Engineered Exosomes as Vehicles for Biologically Active Proteins

Fusion of a WW Tag to Target Proteins Results in Efficient Loading into Exosomes

In the present study, we investigated a new mechanism to load a specific target protein into exosomes. Previously, we found that Ndfip1, an ubiquitin ligase adaptor protein, was able to export binding partners outside of the cell in exosomes.15, 16 One class of Ndfip1 binding proteins found to be exported in exosomes is the Nedd4 family of ubiquitin ligases.¹⁵ Ndfip1 interacts with the WW domains of Nedd4 family ubiquitin ligases through three L-domain motifs (PPxY).¹⁷ On this basis, we hypothesized that adding a WW tag onto a target protein would result in interaction of the tagged protein with Ndfip1, leading to ubiquitination and subsequent export in exosomes. To test this hypothesis, we constructed a plasmid coding for a fusion protein containing two WW regions linked to Cre recombinase (WW-Cre). The recombinase provides a functional reporter for exosome-mediated transfer of the protein into cells harboring floxed fluorescent proteins.

To test whether the addition of a WW tag interfered with Cre activity, we analyzed the efficiency of WW-Cre compared to standard Cre for recombinase activity. We first investigated whether the WW tag interfered with Cre localization when transfected into cells and found no change in WW-Cre subcellular localization compared to Cre (Figure S1). Next, mouse embryonic fibroblasts (MEFs) derived from membrane Tomato/membrane GFP (mT/mG) reporter mice were transfected with both control Cre and WW-Cre to detect recombinase activity. Normally, mT/mG MEFs are red, but after Cre recombination, there is a change to green due to excision of the floxed membrane Tomato (mT) cassette and expression of membrane GFP (mG). Flow cytometry was performed to quantitatively analyze the frequency of recombination between the two forms of Cre. The results showed that WW-Cre was as efficient as Cre after transfection of mT/mG MEFs, indicating that there was no adverse effect of the WW tag on the recombinase protein (Figure 1A; dot plots shown in Figure S2A).

The next set of experiments tested for the possibility of spontaneous Cre protein loading into exosomes without the requirement of Ndfip1 and for the necessity of binding to Ndfip1 before Cre can be exported inside exosomes. In previous studies, expression of Cre in cells has not resulted in Cre protein being exported and detected in exosomes.18, 19 Similarly, we found that transient transfection of Cre into LN18 and HEK293T cells did not result in the transport of Cre protein into exosomes in the presence of either the endogenous or the overexpressed Ndfip1 (Figure 1B, exosome lanes 1 and 2; Figure S2B, exosome lanes 1 and 2). In contrast, the WW-Cre fusion protein can be found in the purified exosomes depending on the level of Ndfip1 present in the cell (Figure 1B, exosome lane 4; Figures S2B and S2C, exosome lane 4), indicating the dependence on Ndfip1 for exosome loading of WW-Cre. To confirm the presence of exosomes, particle size distribution was performed using NanoSight analysis (Figure 1C). A linear sucrose gradient also indicated that the exosome preparation showed the correct density for the fraction, which contained WW-Cre protein and the exosomal marker Tsg101 (Figure S2D). To confirm the trafficking of different WW-tagged proteins into exosomes, we used a WW-mCherry construct. In accordance with the WW-Cre findings in LN18 cells, WW-mCherry was only found in exosomes in the presence of Ndfip1 expression (Figure S2E).

To exclude the function of unspecific release of WW-Cre from the cell, resulting in recombination independent of exosome transfer, we prevented exosome release using the sphingomyelinase inhibitor GW4869.²⁰ Because dying cells release apoptotic bodies that can have a size distribution similar to that of exosomes, we performed cell death assays with different concentrations of GW4869 using phosphatidylinositol (PI) staining combined with flow cytometry. Significant cell death compared to the control was detected at 5 and 10 μM, but not at 2.5 μM, of GW4869 treatment (Figure S3). LN18 cells were co-transfected with both WW-Cre and Ndfip1 and cultured for 2 days with or without 2.5 μM GW4869 before exosomes were purified from the supernatant. In the untreated cells, we observed WW-Cre in the exosome preparation; in contrast, cells treated with GW4869 inhibited the release of WW-Cre into the exosome preparation (Figure 1D). The impaired release of exosomes by GW4869 was confirmed by the reduction of Tsg101 in the exosome fraction. In summary, our results show that a target protein can be labeled with a WW tag, resulting in exosomal export in an Ndfip1-dependent manner.

