Intranasal trimeric sherpabody inhibits SARS-CoV-2 including recent immunoevasive Omicron subvariants

Research reported in this study complies with all relevant ethical regulations, and our animal study protocol has been approved by the Animal Experimental Board of Finland (license number ESAVI/28687/2020.

Phage panning

Sherpabody phage display library (size ~1011 cfu) was obtained from Next Biomed Therapies Oy. To develop sherpabodies specific for SARS-CoV-2 spike receptor binding domain (RBD), phage affinity selection process was conducted using standard solid phase sorting strategy. Specific phage-displayed sherpabodies were selected for by panning against RBD-mouse IgG2a Fc-fusion protein35. Three sequential rounds of off-target depletion and specific panning were performed using a control mFc and specific RBD-mFc, respectively. The immobilized control and target proteins (30 µg/ml in PBS; Maxisorp Immunotubes, Nunc) were sequentially incubated in the presence of infectious naïve sherpabody phage library in 2.5% milk-PBS-0.1%Tween20, for 2 h (RT) and o/n (4 °C), respectively. Non-specific phages were removed by extensive washing (PBS-0.05% PBS-Tween), and the remaining pool of phage were eluted and amplified in E. coli XL1-Blue host cells (Avantor; #AGLS200249) according to standard protocols. The amplified pool of phages was collected, and the process was reiterated over three rounds to enrich phage-displayed sherpabodies specific to the RBD-target protein.

Screening and characterization of RBD-binding clones by ELISA

Representative transformants from the second and third panning rounds were tested for specific binding to the RBD-mFc target protein by enzyme-linked immunosorbent assay (ELISA). Mimicking the affinity selection process, in phage-ELISA the target and control proteins were immobilized on an immunoplate (Maxisorp, Nunc). Each single colony represents progeny from a single E. coli cell that harbours a phagemid expressing a unique sherpabody-pIII fusion protein. Specifcally, phage-ELISA was performed in 96-well Maxisorp microtiter plates (Nunc) coated over night at 4 °C with 100 µl (10 µg/ml in PBS) of target (RBD-mFc) and control proteins (Anti-E-tag antibody; GE Healthcare; cat# 27941201 V; lot# 355351). The wells were washed 3 x with PBS-0.05% Tween20 and blocked with 5% skimmed milk powder in PBS (milk-PBS) for 2 h at RT. Appropriate dilutions of sherpabody-displaying single phage clones were prepared in milk-PBS and incubated with the coated target protein for 1 h at RT followed by washes 3 x with PBS-0.05% Tween20 to remove unbound phage. The detection was performed with HRP-conjugated mouse monoclonal anti-M13 antibody (1:5000; GE Healthcare/Cytiva; Cat# RPN1236; Lot# 385982)), and TMB (3,3’ 5,5’-tetramethylbenzidine) substrate (Thermo Fisher; Cat#34028). The staining reaction was stopped with 1 M sulfuric acid and absorbance measured at 450 nm using Hidex Sense Microplate Reader. The DNA encapsulated by the positive phage clones was then sequenced and translated to determine the sequence of the displayed sherpabody.

Sherpabody production and purification

The discovered unique sherpabodies were cloned and expressed in E. coli as monomeric and/or trimeric GST-fusion proteins using standard protocols (GE Healthcare). 15 mer Gly-Ser, (GGGGS)3, linkers were applied in between each fusion partner. GST-tagged sherpabodies were purified by affinity chromatography using glutathione sepharose according to manufacturer’s instructions (GE Healthcare). The buffer was exchanged to PBS by dialysis (140 mM NaCl, 10 mM phosphate buffer, 3 mM KCl, pH 7.4), and the purified sherpabodies were stored at −80 °C, −20 °C and/or at + 4 °C.


Affinity analysis by semi-quantitative sandwich-ELISA was performed using plastic-coated (o/n, 4 °C) GST-sherpabody fusion protein followed by binding of a concentration series of soluble RBD-His target protein35 (1 h, RT). The washes and detection was performed as described above using monoclonal HRP-conjugated anti-His antibody (1:5000; Invitrogen; Cat# MA1-21315-HRP; Lot# 2312647).

