Genetically Engineered MRI-Trackable Extracellular Vesicles as SARS-CoV-2 Mimetics for Mapping ACE2 Binding In Vivo

The elucidation of viral-receptor interactions and an understanding of virus-spreading mechanisms are of great importance, particularly in the era of a pandemic. Indeed, advances in computational chemistry, synthetic biology, and protein engineering have allowed precise prediction and characterization of such interactions. Nevertheless, the hazards of the infectiousness of viruses, their rapid mutagenesis, and the need to study viral-receptor interactions in a complex in vivo setup call for further developments. Here, we show the development of biocompatible genetically engineered extracellular vesicles (EVs) that display the receptor binding domain (RBD) of SARS-CoV-2 on their surface as coronavirus mimetics (EVsRBD). Loading EVsRBD with iron oxide nanoparticles makes them MRI-visible and, thus, allows mapping of the binding of RBD to ACE2 receptors noninvasively in live subjects. Moreover, we show that EVsRBD can be modified to display mutants of the RBD of SARS-CoV-2, allowing rapid screening of currently raised or predicted variants of the virus. The proposed platform thus shows relevance and cruciality in the examination of quickly evolving pathogenic viruses in an adjustable, fast, and safe manner. Relying on MRI for visualization, the presented approach could be considered in the future to map ligand-receptor binding events in deep tissues, which are not accessible to luminescence-based imaging.


INTRODUCTION
Virus-receptor recognition is the initial step in the infectious cycle and is considered to be a key stage in the induction of viral pathogenesis. 1 Therefore, the elucidation of the interactions of viruses with host cells' receptors is of paramount importance for a better understanding of pathology pathways and for the development of antiviral interventions. For example, it has been shown that the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which has caused the current and prolonged coronavirus disease 2019 (COVID-19) pandemic, specifically attacks cells expressing high levels of angiotensin-converting enzyme 2 (ACE2) receptor. 2 The understanding of the interactions between the receptor binding domain (RBD) of the spike-S protein of the virus with ACE2 3 has resulted in the development of a wide range of efficient therapeutics and vaccines. 4−8 Unfortunately, SARS-CoV-2 has shown an unprecedented ability to rapidly introduce mutations to the spike protein and the RBD for improved affinity and immune evasion, 9−12 which have led to rapid spreading of more transmissible variants and compromised effectiveness of available vaccines. 13 Thus, it is clear that there is a need for the ability to characterize viruses and their evolving mutants quickly and safely and potentially even predict dangerous variants before they emerge. Indeed, in silico 14 and in vitro 15 examination of viruses provides crucial insights into virus-receptor interactions. Nevertheless, these approaches are limited in the study of off-target binding events and are not applicable for spatial and real-time mapping of viral-receptor binding in deep tissues. This calls for a method with the ability to longitudinally and noninvasively monitor and map in vivo viral distribution and receptor binding in a safe and rapid way to enhance our ability to study emerging viruses and assess biological feedback to therapeutics.
Several types of nonviral nanosized formulations have been proposed to elucidate viral-receptor interactions so far, including those for studying SARS-CoV-2. 4,16−18 Among these, extracellular vesicles (EVs) offer several advantages over synthetic nanoparticles. First, as cellular content nanocarriers, 19−21 they are biological substances, suggesting that they can be introduced into the body without leading to the side effects often encountered with synthetic formulations. Second, they can be genetically engineered to present biomolecules on their surface, providing a rapid and general method for the display of peptides. 22−24 As such, EV targetability to a tissue of interest has been enhanced by displaying peptides that are not present on the surface of native EVs. 25−28 Third, EVs share important similarities with enveloped viruses, including comparable sizes and host membrane compositions. 20,29−31 EVs are thus attractive noninfectious vehicles for examination of viral uptake pathways of cellular cargo delivery, 25,32−34 or for the development of EVbased vaccines 35,36 and related adjuvants. 37 For example, EVs presenting the coronavirus S protein or its RBD were proposed as potential vaccines already in the mid-2000s for SARS-CoV 38 as well as for the current SARS-CoV-2 pandemic. 36, 39,40 Moreover, they showed efficiency as decoys for neutralizing antibodies 41 and as systems for targeted delivery of antiviral agents. 42 In addition, the ease at which EVs can be genetically engineered makes these formulations ideal for rapid studies of emerging viral mutations as they appear.
Given that EVs can mimic viruses and can be labeled with imageable material, 27,43,44 EVs can potentially be used for noninvasive in vivo imaging of viral-receptor interactions. In fact, it has been shown that EVs can be fluorescently labeled and imaged in vivo; however, these fluorescent methods are unable to track EVs in deep tissues and offer limited spatial resolution. 45 In contrast, tracking of EVs with three-dimensional imaging modalities (such as CT 43 and MRI 44 ) allows an assessment of their spatial distribution even in deep tissues. In this regard, MRI stands out because of its ability to provide spatial information from the introduced EVs that can be overlaid on high-resolution anatomical images of the same subject, thus avoiding the need to use hybrid multimodal imaging approaches. Here, we show the design, development, and implementation of genetically engineered EVs that display the RBD of SARS-CoV-2 (EVs RBD ) as coronavirus mimetics for studying RBD-ACE2 interactions. Magnetically labeled EVs RBD allow mapping RBD-ACE2 binding in vivo and in realtime using a clinically translatable MRI setup. Moreover, we demonstrate the modifiability of the EV-based formulation by presenting a highly potent mutant of RBD (RBD-62) 46 and of known SARS-CoV-2 variants, Delta and Omicron. The ability to monitor both in vivo biodistribution and the effect of different binding affinities of RBD to ACE2 in a fast and safe way highlights the potential of our approach in prolonged pandemic eras and for the study of other emerging viruses.

