Chemically induced site-specific homing of systemically administered extracellular vesicles

There is a growing interest in therapy applications of cell-free extracellular vesicles (EVs) but systemically administered EVs predominantly accumulate in the liver and thereby limit their applications in extrahepatic tissues. One way of diverting EVs to extrahepatic tissues is by tethering or expressing targeting vectors onto the surface of EVs but this requires chemical or genetic modifications with potential drawbacks. There is an unmet need to direct systemically administered EVs to a desired site without genetic or chemical modifications. Herein, we employed positron emission tomography imaging of 89 Zr-labeled EVs to explore the ability of various substances to alter EV trafficking in the body. Subcutaneous injections of phenothiazine and poly(ethylene) glycol into the hind limb of BALB/c mice significantly increased the accumulation of intravenously administered 89 Zr-labeled EVs to the hind limb. Expanding these findings, we investigated EVs isolated from various sources, the influence of timing of the administration of phenothiazine with respect to labeled EV administration, and the relationship of the trafficking of EVs to the site of phenothiazine administration with overall change in biodistribution. The findings support further exploration of phenothiazine analogs for site specific enhancement of EV trafficking in therapeutic applications.

The term exosome was first coined to define extracellular vesicles (EVs) originating from normal and neoplastic cell lines by selective exfoliation of the plasma membrane of these cells (Trams et al., 1981). The function of these vesicles was not known, but co-incubation of 32 P-labeled cells and harvested exosomes resulted in dephosphorylation of 32 P-labeled membrane proteins of the cells induced by phosphoester hydrolase present on the exosomes, suggesting an interaction between exosomes and cells (Trams et al., 1981). Earlier work by Poste and Nicolson (1980) showed that B16 mouse melanoma cells spontaneously shed intact vesicles. Fusion of these vesicles originating from the highly metastatic B16 subline (F10) to the poorly metastatic B16 subline (F1) using poly(ethylene)glycol and phytohemagglutinin-P significantly increased the metastatic potential of the F1 subline (Poste and Nicolson, 1980). The metastatic capacity of F10 vesicle-modified F1 cells was reversed after the removal of the F10 vesicle components from the plasma membrane (Poste and Nicolson, 1980). These early studies showed that exosomes might be involved in the intercellular communication/interaction with host and target cells.

In addition to intercellular communication, the role of EVs in modifying the originating cells was also identified in maturing reticulocytes (Frazier et al., 1982; Harding et al., 1983; Pan et al., 1985; Johnstone et al., 1987; Johnstone, 1992). During maturation of reticulocytes to erythrocytes, the transferrin receptor density in the cell membrane decreases from 400,000 molecules/cell to <20,000 molecules/cell (Frazier et al., 1982). Harding et al. (1983) tracked the Au–transferrin–transferrin receptor complex in maturing reticulocytes by electron microscopy and showed that the transferrin receptor is externalized into the medium by releasing transferrin receptor-bound vesicles during maturation. Pan et al. (1985) also showed shedding of the transferrin receptor in maturing reticulocytes by 125 I or FITC-labeled anti-transferrin receptor antibody. Later, additional enzymatic activity and presence of transporters, including acetylcholinesterase activity, glucose transporters, nucleoside transporters, Na + -independent amino acid transporters, were identified in vesicles shed from maturing reticulocytes (Pan et al., 1985; Johnstone et al., 1987; Johnstone, 1992).

