The trispecific DARPin ensovibep inhibits diverse SARS-CoV-2 variants

Abstract

The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants with potential resistance to existing drugs emphasizes the need for new therapeutic modalities with broad variant activity. Here we show that ensovibep, a trispecific DARPin (designed ankyrin repeat protein) clinical candidate, can engage the three units of the spike protein trimer of SARS-CoV-2 and inhibit ACE2 binding with high potency, as revealed by cryo-electron microscopy analysis. The cooperative binding together with the complementarity of the three DARPin modules enable ensovibep to inhibit frequent SARS-CoV-2 variants, including Omicron sublineages BA.1 and BA.2. In Roborovski dwarf hamsters infected with SARS-CoV-2, ensovibep reduced fatality similarly to a standard-of-care monoclonal antibody (mAb) cocktail. When used as a single agent in viral passaging experiments in vitro, ensovibep reduced the emergence of escape mutations in a similar fashion to the same mAb cocktail. These results support further clinical evaluation of ensovibep as a broad variant alternative to existing targeted therapies for Coronavirus Disease 2019 (COVID-19).

Fig. 1. Structural modeling of ensovibep.

a, Schematic overview of the ensovibep construct. Protein linkers are depicted as gray dashed lines, and the half-life-extending HSA-binding monovalent DARPins (H1 and H2) are colored yellow. b, Surface representations of the three monovalent DARPin molecules binding to the RBD, with the amino acid residues in the paratope colored according to their biophysical characteristics as indicated. c, cryo-EM density for the SARS-CoV-2 spike ectodomain in complex with the RBD-targeting monovalent DARPin R2, shown as two orthogonal views. The DARPin density is colored magenta, and the three spike protomers are colored light blue, gray and pale pink. d, Zoomed-in view of an RBD bound to DARPin R2 with the cryo-EM density shown semi-transparent. The atomic coordinates for the fitted open RBD (PDB ID: 6XCN) and the DARPin model are overlaid. The atomic coordinates for residues 1–84 of the RBD-bound ACE2 (PDB ID: 6M0J), colored green, are superimposed. e, Pseudo-atomic model of the monovalent DARPin R2 in complex with the RBD, colored pink and gray, respectively. f, Zoomed-in view of the interface between monovalent DARPin R2 and RBD. g, Proposed model of the three covalently linked RBD-targeting monovalent DARPin molecules of ensovibep bound to the trimeric spike protein RBD domains. The three DARPin domains are shown in a rainbow color scheme from the N terminus (blue) to the C terminus (red).

Fig. 2. Potency of ensovibep in neutralizing different SARS-CoV-2 variants.

a, Graph reporting IC₅₀ values (ng ml⁻¹) for ensovibep measured in neutralization assays performed with lentivirus-based or VSV-based pseudoviruses or authentic viruses for the variants indicated. Reference variant is the Wuhan strain for VSV-based pseudovirus, a D614G variant for the lentivirus-based pseudovirus or a patient isolate from the early pandemic for the authentic virus. b, Table of the residues modified in the SARS-CoV-2 spike protein for the different variants tested compared to the Wuhan strain. c, Graph with global frequencies of point mutations in the spike protein of SARS-CoV-2 according to the GISAID database (as of February 2022), including a heat map table with IC₅₀ values for ensovibep, R1, R2, R3, REGN10933 and REGN10987 for all point mutations tested (VSV-based and lentivirus-based pseudovirus assays). Dashed box: mutations in RBD. White fields: data not available.

Fig. 3. Protection against SARS-CoV-2 escape mutations generated over four viral passages.

a, Tabular representation of the CPEs induced by SARS-CoV-2 cultured in the presence of increasing concentrations of monovalent DARPin binder R2, multispecific DARPin antiviral ensovibep and the antibody antivirals REGN10933, REGN10987 and S309 or a cocktail of REGN10933 and REGN10987 through passage 1 to passage 4. Color code represents the highest concentration showing ≥20% CPE, for which the culture supernatants were passaged to the next round and deep sequenced for the identification of potential escape mutations. b, Identification of escape mutations in viral passages using deep sequencing. SARS-CoV-2 virus was serially passaged with the monovalent DARPin binder R2 and ensovibep. To identify putative escape mutations in the spike protein, RNA was extracted and sequenced from the supernatant of wells with the greatest selective pressure showing a substantial CPE. All variants in the spike protein relative to the reference genome (NC_045512.2) are shown. Passage 0 of the virus control corresponds to the inoculum used for all experiments. The color of the fields is proportional to the fraction of the reads containing the respective variant (red = 1.0, white = 0.0).