WW-Cre in Exosomes Is Functionally Active following Uptake by Recipient Cells

The tagged WW-Cre protein in exosomes was then investigated for functional biological activity following uptake in recipient cells. The human glioblastoma cell line LN18 was used as the donor cell for the production of exosomes containing WW-Cre, and the reporter line mT/mG was used as the recipient cell type to monitor the uptake and activity of exosomes containing WW-Cre. In control experiments, LN18 cells were transfected with three varieties of plasmid DNA: Cre, WW-Cre, and Cre plus Ndfip1. mT/mG reporter cells were incubated with purified exosomes for 72 hr to allow for exosome uptake and DNA recombination. The results showed that none of the three conditions resulted in Cre-mediated DNA recombination in mT/mG reporter cells (Figures 2A–2C). In contrast, exosomes derived from LN18 cells transfected with WW-Cre and Ndfip1 showed robust Cre recombination in reporter cells (Figure 2D). Quantification of Cre-mediated recombination in reporter cells showed that only exosomes derived from donor cells with WW-Cre and Ndfip1 transfection reported significant Cre recombinase activity (Figure 2E).

Ndfip1 Is Required for Transport of WW-Cre into Exosomes

Ndfip1 is an adaptor molecule for the Nedd4 family of ubiquitin ligases and can select target molecules for binding and E3 ligase ubiquitination.17, 21, 22 To investigate WW-Cre binding to Ndfip1, both proteins were expressed in HEK293T cells and immunoblotted for Cre following immunoprecipitation using antibodies to Ndfip1. WW-Cre was only detected when Ndfip1 was co-expressed, indicating the direct interaction of the two proteins. WW-Cre was not pulled down when Ndfip1 was not co-expressed, showing the specificity of the interaction between Ndfip1 and WW-tagged protein (Figure 3A).

To further investigate the requirement of Ndfip1 for exosomal transport of WW-tagged proteins, we performed functional assays using MEFs derived from Ndfip1 knockout mice (Figure 3B). Because these assays used endogenous levels of Ndfip1 and exosome production, we performed the experiments using co-culture assays (to increase the sensitivity of the assay) with both Ndfip1^+/+ and Ndfip1^−/− MEFs as the donor cells. Both Ndfip1^+/+ and Ndfip1^−/− MEFs were infected with a lentivirus containing the WW-Cre construct (Figure 3B) 24 hr before being washed and placed in co-culture with the reporter mT/mG cells. Analysis of cells after 72 hr in co-culture showed that only Ndfip1^+/+ MEFs were able to promote recombination in mT/mG MEFs; no recombination was observed in Ndfip1^−/− MEF co-cultures (Figure 3C). These experiments reveal that endogenous Ndfip1 in MEFs was required for the transfer of WW-Cre in exosomes, leading to recombination in co-cultured reporter cells.

Given that Ndfip1 is closely linked to the ubiquitination of target proteins, we next investigated the role of Ndfip1 in mediating the ubiquitination of both Cre and WW-Cre using HEK293T cells and denaturing ubiquitination assays. Expression of Cre protein alone or in the presence of Ndfip1 did not result in ubiquitination of Cre (Figure 3D, lanes 2 and 3). Expression of WW-Cre alone resulted in limited ubiquitination, most likely due to low levels of endogenous Ndfip1 in HEK293T cells (Figure 3D, lane 4). Co-expression of WW-Cre and Ndfip1 resulted in efficient ubiquitination of WW-Cre, with the predominant ubiquitination form observed to be monoubiquitination (Figure 3D, lane 5). These results suggest a mechanism in which Ndfip1 directly interacts with and mediates the ubiquitination of WW-Cre, resulting in exosomal transport of the protein.

