Ethics statement
Animal experiments were licensed by the UK Home Office according to the Animals Scientific Procedures Act 1986 (license PP6047951) and approved by local ethics committees from the University of Cambridge. Human sample collection and analysis were conducted in accordance with the principles of Good Clinical Practice and following approved protocols of the National Institute for Health and Care Research (NIHR) National Bioresource. Samples were collected with the written informed consent of all study participants under the NIHR National BioResource-Research Tissue Bank (NBR-RTB) ethics (REC: 17/EE/0025).
Microfluidic methods
The microfluidic chip was designed using AutoCAD (Autodesk) and printed on a high-resolution photomask (Micro Lithography Services). The channel layout of the microfluidic chip (Supplementary Fig. 1) is available as a.dxf file from https://openwetware.org/wiki/DropBase, to manufacture a chip in-house or by outsourcing. Polydimethylsiloxane (PDMS) microfluidic chips were fabricated using standard soft lithography81. For replica molding, 20 g of PDMS (Sylgard 184; Dow Corning) monomer and curing agent were mixed at a ratio of 10:1, degassed, poured onto the master and solidified overnight at 65 °C. PDMS slabs were cut from the master, inlets and outlets were created using a 1 mm biopsy punch (Kai Medical) and the PDMS slab was permanently bonded to glass slides through plasma treatment (Diener Femto). After bonding, microfluidic channels were flushed with 1% (vol/vol) trichloro(1H,1H,2H,2H-perfluoro-octyl)silane (Sigma-Aldrich) in HFE-7500 (3M Novec) to make them more hydrophobic. Flow rates in microfluidic workflows were controlled using neMESYS 290N syringe pumps (Cetoni) and gas-tight syringes (Hamilton) connected to PTFE tubing (inner diameter 0.5 mm; BOLA). Droplet formation was monitored using a high-speed Mini UX-100 camera (Photron) mounted onto an inverted microscope (Brunel SP981 Microscope). In all experiments, 1% (wt/wt) RAN surfactant (RAN Biotechnologies) in HFE-7500 was used.
Protein expression and purification
VHH and VHH–SNAP
Sequences of VHHs (TP1170 (ref. 26) as anti-mouse κ VHH, HuFab kappa 1 (refs. 27,82) as anti-human κ VHH and HuFab lambda 1 (refs. 27,82) as anti-human λ VHH) were obtained as gene fragments (Integrated DNA Technologies) and cloned into pET28a plasmids (Novagen) using Gibson assembly83 (see Supplementary Table 5 for amino acid sequences of VHH constructs). For the capture of mouse antibodies, we focused on anti-κ VHH, as 95% of mouse B cells express κ light chains84. Proteins were expressed in E. coli BL21(DE3) in 2xYT with 50 µg ml−1 kanamycin. Cultures were grown at 30 °C, protein expression was induced at an optical density at 600 nm (OD600) of 0.6–0.8 with 0.25 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) and proteins were expressed for 5 h at 24 °C. After collection, the cell pellet was resuspended in His-wash buffer (50 mM Tris–HCl (pH 7.5), 300 mM NaCl and 20 mM imidazole) with a protease inhibitor (Protease Inhibitor Cocktail; cOmplete Roche) and 1 µl benzonase nuclease (Sigma-Aldrich). Cells were lysed using 10× BugBuster (Merck). Cleared lysate was added to a Ni-NTA column (Ni-NTA affinity resin; Generon) pre-equilibrated with His-wash buffer. Proteins were eluted with 50 mM Tris–HCl (pH 7.5), 300 mM NaCl and 500 mM imidazole. VHH constructs were not further purified and directly rebuffered into a storage buffer (10% glycerol in PBS) using PD-10 columns (GE Healthcare). To achieve higher purity, a reverse nickel column purification was performed with the VHH–SNAP constructs. Proteins were rebuffered into tobacco etch virus (TEV) protease cleavage buffer (25 mM Tris–HCl, pH 8.0, 150 mM NaCl) and subjected to overnight cleavage at 4 °C with His-tagged TEV protease (Sigma-Aldrich) according to the manufacturer’s instructions. Cleaved proteins were then rerun on a Ni-NTA column to remove the His-tagged TEV protease and the cleaved epitope tag. Flow-through and wash fractions were collected, and the buffer was exchanged for 10% glycerol in PBS using PD-10 desalting columns. Proteins were aliquoted, snap-frozen in liquid nitrogen and stored at −80 °C.
Production of SARS-CoV-2 RBD-CHis for immunization
Synthetic, codon-optimized DNA encoding the RBD of WT SARS-CoV-2 was cloned into the pExp-CHis vector, which introduces a C-terminal octa-His-tag (RBD-CHis; sequences in Supplementary Table 5). RBD-CHis protein was expressed in E. coli BL21(DE3) in 2xYT medium with 100 µg ml−1 ampicillin at 37 °C. Protein expression was induced at OD600 of 0.8–1.0 by the addition of 0.4 mM IPTG and progressed for 16 h. Cells were pelleted by centrifugation and lysed using Avestin EmulsiFlex C-5 homogenizer in 50 mM Tris–HCl (pH 8.0) and 5 mM EDTA supplemented with 250 µg ml−1 of DNase I. Inclusion bodies were isolated by centrifugation and washed first with 50 mM Tris–HCl (pH 8.0), 5 mM EDTA and 1% Triton X-100 and then with 50 mM Tris–HCl (pH 8.0), 1 M NaCl and 5 mM EDTA. Inclusion bodies were solubilized in 50 mM Tris–HCl (pH 8.0), 5 mM EDTA, 10 mM tris(2-carboxyethyl)phosphine (TCEP) and 6 M guanidine hydrochloride. For refolding, the protein solution was added slowly to 50 mM Tris–HCl (pH 8.5), 50 mM ethanolamine, 1 M pyridinium propyl sulfobetaine (PPS), 2 mM EDTA, 2 mM cysteine and 0.2 mM cystine and incubated for 96 h at 4 °C. The refolded protein was first purified by reverse-phase chromatography using a 10 ml source RPC column and applying a 10–90% acetonitrile/0.1% TFA gradient. The peak fractions were pooled, and the protein was lyophilized. Then the protein was dissolved in 20 mM sodium phosphate (pH 7.2), 300 mM NaCl and 0.5 M urea and further purified by size-exclusion chromatography using Superdex 200 Increase 10/300 GL column (Cytiva) equilibrated in 20 mM sodium phosphate (pH 7.2) and 300 mM NaCl.