RESULTS AND DISCUSSION
Genetic Design of SARS-CoV-2 Receptor Binding Domain (RBD) Constructs. Several methods have already been implemented to genetically engineer EVs to display peptides on their surface as, e.g., in fusion with the Lamp2b EVs membrane protein, 33,47 the vesicular stomatitis virus G protein (VSVG), 42 or the transmembrane domain of plateletderived growth factor receptor (PDGFR) when using the pDisplay vector. 27 Although widely used, Lamp2b showed an inability to express the RBD of the SARS-CoV-2 on the membrane efficiently. 42 Starting from the pDisplay vector, which has been extensively used for heterologous expression and surface display of cell receptors in mammalian cells, we first engineered it for efficient expression of viral peptides on the surface of EVs. To this end, and with a purpose to create SARS-CoV-2 mimetics, EV RBD (Figure 1), we constructed a pAGDisplay plasmid through a three-component assembly of fragments from the widely used pDisplay, pLVX-Puro, and pET26b plasmids. There are four main benefits of using our designed pAGDisplay plasmid over other alternatives. First, it uses the puromycin resistance marker, which is a more potent selective antibiotic compared to Geneticin. Second, an IRES sequence, which was introduced downstream from the transmembrane domain, allowed coexpression of the antibiotic resistance gene and the transmembrane domain gene under a single promoter. Third, the fluorescent protein eUnaG2 was fused with the C-terminus of PuroR, allowing for easy detection of transfected cells. Fourth, we introduced an RFnano protein, an MIRFPnano670 derivative (see the Methods section), at the C-terminus of the PDGFβ transmembrane domain to allow efficient selection of RBD-expressing cells using fluorescence-activated cell sorting (FACS). Having designed the pAGDisplay plasmid for efficient and versatile surface display of cell receptors, the Wuhan RBD of SARS-CoV-2 was first cloned at the N-terminus preceding the transmembrane PDGFR domain. Specific tags were added to the obtained constructs for further validation of expression with an ALFA tag as a marker to the RBD construct and a Myc-tag to the control construct (referred to later in the text as noRBD).
Parental Cells Display Functional RBD with Wild Type Affinity to ACE2. Human embryonic kidney 293 (HEK293) cells were transfected with pAGDisplay-RBD or pAGDisplay-noRBD followed by FACS to select cells associated with the  Figure S1). The presence of RBD fused to the ALFA tag on the cell surface was confirmed using a designed anti-ALFA tag nanobody (DnbALFA, see the Methods section) fused with mNeon Green protein (excitation 506 nm, emission 517 nm). RBD-expressing cells incubated with the nanobody showed a bright signal on the cell surface corresponding to the presence of RBD, in contrast to noRBD cells, which showed only a background signal of the eUnaG2 protein present in the cytoplasm (Figure 2a). The binding of the anti-ALFA tag nanobody only to the RBD cells was also confirmed at a single-cell level by flow cytometry (Figure 2b, p < 0.0001 and Figure S2 for FACS data). Moreover, Western blot analysis of lysed cells showed the expression of the RBD-ALFA tag protein only in RBD cells (Figure 2c).
Having demonstrated the presence of RBD on the cell surface, we then tested its functionality. For this purpose, the recombinant extracellular fragment of ACE2 protein (a 723 amino acids length, 83.6 kDa, 7 N-linked glycosylation, theoretical pI of 5.26; Figure S3c) was expressed, purified, and conjugated to a fluorophore CF640R (excitation 642 nm, emission 662 nm) followed by its incubation with either noRBD or RBD cells (Figure 2d−f). Fluorescence microscopy images showed a clear binding of the ACE2 protein only to RBD cells depicted as a higher fluorescent signal compared to controls, confirming the presence of a functional RBD on the cell surface ( Figure 2d). Similarly, flow cytometry analysis showed a significantly higher fluorescent signal in RBD cells compared to noRBD cells (p < 0.001), corresponding to the bound ACE2 protein (Figure 2e and Figure S2 for FACS data). By incubating the cells with a range of different concentrations of fluorescently labeled ACE2 protein and performing a series of FACS experiments to determine its affinity to RBD at the cell surface, an apparent equilibrium dissociation constant (K D ) of 6.7 ± 1.3 nM was determined for RBD cells (Figure 2f, Figure S3a,b). This affinity of RBD expressed at the surface of HEK293 cells to the ACE2 protein is in accordance with the values reported for purified proteins or through a yeast-display assay. 10,46 EVs RBD as Functional SARS-CoV-2 Mimetics That Target ACE2 Receptors. Following the validation of the expression of functional RBD at the surface of HEK293 cells through ACE2 binding assessments ( Figure 2 and Figure S3), secreted EVs of these cells were obtained using a standard method of differential centrifugation of the cells' media (with an average yield of 5 × 10 9 EVs/one million cells) according to MISEV guidelines. 48 The isolated and purified EVs from RBD cells or control cells were referred to as EVs RBD or EVs noRBD , respectively, and were further examined. Western blot analysis confirmed the expression of RBD ( Figure 3a) via detection of the ALFA-tag only in EVs RBD , while CD81 and Alix, typical markers of EVs, 48 were detected in both EVs RBD and EVs noRBD ( Figure 3b). Cryo-transmission electron microscopy (TEM) images revealed the spherical morphology and typical size of vesicles for both EVs noRBD and EVs RBD (Figure 3c). Nanoparticle tracking analysis showed a homogeneous distribution of size, with an average size of 80.9 ± 1.8 nm and 90.4 ± 1.6 nm for EVs RBD and EVs noRBD , respectively (Figure 3d), comparable to the size reported for SARS-CoV-2. 49 Before using them for incubation studies, the stability of the formulations was studied under the conditions used for their long-and short-term storage (4 and −80°C), fluorescent labeling (25°C), and incubation with live cells (37°C), which showed no change in the representation of the targeting peptide ( Figure S5).