Over the years, different vesicles have been identified and classified based on their size and biogenesis (Yuana et al., 2013; Doyle and Wang, 2019). Externalized vesicles (aka extracellular vesicles or EVs) are classified as exosomes, microvesicles, and apoptotic bodies (Doyle and Wang, 2019) (Figure 1). The exosomes, which are the smallest (30–150 nm), are generated by exocytosis, followed by microvesicles (100–1,000 nm) generated by budding, and apoptotic bodies (50–5,000 nm) that are released after programmed cell death (Doyle and Wang, 2019). Advances in proteomics and EV purification techniques have allowed the identification of the unique signals present on the EVs isolated from blood, urine, lymph, and cerebrospinal fluid (Thery et al., 2002). Proteomics data revealed the release of EVs by various cells in culture media and different organs/tissues (Garcia-Martin et al., 2022; Abdelmohsen et al., 2023). These divergent EVs had different functions depending on their source (Raposo et al., 1996; Zitvogel et al., 1998; Fevrier and Raposo, 2004; Tan et al., 2021). For example, exosomes derived from B lymphocytes induce antigen-specific major histocompatibility complex (MHC) class II-restricted T cell responses, suggesting their role in antigen presentation in vivo (Raposo et al., 1996). Similarly, EVs secreted by dendritic cells were found to induce antigen-dependent T-cell-mediated immune responses in mice (Zitvogel et al., 1998). Trams et al. (1981) and MacDonald and Salem (2023) earlier also raised the question of whether the shedding of vesicles and their interaction with a target cell or target organ occur in vivo. Later studies confirmed that EVs engage in intercellular and systemic communication between organs (Fevrier and Raposo, 2004; Tan et al., 2021; Neviani and Fabbri, 2015).

With this view, EVs are identified as being responsible for the paracrine component of the therapeutic effects of stem cells (Vizoso et al., 2017). After realizing the therapeutic role of EVs, interest grew. Several clinical trials use EVs for therapy of various disorders, ranging from orthopedic ailments, acute liver injury, wound healing, traumatic brain injury, acute liver injury, myocardial infarction, hearing loss, acquired optic neuropathies, primary ovarian insufficiency, and chronic kidney disease (Tan et al., 2024).

When employed in therapeutic applications, EVs have been administered systemically and locally (Johnson et al., 2023; Sengupta et al., 2020; Shi et al., 2021). Our group earlier used noninvasive PET imaging to assess the biodistribution of intravenously administered [ 89 Zr]Zr-labeled EVs (hepatocyte-derived) in C57BL/6J mice and identified the liver, spleen, and gastrointestinal tract as the major organs of accumulation at the early time point (0 h) and 4 h post-injection (Dasgupta et al., 2020). Similar observations were made in C57BL/6J mice by Wiklander et al. (2015) and in BALB/c mice by Tolomeo et al. (2023), when they systematically administered fluorescent lipophilic dye-labeled EVs and tracked them by IVIS imaging. Tolomeo et al. (2023) showed accumulation of administered EVs in the liver immediately after intravenous injection, and the liver remained the major accumulating organ for up to 24 h post-injection, assessed by whole body IVIS imaging. Upon dissection, the fluorescence signals were also observed in the spleen and lungs at 24 h post-injection. Wiklander et al. (2015) also showed liver as the major organ accumulating EVs as early as 5 min post-injection, and it remained the major accumulating organ, followed by the spleen, gastrointestinal tract, and lungs for up to 48 h after intravenous injection of EVs.

For EV-based therapies, targeting the liver is straightforward as administered EVs accumulate in the liver, but an intravenous injection of EVs is not a preferred route of administration for targeting extrahepatic tissues or organs (Wu et al., 2022; Williams et al., 2025; Su et al., 2025). To target lungs, administration of EVs via intratracheal injection was successful, showing lungs as the major site of accumulation at 3 h and 24 h post-injection (Tolomeo et al., 2023). However, upon dissection, the presence of fluorescence signals in the liver, spleen, and gut at 24 h post-injection showed accumulation of EVs in these organs as well. To target the brain, intranasal administration of EVs showed administered EVs in the brain at 3 h and 24 h post-injection, but later upon dissection, fluorescence signals pertaining to the EVs were also observed in the dissected heart and the gut at 24 h post-injection (Tolomeo et al., 2023).

EVs could be engineered to hold various targeting moieties through chemical modifications of the EV surface or by genetically engineering the EV-producing cells to express desired targeting moieties on the EV surface for any intended application (Choi et al., 2021; Koh et al., 2023). Technical challenges need to be overcome to develop a standardized and scalable method for isolation of therapeutic EVs (Vader et al., 2016; Allelein et al., 2021; Perocheau et al., 2021; Whitford and Guterstam, 2019; Welsh et al., 2024; Chen et al., 2019; Rezaie et al., 2022). Adding engineered EVs for targeted therapy further steepens the hurdle. It is pivotal to ensure onsite delivery, site-specific accumulation, and a favorable clearance profile of the natural or modified EVs for therapeutic application. Noninvasive imaging using optically active or radiolabeled EVs enables noninvasive tracking of EVs, but the natural propensity of EVs to accumulate in the liver limits their broader application. In this study, we evaluated various classes of compounds to systemically direct EVs to specific extrahepatic locations. Interestingly, we found phenothiazines and poly(ethylene) glycol as two classes of compounds that could direct EVs to a desired site as EV attractants. We provide strong evidence of the feasibility of using these compounds as chemoattractants of systemically administered EVs, as followed by PET imaging.

