Extracellular vesicles engineered to bind albumin demonstrate extended circulation time and lymph node accumulation in mouse models

Abstract

Extracellular vesicles (EVs) have shown promise as potential therapeutics for the treatment of various diseases. However, their rapid clearance after administration could be a limitation in certain therapeutic settings. To solve this, an engineering strategy is employed to decorate albumin onto the surface of the EVs through surface display of albumin binding domains (ABDs). ABDs were either included in the extracellular loops of select EV-enriched tetraspanins (CD63, CD9 and CD81) or directly fused to the extracellular terminal of single transmembrane EV-sorting domains, such as Lamp2B. These engineered EVs exert robust binding capacity to human serum albumins (HSA) in vitro and mouse serum albumins (MSA) after injection in mice. By binding to MSA, circulating time of EVs dramatically increases after different routes of injection in different strains of mice. Moreover, these engineered EVs show considerable lymph node (LN) and solid tumour accumulation, which can be utilized when using EVs for immunomodulation, cancer- and/or immunotherapy. The increased circulation time of EVs may also be important when combined with tissue-specific targeting ligands and could provide significant benefit for their therapeutic use in a variety of disease indications.

Keywords: albumin binding domains; circulation time; extracellular vesicles; lymph node accumulation; tetraspanins.

FIGURE 1

Schematic illustration of the aims, engineering strategy and generation of stable cell lines for the study. (A) Schematic graph of the hypothesis. (B) Strategies to engineer EVs with ABDs. (C) Generation of stable cells for EV production by lentiviral transduction

FIGURE 2

Engineering strategies to generate EVs with albumin binding capacity. (A‐F) Protein expression of transgenes in EV‐producing cells and corresponding isolated EVs (A: mCD81, B: mCD63, C: mCD9, D: hCD81, E: hCD63 and F: hCD9). Cell lysate from 5×10⁵ cells and 1×10¹⁰ EVs were used for Western blotting (WB) assay, respectively. (G‐L). The binding of ABD‐engineered EVs to HSA‐FITC (G: mCD81, H: mCD63, I: mCD9, J: hCD81, K: hCD63 and L: hCD9). (M). Protein expression of mCD63 2nd ABD construct in AEC‐derived EVs. (N). Protein expression of hCD63 2nd ABD construct in AEC‐derived EVs. (O) The binding of mCD63 ABD‐engineered EVs derived from AEC cells with HSA‐FITC. (P) The binding of hCD63 ABD‐engineered EVs derived from AEC cells with HSA‐FITC. (Q) The binding of Lamp2B ABD‐engineered EVs derived from HEK‐293T cells to HSA‐FITC

FIGURE 3

ABD‐engineered EVs bind to HSA in vitro. (A) mCD63‐ABD EVs were bound by HSA as detected by widefield imaging of single vesicles. The upper and lower panels indicated the binding of mCD63‐Nluc EVs and mCD63‐ABD 2nd‐Nluc EVs to HSA, respectively. AF647‐labelled CD9/CD63/CD81 antibodies cocktail was exploited to detect EVs while AF488‐labelled HSA was used to show the co‐localization of EVs with HSA. (B) hCD63‐ABD EVs were bound by HSA detected by widefield imaging of single vesicles. The binding of hCD63‐Nluc EVs and hCD63‐ABD 2nd‐Nluc EVs with HSA was indicated in the upper and lower panels, respectively. (C) ABD EVs bound to HSA as detected by super‐resolution imaging (dSTORM) of single vesicles

FIGURE 4

ABD‐engineered EVs bind to MSA in vivo after injection. (A) Schematic illustration of the animal experimental method. Plasma was harvested 10 min after I.V injection. (B) The binding assay for the injected ABD‐EVs with MSA after injection. FITC‐labelled MSA antibody was used to incubate with the plasma and SEC was performed to get different fractions. (C) Nluc luciferase activity of the injected EVs. (D) Single vesicles bound to MSA after injection detected by widefield imaging. AF647‐labelled CD9/CD63/CD81 antibodies cocktail was exploited to detect EVs whereas FITC‐labelled MSA antibody was used to confirm the co‐localization of EVs with MSA

FIGURE 5

Albumin‐decoration by displaying ABD on the surface of EVs extends circulation time in vivo. (A) Illustration of the animal experimental set‐up with indicated injection routes and time points for sample collection. (B) Screen of best tetraspanin candidates for engineering to extend circulation time of EVs, N = 5 for mCD81, mCD9, hCD81 and hCD9; N = 6 for mCD63; N = 8 for hCD63. The endpoint of this experiment was 270 min after I.V injection. (C‐E) Circulation time extension by mCD63‐ABD engineered EVs after different routes of injections (C: I.V, N = 6, D: I.P, N = 5, and E: S.C, N = 5), endpoint = 270 min. (F) Extension of circulation time by hCD63‐ABD engineered EVs after I.V injection, N = 9, endpoint = 270 min. (G) Lamp2B‐ABD engineered EVs extended their circulation time after I.V injection, N = 5, endpoint = 120 min. (H) Illustration of the animal experimental set‐up for the injection of AEC derived EVs. (I) mCD63‐ABD engineered EVs derived from AEC cells exert circulation time extension after I.V injection, N = 5, endpoint = 120 min. (J) Circulation time extension by hCD63‐ABD engineered EVs derived from AEC cells after I.V injection, N = 5, endpoint = 120 min. (K) Illustration of the animal experimental outline. (L) Circulation amounts of injected mCD63‐ABD engineered EVs derived from HEK‐293 cells after I.V injection in C57BL/6 mice, N = 9, endpoint = 120 min. (M) Plasma concentration of injected mCD63‐ABD engineered EVs derived from AEC cells after I.V injection in C57BL/6 mice, N = 5, endpoint = 120 min. N: number of mice used. * P < 0.05; ** P < 0.01; *** P < 0.001