GST-Sb92/RBD-His/ACE2-mFc sandwich-ELISAs were performed in two ways: first using plastic-coated GST-Sb92 as the capturing reagent (10 µg/ml) and ACE2-mIgG2a as the detection reagent (10 µg/ml) and then vice-versa (10 µg/ml and 5 µg/ml, respectively) to analyse their ability to simultaneously bind to RBD-His35 (2-fold dilution series starting from 10 µg/ml). The complex formation was detected using either HRP-conjugated goat anti-Mouse-IgG (Sigma Aldrich; 1:5000) or mouse monoclonal anti-GST (1:5000; GE Healthcare; Lot# 17007382)) antibodies. Her2-binding GST-Sb1206 served as a negative control.

The specificity (off-target) of TriSb92 binding to RBD was investigated by ELISA. Whole cell lysates of primary human nasal epithelial cells (HNECs; see isolation protocol below) were prepared by lysing 106 cells/ml in ice cold Lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM EDTA) for 30 min, on ice (vortexing every 5–10 min). Nunc Maxisorp 96-well plates were coated with the concentrated cell lysates o/n, + 4 °C (100 µl/well, 1:100 in PBS). Whole cell lysate derived from HNECs transduced with an adenoviral vector directing the expression of cell-surface expressed RBD of SARS-CoV-2 spike protein were used as a positive control. The ELISA was performed similarly as described above using standard protocols.

Neutralization assays

HEK293T (ATCC; Cat# CRL-3216) and HEK293T-ACE2 cells were maintained in DMEM supplemented with 10% fetal bovine serum, 2% L-Glutamine, and 1% penicillin/streptomycin (complete medium). Angiotensin-converting enzyme 2 (ACE2) expressing HEK293T cells were generated by lentivirus-mediated gene transduction. Briefly, pWPI-puro plasmid containing ACE2 cDNA (AB046569.1) was co-transfected with p8.9NdSB and vesicular stomatitis virus G protein (VSV-G) expressing envelope plasmids into HEK293T cells in complete medium using polyethylenimine. The recombinant lentivirus containing supernatant was collected 48 h post-transfection, filtered and used to infect wild-type HEK293T cells. Transduced cells were selected with puromycin.

Luciferase encoding SARS-CoV-2 pseudotyped reporter virus was generated by transfecting HEK293T cells with p8.9NdSB, pWPI-GFP expressing Renilla luciferase, and pCAGGS, an expression vector containing the SARS-CoV-2 S protein cDNA of the Wuhan-Hu-1 reference strain (NC_045512.2). The last 18 amino acids containing an endoplasmic reticulum (ER)-retention signal of the spike protein was removed to enhance transport to the plasma membrane. SARS-CoV-1 pseudovirus was similarly constructed. Pseudovirus stocks were harvested 48 h after transfection, filtered and stored at −80 °C. Inserts for the pseudotyping expression vectors producing spike glycoproteins encoded by the original Wuhan-Hu-1 strain (with D614G mutation) or its variants B.1.351 (Beta), B.1.617.2 (Delta), and B.1.1.529 (Omicron), including the subvariants BA.1, BA.3, BA.2, BA.2.13, BA.4/5, BA.2.3.20, BF.7, BQ.1.1, XBB, BU.1, or SARS-CoV-1 were assembled from synthetic DNA fragments (Integrated DNA Technologies) using Gibson assembly cloning kit (New England Biolabs; Cat# E5510S) and verified by DNA sequencing.

To assess its neutralizing activity TriSb92 was serially diluted in complete medium. 12.5 µl of TriSb92 dilutions were mixed with 37.5 µl of luciferase encoding SARS-CoV-2 pseudotyped reporter viruses in 96-well cell culture plates and incubated at 37 °C for 30 min. After incubation, 20,000 HEK293T-ACE2 cells (in 50 µl) were added to the wells and the plates were further incubated at 37 °C for 48 h. The amount of internalized pseudovirus in infected cells was quantified by measuring luciferase activity using Renilla-GLO assay (Promega; Cat# E2710). The relative luciferase units were normalized to those of control samples. The pseudovirus neutralization experiments were repeated independently three or more times with similar results. The average and standard deviations of representative assays performed in duplicates is shown.