Next, with the aim to examine the binding capabilities of EVs RBD to ACE2-expressing (HEK ACE2 ) as compared to control (HEK) cells, both cell types were incubated with fluorescently labeled EVs for 3 h followed by their examination with fluorescence microscopy ( Figure 3e) and flow cytometry (Figure 3f,g). The expression of the ACE2 receptors in HEK ACE2 cells was confirmed by immunoblotting, FACS analysis, and enzymatically ( Figure S7). Fluorescent images of the examined control cells (HEK) showed no pronounced red fluorescence following their incubation with either EVs RBD or EVs noRBD (Figure 3e, top row). In contrast, a pronounced red fluorescence was obtained for HEK ACE2 after their incubation with fluorescently labeled EVs RBD , confirming their extensive binding to ACE2-expressing cells (Figure 3e, bottom row). Note here that the very low red fluorescence found in control cells after their incubation with the EVs RBD may be in accordance to the expected low expression levels of native ACE2 in HEK293 cells. Flow cytometry analysis showed that EVs RBD had significantly greater binding efficiency to ACE2-expressing cells (HEK ACE2 ) when compared to EV noRBD (p < 0.05), while the uptake into control cells (HEK) was similar for both EVs RBD and EV noRBD (Figure 3f,g and Figure S4 for FACS data). It is important to note that EVs noRBD showed a basal level of accumulation in both HEK and HEK ACE2 cells regardless of the presence of RBD, showing their natural, but still negligible, ability to be internalized or fused to HEK cells ( Figure S4). The results summarized in Figure 3 thus confirm the presence and functionality of the engineered RBD peptide on the EVs RBD surface and its targeting and binding capability to ACE2 receptors expressed on cells. These findings show that the engineered EVs RBD can be used as SARS-CoV-2 mimetics with ACE2 binding capabilities with a similar size and shape to the SARS-CoV-2 virus. Importantly, this EV-based formulation, in contrast to a highly infectious SARS-CoV-2 virus, 50 can be used to study RBD-ACE2 interactions at a single-cell level under standard laboratory conditions without the need for strict safety regulations. This platform may, therefore, expand the research of viral-receptor interactions to research institutes and industrial setups that do not yet have access to dedicated facilities, which must be designated with the highest biosafety level required to work with SARS-CoV-2.
Biodistribution and Targeting of SARS-CoV-2 Mimetics in Mice. Prior to performing in vivo studies, the toxicity effect of the engineered EVs and their biodistribution profiles under the used experimental conditions were examined. No detrimental effect on the viability of recipient cells was found when incubated with EVs RBD (measured by two independent assays, Cell Titer Blue and CKK-8, Figure S6). It is important to mention here that, in order to examine whether there was any effect of the intravesicular content of the EVs RBD , their effect on the viability of the incubating cells was compared to their vesicle-ghost analogues as well as to liposomes, confirming no effect when assessed at the conditions used in this study. These results, which agree with previous studies 51 suggest that there is no pronounced effect on recipient cells when using EVs. The biodistribution of both types of EVs (EVs noRBD and EVs RBD ) was examined in immunocompetent C57BL/6 mice after their intravenous administration, and both showed very similar profiles, with the vast majority of EVs cleared by the spleen and liver as expected ( Figure 4). 24 It should be stressed, however, that further safety and long-term biodistribution examinations of the engineered EVs are required under different experimental conditions and in live animals to examine their maximum tolerated dose (MTD) for a specific application. Using ghosts of the EVs RBD , which should not contain any biomolecular content (proteins or genetic material), may provide an alternative if any safety concern is raised in the future, as these ghosts should preserve the outer membrane feature of the EVs and, thus, their targeting capabilities. Also note that a very recent report showed that lung-derived EVs, 36 which were chemically conjugated to RBD, had no safety concerns of such types of viruslike EVs 52 and strengthen the potentiality of the proposed EVs RBD to be used in vivo in other studies.
The in vivo targetability of the EVs RBD mimetics toward ACE2 receptors was then studied in an animal model. Several studies have already shown the ability of genetically engineered EVs to interact and bind to receptors of interest in vivo, including acetylcholine receptors in the brain, 33 tumor cells, 27 immune cell surfaces, 53 and viral receptors, including ACE2. 42 Since human ACE2 receptors are not naturally present in mice, a few transgenic mice models had been proposed, 54,55 but they are not readily accessible and therefore are not yet extensively used. Therefore, we established a xenograft model to study the in vivo targetability of EVs RBD to ACE2 following their systemic administration. For this purpose, control HEK293 cells (HEK) and ACE2-expressing HEK293 cells (HEK ACE2 ) were first inoculated bilaterally and subcutaneously in two flanks of immunodeficient mice to result in tumor-like tissue appearances at the implantation sites 2 weeks after cell inoculation. In contrast to the transgenic mouse model, 42 the model we used here allows for real-time comparison of the accumulation of EVs following their administration in both the control and ACE2-expressing cells simultaneously in the same animal. Two weeks after cell transplantation, mice were injected, intravenously, with fluorescently labeled EVs, either nontargeted EVs (EV noRBD , group 1, n = 5) or ACE2-targeted EVs (EV RBD , group 2, n = 5) as schematically depicted in Figure 5a. Six hours after intravenous administration of fluorescently labeled (DiR) EVs (3 × 10 11 EVs per mouse), the two types of tumors were excised for further quantitative analysis of their fluorescent signal. Importantly, although introduced systemically (through intravenous injection), a clear and statistically significant difference in the fluorescence of HEK and HEK ACE2 tumors was obtained only after the administration of EVs RBD with a higher accumulation in the ACE2-expressing cells (Figure 5b, c; p < 0.01; n = 5/group). Importantly, no observable difference in the fluorescence of the tumors could be detected after the injection of EVs noRBD , indicating the high targeting ability and enhanced specificity of EVs RBD to the ACE2 receptors in vivo (Figure 5b,c and Figure S9). Moreover, no difference was observed between the accumulation of the two types of EVs (EVs RBD and EVs noRBD ) in other organs ( Figure S9) indicating, once again, the specific targetability of EVs RBD to ACE2-expressing cells even in vivo after systemic administration (Figure 5b,c).