WW-Cre Protein Is Localized inside Exosomes

Exosomes have been shown to promote the transport of proteins both within and on the exterior of the vesicle, with proteins such as α-synuclein observed to be transported in both environments.²³ Transport on the exterior of exosomes does not provide an environment to protect proteins from extracellular proteases and would therefore not be suitable for therapeutic applications. To investigate the exosomal location of WW-tagged proteins, we performed protease digestion assays. Exosomes purified from LN18 cells were treated with proteinase K in the presence or absence of Triton X-100. Proteinase K treatment without Triton X-100 did not result in the degradation of WW-Cre protein, as shown by western blotting, indicating that the tagged protein is contained and protected within the exosome microenvironment (Figure 4A). In contrast, exosomes treated with Triton X-100 and proteinase K showed a loss of WW-Cre protein and the exosome marker Alix (Figure 4A). These results indicate that the WW-tagged Cre protein is localized inside exosomes when secreted from the cell.

Exosomes with WW-Cre Provide a Functional Assay for Exosome Storage

Previous reports have questioned the stability of exosomes under various storage conditions, particularly for long periods.24, 25 These reports rely on western blots to identify protein degradation products but fail to provide additional information on their functional integrity after exosome uptake in recipient cells. The WW-Cre system affords a functional assay to assess the biological activity of exosomes following storage. Western blot analysis showed that WW-Cre protein was stable in exosomes stored below 4°C over 3 days (Figure 4B), similar to other reported findings.²⁵ However, when tested for biological activity using the mT/mG recombinase assay, the stored exosomes reported reduced functional Cre activity compared to fresh preparations of exosomes (Figure 4C). Because the WW-Cre protein was not degraded over these periods (as observed in western blots) (Figure 4B), these findings suggest that following storage, exosomes are compromised in their ability to transfer proteins into recipient cells. Alternatively, the exosomes have increased fragility following storage, rendering them and their contents susceptible to degradation.

Nasal Delivery of Exosomes with WW-Cre into the Brain

Exosomal delivery of specific biomolecules in vivo has predominantly involved the transfer of exosomal RNA between cells.⁸ It remains to be seen whether proteins inside exosomes can also be delivered into the brain with intact biological activity. To test this, exosomes containing WW-Cre were used to test for delivery and recombination inside the brains of transgenic mice carrying the loxP stop cassette linked to tdTomato expression.²⁶ LN18 glioblastoma cells were chosen as the donors for harvesting exosomes due to their reliability in generating biologically active WW-Cre in exosomes in vitro. Control exosomes were harvested from non-transfected LN18 cells, as well as from cells transfected with Cre (not WW-Cre) and Ndfip1. These different categories of exosomes were then delivered to the reporter mouse strain Ai14 using the nasal route previously shown to effectively target exosomes to the brain.²⁷ Exosomes were delivered at post-natal day (P) 7, P10, and P13 before analysis at P28 for tdTomato expression in brain sections. Analysis of control brains for tdTomato expression showed that a small number of cells were able to undergo spontaneous recombination independent of any Cre protein, with exosomes derived from non-transfected LN18 cells, as well as Cre- and Ndfip1-transfected cells showing a low level of recombination in these reporter mice (Figure 5A). By contrast, brains exposed to purified exosomes from cells transfected with both WW-Cre and Ndfip1 showed a significantly increased number of recombined cells (Figures 5A–5C). Cells with positive signals were dispersed throughout the brain, including the olfactory bulb, cortex, striatum, hippocampus, and cerebellum (Figure 5C; distribution of positive cells throughout the brain shown in Figure S4). Counterstaining for neuronal-specific nuclear protein (NeuN) immunoreactivity confirmed the neuronal nature of recipient cells (Figure 5D). In addition, a limited number of microglia (identified by ionized calcium binding adaptor molecule 1 [Iba-1] staining) demonstrated Cre-mediated recombination (Figure 5E). We conclude that nasal delivery of exosomes allows for the transport of functional target proteins to multiple brain regions.