To assess the inhibitory activity of TriSb92 against clinical virus isolates, the neutralization was performed as described above for the pseudoviruses with the following modifications: 40 000 VeroE6-TMPRSS2-H10 cells36 were seeded per well and mixed with virus-inhibitor mixture, which was replaced with complete DMEM 1 h post infection (p.i) at 37 °C. Nine hours later the cells were fixed with 4% paraformaldehyde, permeabilized with 3% BSA-PBS-0.3 % Triton-X100, and labelled with rabbit polyclonal anti-SARS-CoV-2 nucleoprotein (NP; 1:1000; Rockland Immunochemicals; Cat# 200-401-A50; Lot# 200-402-A50) primary antibody and goat Alexa Fluor 488-conjugated (Invitrogen; 1:5000) anti-rabbit polyclonal secondary antibody. The cell nuclei were visualized by Hoechst staining, and the percentage of cells that labelled positive for SARS-CoV-2 NP was quantified using Opera Phenix High Content Screening System (PerkinElmer). Each experimental condition involved five replicate wells, and an average of 15,000 cells (total > 70,00 per condition) per well were analyzed. Neutralization of authentic viruses were repeated with similar results three times for Wuhan, twice for Delta, BA.1, and BA.5 and once for Beta in experiments performed in duplicate and involved counting of approximately 25,000 cells per condition after immunostaining illustrated in the figure insert. The original data are provided as a Source Data file

The clinical isolates used in this study were isolated in Finland from patient nasopharyngeal samples. Beta variant is the isolate hCoV-19/Finland/THL-202101018/2021 (EPI_ISL_3471851), Delta is hCoV-19/Finland/THL-202117309/2021 (EPI_ISL_2557176), Omicron/BA.1 is hCoV-19/Finland/THL-202126660/2021 (pending for the Gisaid ID), Omicron/BA5.1 is hCoV-19/Finland/THL-202213593/2022 (EPI_ISL_13118918, GeneBank ID OP435368). Viruses were propagated in VeroE6-TMPRSS2-H10 cells36. The generation of mouse-adapted maVie16 SARS-CoV-2 is described elsewhere17. All cell lines used had a well-traceable origin and were authenticated in our laboratory by light microscopic morphological analysis only. All cells lines have been tested negative for mycoplasma contamination. No commonly misidentified cells were used in the study.

Cell viability assay

Primary human nasal epithelial cells (HNECs) were isolated by brushing from the nasal cavity of a male donor after informed consent. The isolated cells were cultured and maintained in PneumaCult™-Ex Plus Basal Medium supplemented with 1x PneumaCult™-Ex Plus Supplement, 1x hydrocortisone (200x stock solution 96 µg/ml) and 1x penicillin-streptomycin (all from StemCell Technologies). For cell expansion, the culture flasks and plates were coated with Collagen Solution according to manufacturer’s instructions, and Animal Component-Free Cell Dissociation Kit was used for dissociation and passaging (all from StemCell Technologies).

For the cell viability assay, HNECs were cultured on an opaque-walled 96-well plate (10 000 cells/well) for 24 h, and then treated with a 2-fold concentration series of TriSb92 diluted in complete culture medium (starting from 1 mg/ml) for 24 h at 37 °C. NaN3 (5%) served as a cytotoxic drug control. Cell viability was measured using CellTiter-Glo 2.0 Assay (Promega; Cat# E2710) according to manufacturer’s instructions. Average and standard deviations of a representative assay performed in triplicates were calculated.

Stability of TriSb92 during long-term on-shelf storage

The stability of TriSb92 after long-term storage in different temperatures was investigated using pseudovirus neutralization assay using Wuhan-Hu-1 reference strain. TriSb92 was stored as 1 mg/ml stocks in PBS (140 mM NaCl, 10 mM phosphate buffer, 3 mM KCl, pH 7.4) in room temperature [RT; + 21–25 °C], fridge [+6–8 °C] or freezer [–20 °C or –80 °C] for 1–15 months (mo).