Production of SARS-CoV-2 RBD for crystallization
DNA encoding WT SARS-CoV-2 RBD was cloned into the pExp-His-Zbasic plasmid to obtain a fusion of RBD with an N-terminal His-tag and Zbasic fusion partner followed by a TEV protease cleavage site (His-Zbasic-TEV-RBD; Addgene, plasmid 194999; sequence in Supplementary Table 5). The protein was expressed in E. coli BL21(DE3) in 2xYT media with 100 µg ml−1 ampicillin. Cells were grown to OD600 0.8–1.0, and expression was induced with 0.4 mM IPTG. Cells were pelleted by centrifugation after 5 h of expression at 37 °C and resuspended in 20 mM Tris–HCl (pH 8.0) and 100 mM NaCl. Before lysis by sonication, 250 µg ml−1 of DNase I and 1 mM phenylmethylsulfonyl fluoride (PMSF) were added to the cells. The lysate was cleared by centrifugation, the supernatant was discarded and homogenized inclusion bodies were washed twice—first with 20 mM Tris–HCl (pH 8.0), 10 mM EDTA, 1 mM TCEP and 1 % Triton X-100, and then with 20 mM Tris–HCl (pH 8.0), 1 M NaCl and 1 mM TCEP. The pellet was dissolved in 20 mM Tris–HCl (pH 8.0), 200 mM NaCl, 10 mM imidazole, 1 mM TCEP and 6 M guanidine-HCl, and the insoluble material was removed by centrifugation. Solubilized, denatured protein was purified using a PureCube Ni-NTA agarose column. After loading the protein, the column was washed with ten column volumes (CVs) of 20 mM Tris–HCl (pH 8.0), 500 mM NaCl, 10 mM imidazole, 6 M deionized urea and 0.5 mM TCEP, and the protein was eluted with five CVs of 20 mM Tris–HCl (pH 8.0), 100 mM NaCl, 250 mM imidazole, 6 M deionized urea and 0.5 mM TCEP. The fusion protein was refolded by diluting it 1:20 in a dropwise fashion into refolding buffer (100 mM Tris, 500 mM Arg-HCl, 0.5 g l−1 polyethylene glycol (PEG) 3350, 1 M urea, 2 mM cysteine and 0.2 mM cysteine) with stirring at 4 °C. After 72 h of refolding, the solution was diluted twofold with 50 mM MES–NaOH (pH 6.0) and filtered through a Whatman GF/B microfiber filter. The clarified protein solution was loaded onto a HiTrap SP HP cation-exchange column (Cytiva). The column was washed with five CVs of 20 mM HEPES–NaOH (pH 7.2) and 50 mM NaCl and eluted with a 25 CV linear gradient to 20 mM HEPES–NaOH (pH 7.2) and 1,000 mM NaCl. Peak fractions were pooled, and NHis-TEV protease (produced in-house) was added overnight at 4 °C. Cleaved protein was passed through a Ni-NTA column to remove the His-Zbasic fusion partner, uncleaved fusion protein and NHis-TEV protease. RBD from the flow-through was concentrated using a 10 kDa MWCO Amicon Ultra concentrator and loaded onto HiLoad Superdex 75 pg 16/60 size-exclusion column (Cytiva) equilibrated with 20 mM Tris–HCl (pH 7.2) and 200 mM NaCl. Pooled peak fractions were concentrated, flash-frozen in aliquots and stored at −80 °C.
Production of biotinylated RBD
Synthetic DNA encoding for WT SARS-CoV-2 RBD was cloned into the pExp-His-Zbasic-Avi plasmid to result in a fusion of RBD with an N-terminal His-Zbasic-TEV fusion partner and a C-terminal Avi-tag (His-Zbasic-TEV-RBD-Avi; Addgene, plasmid 195000; sequence in Supplementary Table 5). The Avi-tagged RBD was produced following essentially the same procedure as untagged RBD except for the addition of 5 mM MgCl2, 2 mM ATP, 150 μM biotin and 50 μg ml−1 of biotin ligase (BirA-CHis, produced in-house) during the TEV protease cleavage step for the site-specific biotinylation at the Avi-tag. The reaction proceeded for 16 h at 22 °C with gentle mixing.
Antigen tetramerization
For the streptavidin-RBD complex (strep-RBD) formation, a 10× molar excess of biotinylated RBD-Avi protein was incubated with 0.5 mg ml−1 of streptavidin-Alexa Fluor 488 or streptavidin-Alexa Fluor 647 (BioLegend) for 1 h at room temperature. Streptavidin-RBD complexes were separated by size-exclusion chromatography using a Superdex 200 Increase 10/300 GL column (Cytiva) equilibrated in 20 mM sodium phosphate (pH 7.2) and 200 mM NaCl buffer.
Biotinylated S1 protein was obtained commercially (BioLegend). A 4× molar excess was incubated with streptavidin-Alexa Fluor 647 (0.5 mg ml−1; BioLegend) for 30 min at room temperature directly before use. Practical guidelines on how antigen concentration can affect affinity selection can be found in Supplementary Fig. 9.
Synthesis of BG-agarose
BG-agarose was synthesized by dissolving 500 mg of low-melting-point agarose (Sigma-Aldrich, A5030), 46 mg of 1,1ʹ-carbonyldiimidazole (Sigma-Aldrich) and 236 μl of triethylamine (Sigma-Aldrich) in 50 ml of DMSO (Sigma-Aldrich) and stirring the solution at room temperature under an argon atmosphere (adapted from ref. 85). After 1 h, 250 mg of O-(4-(aminomethyl)benzyl)guanine (Generon) was added, and the solution was stirred overnight under argon. The solution was diluted with 450 ml of warm distilled water and dialyzed against water using snakeskin dialysis tubing (3,500 Da molecular weight cut-off; Thermo Fisher Scientific). After six water exchanges, the product was lyophilized yielding a white, spongy material.
Determination of agarose functionalization efficiency
BG-agarose (1.5%, wt/vol) was dissolved in PBS, melted by heating to 85 °C using a heated shaker and then cooled to 37 °C. BG-agarose beads (25 µm diameter) were generated by microfluidics (flow rates: aqueous phase, 6–7 μl min−1 and oil phase, 16–18 μl min−1). The emulsion was collected on ice to solidify the agarose. BG-agarose beads were demulsified by the addition of 800 μl PBS and 200 μl 1H,1H,2H,2H-perfluoro-1-octanol (PFO; Alfa Aesar), filtered through a 50 μm strainer and washed once in 15 ml FACS buffer (1% ultra-low IgG FBS (Thermo Fisher Scientific) in PBS). Beads (200,000 per condition) were incubated with serial 1:2 dilutions of GFP-SNAP overnight at room temperature, starting from 8.55 × 1015 GFP molecules per condition. Beads were washed twice in FACS buffer, and the GFP fluorescence was analyzed on an Attune Nxt flow cytometer (Thermo Fisher Scientific). Single beads were gated on side scatter versus forward scatter (log scale) and forward scatter height versus forward scatter (log scale). The mean GFP fluorescence and s.d. for each condition were determined in FlowJo (version 10.7; BD Biosciences).
Antibody capture on agarose beads
BG-agarose beads (25 μM diameter, 1.5% (wt/vol) agarose) were generated by microfluidics as described previously and functionalized with 50 µM anti-mouse κ VHH for 1 h or anti-human κ and λ VHH for 15 min at room temperature. For the mouse antibodies, after washing twice in FACS buffer, an anti-streptavidin antibody (mouse IgG2bκ, clone 3A20.2; BioLegend) was added to 106 beads at a concentration of 0.25 mg ml−1 (1:2 dilution) and incubated for 1 h at room temperature. GFP-biotin was tetramerized by incubation with APC-streptavidin (BioLegend) for 1 h at 10× molar excess. The tetramer (428 kDa) was purified from excess GFP-biotin (25 kDa) with a 100 kDa cut-off Amicon Ultra centrifugal filter (Merck). Functionalized agarose beads were washed twice and then incubated with 20 nM of the GFP–streptavidin tetramers and anti-mouse IgG2a/b BV421 antibody (1:10 dilution; clone R2-40, BD Bioscience) for 30 min on ice. After two washing steps in FACS buffer, the beads were analyzed on an Attune Nxt flow cytometer (Thermo Fisher Scientific). For the human antibodies, the beads were washed once in FACS buffer, and 400,000 beads were each incubated with 6.63 nM of either hRBD2, hRBD8 or hRBD10 for 20 min at 37 °C. Beads were stained with strep-RBD Alexa Fluor 647 (1 nM final concentration) for 30 min on ice. After washing once in FACS buffer, the beads were analyzed on a BD LSRFortessa.