Cell Culture

A stable mT/mG reporter MEF line was produced from the global double-fluorescent Cre reporter mouse line mT/mG.³⁷ mT/mG MEFs were cultured from an embryonic day (E) 13.5 embryo and infected with a lentivirus contain the SV40 large T antigen.³⁸ Ndfip1^+/+ and Ndfip1^−/− MEFs have previously been described.³⁹ All cells, including HEK293T, MEFs, and LN18 glioblastoma cells, were cultured in DMEM containing 10% fetal calf serum, 50 units of penicillin, 50 μg/mL streptomycin, and 4 mM L-glutamine (Invitrogen) and were maintained at 37°C with 5% CO₂. Transfection of plasmids was performed using TurboFect Transfection Reagent (Thermo Scientific) according to the manufacturer’s instructions.

Plasmids and Flow Cytometry

FLAG-tagged Ndfip1 (FLAG-Ndfip1) in pcDNA3¹⁷ and His-ubiquitin in pcDNA3⁴⁰ have been previously described. WW-Cre and WW-mCherry in pcDNA3, as well as a WW-Cre lentivirus, were constructed by cloning WW domains 2 and 3 of Nedd4-1 to the N terminus of Cre. The efficiency of the WW-Cre construct for Cre activity was analyzed through comparison with a control Cre expression vector using flow cytometry. Briefly, mT/mG MEFs were transiently transfected using TurboFect Transfection Reagent (Thermo Scientific) with Cre or WW-Cre and cultured for 72 hr. Cells were trypsinized, washed in PBS, and then resuspended in 200 μL PBS before being analyzed using a BD FACSCaliber flow cytometer. Each analysis included 100,000 cells, with a color change from green to red indicating positive recombination.

Isolation of Exosomes and Sucrose Gradient

Cells were cultured for up to 72 hr post-transfection before supernatant was collected. The supernatant was cleared from dead cells and debris by centrifugation for 10 min at 200 g followed by centrifugation for 20 min at 20,000 g. Exosomes were isolated from the supernatant by centrifugation for 70 min at 100,000 g and washed with ice-cold PBS before being centrifuged again for 70 min at 100,000 g. Exosomes used for western blotting were resuspended in 30 μL of radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris [pH 8], 100 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.2% SDS, protease inhibitor cocktail [PIC]) and boiled after the addition of 10 μL Laemmli buffer before application. Exosomes used for uptake or intranasal delivery were resuspended in HEPES buffer at a protein concentration of ∼2 μg/μL. For sucrose-gradient experiments, pelleted exosomes were resuspended in 1 mL of 0.25 M sucrose and PBS before being loaded on top of a linear sucrose gradient (0.25 to 2.5 M). The gradient was centrifuged at 100,000 g for 16 hr before single fractions were collected, diluted in ice-cold PBS, centrifuged at 100,000 g for 70 min, and prepared for western blotting.

Particle Size Analysis

Exosomes were purified from LN18 cells and diluted 50% in PBS buffer. Real-time high-resolution particle detection, counting, and sizing were performed on the NanoSight NS300 following manufacturer protocols (Malvern Instruments). Particle concentration (in particles per milliliter) and size was calculated by the NanoSight system.

Co-culture and Purified Exosome Uptake Assays

mT/mG MEFs were cultured on coverslips in a 12-well plate for 24 hr at 37°C with 5% CO₂ before Ndfip1^+/+ or Ndfip1^−/− MEFs infected with WW-Cre were added and cultured under the same conditions for 72 hr. Each well was washed with ice-cold PBS before being fixed in methanol for 10 min at −20°C. After fixation, coverslips were washed three times with ice-cold PBS; the second wash also contained DAPI for nuclear staining. For exosome uptake assays, exosomes were purified via ultracentrifugation and the exosome pellet was resuspended in HEPES buffer. Exosomes were either directly added after purification (fresh exosomes) or stored in HEPES buffer at −20°C, 4°C, or room temperature (∼21°C) before being added to mT/mG MEFs (250 μg of protein vesicles by Bradford measurement were added in vitro). Cells were cultured for 72 hr after the addition of exosomes to allow for uptake and recombination before coverslip mounting for microscopy.

