SPION, GNP or QD as labels; few researchers have reported the application of exosomes for vehiculating drugs, perhaps due to the limited drug loading.
Despite the advances made in the last decades in the comprehension of the mechanisms by which exosomes are generated and about their functions, either in healthy or diseased conditions, the challenges in the development of efficient GMP protocols for their isolation and characterization are still to be solved. The Minimal Information for Studies of Extracellular Vesicles 2018 (MISEV2018) issued guidelines for nomenclature, parent cell collection and pre-processing, and Ev isolation, concentration and characterization [113
]. However, a lack of uniformity in Ev size classification and characterization is still evident, and guidelines are still missing for Evs drug loading and functionalization.
Another important issue to be considered is the potential immunogenicity of exosomes and other Evs; in this review, we reported several studies employing murine cells as a source of extracellular vesicles, but, in the future, the use of human-derived Evs would be necessary for clinical application. Indeed, while some papers in the literature reported the absence of immunogenic effects of human Evs in mice, no data are available for the contrary.
Finally, deeper insights in the implications of using cancer-derived Evs are needed. In fact, by vehiculating proteins and genetic material deriving from tumor cells, the use of these vesicles as drug carriers might be a double-edge sword.