Biolayer interferometry competition assay
Biotinylated anti-mouse IgG1 antibody (10 µg ml−1; clone RMG1-1, BioLegend) was immobilized on streptavidin (SA) biosensors (Sartorius) for 300 s. Subsequently, 100 nM of mRBD1 antibody (IgG1κ) was captured for 600 s and then probed with either anti-mouse κ VHH–SNAP (100 nM, 300 s) followed by RBD (100 nM, 300 s), or vice versa. Human mAbs (100 nM of hRBD1 (IgG1κ) and hRBD3 (IgG1λ)) were directly immobilized on anti-human IgG Fc-capture (AHC) biosensors (Sartorius) for 600 s and then probed with either anti-human κ or λ VHH (100 nM, 300 s) followed by RBD (100 nM, 300 s), or vice versa. For controls, sensors were incubated in buffer instead of anti-mouse IgG1 (−anti-mouse IgG), buffer instead of recombinant mAb (−mAb), buffer instead of mAb/VHH/RBD (buffer control mouse sample) or buffer instead of VHH/RBD (buffer control human samples). All dilutions were performed in kinetics buffer (0.1% BSA and 0.02% Tween in PBS).
Mouse experiments
C57BL/6 mice (8–14 weeks old) were immunized with EndoFit OVA (Invivogen) or SARS-CoV-2 RBD. For viability and secretion time course experiments, 33–38-week-old nonimmunized mice were used. Antigens were prepared as 2 mg ml−1 sterile-filtered solutions in PBS, mixed with an equal volume of Alhydrogel adjuvant (Invivogen) and incubated on a tube roller for 30 min at 4 °C. Each mouse was injected intraperitoneally with 100 μl of 1 mg ml−1 antigen. For confocal experiments and RBD antibody generation, mice were additionally injected intraperitoneally with 100 μl of 0.5 mg ml−1 OVA or RBD in PBS 21 days after the first immunization. Spleens were collected 10 days after the first immunization, and bone marrow cells were collected 21 days after the first or second immunization.
Extraction of splenocytes and bone marrow
A single-cell suspension of splenocytes was prepared by mashing the spleens through a 100 μm cell strainer into wash buffer (PBS, 0.1% FBS and 2 mM EDTA). Splenocytes were washed in ice-cold wash buffer and then filtered again through a 70 μm strainer. For bone marrow extraction, the femur and tibias were removed from both legs, the ends of the bones were cut open with scissors and the bone marrow was flushed out using a 23-gauge needle and a syringe filled with wash buffer. A single-cell suspension was prepared by filtering through a 100 μm cell strainer. Red blood cells of both spleen and bone marrow were lysed by incubation in 1 ml of RBC lysis buffer (155 mM NH4Cl, 12 mM NaHCO3 and 0.1 mM EDTA) for 5 min at room temperature. Cells were washed once in wash buffer and used for ASC enrichment. All centrifugation steps were carried out at 300g for 5 min at 4 °C.
Mouse ASC enrichment
Plasma cells were enriched using direct magnetic labeling with anti-CD138 (syndecan-1) microbeads (Miltenyi Biotec). Enriched cells were washed once in wash buffer, and an Fc blocking step was performed using the TruStain FcX (anti-mouse CD16/32) antibody (BioLegend).
Mouse ASC encapsulation
BG-agarose (3–6%, wt/vol) was dissolved in PBS, melted by heating to 85 °C using a heated shaker and then cooled to 37 °C. For FACS experiments following OVA immunization, 6% BG-agarose was mixed with anti-mouse κ VHH–SNAP before encapsulation to obtain 100 μM VHH–SNAP and 3% BG-agarose. BG-agarose can be functionalized with VHH–SNAP before or after encapsulation, but in our hands, VHH addition after hydrogel bead formation and demulsification leads to a slightly better signal-to-noise ratio. For all other experiments, a 3% (wt/vol) BG-agarose solution was used directly without prior VHH–SNAP addition. ASCs obtained through magnetic selection (on average >2 × 106 cells per spleen and >1.5 × 106 cells for bone marrow) were washed, resuspended in encapsulation buffer (PBS, 10% ultra-low IgG FBS, 18% OptiPrep, 2 mM EDTA and 0.1% Pluronic F-68) at a concentration of 3 × 107 cells per ml and filtered through a 40 μM strainer. Cells were then mixed with an equal volume of 3% BG-agarose in PBS. The solution was aspirated into microfluidic tubing and encapsulated with flow rates of 6–7 μl min−1 for the aqueous phase (cells in agarose) and 16–18 μl min−1 for the oil phase (mean number of cells per droplet λ = 0.12). During encapsulation, the microfluidic chip was heated to 37 °C using a heated glass microscope plate (Bioscience Tools), although this is not strictly necessary as the low-melting-point agarose will remain liquid for at least 1 h. We routinely ran two microfluidic chips in parallel, enabling encapsulation of up to 1.26 × 107 cells per hour. Alternatively, the mean number of cells per droplet can be increased up to 0.3. The emulsion was collected on ice to solidify the agarose. Hydrogel beads were demulsified by the addition of 800 μl PBS and 200 μl PFO, filtered through a 50 μm strainer and washed once in 15 ml cold FACS buffer. If VHH–SNAP had not been added before encapsulation, the hydrogel beads were resuspended in 50 µM anti-mouse κ VHH–SNAP in PBS, incubated for 15 min at room temperature and then washed once in FACS buffer. For IgG secretion control experiments (designated as −VHH) and viability time course experiments, no VHH–SNAP was added. Hydrogel beads were then resuspended in prewarmed medium at a cell concentration of ≤106 cells per ml (RPMI supplemented with 10% ultra-low IgG FBS, 20 mM HEPES, 50 μM ß-mercaptoethanol, 1 mM sodium pyruvate, 2 mM l-glutamine and 100 U penicillin/streptomycin) and incubated at 37 °C in 5% CO2 for the indicated time frames (1 h and 45 min for antibody selections and confocal experiments). For secretion and viability time course experiments, a fraction of the cells was not encapsulated but kept on ice during the encapsulation/functionalization process and was then incubated and stained in the same way as encapsulated cells. The medium supernatant of samples from the secretion time course was collected for IgG detection by ELISA. All centrifugation steps were carried out at 300g for 10 min at 4 °C.
To reduce potential ‘false positive’ events because of cross-contamination between beads, the sample can be incubated in a larger volume and slowly agitated/rotated during the secretion process. Sample agitation distributes contaminating antibodies across the whole bead population, of which 90% are empty beads and will be gated out during the sort.