Immunoprecipitation Assay and Western Blotting

10-cm dishes containing 2 × 10⁶ HEK293T cells were transfected either with WW-Cre and an empty pcDNA3 plasmid as a control or with WW-Cre and Ndfip1-Flag using TurboFect Transfection Reagent (Thermo Scientific). Cell lysates were prepared 24 hr post-transfection before immunoprecipitation assays were performed with anti-FLAG M2 Affinity Gel according to the manufacturer’s instructions (Sigma-Aldrich). Ubiquitination assay was performed using HisLink beads (Promega) under denaturing conditions using 6 M guanidine HCl/RIPA containing N-ethylmaleimide to inhibit deubiquitinase enzyme activity. After washing the beads three times in guanidine-HCl/RIPA, the fractions were eluted in 300 mM imidazole and boiled after the addition of Laemmli buffer. Cell lysates and isolated exosomes were subjected to 10% SDS-PAGE and western blotting according to standard protocols. Western blots were incubated at 4°C for 16 hr with the indicated primary antibodies and washed three times in Tris-buffered saline T (TBS-T), followed by 1 hr of incubation in secondary antibody at room temperature. Proteins were detected using enhanced chemiluminescence (ECL) reagent according to the manufacturer’s instructions (GE Healthcare) and visualized by exposure to X-ray film.

Intranasal Delivery of Exosomes

All animal procedures were approved by the Florey Institute of Neuroscience and Mental Health Animal Ethics Committee. Ai14 mice were used as a reporter line²⁶ for intranasal delivery of exosomes. LN18 cells, either transfected with WW-Cre and Ndfip1 (WW-Cre + Ndfip1) or Cre and Ndfip1 (Cre+Ndfip1) or as untransfected control (Ctrl), were cultured for 72 hr before exosomes were purified and resuspended in HEPES buffer (protein concentration of 2 μg/μL). Purified exosomes were administered at P7, P10, and P13 to awake mice. The intranasal grip was used to immobilize the mice.⁴¹ A total of 6 μL was administered intranasally as drops with a small micropipette to each side of the nasal cavity per time point (meaning a total of 12 μL per time point, resulting in a total delivery of 72 μg of exosomal protein to each mouse). At P28, mice were transcardially perfused with PBS, followed by a 4% paraformaldehyde solution (pH 7.4). Brain tissue was post-fixed for 2 hr in 4% paraformaldehyde and then cryopreserved in PBS containing 30% sucrose. Brains were embedded in optimal cutting temperature compound (OCT) (Tissue-Tek, Sakura) and kept at −80°C. Brain tissue sections were cut on the coronal plane with a cryostat (14 μm thick), and the tissue sections were stored at −80°C. Analysis for recombined cells was performed on 36 sections per brain, combined with staining for either NeuN or Iba-1. Six sections in six areas (Bregma 4.5, 2.0, 0, −1.5, −2.5, and −5.5), covering areas of the olfactory bulb, cortex, striatum, hippocampus, and cerebellum, were analyzed for recombination.

Antibodies and Chemicals

Antibodies used in this study were rat monoclonal anti-Ndfip1,³⁹ mouse anti-GM130 antibody (BD Biosciences), rabbit anti-Tsg101 (Sigma-Aldrich), rabbit anti-Cre (BioLegend), mouse anti-FLAG M2 antibody (Sigma-Aldrich), mouse anti-Alix (Cell Signaling), mouse monoclonal antibody to NeuN (Chemicon), and rabbit polyclonal antibody anti-Iba-1 (Wako). Secondary antibodies were Alexa Fluor 488-conjugated goat anti-rabbit immunoglobulin G (IgG) (1:500; Molecular Probes), and Alex 488-conjugated goat anti-mouse IgG (1:500; Invitrogen). The horseradish peroxidase (HRP)-conjugated secondary antibodies used were goat polyclonal anti-rabbit (1:10,000; Millipore), goat polyclonal anti-rat (1:10,000; Millipore), and goat polyclonal anti-mouse (1:15,000; Millipore). The reagents used were GW4869 (Sigma), to inhibit exosome release; DAPI (1:10,000; Dako), for nuclear staining; proteinase K (Sigma-Aldrich); Triton X-100 (MP Biomedicals); and trypsin-EDTA (0.05%; Thermo Fisher).