MRI Mapping of Magnetically Labeled SARS-CoV-2 Mimetics.
After showing the specific targetability of EVs RBD to ACE-expressing cells both in vitro ( Figure 3) and in vivo ( Figure 5), we aimed to examine the ability to map the accumulation in their target cells using a noninvasive and three-dimensional imaging modality, such as MRI, which was demonstrated to be applicable for monitoring of magnetically labeled EVs in vivo. 44,56−58 To this end, and to load EVs with superparamagnetic iron oxide nanoparticles (SPIONs), parental HEK-293 cells stably expressing the RBD ( Figure 2) were incubated for 24 h with SPIONs added to their culture medium (40 μg of iron per mL). Then, the incubating medium was replaced with EVs-free medium and the released EVs RBD

ACS Nano
www.acsnano.org Article were collected and purified by differential centrifugation (Figure 6a) and further characterized. Incorporation of SPIONs into the secreted EVs RBD was clearly visualized by cryo-TEM as multiple hypointense clusters inside the lumen of the EVs (Figure 6b). Solutions containing different concentrations of SPIONs-labeled EVs RBD were then examined by

ACS Nano
www.acsnano.org Article MRI and showed concentration-dependent T 2 *-weighted MRI contrast (Figure 6c). Importantly, labeling of EVs with SPIONs did not compromise their size (84.8 ± 4.3 nm vs 94.8 ± 3.4 nm, Figure 6d), charge (−18.8 ± 3.9 mV vs −16 ± 1.0 mV, Figure 6e), or the expression of the RBD (Figure 6f). Then, after confirming a successful labeling of the EVs with SPIONs, we aimed to examine the MRI-detectability of their ACE2 targetability in vivo. To this end, a solution containing 3 × 10 11 SPIONs-labeled EVs RBD was injected systematically through the mouse tail vein. The preferential accumulation of the SARS-CoV-2 mimetics in ACE2-expressing tissue was observed as a reduced MRI signal intensity in T 2 *-weighted images (Figure 6g). To quantify the change in the MRI signal of the two tumorous-like tissues (HEK vs HEK ACE2 ), mice were scanned before and 4 h after EVs administration and the difference in the tissue contrast-to-noise ratio (CNR) obtained before and after injection was calculated. As shown in Figure  6h, a significant (p-value <0.01, n = 5) increase in the CNR difference (before vs after SPIONs-labeled EVs RBD injection) was obtained in the region of interest (ROI) of the ACE2expressing cells compared to control tissue. Moreover, Prussian blue staining of slices of the excised tissues clearly showed iron deposits only at the samples obtained from ACE2-expressing cells (Figure 6i), confirming the delivery of the SPIONslabeled EVs RBD to HEK ACE2 but not to the controls. Injection of SPIONs solution (without EVs) into an additional group of mice showed no difference between HEK and HEK ACE2 tumors ( Figure S10), confirming, once again, the increased accumulation of SPIONs when targeted by EVs RBD (at the same time point of 4 h after tail vein injection).
Note here that other strategies for labeling EVs should be considered in the future to improve their iron content load and, thus, the sensitivity of the approach. Labeling secreted EVs through labeling of their parental cells might have suffered from low labeling capacity, 44,56 and more advanced labeling methods should be developed and employed to allow the detection of lower numbers of targeted EVs in future studies. In addition to MRI mapping of the targetability of magnetically labeled SARS-CoV-2 mimetics to ACE2-expressing cells, our results confirm the successful delivery of an intravesical cargo (nanoparticles cargo) to a target tissue. It should be noted, however, that SARS-CoV-2 is a respiratory virus, and the demonstrated platform would better mimic the virus biodistribution if administered intranasally or intratracheally. Nevertheless, the demonstration that engineered RBD-tagged EVs with encapsulated siRNA can be used to suppress SARS-CoV-2 infection in a transgenic mouse model of ACE2 expression 42 suggests that our platform can be used to deliver several types of cargos in the future, including antiviral or immunosuppressive agents to ACE2-expressing tissue. As such, the proposed SARS-CoV-2 mimetics could be of high importance during the current SARS-CoV-2 pandemic, where efficient and safe treatment strategies against COVID-19 are still needed.

SARS-CoV-2 Mimetics: A Versatile Tool to Study Viral Variants.
As for other RNA viruses as for SARS-CoV-2, the rapid evolution and the introduction of diverse mutations at the RBD manipulates not only the binding affinity to the host cell receptors and the infectivity of the virus but also reduces the efficiency of proposed vaccines and other therapeutics. In that regard, one key feature of the SARS-CoV-2 mimetics presented here is that suspicious mutations or those found in rapidly spread variants of a virus of interest can be easily displayed at the surface of the EVs through a one-step cloning procedure into the pAGDisplay plasmid. This means that the method can provide rapid data on the virus mutations as they appear or, perhaps more importantly, on predicted mutations that might appear in the future. To examine this, and based on our previous experience with predicting SARS-CoV-2 mutants with extremely high affinity to ACE2, we have engineered EVs to express the predicted "62" variant of SARS-CoV-2 RBD (Figure 7, Figure S11 and Supplementary sequences). 46 To this end, the gene encoding for the RBD sequence of the predicted "62" variant of SARS-CoV was designed and cloned into a pAGDisplay, and HEK-293 cells were transfected to obtain a stable cell line (HEK RBD-62 ). After EVs isolation, the difference in the binding capabilities of EVs RBD to EVs RBD-62 to ACE2 receptors expressed at the surface of live cells was examined using flow cytometry. As shown in Figure 7, and as expected from the extremely high affinity of RBD-62 to ACE2, 46 HEK ACE2 cells incubated with engineered EVs RBD-62 showed a 4 times higher fluorescence compared to that obtained when the same cells were incubated with EVs RBD (the so-called Wuhan variant). In contrast, EVs with RBD of Delta

ACS Nano www.acsnano.org Article
(EVs RBD-Delta ) or Omicron (EVs RBD-Omicron ) mutations, although they showed a higher accumulation in HEK ACE2 cells compared to EVs noRBD , did not show significantly higher fluorescence when compared to cells incubated with the Wuhan RBD (EVs RBD , Figure S11). These results may imply only a small increase in the binding capabilities of RBD-Delta and RBD-Omicron to the ACE2 receptor compared to that of Wuhan RBD. 59,60 Importantly, all these experiments (Figure 7 and Figure S11) were performed using a single concentration of EVs and may not reflect the difference in the ACE2 binding affinity of each of the examined type of EVs. Such a determination would require a series of concentrations of the incubating EVs in the same manner as that performed for the data shown in Figure 2f. Nevertheless, the results shown in Figure 7 demonstrate that the proposed genetically engineered EVs, which are here used as coronavirus mimetics, can be used as a reliable platform for fast and safe examination of evolved and predicted mutations of SARS-CoV-2, with the potential to be extended to other viruses.