Staining of encapsulated mouse ASCs for flow cytometry
After incubation, samples were washed once in FACS buffer. For viability time course experiments, hydrogel beads and non-encapsulated ASCs were first stained with Calcein AM (Invitrogen) according to the manufacturer’s instructions. Samples were incubated with anti-CD138 BV785, anti-IgG1 PerCP-Cy5.5 and anti-CD45R/B220 AF647 for 30 min on ice and then washed once in FACS buffer. Before the acquisition, DAPI (2.5 µg ml−1; Sigma-Aldrich) was added to each tube. For secretion time course experiments, hydrogel beads and non-encapsulated ASCs were stained with anti-CD138 phycoerythrin (PE), anti-IgG1 BV421, anti-IgM BV605 and anti-CD45R/B220 BV785 for 30 min on ice and then washed once in FACS buffer. Before the acquisition, helix green (2.5 µM; BioLegend) was added to each tube.
For OVA antibody selection experiments, hydrogel beads were stained with anti-CD138 PE, anti-FLAG PE-Cy7, OVA-Alexa Fluor 647 (3 μg ml−1; Thermo Fisher Scientific), anti-IgG1 BV421, anti-IgG2a/2b BV421, anti-IgG2a BV421, anti-IgG3 BV421, anti-IgM BV605 and anti-CD45R/B220 BV785 for 30 min on ice and then washed once in FACS buffer. Before the acquisition, 5 μl of 7-aminoactinomycin D (7-AAD, 50 µg ml−1; BioLegend) or helix green (1 µM) were added to each tube. For RBD experiments, hydrogel beads were stained with strep-RBD Alexa Fluor 647 (40 nM), anti-IgG1 BV421, anti-IgG2a/2b BV421, anti-IgG2a BV421, anti-IgG3 BV421, anti-IgM BV605 and anti-CD138 BV785 for 30 min on ice and then washed once in FACS buffer. Before the acquisition, helix green (1 µM) was added to each tube. For antibody clones and dilutions, see Supplementary Table 6.
Analysis of encapsulated mouse ASCs by confocal microscopy
Hydrogel beads were stained with OVA-Alexa Fluor 555 (1.5 μg ml−1; Thermo Fisher Scientific), anti-CD138 Alexa Flour 647 (1:80 dilution; BioLegend, clone 281-2), donkey anti-mouse IgG Alexa Fluor Plus 405 (Thermo Fisher Scientific, A48257; 1:200 dilution) for 30 min on ice and then washed once in FACS buffer. For confocal imaging, 5 µl of unfixed, unpermeabilized encapsulated cells were pipetted into microscopy-compatible 35 mm dishes (ibidi, 81156) and overlayed with 500 µl anti-evaporation oil (ibidi, 50051). Imaging was performed on a Leica TCS SP8 confocal microscope. Fluorophores were excited with a 405 nm, a 552 nm and a 638 nm laser, respectively. Images were analyzed with Fiji86. The plot profile function was used to determine the signal distribution across the hydrogel beads.
Analysis and sorting of mouse ASCs by flow cytometry
For viability and secretion time course experiments, samples were analyzed on a BD LSRFortessa. Before sorting, hydrogel beads were filtered through a 50 μm strainer. Hydrogel beads were sorted on a four-laser BD FACSAria III using a 100 μm nozzle into 96-well plates (Frame Star, 4titude) that contained 4 μl of lysis buffer (3 U μl−1 RNAseOUT RNase inhibitor and 1 mM DTT in 0.5× PBS87). Plates were spun down for 30 s at 400g immediately after the sort, placed on dry ice and then stored at −80 °C until processing. Hydrogel beads were gated to contain live ASCs (CD138+) that were IgG-positive but IgM-negative. The antigen signal (OVA or RBD) of hydrogel beads that were IgG−/IgM− was used as guidance for setting the antigen-positive gate. For OVA experiments, an anti-FLAG antibody was used to identify cells encapsulated into the functionalized BG-agarose hydrogel as the VHH–SNAP construct contained a FLAG tag. The secretion control sample, to which no VHH–SNAP–FLAG was added, was therefore gated as FLAG−.
PBMC isolation from peripheral blood
Peripheral blood (40 ml) was collected from two study participants in lithium heparin tubes. PBMCs were isolated by density gradient centrifugation using Histopaque-1077 (Sigma-Aldrich). Blood was diluted 1:1 with PBS (room temperature), and 35 ml of blood was layered onto 15 ml of Histopaque and centrifuged (acceleration: 1 and deceleration: 0, 800g for 20 min at room temperature). The buffy coat was removed, and the PBMCs were washed twice in ice-cold wash buffer (0.1% FBS and 2 mM EDTA in PBS). PBMCs were counted, and one half was used directly for encapsulation experiments while the other half was resuspended in FBS and then mixed 1:1 with freezing medium (50% FBS and 10% DMSO in RPMI). PBMCs were put into a freezing container (cooling rate: −1 °C min−1) and stored at −80 °C before transfer to liquid nitrogen. Unless stated otherwise, all centrifugation steps were carried out at 400g for 10 min at 4 °C.
PBMC stimulation
Normal cryopreserved human PBMCs (10 million cells per vial; Lonza, 4W-270) were thawed quickly at 37 °C and added dropwise to 10 ml of prewarmed PBMC medium (RPMI 1640 supplemented with 10% FBS, 2 mM GlutaMAX, 1 mM sodium pyruvate, 10 mM HEPES (pH 7.4), 1× MEM nonessential amino acids (Sigma-Aldrich), 50 μM ß-mercaptoethanol, 0.25 μg ml−1 amphotericin B and 100 U ml−1 penicillin/streptomycin) with 100 μg ml−1 DNase (STEMCELL Technologies). Cells were spun down at 500g for 10 min at 4 °C, resuspended in 2 ml of medium and rested for 1 h in the incubator (37 °C in 5% CO2). Stimulation medium (2× concentration) was prepared by adding 400 ng ml−1 HA-tagged rhCD40L (R&D Systems) and 100 ng ml−1 anti-HA tag antibody (R&D Systems, clone 543851) to PBMC medium and incubating for 15 min at room temperature to cross-link the proteins. Afterward, 100 ng ml−1 human interleukin (IL)-21 (Peprotech) and 20 μg ml−1 anti-IgM antibody (polyclonal F(ab′)2 fragment goat anti-human; Jackson ImmunoResearch, AB_2337553) were added. Cells were washed once in magnetic-activated cell sorting (MACS) buffer (2 mM EDTA, 0.5% BSA in PBS), and B cells were negatively selected with the Human Pan-B Cell Isolation Kit (Miltenyi Biotec). Enriched B cells were spun down and resuspended in PBMC medium supplemented with 80 μg ml−1 apotransferrin (Sigma-Aldrich) at 106 cells per ml. Cells were then mixed 1:1 (vol/vol) with stimulation medium and incubated for 96 h.
Encapsulation of stimulated PBMCs
A preheated 6% (wt/vol) BG-agarose solution in PBS was mixed 1:1 (vol/vol) with anti-human κ VHH–SNAP (final concentration 100 μM) and left at 37 °C for 15 min before use. Cells were washed once in MACS buffer, and CD38+ cells were positively selected using the CD38 MicroBead kit (Miltenyi Biotec). CD38+ cells were washed once, and a blocking step using FcR Blocking Reagent (Miltenyi Biotec) was performed. Cells were then resuspended in encapsulation buffer and filtered through a 40 μM strainer to obtain a concentration of 3 × 107 cells per ml. Cells were then mixed with an equal volume of VHH-functionalized 3% BG-agarose in PBS, encapsulated into agarose and demulsified as previously described. Beads were filtered through a 50 μm strainer and washed once in 15 ml cold FACS buffer. Afterward, hydrogel beads were resuspended in prewarmed PBMC medium (supplemented with 10% ultra-low IgG FBS instead of normal FBS) and incubated for 1 h and 45 min at 37 °C in 5% CO2. All centrifugation steps were carried out at 300g for 10 min at 4 °C.