General consideration

EVs derived from SK-N-SH neuroblastoma cells (NBP2-49852), PC-3 prostate cancer cells (NBP2-49856), A549 lung cancer cells (NBP3-11645), B16F10 melanoma cells (NBP2-49866), and U87MG glioblastoma cells (NBP2-49844) were obtained from Novus Biologicals, LLC, Centennial, CO, United States. Lactated Ringer’s solution was obtained from B. Braun Medical Inc., Bethlehem, PA, United States. Methylene blue, azure B, thionin, toluidine blue, patent blue, bromophenol blue, and PEG-400 (PEG) were obtained from Sigma-Aldrich, St. Louis, MO, United States.

Isolation of EVs

EVs were isolated from the mesenchymal stem cells using a qEV exosome isolation kit (iZON Science, Medford, MA, United States) according to the manufacturer’s protocol and standard isolation procedures (Zhang et al., 2022). The culture medium from mesenchymal stem cell cultures was collected and centrifuged at 2,500×g for 20 min. The supernatant after centrifugation was collected and concentrated using a 100 kDa molecular weight cut-off (MWCO) Amicon filter (Millipore Sigma, Burlington, MA, United States). The concentrated medium was transferred to a qEV column. The fraction containing EVs was collected and concentrated using a 10 kDa MWCO Amicon filter (Millipore Sigma, Burlington, MA, United States).

Radiosynthesis of [ 89 Zr]Zr-p-isothiocyanatobenzyl-desferrioxamine ([ 89 Zr]Zr-DBN)

Cyclotron-produced 89 Zr was purified as [ 89 Zr]Zr-chloride ([ 89 Zr]ZrCl 4), as previously described (Holland et al., 2009; Pandey et al., 2014; Pandey et al., 2016; Pandey and DeGrado, 2021; Sharma and Pandey, 2023; Larenkov et al., 2019) and 1-(4-isothiocyanatophenyl)-3-[6,17-dihydroxy-7,10,18,21-tetraoxo-27-(N-acetylhydroxylamino)-6,11,17,22-tetraazaheptaeicosine]thiourea (referred to as DFO-Bn-NCS in Figure 2) was used to synthesize [ 89 Zr]Zr-DBN, as reported in our earlier work (Pandey et al., 2022; Bansal et al., 2015; Bansal et al., 2022; Bansal et al., 2025).

Radiolabeling of EVs with [ 89 Zr]Zr-DBN

Aliquots of 200–500 μg EVs in 500 µL phosphate-buffered saline (PBS) at pH 7.5 were taken in an Eppendorf tube, and 1–2 µL of 1 M sodium carbonate (Na 2 CO 3) was added to obtain a pH of 7.5–8.0. After adjusting the pH, ∼50 µL of [ 89 Zr]Zr-DBN solution (pH 7.5–8.0) (16.65 ± 11.26 MBq or 449.99 ± 304.44 µCi, n = 11) was added, and the final pH remained at 7.5–8.0. The radiolabeling of EVs was achieved by incubating the mixture for 60 min at 37 °C in a thermomixer (Eppendorf Inc., Enfield, CT, United States) (Figure 2). Radiolabeled EVs were purified using a Sephadex G50-based size exclusion column with a 15 mL bed volume in PBS, which was prepared as described previously by Bansal et al., 2025. The radiochemical purity (%) of the radiolabeled EVs was assessed by r-TLC using 20 mM sodium citrate (pH 5.0) and methanol (1:1) as the mobile phase (equations shown below). In the r-TLC, the [ 89 Zr]Zr-EVs showed R f = 0.032–0.039 (Supplementary Material Figure S1A), and [ 89 Zr]Zr-DBN showed R f = 0.720–0.799 (Supplementary Material Figure S1B).

