A pan-SARS-CoV-2-specific soluble angiotensin-converting enzyme 2-albumin fusion engineered for enhanced plasma half-life and needle-free mucosal delivery

Abstract

Immunocompromised patients often fail to raise protective vaccine-induced immunity against the global emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants. Although monoclonal antibodies have been authorized for clinical use, most have lost their ability to potently neutralize the evolving Omicron subvariants. Thus, there is an urgent need for treatment strategies that can provide protection against these and emerging SARS-CoV-2 variants to prevent the development of severe coronavirus disease 2019. Here, we report on the design and characterization of a long-acting viral entry-blocking angiotensin-converting enzyme 2 (ACE2) dimeric fusion molecule. Specifically, a soluble truncated human dimeric ACE2 variant, engineered for improved binding to the receptor-binding domain of SARS-CoV-2, was fused with human albumin tailored for favorable engagement of the neonatal fragment crystallizable receptor (FcRn), which resulted in enhanced plasma half-life and allowed for needle-free transmucosal delivery upon nasal administration in human FcRn-expressing transgenic mice. Importantly, the dimeric ACE2-fused albumin demonstrated potent neutralization of SARS-CoV-2 immune escape variants.

Keywords: ACE2; FcRn; SARS-CoV-2; half-life; mucosal delivery.

Fig. 1.

Design of human albumin-fused dimeric ACE2 with favorable pharmacokinetic and transmucosal transport properties. A) Illustration of the dimeric ACE2-albumin fusions (300 kDa), in which truncated dimeric human ACE2 (residues 18–740) is genetically fused to the N-terminal end of full-length human WT or QMP albumin. The extracellular part of ACE2 includes the PD and the neck region of CLD, where dimerization occurs. The three subdomains DI, DII, and DIII of albumin are indicated. B) Representative analytical SEC profile and C) nonreducing SDS-PAGE gel of noncovalent dimeric ACE2-WT and ACE2-QMP (300 kDa), in comparison to full-length WT human albumin (67 kDa) and human IgG1 (146 kDa). D) Illustration of the ELISA setup used to evaluate binding to human FcRn, in which a His-tagged human FcRn is captured on a coated human IgG1 mutant variant (MST/HN), followed by binding of the dimeric ACE2-albumin fusion and detection with an ALP-conjugated anti-human albumin antibody. E and F) Representative ELISA results showing the pH-dependent binding of unfused ACE2, ACE2-WT, and ACE2-QMP to human FcRn at E) pH 5.5 and F) pH 7.4 (mean ± SD, n = 2). G) Illustration of the HERA with adherent HMEC-1-FcRn cells used to study cellular recycling. H) The recycled amounts of ACE2-WT and ACE2-QMP, in comparison with unfused ACE2 (mean ± SD, n = 3; 4 independent experiments). I) Illustration of the in vivo experimental setup used to determine plasma half-life in homozygous human FcRn Tg32 mice by IV injection into the lateral tail vein. Isolated plasma samples from blood were collected over a period of up to 7 days. J) Representative elimination curves of ACE2-WT and ACE2-QMP post-IV administration. The data are presented as the percentage of ACE2-albumin fusions remaining in plasma compared with day 1 (mean ± SD, n = 5), and the average plasma half-life ( $t_{1/2\beta }$) (mean ± SEM, n = 5). K) Illustration of the Transwell assay with polarized epithelial cell monolayers of MDCKII-FcRn cells used to study apical-to-basolateral transcytosis. L) The transcytosed amounts of ACE2-WT and ACE2-QMP, in comparison with unfused ACE2 (gray) (mean ± SD, n = 4; 2 independent experiments). M) Illustration of the in vivo experimental setup used to evaluate pulmonary delivery in homozygous human FcRn Tg32 mice by IN administration into the nostrils. Isolated plasma samples from blood were collected over a period of up to 4 days. N) The amount of ACE2-WT and ACE2-QMP detected in plasma at 8, 24, 48, 72, and 96 h post-IN administration, and the respective AUC from a representative experiment (mean ± SD, n = 5). Unpaired two-tailed t-test was used for statistical analysis, where *P = 0.0435, **P = 0.0043, ***P = 0.0001, and ****P < 0.0001. The illustrations were created with BioRender.com.