Staining of encapsulated normal PBMCs for FACS/microscopy
After incubation, hydrogel beads were washed once in FACS buffer and then stained with anti-CD14 VioBlue, anti-CD3 VioBlue, anti-IgD VioBlue, anti-CD19 VioBright-FITC, anti-CD27 APC-Vio770, anti-CD20 PE-Vio770, anti-CD38 APC, anti-IgG Fc PE, anti-IgM BV711 and anti-FLAG Alexa Fluor 594. The sample was incubated on ice for 30 min and washed once in FACS buffer. DAPI (1:50,000) was added before analysis on an Attune NxT flow cytometer. Hydrogel beads were examined for APC (anti-CD38) and PE fluorescence (IgG secretion) under an EVOS FL fluorescence microscope (Thermo Fisher Scientific) using filters for Cy5 (ex, 628/40 nm and em, 692/40 nm) and RFP (ex, 531/40 nm and em, 593/40 nm), respectively. For antibody clones and dilutions, see Supplementary Table 6.
Encapsulation of PBMCs post vaccination
Blood was collected from study participants 7–9 days after the second BNT162b2 vaccine dose. Freshly isolated PBMCs were used directly for B cell enrichment. Cryopreserved PBMCs were thawed quickly at 37 °C and added dropwise to 10 ml of prewarmed medium (RPMI 1640 supplemented with 10% ultra-low IgG FBS, 2 mM GlutaMAX, 1 mM sodium pyruvate, 10 mM HEPES (pH 7.4), and 100 U ml−1 penicillin/streptomycin) with 100 μg ml−1 DNase (STEMCELL Technologies). Cells were spun down (400g for 10 min at 4 °C) and washed once in MACS buffer. B cells were negatively enriched with the Human Pan-B Cell Isolation Kit (Miltenyi Biotec). On average, 5–8 × 106 cells were obtained per purification. Enriched B cells were spun down, and a blocking step using FcR Blocking Reagent (Miltenyi Biotec) was performed. Cells were resuspended in encapsulation buffer and filtered through a 40 μM strainer to obtain a concentration of 3 × 107 cells per ml. Cells were mixed with an equal volume of 3% BG-agarose in PBS and encapsulated and demulsified as previously described for mouse ASC encapsulation (λ = 0.12; two microfluidic chips run in parallel). Hydrogel beads were filtered through a 50 μm strainer and washed once in 15 ml cold FACS buffer. Beads were resuspended in PBS with anti-human VHH-SNAPs (anti-κ and anti-λ; 50 µM of each) and incubated for 15 min at room temperature. Hydrogel beads were washed once in cold FACS buffer and resuspended in prewarmed medium supplemented with 10 ng ml−1 human IL-6 and incubated for 1 h and 45 min at 37 °C in 5% CO2.
Staining and sorting of encapsulated PBMCs post vaccination
For sorting with RBD, hydrogel beads were washed once in FACS buffer and stained with strep-RBD Alexa Fluor 647 (40 nM), anti-IgG Fc PE, anti-IgM BV785, anti-CD38 BV421 and anti-CD20 BV510 for 30 min on ice. Beads were washed once in FACS buffer, and helix green (1 µM) was added. For sorting with RBD and S1, hydrogel beads were washed once in FACS buffer and then stained using strep-S1 Alexa Fluor 647 (40 nM), strep-RBD Alexa Fluor 488 (40 nM), anti-IgG Fc PE, anti-IgA PE, anti-IgM BV785, anti-CD38 BV510 and anti-CD20 PE-Cy7 for 30 min on ice. Hydrogel beads were washed once in FACS buffer, and DAPI (Thermo Fisher Scientific; 1:50,000) was added to the sample. Beads were (index) sorted on a four-laser BD FACSAria III using a 100 μm nozzle into 96-well plates (Frame Star, 4titude) that contained 6 μl of lysis buffer (5% PEG 8000, 0.1% Triton X-100, 0.5 U μl−1 RNAseOUT, 0.5 μM OligoT30VN, 0.5 μM random hexamers and 0.5 mM/each dNTP). The antigen signal of hydrogel beads that showed neither the IgG/IgA nor the IgM signal was used as guidance for setting the antigen-positive gate. Plates were spun down for 30 s at 400g immediately after the sorting, placed on dry ice and then stored at −80 °C until processing. For antibody clones and dilutions, see Supplementary Table 6.
Reverse transcription, PCR and sequencing (mouse)
Plates were thawed on ice for 5 min and spun down at 400g for 1 min at 4 °C. For lysis and reverse transcription, a combination of previously published protocols87,88 was used together with the template-switch oligo from the 10X Chromium Next GEM Single Cell V(D)J Reagent Kit v1.1 (without barcode and unique molecular identifier). See Supplementary Table 7 for primer sequences. Per well, 7 μl of lysis and annealing mix (0.6% IGEPAL, 5.5 mM random hexamer primers, 3.99 mM anchored oligo-dT and 0.9 U μl−1 RNaseOUT) were added, and the plate was incubated at 65 °C for 5 min. Plates were placed on ice for 5 min, and 10 μl of reverse transcription mix was added to each well (15% PEG 8000, 2× Maxima RT buffer, 2 mM dNTPs, 4 μM template-switch oligo (10X_TSO) and 4 U μl−1 Maxima H Minus Reverse Transcriptase). Plates were spun down and transferred to a thermocycler for reverse transcription (90 min at 42 °C, five cycles of 2 min at 50 °C followed by 2 min at 42 °C). Plates containing cDNA were stored at −20 °C. Leader peptide and variable regions of both the heavy chain and κ light chain were then amplified in two rounds of semi-nested PCR. For the first PCR, 8 μl of cDNA reaction mix was added to 17 μl of PCR mix using KAPA HiFi HotStart polymerase and KAPA HiFi Fidelity Buffer according to the manufacturer’s instructions (primer concentrations: 0.3 μM fw_TSO_PCR1, 0.3 μM mIgG12abc, 0.1 μM mIgG2b, 0.1 μM mIgG3-1, 0.1 μM mIgG3-2 and 0.2 μM mκ). The PCR was performed at 95 °C for 3 min, followed by 20 cycles of 20 s at 98 °C, 30 s at 67 °C and 70 s at 72 °C and a 5 min final elongation at 72 °C. From the first PCR reaction, 5 μl was transferred to the second PCR (final volume, 25 μl) for 22 amplification cycles. DNA obtained from anti-OVA hits was cut out of agarose gels, and amplicons from heavy and light chains were purified separately by gel extraction (Zymoclean Gel DNA Recovery Kit, Zymo Research).