The radiolabeling efficiency (%) of EVs derived from different sources was calculated as shown below:

Animals

BALB/c mice were used and obtained from the Jackson Laboratory (Bar Harbor, ME, United States). At the time of injection, animals in every group had a 1:1 male-to-female ratio with an average age of 9–11 weeks old and average body weights of 22.76 ± 2.18 g (n = 40).

PET imaging and ex vivo biodistribution studies

[ 89 Zr]Zr-EVs (0.52 ± 0.28 MBq, or 14.2 ± 7.6 µCi) (n = 40) were administered via tail vein injection under brief anesthesia (1–2% isoflurane). The EV dose injected in each animal was normalized to the protein amount (∼25 µg). For PET imaging studies, each animal underwent a 10-min static PET scan at 1 h or 24 h after the injection of radiolabeled EVs. To test the effect of chemical agents on the subsequent biodistribution of the radiolabeled EVs, each mouse was re-anesthetized and ∼50 µL of a) 1% methylene blue (MB) formulated in lactated Ringer’s solution, b) 1% azure B formulated in lactated Ringer’s solution, c) 1% thionin formulated in lactated Ringer’s solution, d) 1% patent blue formulated in lactated Ringer’s solution, e) 0.4% bromophenol blue formulated in lactated Ringer’s solution, f) 0.5 µg PEG-400 (PEG) formulated in lactated Ringer’s solution, or g) 100% lactated Ringer’s solution (control) was injected subcutaneously in the foot pad of a hind limb. The mouse was then allowed to recover, and 1 h later, the animal underwent a 10-min static PET scan. The PET scans were performed on a small animal micro-PET/X-ray system, Genesys 4 (Sofie BioSystems, Dulles, VA, United States). In a separate study, to test whether pre-injection of 1% MB can direct EVs, 1% MB was injected into one hind limb at 10 min or 30 min prior to injection of the radiolabeled EVs. PET scans were performed at 1 h after injection of the radiolabeled EVs. The concentrations of MB and PEG were selected based on the 1% USP grade methylene blue standard and the PEG specific solubility profile.

The PET images were reconstructed using the Genesys 4 imaging software installed on the imaging system. The reconstructed PET images were visualized in the coronal and sagittal planes and scaled to the SUV using MIM 7.2.7 image analysis software (MIM Software Inc., Cleveland, OH, United States). The SUV threshold on the PET images was adjusted to delineate organs and regions of interest (ROIs).

ROIs were manually drawn across various PET image planes using a 3D brush, and the mean radioactivity concentration value in kBq/g was calculated and converted to µCi/g. After completion of imaging, the animals were euthanized by cervical dislocation, and organs/tissues were collected, weighed, and counted for radioactivity using a gamma counter for assessing the ex vivo biodistribution of [ 89 Zr]Zr-EVs. The radioactivity concentration (µCi/g) was calculated in different organs/tissues using weight and radioactivity measurements of excised organs/tissues. The SUV mean values from PET imaging and ex vivo biodistribution data were calculated using decay-corrected activity values in the following formula:

Clodrosome® or encaposome® treatment

To test the effect of pretreatment of mice with clodronate-encapsulated liposome (Clodrosome®) or control liposomes (Encaposome®) on EV trafficking, macrophages in BALB/c mice were depleted by injecting 40 mg/kg Clodrosome® or Encaposome® via tail vein injection every other day for 6 days, as per the manufacturer’s instructions (Encapsula NanoSciences LLC, Brentwood, TN). After the 6th day, radiolabeled EVs were injected into the Clodrosome®- or Encaposome®-treated mice. This was followed by subcutaneous injections of either 1% MB or control lactated Ringer’s solution in the center of the foot pad of a hind limb. A PET image was performed on the animal 1 h after the injection of MB, followed by euthanization. The organs and tissues were extracted following euthanization for ex vivo biodistribution analysis.

Statistics

The uptake values assessed by PET imaging and ex vivo biodistribution analysis were compared using the T.TEST function of the Microsoft Excel program (Microsoft Corporation, Redmond, WA) for unpaired student’s t-test analysis. The differences in uptake values were considered statistically significant when the p -value <0.05. The data were visualized in terms of graphs using GraphPad Prism version 10.0.3 for Windows (GraphPad Software, Boston, MA).

