Tracking Radiolabeled Endothelial Microvesicles Predicts Their Therapeutic Efficacy: A Proof-of-Concept Study in Peripheral Ischemia Mouse Model Using SPECT/CT Imaging

Microvesicles, so-called endothelial large extracellular vesicles (LEVs), are of great interest as biological markers and cell-free biotherapies in cardiovascular and oncologic diseases. However, their therapeutic perspectives remain limited due to the lack of reliable data regarding their systemic biodistribution after intravenous administration. Methods: Applied to a mouse model of peripheral ischemia, radiolabeled endothelial LEVs were tracked and their in vivo whole-body distribution was quantified by microSPECT/CT imaging. Hindlimb perfusion was followed by LASER Doppler and motility impairment function was evaluated up to day 28 post-ischemia. Results: Early and specific homing of LEVs to ischemic hind limbs was quantified on the day of ischemia and positively correlated with reperfusion intensity at a later stage on day 28 after ischemia, associated with an improved motility function. Conclusions: This concept is a major asset for investigating the biodistribution of LEVs issued from other cell types, including cancer, thus partly contributing to better knowledge and understanding of their fate after injection.


Introduction
Critical limb ischemia (CLI) is an advanced form of peripheral artery disease. CLI has a growing incidence, from 500 to 1000 new cases per million every year, in Western Europe and North America [1]. Despite progresses in public health, hygiene, cardiovascular events prevention, or drugs and medical devices developments, CLI remains associated with a decreased quality of life and a high morbidity and mortality worldwide, being responsible for a high rate of amputation [2]. Vascular regenerative medicine has a key role to play in amputation prevention: cell therapies based on endothelial progenitors or mesenchymal stem cells display interesting properties towards ischemic diseases, mainly granted for factor release, immunomodulation, and inflammatory capacity, among other properties [3]. However, many clinical trials are necessary before achieving a specific, safe and effective

Production of Endothelial LEVs
Methods for LEV production and purification followed the latest recommendations from the International Society for Extracellular Vesicles [13] and were submitted to the Transparent Reporting and Centralizing Knowledge in Extracellular Vesicle research consortium (EV-TRACK ID: EV200112) [26,27].

Tunable Resistive Pulse Sensing (TRPS)
TRPS was performed using a qNano Gold TRPS measurement instrument (Izon, Oxford, UK) and CPC400 calibration beads with a mean diameter of 350 nm as calibration standard, following the manufacturer's instructions [30]. Samples were diluted in PBS or HEPES buffer solutions in a small sterile tube and analyzed using a 200-1000 nm NP400 nanopore (Izon, Oxford, UK) at a stretch of 43-45 mm. Voltage was set on 0.30-0.50 V to achieve a stable 110-130 nA current and a 1.4-2.0 kPa pressure, with root mean square noise below 10 pA. Blockade counts setting in this study was fixed at minimum of 500 vesicles count for each, and each sample was analyzed in duplicate. Data were collected and analyzed using Izon Control Suite software v3.3.3.2001 (Izon, Oxford, UK).

Transmission Electron Microscopy (TEM)
LEV pellets were fixed in 2% paraformaldehyde, 2.5% glutaraldehyde overnight, post-fixed in 2% osmium for 1 h on ice, dehydrated in gradient series of acetone baths, and embedded in epoxy resin. Pellets were sectioned on an UC7 ultra-microtome (Leica, Wetzlar, Germany), and sections were contrasted with aqueous uranyl acetate 1% (10 min) and lead citrate (4 min). The grids were observed at 80 kV on a FEI Morgagni transmission electron microscope (ThermoFisher, Waltham, USA) and images were acquired using a MegaView3 camera (Emsis, Muenster, Germany).

Purification of Radiolabeled LEVs from Free [ 99m Tc]Tc-AnnV Radiotracer
To further purify radiolabeled [ 99m Tc]Tc-AnnV-LEVs from free [ 99m Tc]Tc-AnnV radiotracer, the elution profile from qEV single-use SEC of radiolabeled [ 99m Tc]Tc-AnnV-LEVs was first compared to that of non-radiolabeled LEVs. Three samples containing [ 99m Tc]Tc-AnnV-LEVs and 3 samples containing non-radiolabeled LEVs were loaded on qEV singleuse SECs and eluted in the same operating conditions as for LEV purification described supra. LEVs were quantified in each eluted fraction by flow cytometry as described supra.
Then, 3 samples containing [ 99m Tc]Tc-AnnV-LEVs and 3 samples containing free [ 99m Tc]Tc-AnnV only were loaded on qEV single-use SECs and eluted in the same operating conditions as for LEV purification described supra. The activity in V 1 was measured in a dose calibrator (Scintidose, LemerPax, La Chapelle sur Erdre, France).