CONCLUSION
In summary, we showed the design and implementation of genetically engineered EVs as SARS-CoV-2 mimetics (EVs RBD ), which efficiently bind the ACE2 receptor both in vitro and in vivo, even after intravenous systemic delivery. The ability to load EVs RBD with MRI-trackable material (SPIONs) allowed mapping of their targetability in live subjects and, thus, could be applied to study viral-receptor interactions in deep tissues, such as the lungs, which might have played a significant role in viral infectivity and are not accessible to luminescentbased imaging approaches. Having demonstrated the modifiability of the approach and proposing EVs that mimic additional variants of SARS-CoV-2 show its potential to be applied for currently evolving or predicted mutants in a safe and reproducible manner without the risk of using infectious material. We envision that the demonstrated strategy could be employed to study other virus-receptor interactions, beyond SARS-CoV-2-ACE2 demonstrated here, by simply engineering a tailored peptide on the EVs surfaces and using them in a wide spectrum of viral-induced pathologies. Finally, with the increasing interest in EVs research, beyond viral mimetics, including their use as cellular-derived nanocarriers, the reported visualization strategy could be of high cruciality to enrich our understanding of EV targetability to the organ of interest in the context of a live organism.

DNA Manipulations.
All DNA fragments were PCR amplified using KAPA HiFi HotStart ReadyMix (Roche, Switzerland) and purified by NucleoSpin Gel and a PCR Clean-up Kit (Nachery-Nagel, Germany). Restriction-free cloning PCR reactions (50 μL, KAPA HiFi HotStart ReadyMix) were composed of 100−200 ng of cloned DNA fragment and 20 ng of destination plasmid. The assembly PCR reactions consisted of 30 cycles of 1 min annealing, at 60°C, and 6 mins of polymerization and 20 s at 98°C for denaturation. The template DNA was removed by using DpnI type IIM restriction enzyme (NEB, USA) at 37°C (1−2 h), and 0.9 μL from the reaction mix were electroporated to E. coli Cloni 10G cells (Lucigen, USA). Colony PCR and sequencing were used for analysis and verification.
pAGDisplay Vector Construction. pDisplay Mammalian Expression vector was purchased from Invitrogen (V66020). The pAGDisplay vector backbone was assembled by combining a pET28b fragment bearing KanR and origin of replication, a pLVX vector fragment bearing WPRE, PuroR, and IRES sequences, and a pDisplay CMV promoter with a PDGFRβ expression cassette by a restrictionfree three-component assembly. 61 In the subsequent restriction-free cloning step, the PuroR was fused with eUnaG2 fluorescent protein at the C-terminus. 62,63 The full-length pAGDisplay was sequenced to verify its correct assembly.
pAGDisplay Modification with RFnano Near-Infrared Fluorescence Protein. To increase our spatial resolution, we introduced, by restriction-free cloning, a modified near-infrared MIRFP670nano, 64 an RFnano protein, as a cytoplasmic domain at the C-terminus of PDGFRβ. The preparation and design of RFnano is the subject of a publication currently in preparation. Briefly, the PROSS and Rosettabased stabilization design was combined with S. cerevisiae EBY100 expression and sorting to obtain a brighter signal and the same spectral parameters. In total, 23 mutations were introduced into the protein.
Protein Production, Purifications, and Labeling Procedures. The designed ALFA-tag binding nanobody (DnbALFA) and its mNeonGreen fusion protein were expressed by using expression plasmid pET28bdSUMO 65 and E. coli BL21(DE3) cells as described previously. 63 Briefly, 200 mL of 2YT medium (16 g of tryptone, 10 g of yeast extract, and 5 g of NaCl, pH 7) was inoculated (1%), grown to the OD600 = 0.6 (37°C), and the expression was initiated by the addition of IPTG to a final concentration of 0.5 mM. The expression continued for the next 16 h at 20°C. Expressed cells were washed (50 mM Tris-HCl, 200 mM NaCl buffer, pH 8), disintegrated by sonication, immobilized on Ni-NTA agarose (PureCube Ni-NTA agarose, Cube Biotech, Germany), and cleaved on a column by using bdSUMO protease. 66 The eluted fraction was further purified by size exclusion chromatography on HiLoad 26/600 Superdex 75 using a GE AKTA Purifier FPLC system. The soluble his-tagged peptidase domain of ACE2 protein (Q18−S740), inserted in pHLsec plasmid, was expressed in an Expi293F cell system with an ExpiFectamine 293 Transfection Kit (ThermoFisher, USA) according to the manufacturer's protocol and purified as described previously. 46 The ACE2 or DnbALFA proteins were labeled by CF640R succinimidyl ester dye (Biotium, USA, catalog no. 92108) in 0.1 M bicarbonate buffer using a 1:4 protein to dye ratio. The reaction was continued for 1 h at room temperature, and subsequently, the free dye was removed by dialysis (GeBAflex-Midi Dialysis Tubes, 8 kDa MWCO, Geba, Israel) against PBS buffer (16 h, 4°C).