Fig. 4. Clinical parameters of SARS-COV-2-infected and treated Roborovski dwarf hamsters.

a, Design of the Roborovski dwarf hamster study. Animals were infected on day 0 with 10⁵ PFU of SARS-CoV-2 Alpha (B.1.1.7) variant. Treatment was administered either directly after infection (0 hours p.i.) or 1 day after infection (24 hours p.i). For each treatment group, 12 animals were injected i.p. with 10 mg kg⁻¹ of ensovibep, 10 mg kg⁻¹ of mAb cocktail (5 m kg⁻¹ of REGN10933 and 5 mg kg⁻¹ of REGN10987) or PBS (placebo). Additionally, a group of six non-infected and non-treated control animals was included as comparators for the infected and treated groups. Daily measurement of body weights and temperatures as well as observation of clinical symptoms was undertaken. Animals were sacrificed on day 3 or day 5 p.i. or immediately once an individual animal reached a defined humane endpoint. b, Survival of animals for 5 days p.i. Animals that had to be euthanized according to defined humane endpoints were considered as non-survived. Body weights (c) and body temperatures (d) throughout the study duration. Data points show mean ± s.d. of the following number of animals analyzed per treatment group at 0/1/2/3/4/5 days p.i.: ensovibep 0 hours: n = 10/10/10/10/5/5; mAb cocktail 0 hours: n = 9/9/9/9/5/5; ensovibep 24 hours: n = 10/10/10/10/5/5; mAb cocktail 24 hours: n = 12/12/12/7/6/6; placebo, infected: n = 12/12/12/7/4/4; placebo, non-infected: n = 6/6/6/6/3/3. Some animals were excluded from the final analysis due to low drug exposure, likely due to a failure of i.p. injections. These animals were excluded from all analyses. Lines connecting dots are interrupted for any change in animal numbers between consecutive days. Because a considerable number of animals in the mAb cocktail and placebo groups reached defined humane endpoints by day 2 p.i., this day is zoomed-in. Red symbols: animals taken out of the study at day 2 due to severe clinical symptoms. Data are represented by the median and values for individual animals. Number of biologically independent animals: ensovibep 0 hours: n = 10; mAb cocktail 0 hours: n = 9; ensovibep 24 hours: n = 10; mAb cocktail 24 hours: n = 12; placebo, infected: n = 12; placebo, non-infected: n = 6. NS, not significant.

Fig. 5. Virology of SARS-COV-2-infected and treated Roborovski dwarf hamsters.

a, qPCR analysis of virus gRNA copy numbers in oropharyngeal swabs and lung homogenates at day 2/3 or day 5 p.i. b, Titration of replication-competent virus from lung homogenates as plaque assay on Vero E6 cells at day 2/3 or day 5 p.i. Red symbols: animals taken out of the study at day 2 due to severe clinical symptoms. Orange symbols: animals taken out of the study at day 3 due to severe clinical symptoms. Data are represented by the median and values for individual animals. Number of animals analyzed per treatment group at day 2/3; day 5 p.i.: ensovibep 0 hours: n = 5; 5 / mAb cocktail 0 hours: n = 4; 5 / ensovibep 24 hours: n = 5; 5 / mAb cocktail 24 hours: n = 6; 6 / placebo, infected: n = 8; 4 / placebo, non-infected: n = 3; 3. Some animals were excluded from the final analysis due to low drug exposure, likely due to a failure of i.p. injections. These animals were excluded from all analyses. Statistics: two-tailed Mann–Whitney test. P values: not significant (NS) = P > 0.05; *a: 0.0295; *b: 0.0162; *c: 0.0159; *d: 0.0159; *e: 0.0317; *f: 0.0100; *g: 0.0317; *h: 0.0159; *i: 0.0159; *k: 0.0190; *l: 0.0159; **a: 0.0016; **b: 0.0040; **c: 0.0016; **d: 0.0159; **e: 0.0079; **f: 0.0079; **g: 0.0048; ***a: 0.0007; ***b: 0.0008; ***c: 0.0008; ***d: 0.0007.

Fig. 6. Lung histopathology of Roborovski dwarf hamsters at 2 days or 3 days p.i.