Stability of Radiolabeled LEVs in Serum
A 40 µL sample of purified radiolabeled LEVs was mixed with 60 µL of human serum and incubated for 0 or 30 min at 37 • C (n = 6 each). The total 100 µL were then re-purified by SEC. The activity in V 1 fraction was measured in a dose calibrator (Scintidose, LemerPax, La Chapelle sur Erdre, France).

In Vivo Experimentations
Procedures using animals were approved by the Institution's Animal Care and Use Committee (Project #14177, CE71 Aix-Marseille University) and were conducted according to the 2010/63/EU European Union Directive and following the ARRIVE 2.0 guidelines [31], by qualified and trained operators in an accredited laboratory (A-13-055-32). A total of 20 female 7-week-old BALB/c mice (Janvier Labs, France) were housed in enriched cages placed in a temperature-and hygrometry-controlled room with a daily monitoring, fed with water and commercial diet ad libitum, and weighed once a week. No animal was excluded during the 28-day follow-up. In vivo experimentations are summarized in experimental paradigm ( Figure 1). MicroSPECT/CT was performed 30 min after the injection of radiolabeled compounds. All the mice were followed up by LASER Doppler for their hind limb perfusion on days 1, 3, 4, 7, 14, 21, and 28. A motility impairment score was calculated on day 28.

Mouse Model of Hind Limb Ischemia Induction and Follow-Up
Unilateral hind limb ischemia was performed after femoral artery excision as previously described [32]. LASER Doppler perfusion imaging (Perimed, Craponne, France) was performed to quantify hind limb perfusion on a 37 • C heated bed under isoflurane anesthesia (induction at 5%, maintenance at 1.5% in air, Iso-vet, Piramal). Each LASER Doppler acquisition lasted 120 s and was repeated 3 times for each animal. On day 0, LASER Doppler was used to check and quantify the induced perfusion defect in the right hind limb, allowing the constitution of 2 homogeneous groups of 10 mice accordingly, for subsequent experiments. Hind limb perfusion was then quantified on days 1, 3, 7, 14, 21, and 28 post-ischemia. Results were expressed as a mean ± SD ratio of ischemicto-contralateral (i/c) hind limb blood flow, and graphically represented as a mean ± SD reperfusion ratio to day 0. A motility impairment score, inspired by Suffee et al., and as previously published [32,33], was calculated for each mouse on day 28, as follows: 1-unrestricted active movement; 2-restricted active foot; 3-use of the other leg only; 4-leg necrosis; 5-self-amputation.

Quantification of the In Vivo Biodistribution of Radiolabeled LEVs by Isotopic Imaging
Micro-single photon emission computed tomography coupled with microtomodensitometry (microSPECT/CT) imaging sessions were performed on a NanoSPECT/CT+ camera (Mediso, Budapest, Hungary). One group received an injection of radiolabeled LEVs ([ 99m Tc]Tc-AnnV-LEVs, 2.0 ± 0.5 × 10 6 LEVs/2.0 ± 0.4 MBq/150 µL, n = 10) in the caudal vein. The other group was injected in the caudal vein with a solution containing only the vehicle (150 µL binding buffer as described in Section 2.2.1, n = 10). A subgroup of the control group received an injection in the caudal vein of 7.1 ± 0.7 MBq/50 µL [ 99m Tc]Tc-AnnV (n = 3). All the 20 mice, whether injected with radioactive materials or not, were anesthetized under isoflurane (induction at 5%, maintenance at 1.5%) on a heated bed, to be equally exposed to anesthesia. Subsequent experiments and data analysis were performed by blind operators. Only the 3 mice injected with free [ 99m Tc]Tc-AnnV and the 10 mice injected with [ 99m Tc]Tc-AnnV-LEVs underwent a 15 min whole-body microSPECT/CT acquisition starting 30 min after the injection. The animals were constantly monitored for breathing during the acquisition. Quantitative region of interest analysis of the SPECT signal was performed using Invivoscope software (Invicro, Boston, MA, USA) to quantify the tissue uptake of [ 99m Tc]Tc-AnnV-LEVs and that of free [ 99m Tc]Tc-AnnV. Tissue uptake values were expressed as a mean ± SD percentage of injected dose per cubic millimeter of tissue (%ID/mm 3 ) as recommended by the AQARA requirements [34]. The SPECT signal quantifications expressed as percentage of injected dose can be found in Supplementary Figure S1.