Cell Culture and Transfection. HEK293, HEK293T, ACE2expressing HEK293T cells, and stable pAGDisplay-RBD cell lines were cultured at 37°C in a 5% CO 2 atmosphere in DMEM medium (4.5 g/L glucose, L-glutamine; Gibco, USA) supplemented with 10% fetal bovine serum (FBS) and 1% glutamine (4 mM). Cells were passaged 2−3 times per week using trypsin EDTA solution A (Biological Industries, USA) for cell detachment. The stably expressing ACE2 HEK293T cell line was kindly obtained from the lab of Dr. Ron Diskin (Weizmann Institute of Science) and kept under puromycin antibiotics (0.5 μg/mL, Invitrogen, USA).
HEK-pAGDisplay-RBD Stable Cell Line Generation. The pAGDisplay-based plasmids expressing RBD (1 μg of DNA) were transfected in a 60 mm culture dish with 80% confluent HEK293 cells by a JetPrime transfection reagent (Polyplus, France) according to the manufacturer's protocol. After transfection (24 h), cells were transferred to a 150 mm culture dish. Subsequently, 48 h after transfection, the media was replaced by fresh DMEM medium supplemented with 10% FBS and 1 μg/mL puromycin (Invitrogen, USA). Puromycin-resistant cells were selected for 1 week with regular replacement of cell media. Stably transfected cells associated with the top 1% green fluorescence signals (Puro-eUnaG2), were sorted out from the population using the S3e Cell Sorter device (Bio-Rad, USA) and further subcultured to single colonies.
Characterization of Parental Cells. The presence of the RBD on the cell surface was measured by fluorescence-activated cell sorting (FACS). The cells were cultured until full confluence, then they were detached from the plates by PBS and added to the Eppendorf tubes. After spinning down (5 min at 500g), the cell pellet was resuspended in a labeling solution containing fluorescently labeled, purified ACE2 protein in PBS supplemented with 2% FBS. The cells were labeled for ACS Nano www.acsnano.org Article 1 h on ice and then washed two times with 1 mL of PBS (spun down each time for 5 min at 500g). After the last wash, the cells were resuspended in PBS with 2% FBS solution, filtered by 0.45 μm filters, and added to the FACS tubes. The fluorescence was measured by the LRSII FACS machine (BD Biosciences, USA) and analyzed by the FlowJo software. Preparation of Cell Ghost Vesicles. Vesicles composed of plasma membranes of cell ghost were prepared as previously described, 67 with some modifications. Briefly, approximately 1 × 10 7 RBD cells were centrifuged at 500g for 10 min, washed once with phosphate buffered saline (PBS), and suspended in 0.06% w/v sucrose in 0.25× PBS, supplemented with 1% v/v penicillin streptomycin antibiotics and 0.5% v/v protease inhibitor cocktail, on a shaker overnight at room temperature. The resulting cell suspension was subsequently centrifuged at 6000g for 10 min and suspended in 0.06% w/v sucrose in 1× PBS on a shaker overnight at room temperature. Subsequently, the cell suspension was centrifuged again at 6000g for 10 min and suspended in PBS −/−, and extruded sequentially through 10, 5, 1, and 0.1 μm polycarbonate membrane filters (Whatman) using a mini extruder (Avanti Polar Lipids) 5 times each filter.
Preparation of Large Unilamellar Vesicles (LUVs). LUVs were prepared composed of DOPC. Lipid solution in chloroform was placed in a glass vial, and the organic solvent was evaporated by 12 h of vacuum pumping. The lipid film was then hydrated with PBS −/− to reach the desired concentration and gently vortexed. The resulting MLV suspension was then sonicated for 10 min to disperse larger aggregates and the liposomal suspension was extruded 21 times through polycarbonate filters (100 nm pore size, Avanti Polar Lipids) using a mini-extruder (Avanti Polar Lipids). Size and concentration were verified using NTA and the liposomal suspension was used within 2 weeks from extrusion.
The extruded sample is then pipetted on the top of a 10−50% Optiprep band and ultracentrifuged at 100 000g for 2 h at 4°C using a SW41 rotor. The ghost vesicles are then collected from the interface of the gradient and utilized for further experiments.
EV Isolation. EV-depleted medium was prepared by two rounds of ultracentrifugation (100 000g, 16 h) of DMEM with 20% FBS, diluted to 10% FBS, and supplemented with glutamine. For EVs isolation, cells were cultured in EV-free medium for 48−72 h, then the medium was collected and processed by differential centrifugation (400g, 10 min; 2000g, 10 min; 10 000g, 30 min, all at 4°C). The final supernatant was collected and EVs were pelleted at 100 000g at 4°C for 4 h in an Optima ultracentrifuge using the Beckman Ti45 rotor (Beckman Coulter, USA). The EVs pellet was washed with PBS and resuspended in 0.22 μm filtered PBS. EVs were isolated from the RBD-transfected cells (EV RBD ) and control cells (EV noRBD ).
EV NTA Analysis. The size and concentration of EVs diluted in PBS (1:1000 or 1:5000) was measured by nanoparticle tracking analysis using the NanoSight system (Malvern Instruments, U.K.) with a 405 nm laser by acquiring five, 1 min videos at the camera level 16. Threshold 5 was used for the analysis in all samples. Protein content was measured by the BCA (cells) or microBCA (EVs) protein assay according to the manufacturer's instructions (Sigma-Aldrich, USA).
