Ethics statement
The animal work was conducted with the approval of the regional Ethical Committee on Animal Experiments (Stockholms Norra Djurförsöksetiska Nämnd). All animal procedures were performed according to approved guidelines, following permits #18427-2019 and #10895-2020.
Animals’ immunization and sample collection
One male (I10) and one female (H03) rhesus macaque (Macaca mulatta) of Chinese origin, 4-5 years old, were housed at the Astrid Fagraeus Laboratory at Karolinska Institutet. Housing and care procedures complied with the provisions and general guidelines of the Swedish Board of Agriculture. The facility has been assigned an Animal Welfare Assurance number by the Office of Laboratory Animal Welfare (OLAW) at the National Institutes of Health (NIH). The macaques were housed in groups in 14 m3 enriched cages. They were habituated to the housing conditions for more than six weeks before the start of the experiment and subjected to positive reinforcement training to reduce the stress associated with experimental procedures. The macaques were weighed at each sampling. All animals were confirmed negative for simian immunodeficiency virus (SIV), simian T cell lymphotropic virus, simian retrovirus type D and simian herpes B virus. All immunizations and samplings were performed under sedation with 10-15 mg/kg ketamine (Ketaminol 100 mg/ml, Intervet, Sweden) administered intramuscularly (i.m.). For macaque immunizations, stabilized spike trimer (100 μg) was mixed in 75 μg of Matrix-M (Novavax AB). Macaques were immunized intramuscularly (i.m.) with half of the dose administered in each quadricep at weeks 0, 4 and 9 and weeks 30 (only I10) and 31 (only H03). Blood samples were collected pre-immunization and at weeks 2, 6, 11, 19, 23, 27, 32 (only I10) and 46 (only I10). Draining iLNs were collected at weeks 6 (right side), 11 (left side), 32 (only H03, at termination) and 46 (only I10, at termination). BM samples were collected at weeks 6 and 11 by aspiration from the humerus and at weeks 32 (only H03, at termination) and week 46 (only I10, at termination) by collection and washing of the femur. Spl samples were collected at weeks 32 (only H03, at termination) and week 46 (only I10, at termination). perLN, axLN and mesLN samples were collected at week 32 (only H03, at termination).
Recombinant spike and RBD
Recombinant RBD (encoding a C-terminal His-tag) were synthesized (Integrated DNA Technologies) and cloned into a mammalian expression vector (pcDNA3.1), using a Gibson Assembly Mastermix (New England Biolabs). Spike ectodomain were produced in the format of prefusion stabilized with 2 prolines for immunization as reported in Mandolesi et al.23 and prefusion stabilized with 6 prolines (HexaPro) for ELISAs and Cryo-EM structure resolution as reported in Sheward et al.65. Spike and RBD were produced by the transient transfection of Freestyle 293-F cells using FreeStyle MAX reagent (Thermo Fisher) and polyethylenimine (PEI), respectively. The HIS-tagged Spike ectodomain and RBD were purified from filtered supernatant using nickel IMAC resin (HisPur Ni-NTA, Thermo Fisher Scientific) followed by size-exclusion chromatography on a Superdex 200 (Cytiva) in PBS.
Biotinylated spike probe construction
The spike probe was produced by conjugating biotinylated HexaPro spike to streptavidin-allophycocyanin (SA-APC) (Invitrogen). The biotinylated spike was produced using NHS-Chromalink-Biotin (Solulink) with an average ratio of 3:1 (biotin:spike). The labeled spike was purified on PD-10 desalting columns. For the conjugation, 10 μg of spike protein was incubated with 1 μl of SA-APC (1mg/ml) for 20 minutes on a shaker at 4°C. The process was repeated until a total of 5 μl SA-APC had been added.