Statistical Analysis
Data were analyzed using Prism v9.1 software (GraphPad, San Diego, CA, USA). The collected activities from SEC in V 1 and the radiolabeling stability over time in serum were analyzed using an unpaired t-test. The differences between quantified activities in each organ or hind limb from each condition were compared using a two-way ANOVA followed by Šídák's multiple comparison post hoc test. Animal weight and LASER Doppler over time were analyzed with a two-way repeated-measures ANOVA followed by a Šídák's multiple comparisons post hoc test. Correlation between LEV homing and late LASER Doppler data was tested using Pearson R test after validating data for normality. Sample size was calculated using the BiostaTGV online tool (https://biostatgv.sentiweb.fr, accessed on 23 March 2018). Samples distributions were tested for normality using the Shapiro-Wilk test. Unless indicated otherwise, data were expressed as mean ± SD values, p < 0.05 indicating statistical significance.

Characterization of Produced Endothelial LEVs
Biophysical characterization by TRPS showed approximately 60% of the LEVs ranged between 250 and 400 nm (Figure 2A). Western blot analysis identified protein markers of LEVs ( Figure 2B). CD63 and CD81 tetraspanins, predominantly associated with late endocytic organelles and the smallest LEV types [13,35], were present, respectively, as 40-50 kDa and 25 kDa bands. CD51 integrin (alpha-V) was present, whereas β 3 -integrin was abundant, as previously described [36]. Caveolin-1, identified as a major component of LEVs and involved in diverse protein trafficking pathways [13,37,38], was abundantly present, as were actin and tubulin. Albumin, a major constituent of non-LEV structures [13], appeared as a weak band, indicating the low level of major cellular contaminants on the isolated LEV preparations. TEM analysis showed circular LEVs with typical bilayer membrane with a mean size of 434.5 ± 124.20 nm (n = 2) ( Figure 2C). Isolated, SEC-purified LEV fractions from a single donor were characterized by flow cytometry regarding the exposure of phosphatidylserine and markers of cellular origin. The expression of these markers was also evaluated, and LEVs were stained with a combination of antibodies directed against endothelial markers (CD31 and CD146), as well as the activation marker ICAM1/CD54. Approximately half of the HUVEC-derived LEVs stained for CD31, while most stained strongly for CD146 and for ICAM1/CD54 ( Figure 2D). Altogether, these data confirmed the endothelial cell origin of LEVs.

Endothelial LEVs Preferentially Homed to the Ischemic Hind Limb
MicroSPECT/CT biodistributions of free [ 99m Tc]Tc-AnnV and [ 99m Tc]Tc-AnnV-LEVs 30 min after injection were overall highly significantly different (two-way ANOVA ** p = 0.0019, Figure 4A,B), especially in the liver, in the kidneys, in the heart, in the lungs, and in the spleen (Table 1, Figure 4B). A significantly higher SPECT signal quantification was found with [ 99m Tc]Tc-AnnV-LEVs in the ischemic hind limb compared with the contralateral hind limb 30 min after injection (* p = 0.0090); whereas, no significant difference was found between SPECT signal quantifications in the ischemic and in the contralateral hind limbs with free [ 99m Tc]Tc-AnnV (p = 0.9722). A significantly higher SPECT signal quantification was found in the ischemic hind limb with [ 99m Tc]Tc-AnnV-LEVs compared with that of free [ 99m Tc]Tc-AnnV (** p = 0.0013, Table 1, Figure 4C).