Binding Assays and Affinity Curve Determination Using Cell-Display. The HEK-pAGDisplay-RBD stable cells grown to 80% confluency were gently detached by Accutase (1/2 solution in PBS, 3 min, Sigma-Aldrich catalog no. A6964) and washed by PBS with 2 g/ L BSA (PBSB). Aliquots of detached cells (approximately 2 × 10 6 ) were incubated with a series of CF640R succinimidyl ester-labeled (Biotium, USA, catalog no. 92108) ACE2 extracellular domain (AA Q18−S740) solutions (0.2−83 nM) in 1 mL of PBSB for 2 h on ice. After the incubation, cells in aliquots were separated by centrifugation (500g, 5 min), washed, and resuspended in ice-cold PBSB, passed through a cell strainer nylon membrane (100 μM, SPL Life Sciences, Korea), and analyzed. The expression (eUnaG2-Puro signal, FL-1, excitation 498 nm, emission 527 nm) and binding (CF640R-ACE2, FL-4, excitation 642 nm, emission 662 nm) signals were recorded for 10k gated single-cell events by an S3e Cell Sorter (BioRad, USA). Mean FL-4 fluorescence signal values of RBD + cells, subtracted by RFnano and the nonspecific signal of the RBD-population, were used to determine the K D of binding constants using a noncooperative Hill equation and a nonlinear least-squares regression using Python 3.7. 46 Cryo-Transmision Electron Microscopy (TEM). Sample Preparation. Lacey carbon EM grids were glow-discharged (30 s, 25 mA) in a Pelco EasiGlow system. A volume of 3.5 μL of EVs resuspended in PBS −/− at ∼10 11 particles/mL concentration were applied onto the EM grid and the grid was incubated in the humidity chamber of Vitrobot Mark IV instrument (Thermo Fisher Scientific, USA) for 5 min at 100% humidity and at room temperature, followed by blotting (4.0 s and −10 Blot Force), and plunge-freezing into precooled liquid ethane. Samples were imaged using a Talos Arctica G3 TEM/STEM microscope (Thermo Fisher Scientific, USA), equipped with a OneView camera (Gatan, USA) at an accelerating voltage of 200 kV using SerialEM. 68 Images were recorded at ×73 000 magnification (calibrated pixel size 0.411 nm) with a −3.5 μm defocus.
In Vitro Binding of EVs RBD to the ACE2 Receptor. Flow Cytometry. To confirm binding and uptake of EV RBD in ACE2expressing cells, isolated EVs RBD or EV noRBD were fluorescently labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide (DiR) (ThermoFischer Scientific, D12731) at a concentration of 15 μM by incubation for 1 h at room temperature. EVs were then washed three times with the VivaSpin centrifugation filters, 5 min at 14 000g, and 2 min at 1000g in a reverse position. HEK293T or ACEexpressing HEK293T cells were incubated with labeled EVs RBD or EV noRBD in EV-depleted DMEM for 3 h. Then, the cells were detached from the plate by PBS, washed twice with PBS in the Eppendorf tubes (spun down five min at 500g), filtered by 0.45 μm filters, and added to the FACS tubes in FACS solution (PBS with 2% FBS). The fluorescence signal (APC-Cy7 filter) of cells was measured using a LSRII cell analyzer flow cytometer (BD Biosciences, USA) and analyzed by the FlowJo software.
Fluorescence Microscopy. For visualization by fluorescence microscopy, cells were seeded on 14 mm-diameter coverslips in a 24-well plate. The wells were coated with fibronectin by incubation for 45 min. Then, fibronectin was removed and the wells were washed with PBS before seeding the cells. For uptake of EVs in cells, the DiDlabeled EVs were dissolved in exosome-free medium and incubated with cells for 3 h. Then, the medium was removed, and cells were washed two times with PBS. For staining of cell nuclei, DAPI was dissolved in 2.5% formaldehyde and the cells were incubated in the solution for 20 min. After two washes, the coverslips were carefully transferred and mounted on glass slides and the fluorescence was visualized using a wide-field microscope (Leica DMI8).
Detection of ACE2 in Cells. For confirmation of the presence of ACE2 protein in the ACE2-expressing cells, ACE2-expressing HEK293T and HEK293T control cells were lysed and the activity of the ACE protein was assessed by the SensoLyte 390 ACE2 Activity Assay Kit (AnaSpec, USA) according to the manufacturer's instructions. The fluorescence (excitation 330 nm, emission 390 nm) of the final product was read within 5−35 min using a plate ACS Nano www.acsnano.org Article reader. The expression of ACE2 protein in cells was also confirmed by Western blot analysis, as described above, using a rabbit monoclonal ACE2 antibody (ab239924; 1:1000; Abcam, USA). For FACS analysis, the cells were seeded in a 6-well plate. Then the cells were detached from wells by PBS and incubated with ACE2 primary antibody (ab239924; 1:1000; Abcam, USA) for 1 h on ice, washed with PBS and then incubated with antirabbit fluorescently conjugated secondary antibody (Alexa Fluor 647 nm) for 1 h on ice. The cells were then washed with PBS, filtered with 0.45 μm filters, and resuspended in PBS with 2% FBS. The fluorescence signal (APC filter) of cells was measured by a FACS machine (LRS-II) and analyzed by the FlowJo software. Toxicity Evaluation. For the evaluation of the toxicity of the vesicles, HEK293T cells were seeded in a 96-well plate a day before the addition of EVs RBD or EVs noRBD . The cells were incubated with EVs, vesicle ghosts, and LUVs of different concentrations (0.6 × 10 9 to 2.5 × 10 9 EVs/well, n = 5 biologically independent replications for each sample) for 4 h. The viability of the cells was measured with two independent assays, CCK-8 and Cell Titer Blue according to the manufacturer's instructions. 50% DMSO was used as a positive control to represent a cell toxicity scenario.