Animal studies

Balb/c mice (Envigo) were transported to the University of Helsinki (Finland) biosafety level 3 (BSL-3) facility and acclimatized to individually ventilated biocontainment cages (ISOcage; Scanbur) for seven days with ad libitum water and food (rodent pellets). The mice were kept 4–6 h in the light during the day time, and the rest of the day i.e., 18–20 h in the dark. The temperature and humidity in the BSL-3 laboratory and cages was kept between 21–23 °C and 25–35%, respectively. After the acclimatization period, 9-week old female Balb/c were placed under isoflurane anesthesia and intranasally inoculated with 25 µl per nostril of TriSb92 (25 or 2.5 µg/nostril). The mice were challenged by infection with 20 µl of SARS-CoV-2 B.1.351 clinical isolate (2 × 105 PFU) or with mouse-adapted maVie16 SARS-CoV-217 (0.5–1 x 106 PFU). Immediately following the inoculation, the isoflurane was switched off and the animals were held in an upright position for a few seconds to allow the liquid to flush downwards in the nasal cavity. All mice were weighed on a daily basis, and their wellbeing was carefully monitored throughout the experiment for signs of illness (changes in posture or behaviour, rough coat, apathy, ataxia and weight loss), but none of the mice showed any clinical signs. Euthanasia was performed 2 days post infection under terminal isoflurane anesthesia with cervical dislocation. All animals were dissected immediately after death and the right lungs collected for virological examination. The left lung, remaining thoracic organs and heads (animals from the first experiment) were fixed in 10% buffered formalin for 48 h and stored in 70% ethanol for histological and immunohistochemical examinations.

RNA isolation and RT-PCR

RNA was extracted from the right lung of the mice using Trizol (Thermo Scientific) according to the manufacturer’s instructions. Isolated RNA was directly subjected to one-step RT-qPCR analysis using TaqMan fast virus 1-step master mix (Thermo Scientific) and AriaMx instrumentation (Agilent) as described previously for E and subE genes37. The RT-qPCR for actin was conducted as previously described in ref. 38. Relative quantification of actin-normalized viral RNA levels was achieved by the comparative Ct method39 using the average of non-treated animals as reference.

Histology and immunohistochemistry

The left lung, remaining thoracic organs and heads were trimmed for histological examination. Heads were sawn longitudinally in the midline using a diamond saw (Exakt 300; Exakt) and gently decalcified in RDF (Biosystems) for 5 days at room temperature and on a shaker. Tissues were routinely paraffin wax embedded. Consecutive sections (3–5 µm) were prepared from lungs and heads and routinely stained with hematoxylin-eosin (HE) or subjected to immunohistochemistry for the detection of SARS-CoV-2 antigen, as previously described in ref. 16.

Recombinant SARS-CoV-2 S trimer production and complex formation

To express the SARS-CoV-2 S, a gene encoding for the prefusion stabilized S protein ectodomain19 was produced as synthetic cDNA (GeneArt, Life Technologies). The cDNA template encoded for the residues 14 − 1208 of the original Wuhan-Hu-1 strain S protein (NCBI Reference Sequence: YP_009724390.1) with prefusion-stabilizing proline substitutions at residues 986 and 987, an abrogated furin S1/S2 cleavage site with a GSAS substitution at residues 682–685, and a C-terminal T4 fibritin trimerization motif followed by an HRV3C protease cleavage site, SpyTag003, and 8xHisTag. The gene was cloned into the mammalian expression vector pHLsec (Adgene) and transfected into Expi293F™ (Thermo Fisher Scientific) suspension cells at a density of 3 × 106 cells per ml using the ExpiFectamine™ 293 Transfection Kit (Thermo Fisher Scientific). Following 6 days of cultivation on an orbital shaker a 36.5 °C and 5% CO2, the S protein containing supernatant was collected, clarified by centrifugation, and filtered through a 0.45 µM filter. Imidazole was added to the supernatant to 3 mM final concentration, and SARS-CoV-2 S protein was purified from the supernatant by immobilized nickel affinity chromatography with a 1 ml HisTrap excel column (Cytiva) using 300 mM imidazole for elution. S-protein containing eluate was concentrated and buffer exchanged to 10 mM Tris pH 8 + 150 mM NaCl buffer using an Amicon Ultra centrifugal filter (MWCO 100 kDa, Millipore). Prior to grid preparation, pure TriSb92 was added to a purified S-trimer aliquot at 1.5x molar excess, and the complex was incubated on ice for 15 min.