FIGURE 6

Albumin‐binding EVs accumulate in ILNs and tumours. All graphs depict Nluc luciferase activity in RLU/mg of harvested ILNs or tumours. (A) Nluc activity in ILNs at 4.5 h after I.V injection of mCD63‐ABD EVs derived from HEK‐293T cells in NMRI mice, N = 5. (B) 2 h after I.V injection of mCD63‐ABD EVs derived from HEK‐293T cells in C57BL/6 mice, luciferase activity was evaluated in ILNs, N = 9 for control group and N = 8 for ABD group. (C) ILNs were harvested at 4.5‐h time point after I.V injection of hCD63‐ABD EVs derived from HEK‐293T cells in NMRI mice and then Nluc activity was evaluated, N = 8. (D) Luciferase activity determined in ILNs after 2 h of I.V injection of mCD63‐ABD EVs derived from AEC cells in NMRI mice, N = 5. (E) hCD63‐ABD EVs derived from AEC cells were injected (I.V) into NMRI mice for 2 h, and then Nluc activity in ILNs was measured, N = 5. (F) Harvested ILNs in NMRI mice were subjected to luciferase activity measurements after 4.5 h of I.P injection of mCD63‐ABD EVs derived from HEK‐293T cells, N = 5. (G) Nluc activity in B16F10 melanoma after 2 h of I.V injection of mCD63‐ABD EVs derived from HEK‐293T cells in C57BL/6 mice, N = 5. (H) B16F10 melanomas were harvested after 2 h of I.V injection of mCD63‐ABD EVs derived from AEC cells in C57BL/6 mice and then Nluc luciferase activity was evaluated, N = 3. N: number of mice used. * P < 0.05; **** P < 0.0001

FIGURE 7

Albumin‐decorated EVs accumulate across organs. (A) Biodistribution of mCD63‐ABD EVs derived from HEK‐293T cells at 4.5‐h time point after I.V injection in terms of RLU/mg in NMRI mice, N = 6. (B) Different organ‐distribution of the mCD63‐ABD EVs derived from HEK‐293T cells after 4.5 h of I.P injection in terms of RLU/mg in NMRI mice, N = 5. (C) Nluc signals from a variety of organs after 4.5 h of S.C injection of mCD63‐ABD EVs derived from HEK‐293T cells in terms of RLU/mg in NMRI mice, N = 5. (D) I.V injection of mCD63‐ABD EVs derived from HEK‐293T cells for 2 h and then biodistribution was evaluated in terms of RLU/mg in C57BL/6 mice, N = 9 for control group and N = 8 for ABD group. (E) Organ and plasma distribution of the mCD63‐ABD EVs derived from HEK‐293T cells after 4.5 h of I.V injection in terms of percentage of injected EVs in NMRI mice, N = 6. (F) After 4.5 h of I.P injection of the mCD63‐ABD EVs derived from HEK‐293T cells, biodistribution of EVs was evaluated in terms of percentage of injected EVs in NMRI mice, N = 5. (G) Biodistribution of the mCD63‐ABD EVs derived from HEK‐293T cells after 4.5 h of S.C injection in terms of percentage of injected EVs in NMRI mice, N = 5. (H) mCD63‐ABD EVs derived from HEK‐293T cells were injected (I.V) into C57BL/6 mice for 2 h, and then biodistribution of EVs was demonstrated in terms of percentage of injected EVs in, N = 9 for control group and N = 8 for ABD group. N: number of mice used. * P < 0.05; ** P < 0.01; *** P < 0.001

FIGURE 8

Delayed clearance of albumin‐decorated EVs demonstrated by live imaging based on Tluc activity. (A) Dynamic change of luciferase activities in mice injected into mCD63‐ABD‐Tluc EVs in NMRI mice. The upper and lower panels indicated the dorsal and ventral dynamic images, respectively. The monitoring of the luciferase activity started from 10 min after injection and recorded every 5 min. (B) luciferase activity dynamic change in mice injected with mCD63‐ABD‐Tluc EVs in BALB/c mice. Ventral position images were shown. (C) 3D imaging showing the biodistribution of the mCD63‐ABD‐Tluc EVs at 10 min after injection in NMRI mice. Live imaging combined with computed tomography (CT) scanning was used to demonstrate the luciferase signals in different organs