Production of Magnetically Labeled EVs. HEK293 cells stably expressing RBD or without RBD (control) were seeded into 15 cm culture plates at 50% confluency. The next day, SPIONs (Molday ION Dye Free; 2 mg Fe/mL; BioPal, USA, CL-50Q02-6A-0) were added to the culture medium at a final concentration of 40 μg/mL. Then, 24 h after, medium was discarded, the cells were washed three times with 10 mL of PBS, and EV-depleted medium was added to the cells followed by another 24-h incubation. Afterward, EVs were isolated from the medium according to the standard protocol and resuspended in PBS.
Zeta Potential Measurement. The zeta-potential of the EVs diluted in PBS (10 8 EVs/mL) was determined using a zeta-potential analyzer Malvern Zetasizer (Malvern Panalytical Ltd., U.K.) according the manufacturer's instructions.
Animal Model. Immunodeficient Hsd:Athymic Nude-Foxn1nu mice and C57BL/6 mice (Envigo) were used in the experiments. All animal studies were approved in accordance with the Weizmann Institute's Animal Care and Use Committee (IACUC) guidelines and regulations (approval number 00580120-3). All animals were kept in a daily controlled room at the Weizmann Institute of Sciences animal facility with a surrounding relative humidity level of 50 ± 10% and a temperature of 22 ± 1°C, with a 12/12 cycle of dark and light phases. Subcutaneous xenograft tumors were induced by injection of HEK293T (left side) and ACE2-expressing HEK293T cells (right side) under the skin above the mouse flank. Mice were kept under isoflurane anesthesia during the whole procedure. Each mouse received 10 mil of each cell type in 150 μL of PBS. The tumors were allowed to grow for 2−3 weeks until they reached a sufficient size for imaging (diameter around 0.5−1 cm). If the tumors reached more than 1 cm or the tumor size was not equal, the animals were removed from the experimental groups.
Magnetic Resonance Imaging. MRI experiments were conducted on a horizontal 15.2 T horizontal scanner (in vitro and in vivo) using a 1 H volume radiofrequency coil with a 23 mm diameter (Bruker BioSpin, Germany).
Phantom Measurements. Different concentrations of isolated EVs were added to glass tubes and imaged using a Rapid Acquisition with Relaxation Enhancement (RARE) sequence with the following parameters: time of repetition (TR) 2800 ms; echo time (TE) 42 ms; and resolution 0.23 × 0.23 × 1 mm 3 . MR images were analyzed by the ImageJ software or by a custom-made script written in MATLAB (MathWorks, USA).
In Vivo MRI of Mice. Prior to MR imaging, a cannula was inserted into a mouse tail. Mice with induced tumors were then measured on a 15.2 T MRI scanner before and after injection of SPIONs-labeled EVs RBD at a dose of 3 × 10 11 EVs in 100 μL of PBS administered through a long tube without changing the position of the animal (n = 5). In a control experiment, only SPIONs solution (1 mg of iron per kg) was injected into the tail vein of mice (n = 4). The mice were measured under general isoflurane inhalation anesthesia (5% induction, 1% maintenance) up to 4 h after EVs injection using a gradient echo Fast Low Angle Shot (FLASH) with the following parameters: TR = 300 ms; TE = 2 ms; resolution of 0.23 × 0.23 × 0.7 mm 3 . After tuning and matching to the 1 H frequency, shimming of the magnetic field, B 0 correction, both axial and coronal images of the tumor area were measured followed by analysis in the ImageJ software. Regions of interest (ROI) were outlined around each tumor, muscle, and a noise area outside of the animals. The contrast-to-noise ratios (CNR) were calculated as follows: The CNR difference was calculated as a percentage of CNR before and after EV injection. Fluorescence Imaging. Prior to imaging, mice with induced subcutaneous tumors (n = 5 for each formulation) or C57BL/6 mice (n = 4 for each formulation) were retro-orbitally injected with DiRlabeled EVs RBD or EVs noRBD (dose of 3 × 10 11 EVs in 100 μL of PBS). Six hours after injection, the mice were intracardially perfused with 2.5% formaldehyde solution under general anesthesia induced by ketamine (80 mg/kg) and medetomidine (0.6 mg/kg). Organs (liver, kidneys, spleen, intestine, heart, lungs, brain) and tumors were excised, fixed with 2.5% formaldehyde solution overnight, and then kept in PBS. A day after, fixed organs were measured with IVIS Lumina XR optical imager (PerkinElmer, USA) with the FOV and exposure time of 1 s (liver), 10 s (kidney, lungs, spleen, heart), 30 s (brain), or 60 s (tumors). For intravital microscopy, an Olympus microscope MVX was used with exposure times of 700 ms (Cy7 filter for detection of the DiR signal).
Histology. The fixed tumor tissues were kept in 1% formaldehyde solutions and then embedded in paraffin blocks according to a standard protocol. Tissue slices were cut on a microtome and stained by hematoxylin & eosin and Prussian blue. The slides were imaged on a Leica DMI8 wide-field fluorescent microscope using a standard bright-field filter.
Statistical Analysis. All numerical data are presented as mean ± standard deviation (s.d.). Statistical analysis was performed using GraphPad Prism 8.0 software (GraphPad Software Inc., USA). Comparison of two groups was analyzed by a two-tailed Student's t test. A p-value of 0.05 and below was considered significant: *p-value <0.05, **p-value <0.01, ***p-value <0.001, and ****p-value <0.0001.
Confocal images of RBD cells, FACS data and analyses, ACE2 binding assay, stability determination of the EV formulations, results of toxicity assays, Western blots, fluorescence images of tumors, results of biodistribution experiments, MRI data, and analysis of mice injected with SPIONs; uptake assay of EVs displaying the RBD of different SARS-CoV-2 mutations; supplementary sequences (nucleotide sequences) of the used RBDs; and supplementary plasmid map of the pAGDisplay-RBD construct used in this study (PDF)