Cryo-EM grid preparation, data acquisition and data processing

A 3 μl aliquot of a pure, prefusion SARS-CoV-2 S-trimer (0.3 mg/ml) mixed with TriSb92 (0.05 mg/ml) was applied on Quantifoil 1.2/1.3 grids (1.2 μm hole diameter, 200 mesh copper) that had been glow discharged in a plasma cleaner (PDC-002-CE, Harrick Plasma) for 30 s. The grids were blotted for 8 s and plunged into liquid ethane using a vitrification apparatus (Vitrobot, Thermo Fisher Scientific). Data were collected on a Titan Krios transmission electron microscope (Thermo Fisher Scientific) equipped with Gatan K2 direct electron detector using electron exposure of 55 e/Ų per image at a nominal magnification of 165,000x, resulting in a pixel size of 0.82 Å (Table S1). Data were processed in cryoSPARC40 (Fig. S5). Contrast transfer function parameters were estimated by CTFFIND441. Particles were picked with Topaz42, using a subset of S-trimer particles first pruned by 2D classification as the training set. After another round of 2D classification to discard false positive particles, an initial volume of the S trimer was calculated ab initio. After 3D refinement of particle poses and shifts, local motion correction, extraction of particles with box size of 512×512 and second round of refinement with C3 symmetry, the symmetry was expanded, and the particles were subjected to 3D variability analysis43 which focused to the RBD region. This revealed three main classes of particles: all RBDs up, all RBDs down, and a mixed up and down population (Fig. S5, Fig. S6). Maps of the all up and all down classes were reconstructed with non-uniform refinement. A map for all RBDs down (reconstruction B) class was calculated applying C3 symmetry, after removing the symmetry copies created by symmetry expansion. For all RBDs up map, no symmetry was applied (reconstruction A). Instead, the RBD region density was further refined by local, focused refinement using particles where the bulk of the S-trimer density had been subtracted. The resulting maps were filtered to local resolution for further analysis. CryoEM data collection and processing statistics are presented in Table S1.

Fitting SARS-CoV-2 S and Sb92 into the cryo-EM reconstructions

A molecular model of the SARS-CoV-2 S-trimer, derived from PDB 7KMS44, was docked into our reconstruction A of the SARS-CoV-2 spike using the fitmap function in UCSF ChimeraX45. In short, a map was simulated for the S-trimer structure to the resolution of 2.9 Å, to match the resolution of the spike reconstruction, and fitting was performed using fitmap global search. The top solution yielded a map-to-map correlation score of 0.78, confirming that, similar to structure 7KMS, our reconstruction of the S-trimer displays a 3-up RBD conformation. We observed that the innate flexibility of RBDs resulted in weaker density features in RBD regions of the full spike reconstruction, and so, RBDs were modelled into locally refined RBD-region density (Figs. S5, S8). First, an RBD was separated from the 7KMS spike structure in Coot46 and fitted into the locally-refined RBD region reconstruction using fitmap in ChimeraX, yielding a correlation score of 0.86. Furthermore, additional density not comprised of the S-protein was observed adjacent to each RBD (Fig. 6). To confirm whether this density was comprised of Sb92, the cryoEM reconstruction was segmented around the RBD using color zone (coloring radius of 2.5 Å) and split map functions in Chimera, followed by fitting a model of Sb92, predicted by I-TASSER21 to high confidence (C-score 0.51) and geometry optimized in Phenix47, into the additional density with a fitmap global search. The top solution placed Sb92 into the density adjacent to RBDs with a correlation score of 0.78. Furthermore, this sherpabody placement reveals that the RT-loop, which contains antigen-binging mutations in Sb92 (Fig. 1), forms a key interface with the RBD (Fig. S8). Residues comprising the RT-loop were refined into the density using ISOLDE48 within ChimeraX. Finally, three copies of the RBD−Sb complex were fitted back into the reconstruction of the full spike (reconstruction A) and connected to the rest of the fitted spike derived from 7KMS in Coot, completing the model of the SARS-CoV-2 spike in complex with Sb92.

Our second reconstruction B (Fig. S7), was validated as a closed conformation by fitting a SARS-CoV-2 structure (PDB 6ZP049, presenting a 3-down RBD conformation) into the density. Fitmap global search was performed between a map simulated for structure 6ZP0 at resolution of 3.1 Å, and our cryoEM reconstruction B, yielding a map-to-map correlation score of 0.76 and leaving no unexplained density.

Molecular graphics and protein interface analysis

Molecular graphics images were generated using PyMOL (The PyMOL Molecular Graphics System, Version 2.5.0, Schrödinger, LLC) and C5151himeraX v1.5, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California45. Residues comprising the interface of SARS-CoV-2 S with ACE2 receptor and inhibitors, including TriSb92 and therapeutic monoclonal antibodies, were identified using the PDB ePISA server.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Read more here: Source link