Fig. 2.

Effective binding and blockage of SARS-CoV-2 require ACE2 dimerization. A) Illustration of the SARS-CoV-2 covered by spike proteins made up of the S1 subunit with RBD and the S2 subunit. B) Illustration of the ELISA setup used to evaluate viral binding, in which a dimeric ACE2-albumin fusion is captured by the SARS-CoV-2 spike or derived RBD, and detected with an alkaline phosphayase (ALP)-conjugated anti-human albumin antibody. C and D) Representative ELISA results showing the binding of ACE2-WT and ACE2-QMP to recombinant C) spike and D) derived RBD of SARS-CoV-2 (Wuhan), in comparison with CD147-QMP (mean ± SD, n = 2). E) Illustration of the binding of an ACE2-albumin fusion to the spike proteins displayed on the viral surface and blocking of viral entry to cells expressing ACE2 and TMRPSS2. F and G) Representative neutralization experiments showing the capacity of ACE2-WT and ACE2-QMP to block cellular infection of F) Wuhan SARS-CoV-2 spike-pseudotyped lentivirus expressing GFP to 293T-ACE2-TMPRSS2 cells, and G) live virus to Vero E6 cells, in comparison with CD147-QMP (mean ± SD, n = 2). H) Illustrations of the albumin-fused dimeric ACE2 (residues 18–740; 300 kDa) compared with monomeric ACE2 lacking CLD (residues 18–615; 136 kDa). I) Representative analytical SEC profile of dimeric and monomeric ACE2-albumin fusions. J) ELISA results showing the binding of dimeric and monomeric ACE2-albumin fusions to Wuhan SARS-CoV-2 spike (mean ± SD, n = 2). K) Representative pseudovirus neutralization and L) live virus neutralization experiments of monomeric ACE2(18–615)-QMP and dimeric ACE2-QMP (mean ± SD, n = 4 or 2, respectively). Unpaired two-tailed t-test was used for statistical analysis, where ns = not significant, **P = 0.0043, and ****P < 0.0001. The illustrations were created with BioRender.com.

Fig. 3.

Engineered ACE2-YTY fused albumin potently blocks cellular infection against SARS-CoV-2 mutant variants. A) Overview of SARS-CoV-2 mutant variants used in the study. B and C) Representative ELISA results showing the binding of dimeric ACE2-QMP to recombinant B) spike and C) derived RBD of SARS-CoV-2 variants (mean ± SD, n = 2). D) Crystallographic illustration of the YTY-containing ACE2 bound to SARS-CoV-2 RBD. YTY mutations (T27Y, L79T, and N330Y) are shown as yellow spheres. The figure was made using PyMOL with the crystal structure data deposited (PDB ID: 7U0N). E and F) Representative ELISA results showing the binding of QMP albumin-fused dimeric ACE2 and ACE2-YTY to recombinant E) spike and F) RBD of the Wuhan SARS-CoV-2 (mean ± SD, n = 2). G) Pseudovirus neutralization experiment showing the capacity of dimeric ACE2-QMP and ACE2-YTY-QMP to block cellular infection of Wuhan spike-pseudotyped lentivirus to 293T-ACE2-TMPRSS2 cells (mean ± SD, n = 2, 3 independent experiments). H and I) Representative ELISA results showing the binding of ACE2-QMP and ACE2-YTY-QMP to recombinant spike of the H) B.1.617.2 and (I) B.1.529 variants. J and K) Pseudovirus neutralization experiments of dimeric ACE2-YTY-QMP against the J) B.1.617.2 and K) B.1.1.529 variants, in comparison with ACE2-QMP (mean ± SD, n = 2, 2 independent experiments). L–N) Representative live virus neutralization experiments of dimeric ACE2-YTY-QMP against the Omicron subvariants L) BA.5, M) BQ.1.1, and N) XBB, in comparison with unfused ACE2 and ACE2-QMP (mean ± SD, n = 3). The illustrations were created with BioRender.com.