a–e, Lungs of untreated hamsters at day 2/3 p.i. developed marked inflammation. a, Whole slide scan revealing consolidation of approximately 60% of the left lung. b, Untreated hamsters had moderate necro-suppurative and hyperplastic bronchiolitis with intraluminal accumulation of neutrophils and cellular debris (asterisk) as well as neutrophils transmigrating through the bronchial epithelium into the lumen (arrowhead). The lung parenchyma presented with a patchy distribution of acute necrosis (c, asterisk) with microvascular thrombosis (arrowheads) or with areas of dense infiltration by macrophages and neutrophils (d). e, Pulmonary blood vessels had mild to moderate endothelialitis. f–i, Lungs of hamsters treated with ensovibep on the day of infection developed moderately less consolidation of their lungs (f). g, Bronchiolitis was milder with less inflammatory cell infiltrate compared to the untreated group. Neutrophils were mostly absent. h, Alveolar walls were only moderately expanded by neutrophils and macrophages with less alveolar edema compared to untreated hamsters. i, Endothelialitis was virtually absent with marginating neutrophils as only immune cells interacting with the vascular lining. j–m, Hamsters treated with the antibody cocktail at the day of infection developed lesions that were similar to those as described for the ensovibep-treated group. n–w, In contrast, lungs of hamsters treated at 1 day p.i. had lesions similar to the untreated hamsters at that time, regardless of their treatments. o,t, Both treatment groups developed moderate bronchiolitis similarly to the untreated group. p,u, Interstitial (asterisks) and alveolar (arrowheads) infiltration with neutrophils and macrophages with variable necrosis of alveolar epithelial cells. Additional lesions in both treatment groups included moderate to marked alveolar edema (q, asterisk), here shown for the ensovibep group, and moderate interstitial edema (v, asterisk), here shown for the antibody group. r,w, Both treatment groups developed moderate endothelialitis with monomorphonuclear infiltrates underneath detached endothelial cells, similarly to the untreated group. Scale bars: a, f, j, n, s, 1 mm; b, g, k, o, t, 50 µm; c, d, h, l, p, q, u, v, e, i, m, r, w, 20 µm.

Extended Data Fig. 1. Neutralization of Omicron BA.1 SARS-CoV-2 Variant by Antivirals.

Titration curves (mean ± SEM) and IC₅₀ values (mean ± CI at 95%) for VSV-pseudotype neutralization assays with wild-type and two different Omicron BA.1 variant spike proteins containing either an arginine or a lysine in position Q493. Ensovibep was tested together with a panel of clinically validated monoclonal antibodies. The table provides the numeric IC50 values as well as the fold change towards the wild-type values.

Extended Data Fig. 2. Scoring of Histopathologic Lesions.

Histopathologic lesions were scored semi-quantitatively and scores plotted as graphs for histologic signs of general inflammation and histologic parameters of bronchiolar, alveolar and vascular lesions at day 2/3 p.i (a) or day 5 p.i. (b). Data are represented as mean values +/- SEM. Number of animals analyzed per treatment group at days 2/3; day 5 p.i.: Ensovibep 0 h: n = 5; 5 / mAb cocktail 0 h: n = 4; 5 / Ensovibep 24 h: n = 5; 5 / mAb cocktail 24 h: n = 6; 6 / Placebo, infected: n = 8; 4 Placebo, non-infected: n = 3; 3. The rational for excluding animals is the identification of animals with low drug exposure, likely due to a failure of i.p. injections.

Extended Data Fig. 3. Transcriptome Analysis of Animals Treated 24 h p.i. in Comparison to Placebo Group.

Extended Data Fig. 3: a) SARS-CoV-2 relExp: Comparison of the relative expression of total canonical junction-spanning viral mRNA transcripts, compared to the total genomic transcripts, among dwarf hamsters infected with SARS-CoV-2 after 2/3 and 5 days, and receiving therapeutically 24 h p.i. ensovibep, mAb cocktail or a PBS solution. Values are shown in log10 scale for both time-points. Red symbols: animals taken out of the study at day 2 due to severe clinical symptoms. Orange symbols: animals taken out of the study at day 3 due to severe clinical symptoms. Data is represented by the median and values for individual animals. Number of animals analyzed per treatment group at days 2/3; day 5 p.i.: Ensovibep 24 h: n = 5; 5 / mAb cocktail 24 h: n = 6; 6 / Placebo, infected: n = 7; 4. The rational for excluding animals in the Ensovibep 24 h group is the identification of animals with low drug exposure, likely due to a failure of i.p. injections. In addition, one sample in the PBS group did not yield enough RNA and was excluded. Statistics: two-tailed Mann-Whitney Test. P-values: ns = P > 0.05; *a: 0.0159; **a: 0.0082; **b: 0.0095. b-d) Heatmaps of differently expressed genes in the lung of dwarf hamsters after infection with SARS-CoV-2 and treatment with ensovibep or a mAb cocktail. Genes related to cytokine-mediated signaling pathway (b), type I IFN signaling (c), cellular-response to IFNy and pro-inflammatory cytokines (d) were selected, as previously described by Winkler et al. (2020) (10.1038/s41590-020-0778-2). Columns represent samples and rows genes. Shown are z-scores of DESeq2-normalized data and color scale ranges from blue (10 % lower quantile) to red (10 % upper quantile) of the selected genes.