For RBD experiments, PCRs were split after the first PCR, and heavy and light chains were amplified separately using final primer concentrations of 0.3 μM for the reverse κ primer and the same conditions for the heavy-chain primers as for the first PCR. Amplicons from anti-RBD hits were purified by SpeedBead (Sigma-Aldrich) purification. PCR reactions that resulted in sufficient material (>100 ng determined by absorbance measurement at 260 nm) were subjected to Sanger sequencing. We found that sequencing efficiency was suboptimal with this protocol and recommend following a published protocol that does not use a template-switch oligo87. We also advise decreasing the amount of PFO for demulsification or using chemical-free breaking of the droplet emulsion, as PFO can impact the efficiency of enzymatic reactions89.
Reverse transcription, PCR and sequencing (human)
Reverse transcription in each well was performed based on a modified Smart-seq3 protocol using a template-switch approach90. Specifically, all volumes for the lysis and reverse transcription steps were doubled, and random hexamers (0.5 μM final concentration) were added to the lysis buffer mix. All other steps were performed according to the published protocol. All oligonucleotides are listed in Supplementary Table 8. For anti-RBD antibodies, 12 μl of PCR 1 mix (final concentrations in KAPA HiFi Fidelity Buffer: 0.3 mM/each dNTPs, 0.5 mM MgCl2, 0.5 μM fw_Read1/2 primer and 0.1 μM of each reverse primer (rv_IgG_HC_PCR1, rv_kappa_PCR1, rv_lambda_PCR1), 0.02 U μl−1 KAPA HiFi HotStart polymerase) were added to each well after reverse transcription. PCR reactions were performed at 95 °C for 3 min, 20 cycles of 20 s at 98 °C, 30 s at 65 °C and 2 min 30 s at 72 °C and a 5 min final elongation step at 72 °C. For the semi-nested second PCR, 5 μl were transferred from the first PCR, and PCRs were performed using KAPA HiFi HotStart polymerase and KAPA HiFi Fidelity Buffer. Because we initially planned to use Illumina sequencing76, we performed two sets of second PCRs, one that introduces the full-length Illumina Read 1 or Read 2 at the 5′ end and Read 2 or Read 1 at the 3′ end. For this reaction, fw_R1_PCR2 was combined with rv_R2 primers, and fw_R2_PCR2 was combined with rv_R1 primers at 0.3 μM final concentration each in 25 µl. PCR reactions were performed as previously but with 22 amplification cycles. PCR products were purified using SpeedBeads (Sigma-Aldrich) and eluted with 15 μl of nuclease-free water. For the addition of barcode indices and Illumina P5/P7 adapters, a third PCR was performed. For PCRs with 5′ Read 1, i5 primers (P5_1–P5_96, used as well barcodes) were used as forward primers, and i7 primers (P7_1–P7_20, used as plate barcodes) were used as reverse primers. For PCRs with 5′ Read 2, i5 primers were used as reverse primers and i7 primers as forward primers. PCRs were performed using KAPA HiFi HotStart polymerase and KAPA HiFi Fidelity Buffer in a 25 μl volume, with 0.3 μM of primers and 2 μl of PCR 2 as template with annealing at 65 °C and 22 cycles of amplification. PCR products were purified using SpeedBeads (Sigma-Aldrich). PCR reactions that resulted in sufficient material (>100 ng determined by absorbance measurement at 260 nm) were subjected to Sanger sequencing. We found that this protocol did not yield high sequencing efficiencies (≅17%) and therefore decided to amplify the cDNA from RBD/S1 hits with a different primer set that does not rely on template switching91. After reverse transcription, the cDNA was diluted by the addition of 27.4 µl of nuclease-free water. For the first PCR reaction, 4 µl of diluted cDNA were mixed with 21 µl of PCR 1 mix (final concentrations in KAPA HiFi Fidelity Buffer: 0.3 mM/each dNTP, 0.5 mM MgCl2, 0.2 µM oPR_IGHV mix, 0.2 µM oPR_IGKV mix, 0.2 µM oPR_IGLV mix, 0.2 µM oPR_1st_rv mix and 0.02 U μl−1 KAPA HiFi HotStart polymerase). PCR reactions were performed at 95 °C for 3 min, followed by 50 cycles of 30 s at 98 °C, 30 s at 62 °C and 45 s at 72 °C and a 5 min final elongation step at 72 °C. Three separate, semi-nested second PCRs were performed for the amplification of heavy chains, κ and λ light chains. For the second PCR, 4 µl of the first PCR was used directly and mixed with 21 µl of PCR 2 mix (final concentration in KAPA HiFi Fidelity Buffer: 0.3 mM/each dNTP, 0.3 µM forward primer mix (oPR_IGHV mix, oPR_IGKV mix or oPR_IGLV mix) with the respective 0.3 µM reverse primer mix (IgG/IgA rv mix, 3′ Cκ 494 (κ) or 3′ XhoI Cλ (λ)), 0.02 U μl−1 KAPA HiFi HotStart polymerase) with the same conditions as PCR 1. PCR products were purified using SpeedBeads (Sigma-Aldrich) and sequenced by Sanger sequencing, leading to an increase in matched heavy- and light-chain sequences (over 40%).
Bioinformatic analysis
Antibody sequences were annotated using international ImMunoGeneTics information system (IMGT)/V-Quest from the IMGT database92. All clonotyping was performed using an in-house script (code available in the Supplementary Information), with antibodies greedily clustered if they had a common IGHV gene assignment, the same length CDR3 and within 80% CDR3 amino acid identity of the cluster representative. The CoV-AbDab35 (February 2022 version) was downloaded as a set of anti-coronavirus reference antibodies with functional annotations. Clonotyping was then applied to functionally cluster input files containing each set of ASC-derived antibodies concatenated with all CoV-AbDab antibodies with a complete IGHV and CDRH3 label. An alternative clustering method based on grouping antibodies by similar predicted 3D structure (SPACE) was also applied51, using a threshold structural similarity of 0.75 Å. This was performed over each set of ASC-derived antibodies coupled with the subset of antibodies in CoV-AbDab with full variable heavy (VH) and variable light (VL) chain sequences (as sequence coverage over all six CDRs is required by SPACE).
Cloning of antibodies
For mouse antibodies, primers were designed to amplify the variable regions, including the leader peptides. To generate PCR amplicons, Q5 Polymerase High-Fidelity 2X Master Mix (New England Biolabs (NEB)) was used according to the manufacturer’s protocol using the second sequencing PCR as a template. For anti-OVA antibodies, variable regions and leader peptides were cloned into pVITRO1 (Invivogen) plasmids that contained the constant regions for mouse IgG1 κ heavy and light chains using Gibson assembly83. For anti-RBD antibodies, standard restriction cloning using the type IIS restriction enzyme BsmBI-v2 (NEB) was used. Antibodies were excluded from further testing if the plasmid sequencing did not match the expected sequence.
For human antibodies, primers were designed to amplify the variable regions, excluding the leader peptides. To generate PCR amplicons, Q5 Polymerase High-Fidelity 2X Master Mix (NEB) was used with the second or third PCR as a template. Variable regions were cloned into pVITRO1 plasmids that already contained the constant regions for human IgG1 heavy and light chains and leader peptides (κ or λ; Addgene, plasmids 61883 and 50366)93 using Gibson assembly83. For amino acid sequences of the variable regions of the antibodies, see Supplementary Tables 2 and 4.