Considering the therapeutic potential of EVs and their predominant accumulation in the liver after systemic administration, it would be impactful to identify interventions that can direct EVs to extrahepatic sites. An initial serendipitous observation in our laboratory led us to embark on the exploration of various compounds as site-specific chemoattractants of EVs. To evaluate EV trafficking in the body, we utilized our in-house developed 89 Zr-radiolabeling methodology of EVs, followed by PET imaging of the 89 Zr distribution over time after their IV administration in mouse models. The chemical agents, their structures, and the level of EV response to chemically induced site-specific homing are listed in Table 1. To effectively evaluate the potential of directing or attracting properties of the compounds listed below (Table 1), we first radiolabeled EVs derived from various types of cells. We evaluated their normal distribution post-intravenous injection in one cohort of normal animals, followed by another cohort of normal animals intravenously injected with the same batch of radiolabeled EVs but also injected subcutaneously with nanomoles to micromoles of the compounds listed in Table 1 at a site of intended accumulation of EVs. The chemical agents were tested both before and after intravenous injection of radiolabeled EVs.

Direct radiolabeling of extracellular vesicles with [ 89 Zr]Zr-DBN

The EVs derived from various sources were successfully radiolabeled with our in-house developed [ 89 Zr]Zr-DBN and purified using a Sephadex G50 size exclusion column (Bansal et al., 2015; Pandey et al., 2022; Bansal et al., 2025). The [ 89 Zr]Zr-EV fraction with the highest radioactivity had >98% radiochemical purity as assessed by r-TLC and showed a range of radiolabeling efficiency (%) from 16% to 24% (Supplementary Material Table S1).

Evaluation of chemically induced site-specific homing of extracellular vesicles via PET imaging and ex vivo biodistribution

Two groups of animals were used to evaluate each chemical test agent. One group was used as a control without administration of the test compound, and the other group was used as a test group with administration of the test compound. For each compound, animals (n = 10) were equally divided into two groups, 1 h after administration of [ 89 Zr]Zr-EVs. One group (control, n = 5) received a subcutaneous injection of lactated Ringer’s solution in the left hind limb, and the other group (MB, n = 5) received a subcutaneous injection of the test compound, MB. PET images were acquired 1 h after administration of the chemoattractant test compounds. Figure 3 shows the results obtained for experiments with methylene blue (MB) as a chemoattractant. The PET images (Figure 3A) were analyzed, and the uptake of [ 89 Zr]Zr-EVs in various organs was assessed in terms of standardized uptake value (SUV mean) and is presented in Figure 3B. In both groups, the major organs accumulating the administered [ 89 Zr]Zr-EVs were the heart, liver, lung, and spleen, and some of the test compound accumulated in the gastrointestinal tract (Figure 3; Supplementary Material Table S2). It was a key finding that MB injection in one of the hind limbs in the MB group caused the PET signal (from [ 89 Zr]Zr-EVs) to markedly increase near the MB injection site, which was absent in the lactated Ringer’s solution injection site in the control group (Figure 3B; Supplementary Material Table S2). The accumulation over the hind limb region increased approximately 7.6-fold (p < 0.05). In the MB group, there was a trend (not statistically significant) for decreasing uptake or retention of [ 89 Zr]Zr-EVs in lungs and liver, and a statistically significant decrease in retention in the heart and spleen compared to the control group (Figure 3; Supplementary Material Table S2). This suggested that MB injection caused [ 89 Zr]Zr-EVs to move from the major accumulating organs to the chemoattractant injection site.

The accumulation of EVs at the MB injection site was also checked when 1% MB was injected 30 min (n = 3) and 10 min (n = 3) before the injection of [ 89 Zr]Zr-EVs (Figure 4; Supplementary Material Table S2). These data were compared with the previously shown data with 1% MB administration 1 h post-administration of the [ 89 Zr]Zr-EVs. Based on PET imaging analysis, the radioactivity of [ 89 Zr]Zr-EVs was reduced in the heart, spleen, and liver of the animals injected with MB 30 min before injection of [ 89 Zr]Zr-EVs compared to when animals were injected with MB 1 h after injection of [ 89 Zr]Zr-EVs (Figures 4A,B; Supplementary Material Table S2). When MB was injected 10 min before the injection of [ 89 Zr]Zr-EVs, only the heart and spleen showed a decrease in activity relative to when animals were injected with MB 1 h after the injection of [ 89 Zr]Zr-EVs (Figures 4A,B; Supplementary Material Table S2). This suggests that the heart, spleen, and liver may be the sites of initial [ 89 Zr]Zr-EV accumulation, from which [ 89 Zr]Zr-EVs are redistributed to the site of the MB injection. Additionally, it was identified that pre-injection of MB to a site as early as 10 min before administration of [ 89 Zr]Zr-EVs can also preferentially direct [ 89 Zr]Zr-EVs to the site of the MB injection.