Discussion
A simple method was designed for radiolabeling endothelial LEVs using [ 99m Tc]Tc-AnnV. In a preclinical model of hind limb ischemia, early and specific homing of radiolabeled LEVs to the ischemic hind limb were quantified on the day of ischemia and positively correlated with reperfusion intensity at a later stage on day 28 after ischemia, associated with an improved motility function.
SEVs are described as being very quickly cleared from the plasma, within the first 30 min following their injection [37,39]. By extrapolation to LEVs and to avoid the potential risk of quantifying an unspecific signal related to LEV degradation and free [ 99m Tc]Tc-AnnV-128 tissue accumulation at later time points, the radiolabeling stability was validated up to 30 min after incubation in vitro and the quantification of microSPECT/CT signal in main organs and hind limbs was performed up to 30 min after injection. In line with previous reports on the biodistribution of extracellular vesicles with other techniques, [ 99m Tc]Tc-AnnV-LEVs were mainly distributed in organs of the mononuclear phagocyte system with highest accumulation in the liver, secondarily in kidneys, spleen, and lungs.
Remarkably, the [ 99m Tc]Tc-AnnV-LEVs SPECT signal quantified in the ischemic hind limb was 4.5 times higher than that of free [ 99m Tc]Tc-AnnV in the same hind limb, and 2 times higher than that of [ 99m Tc]Tc-AnnV-LEVs in the contralateral hind limb. Of note, apoptosis, that can be imaged with [ 99m Tc]Tc-AnnV radiotracer, occurs in later days in the chronology of ischemic hind limb model and is very unlikely to generate a [ 99m Tc]Tc-AnnV uptake on the day of ischemia [38,40]. Most importantly, despite an unfavorable input function in the ischemic hind limb on the day of ischemia, a quick and significant homing of LEVs was quantified in the ischemic region. As for many cell-based therapies, the very small proportion of injected radiolabeled LEVs reaching the ischemic site (<1% of the injected dose) was nevertheless sufficient to induce significant therapeutic effects [41,42]. The administration route directly influences the biodistribution of extracellular vesicles and constitutes an interesting avenue for refinement; Wiklander et al. reported subcutaneous and intraperitoneal delivery routes associated with half less accumulation of the extracellular vesicles in the liver compared with intravenous delivery route [39]. Indeed, diffusion and transport of extracellular vesicles is influenced by their environment and the specific shear stress in the impaired locoregional blood circulation passively affects their uptake [43,44].
Most outstandingly, no later than 7 days after the insult, LEVs enabled a significant earlier and higher vascular recovery. Besides, the more LEVs were addressed to the site of ischemia on the day of their injection, the better the vascular recovery 28 days after was, corroborating the importance of promoting early angiogenesis in post-ischemic regenerative therapies [45]. Preclinical ex vivo evidences of post-ischemic recovery enhancement following LEV injection are in line with observations in the literature [46][47][48]. The therapeutic properties of endothelial LEVs rely not only on their origin, but also on the considered animal disease model or the patient's condition [16,49]: for instance, extracellular vesicles from human adipose tissue were also recently reported with pro-angiogenic properties and could have a high potential for therapeutic use in ischemia [50]. Considering the inner composition and phenotype of LEVs, but also considering the route of administration, each LEV subtype is expected to demonstrate a different biodistribution profile (liver retention and non-specific accumulation in healthy tissues) linked to various homing and various therapeutic effects (uptake in target tissues) [39]. Consequently, such a biodistribution study should be repeated for each LEV subtype, testing for a positive correlation between their early homing and the later therapeutic effects, for each disease model or pathological condition. Of note, the therapeutic performance of LEVs related in this study might be underestimated due to the use of human cell-derived LEVs instead of mice cell-derived LEVs, mainly because the primary objective was to set up and demonstrate the feasibility of a radiolabeling method for evaluating the biodistribution of human LEVs in clinical trials. The absence of a LEV co-labeling method and of a supplementary method to radionuclide tracking could be considered as a limitation of our study, although hardly technically feasible in vivo regarding the short life of LEVs. Notably, as this LEV radiolabeling method with [ 99m Tc]Tc-AnnV relies on an industrial GMP-manufactured radiotracer, this radiolabeling method could be easily transferable to other LEV subtypes and most importantly, would be an asset as companion tool for emerging therapies based on extracellular vesicles at the clinical stage.

Conclusions
This work reported an innovative method to radiolabel endothelial LEVs enabling quantification of their in vivo biodistribution after systemic injection. Applied to a mouse model of critical hind limb ischemia, microSPECT/CT imaging enabled the quantification of an early and specific homing of LEVs to ischemic tissues that correlated with reperfusion intensity 28 days after ischemia. LEV injection was also associated with an enhanced motricity on day 28. This concept could be a major asset for investigating the biodistribution of LEVs issued from other cell types, including cancer, thus partly contributing to better knowledge and understanding of their fate after injection.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/pharmaceutics14010121/s1. Figure S1: SPECT signal quantification in organs and hind limbs expressed as percentage of the injected dose (%ID). Figure S2: Angiogenic activation in a mouse model of hindlimb ischemia treated by LEVs or vehicle. Figure S3: Quantitative analysis of LASER Doppler signal expressed as ischemic-to-contralateral muscle ratio (%, mean ± sd) from day 0 to day 28.