Antibody expression and purification
FreeStyle 293-F cells (Thermo Fisher Scientific) were maintained in shaking flasks in Freestyle 293 Expression Medium at 0.4–2 × 106 cells per ml (incubator conditions: 37 °C, 70% humidity, 8% CO2 and 125 rpm). Per 106 cells, 1.2 μg of DNA was mixed with 2.4 μl of 1 mg ml−1 polyethylenimine (linear PEI, MW 25,000, Polysciences) in one-tenth of the final culture volume of medium. The mixture was vortexed for 15 s followed by 15 min incubation at room temperature and added to a culture of 106 cells per ml. Valproic acid (2-propyl-pentanoic acid; Sigma-Aldrich) was added 4 h post transfection to a final concentration of 3.5 mM. The culture supernatant was collected 1 week after transfection by centrifugation for 15 min at 4,000g, sterile filtered and stored at 4 °C. The supernatant was added to a gravity column containing 0.5 ml of Pierce Protein G Agarose (Thermo Fisher Scientific) pre-equilibrated with PBS. The column was washed with ten CVs of PBS, and the antibodies were eluted in several fractions of 250 μl with elution buffer (10 mM glycine, pH 2.7) into collection tubes containing 12.5 μl of neutralization buffer (1 M Tris–HCl, pH 9.0). The fractions containing protein (measured by absorbance at 280 nm) were pooled, and the buffer was exchanged to PBS. The purity of the proteins was confirmed by SDS–PAGE. Antibodies were excluded from further testing if it was not possible to express the antibody in a single attempt.
ELISA
Anti-mouse IgG ELISA
The concentration of mouse IgG in the supernatant of encapsulated and non-encapsulated samples was determined with the IgG (total) Mouse Uncoated ELISA Kit (Invitrogen, 88-50400; assay range 1.56–100 ng ml−1). The assay was performed with 50 µl of undiluted supernatant, a standard concentration range of 0.78–100 ng ml−1 IgG, and with medium used as a negative control. Absorbance was read at 450 nm in duplicate in a Varioskan LUX plate reader (Thermo Fisher Scientific). The standard curve was created in GraphPad Prism using the Sigmoidal, 4PL, X is log(concentration) function. The mean signal of negative control wells was subtracted from all measurements.
Anti-RBD/S1 ELISA
Microtiter plates (384 well Nunc MaxiSorp, flat bottom; Thermo Fisher Scientific) were coated with 20 μl of 2 μg ml−1 RBD or S1 in PBS overnight at 4 °C. Plates were washed twice with 90 μl of PBS-T and incubated for 1 h at room temperature with 40 μl of blocking buffer (2.5% BSA in PBS-T). Plates were washed twice with PBS-T. Antibody samples were diluted in 1% BSA in PBS-T (final concentrations: 400 nM, 200 nM, 20 nM, 2 nM, 0.2 nM, 0.02 nM, 0.002 nM and 0.0002 nM), and 20 μl of the sample was added per well followed by incubation at room temperature for 1 h. Commercial anti-RBD antibodies (mouse IgG1κ, clone 1035753 (R&D Systems) or human IgG1κ, clone AM001414 (BioLegend)) and isotype controls (mouse IgG1κ, clone MG1-45 (BioLegend)) or human IgG1κ, clone QA16A12 (BioLegend)) were used as controls. Plates were incubated for 1 h at room temperature, washed four times with PBS-T and 20 μl of polyclonal HRP-conjugated detection antibodies (goat anti-mouse (Thermo Fisher, G-21040) at 1:1,000 dilution or goat anti-human (Abcam, ab7153) at 1:10,000 dilution) were added to each well and incubated for 30 min (anti-mouse) or 45 min (anti-human) at room temperature. The plate was washed three times with PBS-T and one time with PBS. 3,3′,5,5′-tetramethylbenzidine (Thermo Fisher Scientific; 20 μl per well) was added followed by the addition of 0.16 M H2SO4 (20 μl per well) after 5 min. Absorbance was read at 450 nm and 540 nm in a Tecan Infinite 200 PRO plate reader. The absorbance at 540 nm was subtracted from the signal at 450 nm. To obtain EC50 values, titration curves were plotted as absorbance versus log (antibody concentration) and then analyzed by nonlinear regression using the Sigmoidal, 5PL, X is log(concentration) function in GraphPad Prism94. For the purpose of visualization, mouse antibodies with no quantifiable ELISA signal at the tested concentrations were assigned an arbitrary EC50 of 1,000 nM. Mouse antibodies that did not show binding to RBD under these conditions were also tested in an anti-streptavidin ELISA (under the same conditions as described previously except that plates were coated with 20 μl of 2 μg ml−1 streptavidin). Under the tested conditions, no binding to streptavidin was observed.
Determination of VHH and antibody binding affinities
Binding affinities were determined using the Octet RED96 BLI system. VHH binding affinities were measured by immobilizing recombinant mouse and human isotype control antibodies (10 μg ml−1) of human IgG1λ (clone AbD00264_hIgG1, BioRad) or mouse IgG1κ (clone P3.6.2.8.1, Thermo Fisher Scientific) on preconditioned anti-human or anti-mouse IgG Fc-capture biosensors (AMC and AHC; Sartorius) for 600 s. Control sensors were incubated in kinetics buffer (0.1% BSA and 0.02% Tween in PBS) during the antibody loading step. Biosensors were then probed with different concentrations of VHH. Mouse mAbs (7.5 μg ml−1 anti-OVA and 5 μg ml−1 anti-RBD) were captured on preconditioned anti-mouse IgG Fc-capture biosensors (AMC; Sartorius) for 600 s. Human mAbs (5 μg ml−1 or 7.5 µg ml−1) were captured on preconditioned anti-human IgG Fc-capture biosensors (AHC; Sartorius) for 600 s. Control sensors were incubated in kinetics buffer during the antibody loading step. Biosensors were then probed with different concentrations of the corresponding antigens. All dilutions were performed in kinetics buffer. Sensors were regenerated with 10 mM glycine (pH 1.7). Data were analyzed using Octet data analysis software using a 1:1 binding model. For the purpose of visualization, one human antibody with no quantifiable KD at the tested concentrations was assigned an arbitrary KD value of 10−7 M. For individual binding curves of human RBD and S1 binders, see Supplementary Fig. 2.
Live virus neutralization assays
The SARS-CoV-2 viruses used in this study were a WT (lineage B) isolate (SARS-CoV-2/human/Liverpool/REMRQ0001/2020), a kind gift from I. Goodfellow (University of Cambridge), isolated by L. Turtle (University of Liverpool), D. Matthews and A. Davidson (University of Bristol)95,96, and an Omicron (lineage B.1.1.529) variant, a kind gift from R. Gupta (University of Cambridge)97. Half-maximal inhibitory concentrations (IC50) for indicated mAbs were measured essentially as previously described37,38,39. In brief, luminescent reporter cells expressing SARS-CoV-2 papain-like protease-activatable circularly permuted firefly luciferase (FFluc) were seeded in flat-bottomed 96-well plates. HEK293T-ACE2-30F-PLP2 cells (clone B7; available from the National Institute for Biological Standards and Control, 101062) were used to test mouse antibodies37, and A549-ACE2-TMPRSS2-30F-PLP2 cells (clone E8) were used to test human antibodies97. The next day, SARS-CoV-2 viral stock (multiplicity of infection (MOI) = 0.01) was pre-incubated with a threefold dilution series of each antibody for 1 h at 37 °C, then added to the cells. Twenty-four hours post infection, cells were lysed in Bright-Glo Luciferase Buffer (Promega) diluted 1:1 with 1% NP-40 in PBS, and FFluc activity was measured by luminometry. Experiments were conducted in duplicate or triplicate, as indicated. To obtain IC50 values, titration curves were plotted as FFluc versus log (antibody concentration) and then analyzed by nonlinear regression using the Sigmoidal, 4PL, X is log(concentration) function in GraphPad Prism. IC50 values were quantified when (1) at least 50% inhibition was observed at the highest antibody concentration tested (100 μg ml−1), and (2) a sigmoidal curve with a good fit was generated. For purposes of visualization and ranking, antibodies with no quantifiable neutralizing capacity were assigned an arbitrary IC50 of 1,000 μg ml−1. For individual neutralization curves, see Supplementary Fig. 3.