For ex vivo biodistribution analysis, the animals in the control and MB group were euthanized after imaging, and organs/tissues were extracted for counting radioactivity and assessing SUVs in various organs. The ex vivo biodistribution analysis of major organs (Figure 5A; Supplementary Material Table S3) and excretory organs (Figure 5C; Supplementary Material Table S3) provided us a clearer picture. On comparing the control and MB groups (1 h after injection of [ 89 Zr]Zr-EVs) in the ex vivo biodistribution data, we found that the uptake or retention of [ 89 Zr]Zr-EVs in most of the major and excretory organs decreased in the MB group (1 h after injection of [ 89 Zr]Zr-EVs) vs. the control group. Of the major organs, the decrease was found to be statistically significant in the lungs and the blood pool. This showed that the subcutaneous injection of MB in the hind limb in the MB group (1 h after injection of [ 89 Zr]Zr-EVs) caused movement of [ 89 Zr]Zr-EVs in circulation to the injection site. The directing of [ 89 Zr]Zr-EVs from circulation to the MB injection site occurred within 2 h of injection of [ 89 Zr]Zr-EVs, which suggests that [ 89 Zr]Zr-EVs were in circulation in this time frame. It is known that systemically administered EVs in blood circulation are taken up by liver in the first hour after administration and their distribution in lungs and spleen peaks between 2 h and 12 h post-injection, and the EV levels remain high in liver for up to 2 days post-injection, which is within the time frame of our observations (Kang et al., 2021; Saunderson et al., 2014). Hence, it is highly probable that [ 89 Zr]Zr-EVs in the circulation, spleen, and liver are sequestered to the injection site in this time frame. The blood pool as a source of [ 89 Zr]Zr-EVs is also plausible because the heart signal (assessed by quantifying PET signal in myocardium) decreased in the MB group vs. control group based on imaging (Figure 3B), and when the heart was excised and perfused to remove blood for ex vivo biodistribution, the difference in heart (MB vs. control) was no longer statistically significant (Figure 5A; Supplementary Material Table S3). On comparing the ex vivo biodistribution of [ 89 Zr]Zr-EVs within the MB group, we found that there was significantly lower activity in the blood and liver of animals injected with MB 1 h after injection of [ 89 Zr]Zr-EVs compared to animals injected with MB 30 min and 10 min prior to injection of [ 89 Zr]Zr-EVs (Figure 5A). We believe that lower blood activity could be attributed to the time difference in MB administration, and lower uptake in the excised portion of the liver could be due to the heterogeneous uptake of the [ 89 Zr]Zr-EVs in the liver, as clearly evident in liver PET images.

It was interesting to observe that other than the blood pool, the PET signal of [ 89 Zr]Zr-EVs was also decreased in the lungs (based on ex vivo measurement) and spleen (based on PET imaging measurement) in the MB-treated group compared to the control (Figures 3B, 5A). Two possible scenarios may explain the origin of [ 89 Zr]Zr-EVs accumulating in the chemoattractant-administered region based on these observations: 1) the [ 89 Zr]Zr-EVs remain in the circulation at the time of chemoattractant injection, or 2) the [ 89 Zr]Zr-EVs backflux from the lungs and spleen into the circulation as the [ 89 Zr]Zr-EVs are redistributed to the chemoattractant injection site.