Single cell sorting and sequencing
Single cells were sorted with a four-laser FACSAria cell sorter (BD Bioscience) by gating Aqua Live/Dead stain (Life Technologies)–negative, CD3−, CD14−, CD20+, spike+ cells. Spike probes were produced by conjugating a fluorophore via biotinylation as described above. Prior to sorting, TotalSeq-C hashing barcodes for downstream sample identification were added to the following samples collected at week 32: one draining iLN-R1, another distinct draining iLN-R2, spleen and pooled non-draining LNs, LNOther (mesLN, axLN and perLN) obtained from H03. PBMCs from blood samples obtained from I10 were also labeled with a unique TotalSeq-C barcode. The sorted cells were processed with the 10X Chromium, sequencing ‘5’ V(D)J’ enriched libraries, and TotalSeq-C feature barcode libraries. Rhesus Macaques are not a species directly supported by 10X, so we adapted the “Human” 10X kit by spiking-in a mix of IG constant-region primers from Brochu et al.66 during the two enrichment PCR steps (SI. Table 1). Briefly, 5 µl of each primer (100 µM) were added to MasterMix1 and MasterMix2, for enrichment steps 1 and 2, and diluted to a final volume of 50 µl. Each enrichment steps were spiked with 5 µl of MasterMix after removal of 5 µl nuclease free water. VDJ read assembly was run in CellRanger (v3.1.0) “de novo” mode, to avoid reference bias, since CellRanger lacks a Rhesus Macaque reference database. The filtered contigs files were then assigned to individualized H03 or I10 IGH, IGK, and IGL database. Non-productive sequences or sequences without CDR3 identification were removed. Subsequently, we filtered cells with multiple IGH, IGK or IGL assignments and cells with presence of both IGK and IGL. Finally, we set a strict threshold for hashing barcode assignment. The sequences with maximum value of counts per hash lower than 20 were classified as “Low_counts”. Hashes accounting for >= 60% of all counts were selected for assignment. Otherwise, the sequences were classified as “Unassigned”.
HC NGS Library preparation
IgG, multiplexed IgG and IgA, and IgM libraries were prepared following the 5′MTPX protocol described in Vázquez et al. (2019)67 and Vázquez et al. (2021)42. Briefly, cDNA synthesis was performed using the Sensiscript RT kit (QIAGEN), 200 ng of total RNA and isotype specific primer containing 21 nt of semi-structured UMI and the Illumina Read2 sequence for 1 h at 37°C. The cDNA synthesis product was purified (GeneJET PCR purification Kit, Thermo Scientific). The amplification of the second cDNA strand was performed separately with a mix of 5′ forward primers targeting the leader sequence of IGHV genes, reported in Vázquez et al. (2021)42, and Read2U using KAPA HiFi Hotstart ReadyMix system (Kapa Biosystems). Indices were introduced in a 10 cycle PCR reaction. The libraries were sequenced with 15% PhiX174 DNA and the Illumina Version 3 (2 × 300 bp) sequencing kit using a MiSeq 3x300 platform (Illumina).
Individual germline database generation
Individualized germline databased were obtained from IgM, IgK and IgL libraries synthesized from blood samples obtained before the vaccination regimen were used to generate individual germline repertoire via inference analysis performed with the IgDiscover24 pipeline version v1.0.0 (https://gitlab.com/gkhlab/igdiscover22) with default settings. The reference database for IgH genes was obtained from KIMDB42 (http://kimdb.gkhlab.se/) by pooling both rhesus and cynomolgus databases and removing duplicates. The reference databases for IgK and IgL genes were obtained from KIMDB by pooling both rhesus and cynomolgus databases and removing duplicates.
Single cell paired V(D)J sequences animal and compartment assignment
V(D)J sequences from single cell analysis were initially assigned to a compartment via TotalSeq-C hashing barcodes count. With counts equal to or over 20, a threshold of 60% was used to determine reliable assignment to one compartment. The rest of the sequences were divided into two groups, low_count and unassigned, based on TotalSeq-C count lower than 20 or failure to pass the 60% threshold, respectively. To determine whether these sequences belonged to H03 or I10, they were assigned to each animal individualized databases and underwent lineage tracing analysis. Only sequences traced to lineages within only one animal were retained and used for downstream analysis.