Epitope binning of mAbs
For in-tandem assays of mouse antibodies, biotinylated RBD (10 μg ml−1) was immobilized on streptavidin BLI biosensors (Sartorius) for 300 s. The biosensors were then probed with 100 nM of the first antibody for 600 s followed by 100 nM of the second antibody for 300 s. Independent binding of both clones was verified with a buffer control. For classical sandwich assays of human antibodies, 10 µg ml−1 of the first antibody was immobilized on IgG Fc-capture biosensors for 600 s. After blocking of free capture sites on the biosensors with isotype control (10 µg ml−1; human IgG1κ, clone QA16A12; BioLegend) for 600 s, sensors were incubated with 20 nM RBD for 300 s. Biosensors were then probed with the second antibody (10 µg ml−1) for 300 s. All dilutions were performed in kinetics buffer (0.1% BSA and 0.02% Tween in PBS). For individual binding curves, see Supplementary Fig. 4.
Purification of mRBD2 Fab and RBD–Fab complex
DNA encoding for the heavy and light chains of the Fab region of mRBD2 was cloned into a bi-cistronic expression cassette in the modified pExp vector. The light chain was under the control of a T7 promoter followed by a short intergenic region containing a favorable ribosomal binding site and the heavy chain with an added C-terminal His-tag. The pExp-Fab expression vector was cotransformed with plasmid pRARE2 (extracted from Rosetta2 (DE3) cells; Novagen) into Shuffle T7 Express cells (NEB). Cells were grown in 2xYT media with 100 µg ml−1 ampicillin and 15 µg ml−1 chloramphenicol to OD600 0.8–1.0, and protein production was induced with 0.4 mM IPTG and proceeded for 16 h at 20 °C. Cells were collected, resuspended in lysis buffer (20 mM HEPES–NaOH (pH 7.2), 300 mM NaCl, 20 mM imidazole, 250 µg ml−1 of DNase I and 1 mM PMSF) and then lysed by sonication. Lysate was cleared by centrifugation, and protein was purified by Ni-affinity chromatography using PureCube Ni-NTA agarose resin. After washing the column with 20 CV of 20 mM HEPES–NaOH (pH 7.2), 300 mM NaCl and 20 mM imidazole, the protein was eluted in 5 CV of 20 mM HEPES–NaOH (pH 7.2), 300 mM NaCl and 250 mM imidazole. Next, the buffer of the eluted protein was exchanged to 20 mM HEPES–NaOH (pH 7.2) and 20 mM NaCl using a HiPrep 26/10 Desalting column (Cytiva), and the Fab was loaded onto the HiTrap Capto SP ImpRes column (Cytiva). The protein was eluted in a linear gradient of 20–1,000 mM NaCl buffer containing 20 mM HEPES–NaOH (pH 7.2). Peak fractions were combined, and the protein was concentrated using 30 kDa MWCO Amicon Ultra concentrators and further purified by size-exclusion chromatography using the HiLoad Superdex 75 pg 16/60 column (Cytiva) equilibrated in 20 mM Tris–HCl (pH 7.2) and 250 mM NaCl. For the RBD–Fab complex formation, an excess of RBD was incubated with concentrated purified Fab for 1 h at 22 °C, and then the complexes were separated by size-exclusion chromatography using the HiLoad Superdex 75 pg 16/60 column (Cytiva) equilibrated in 20 mM Tris–HCl (pH 7.2) and 250 mM NaCl. The RBD–Fab protein complex in peak fractions was concentrated to 10 mg ml−1 using 30 kDa MWCO Amicon Ultra concentrators and used for crystallization trials.
RBD-mRBD2 Fab crystallization and structure determination
Crystallization of the RBD–Fab complex was performed in sitting drops using the mosquito liquid dispensing robotic system (SPT Labtech) by mixing 200 nl of protein solution with 200 or 100 nl of precipitant in 96-well 2-drop MRC crystallization plates. Crystals appeared in several conditions of the Ligand-Friendly Screen (Molecular Dimensions) after 3–8 days at 16 °C; they were cryo-cooled in liquid nitrogen without the need for additional cryoprotectant. The crystals resulting in the presented structural data grew in a condition containing 100 mM Bis-Tris propane (pH 6.5), 200 mM potassium thiocyanate, 20% wt/vol PEG 3350 and 10% vol/vol ethylene glycol. X-ray diffraction data were collected at Diamond Light Source on beamline I04 and processed by auto-processing pipelines using autoPROC and Staraniso (Global Phasing). The structure was solved using molecular replacement (Phaser98) with structures of the RBD (Protein Data Bank (PDB): 6YM0) and a Fab fragment (PDB: 3BGF) as search models. The structure was refined with phenix.refine99 and autoBuster (Global Phasing) with manual fixing using Coot100. The final structure (PDB: 8BE1) contains two RBD:Fab complexes per asymmetric unit. As the complex made from chains D, E and F is likely affected by crystal packing (ca. 14° shift in the RBDs relative to the Fab; Supplementary Fig. 5), we used the other complex for analysis. Crystallographic data collection, data reduction and refinement statistics are shown in Supplementary Table 9.
Data analysis
Flow cytometry data were analyzed using FlowJo (version 10.7). Analysis of Sanger sequencing reads was performed using Geneious Prime (version 2019.2.1; Biomatters). Antibody sequences (V gene, J gene, nt identity, CDR3) were analyzed with IMGT/V-Quest from the IMGT database92. All other bioinformatics were performed using Python. Clonotyping was performed using an in-house script (code available in Supplementary Information). Structural clustering was performed using the SPACE algorithm51 (codebase released with the original paper at https://doi.org/10.1371/journal.pcbi.1009675). All BLI data were analyzed using Octet data analysis software (version 11.0.0.4). Microscopy data were analyzed using Fiji86. X-ray diffraction data were processed using autoPROC (version 1.0.2) and Staraniso (version 1.0.2) (Global Phasing). XIA2 (version 3.8.dev0) and DIALS (version 3.dev.620-g8f4b6839d) were used for crystallographic diffraction data processing. Phenix (version 1.16), CCP4 package (version 7.1), Buster (version 2.10.4) and Coot (version 0.9.8.1) were used for structure determination. PyMOL (version 2.5.0) was used for structure analysis and to generate structure figures. Additional data were analyzed using GraphPad Prism 6 and Microsoft Excel. Adobe Illustrator (2015) was used to create all other figures and illustrations.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