Sequestration of EVs from blood circulation by the lungs, liver, and spleen can also be attributed to resident macrophages in these organs (Imai et al., 2015). To test whether depletion of macrophages affects MB mediated directing of [ 89 Zr]Zr-EVs in any way, we exposed the animals to standard Clodrosome® or Encaposome® (control) treatment using a kit (as mentioned in the methods section) as per the manufacturer’s instructions (Naito et al., 1996; van Rooijen et al., 1989). Clodrosome® is a clodronate-encapsulated multilamellar liposome suspension, whereas Encaposome® (control) is a multilamellar liposome suspension without clodronate. We compared the distribution of intravenously administered [ 89 Zr]Zr-EVs after MB injection in Clodrosome®-treated (n = 4) vs. Encaposome®-treated (control) (n = 5) and untreated mice (n = 5) (Figures 6A,B, 7; Supplementary Material Tables S4, S5). The analysis of PET images after subcutaneous injections of MB in these three cohorts showed that the SUV mean of [ 89 Zr]Zr-EVs in the MB injected hind limb was higher in the Clodrosome®-treated cohort than in the untreated cohort (Figure 6B; Supplementary Material Table S4). In contrast, there was no statistical difference in the SUV mean in the MB injected hind limb of Encaposome®-(control) treated and untreated cohort (Figure 6B; Supplementary Material Table S4). An increase in bioavailability of [ 89 Zr]Zr-EVs in the Clodrosome®-treated cohort due to depletion of EV scavenging macrophages could explain the reason for the higher signal in the MB injected hind limb in the Clodrosome®-treated cohort compared to the untreated or Encaposome®-(control) treated cohort. Based on PET image analysis, other than the hind limb, only the spleen showed a difference in uptake of [ 89 Zr]Zr-EVs, while no differences were observed in the heart, lungs, and liver of all three cohorts. There was higher uptake of [ 89 Zr]Zr-EVs in the spleens of mice treated with Clodrosome® or Encaposome® (control) vs. untreated mice, which suggests that uptake of EVs in the spleen is not entirely dependent on the activity of macrophages within 2 h of injection of [ 89 Zr]Zr-EVs.

Following imaging, the animals were euthanized, and organs/tissues were collected for ex vivo biodistribution analysis (Figure 7; Supplementary Material Table S5). Ex vivo biodistribution analysis showed no difference in the uptake of [ 89 Zr]Zr-EVs in the major organs (heart, lung, spleen, eyes, and bone), excretory organs (intestine, bladder, stomach, and kidney), blood, brain, and muscle in the three cohorts (Figure 7; Supplementary Material Table S5). Contrary to PET image analysis, the ex vivo biodistribution analysis showed no statistical difference in the SUV mean of [ 89 Zr]Zr-EVs in the MB-injected hind limb of the untreated, Encaposome®-treated, and Clodrosome®-treated cohorts. Uptake of [ 89 Zr]Zr-EVs in the MB-administered hind limb region was elevated in all three cohorts, with no significant difference in uptake level between them. The reason for the lack of difference in the Clodrosome®- and Encaposome®-treated groups relative to the control is not understood. One possible explanation is that there is a tissue sampling issue as the signal of [ 89 Zr]Zr-EVs was distributed diffusely in the hind limb, and the excision of the hind limb resulted in leaking and hence incomplete collection of all the directed [ 89 Zr]Zr-EVs from the hind limb. The PET images were acquired from an intact animal, so there was no loss of the signal from the hind limb, and hence, the differences were evident in the PET images.

All the previous results involved subcutaneous injection of MB after 1 h of intravenous injection of [ 89 Zr]Zr-EVs. We then tested whether administered [ 89 Zr]Zr-EVs can be directed to the hind limb upon injection of MB 24 h after intravenous injection of [ 89 Zr]Zr-EVs (Figure 8). Based on PET images acquired 1 h after injection of MB (Figures 8A, B; Supplementary Material Table S6), it was encouraging to observe that MB was able to direct [ 89 Zr]Zr-EVs to the hind limb even 24 h after intravenous injection of [ 89 Zr]Zr-EVs. The lack of obvious bone uptake of 89 Zr indicated that the labeling of the EVs remained stable over the 24 h period because free 89 Zr would be redistributed to bones.

On comparing the biodistribution profiles of [ 89 Zr]Zr-EVs at 1 h and 24 h post-injection based on PET imaging, it was observed that the [ 89 Zr]Zr-EVs signal decreased in the heart, lung, liver, spleen and hind limb with time in the control group that received the injection of lactated Ringer’s solution in the hind limb (Figure 9A; Supplementary Material Tables S2, S6). Low signals were observed in the lungs and spleen at the 24 h time point (Figures 8A, 9). This shows that with time, [ 89 Zr]Zr-EVs are being cleared from the body. In the MB group, similar observations were made in the heart, lungs, spleen, and liver (Figure 9B; Supplementary Material Tables S2,S6).

In the case of the hind limb injected with MB, the PET signal was lower when MB was used to direct EVs 24 h after injection of [ 89 Zr]Zr-EVs than when MB was used to direct EVs 1 h after injection of [ 89 Zr]Zr-EVs (Figure 9; Supplementary Material Table S6). This could be attributed to a lower level of non-sequestered “labile” [ 89 Zr]Zr-EVs in the body at 24 h post-injection than at 1 h post-injection of [ 89 Zr]Zr-EVs, consistent with a lower amount of [ 89 Zr]Zr-EVs in the lungs and spleen pool, as per PET imaging analysis, that could have been used as a source of EVs directed to the hind limb.

Different dyes such as patent blue, bromophenol blue, azure B, and thionin were tested. Among these, only azure B and thionin were able to direct the systemically administered [ 89 Zr]Zr-EVs to the hind limb, although not to the same degree as MB. The successful directing of [ 89 Zr]Zr-EVs by azure B and thionin can be observed in the PET images (Figure 10A; Supplementary Material Table S7) and the corresponding SUV mean graph (Figure 10B; Supplementary Material Table S7). Apart from dyes, subcutaneous injection of poly(ethylene) glycol was also able to direct the [ 89 Zr]Zr-EVs to the hind limb (Figure 10; Supplementary Material Table S7). The commonality between MB, azure B, and thionin is the phenothiazine core chemical structure, which could be contributing to the EV-directing properties.

To this point, we established that MB served as a powerful chemoattractant of [ 89 Zr]Zr-EVs, as tested in EVs isolated from stem cells. We then tested whether EVs from sources other than stem cells could show the same effect. EVs derived from prostate cancer cells (PC-3), neuroblastoma cells (SK-N-SH), lung cancer cells (A549), melanoma cells (B16F10), and glioblastoma cells (U87MG) were tested. Of these, EVs derived from PC-3, SH-N-SH, and A549 responded to subcutaneous injection of MB in the hind limb (Figure 11A; Supplementary Material Table S8), but EVs from B16F10 and U87MG did not respond (Figure 11B; Supplementary Material Table S8). It is not clear why EVs from B16F10 and U87MG did not respond, but their biodistribution profiles were very different from those of the EVs from PC-3, SH-N-SH, and A549 cells. There was less uptake in the heart and more uptake in the liver and spleen of EVs from B16F10 and U87MG after systemic administration of radiolabeled forms compared to the uptake patterns of EVs from PC-3, SH-N-SH, and A549 cells (Figure 11B; Supplementary Material Table S8).

Plausible mechanism: We hypothesize that changes in vascular permeability induced by MB and PEG at the injection site may have triggered cellular communication pathways that subsequently facilitated the localization of systemically administered EVs to the region of MB/PEG administration. An independent and dedicated study is necessary to establish a well-defined mechanism underlying chemically induced EV redistribution, including potential variability among EVs isolated from different cellular sources, in their response to MB.

EVs from multiple sources were successfully isolated and radiolabeled with the PET radioisotope 89 Zr. Our [ 89 Zr]Zr-DBN-based radiolabeling strategy enabled stable incorporation of the radioisotope into EVs, allowing real-time tracking of intravenously administered EVs in mice for up to 24 h using PET imaging. Through noninvasive PET imaging, we identified two classes of compounds, phenothiazines (e.g., methylene blue) and poly(ethylene glycol), which both effectively direct EV accumulation to specific target sites. These EV-attractant compounds can be used to reroute systemically administered EVs away from hepatic uptake and toward extrahepatic tissues or other desired anatomical locations.

Overall, this work demonstrates a straightforward and generalizable approach for directing EVs to target tissues without requiring complex EV engineering. This strategy enables precise delivery of EV-based therapeutics to maximize efficacy, reduce off-target effects, and ultimately enhance their potential for clinical translation.