Lineage tracing analysis
For each animal, paired V(D)J sequences obtained from single cell analysis were processed using the IgBlast module from IgDiscover for assignment of individualized V(D)J germline repertoire for HC and LC. Bulk IgG and multiplexed IgG and IgA libraries were initially processed with IgDiscover for sequence assignment to individualized VDJ germline repertoire and selection of IgG reads. We then performed a denoising step to remove sequencing errors using the Fast Amplicon Denoising (FAD) tool described in Kumar et al. (2019)68 and removed chimeric sequences (likely due to PCR recombination between unrelated antibodies) using a hidden Markov model designed for this purpose (https://github.com/MurrellGroup/CHMMera/). Lineages were identified using the IgDiscover clonotypes module by combining HC sequences from single cell and bulk sequencing. Lineages were defined by identical V and J allele assignments, identical CDR3 lengths, and placement into the same single-linkage cluster of HCDR3s with a 0.8 nucleotide identity cutoff. If single cell sequences with different LC assignments were assigned to the same lineage, the lineage was redefined by also considering same V and J allele assignment of the LC and same CDR3 length. In case of divergent LCs, lineage assignment was determined by minimum Levenshtein distance to a single cell HC sequence as reference. Sequences from each lineage were then aligned using MAFFT v7.49069. The alignment was used as input to FastTree70 (compiled with the double-precision flag) to compute maximum-likelihood phylogenetic trees. The function phylo::reroot from R package “phytools v1.2-0” was used to root the tree to its germline sequence. The ancestral HCDR3 germline sequence, used for visualization, was inferred using the package MolecularEvolution.jl (https://github.com/MurrellGroup/MolecularEvolution.jl).
Monoclonal antibody generation
VDJ sequences from single cell analysis were re-adapted for Gibson assembly cloning by inserting complementary overlapping sequences to the expression vectors human IgHγ1 (80795), Igκ1 (80796) or Igλ2 (99575) leader and constant regions71. The reaction was carried out by mixing 50 ng of digested vector, 30 ng of VDJ insert and 10 μl of 2X Gibson Assembly Master Mix (New England BioLabs) in a 20 μl reaction and incubating for 1h at 50°C. The reaction product was transformed into XL10-Gold Ultracompetent cells (Agilent) and cultures were scaled up to obtain a suitable quantity of plasmid for expression in HEK293F cells. Before transfection, the presence of the insert and in the correct reading frame was verified by Sanger sequencing. Transfection was carried out by adding a transfection mix to 1.2-1.5 x 106/ml HEK293F cells. Mix for a 30 ml reaction was prepared by mixing 18 ng of HC vector, 18 ng of LC vector and 50 μl of FreeStyle™ MAX Reagent and Opti-MEM in a final volume of 600 μl. The supernatant was harvested after 7 days and mAbs were purified using gravity driven column purification with Protein G Sepharose (Cytivia). From a volume of 30 ml of supernatant, 200 μl protein G Sepharose were loaded in a column and supernatant was applied to the column 5-6 times. After washing with PBS, elution was carried out with 2 ml of elution buffer (0.1M Glycine.HCl, pH 2.7) into 300 μl of neutralization buffer (1M Tris-HCl, pH 9). The solution was then diluted in PBS and concentrated using Pierce™ Protein Concentrators with a cutoff of 30 kDa.
mAb ELISA
ELISA plates were coated with 100 μL of prefusion-stabilized spike or RBD protein at a concentration of 1 μg/ml and blocked for 1 hour at room temperature with 200 μL blocking solution containing 5%(w/v) non-fat milk powder in 1x PBS. MAbs were serially diluted in blocking solution starting from 5 μg/ml were added and incubated for 2 hours at room temperature. Plates were washed 6 times with PBS-T and antibody-antigen interaction was detected using 100 μL with HRP-conjugated anti-human Fcγ Ab (Jackson ImmunoResearch) diluted to 1:10 000 in PBS-T. Plates were washed 6 times with PBS-T, developed using 100 μL of 3,3′,5,5″-tetramethylbenzidine (TMB) substrate solution (Invitrogen) per well and stopped using 100 μL of 1M sulphuric acid per well. OD was read at 450 nm in an Asys Expert 96 plate reader (Biochrom). EC50 titers were calculated from the posterior median value midway between the plate minimum and maximum. All experiments were performed in triplicates.
Pseudovirus neutralization assay
Pseudovirus neutralization assays were performed as described previously8,9. Briefly, pseudoviruses sufficient to produce approximately 100 000 relative light units were incubated with serial three-fold dilutions for 60 min at 37°C in a black-walled 96-well plate. 10 000 HEK293T-ACE2 cells were then added to each well, and plates were incubated for 48 h. Luminescence was measured using Bright-Glo Luciferase Assay System (Promega, Madison, WI, USA) on a GloMax Navigator Microplate Luminometer (Promega). All fold-changes reported use titers from neutralization assays run in parallel. IC50 titers were interpolated as the mAb concentration at which relative light units (RLUs) were reduced by 50% relative to the mean of 8 control wells in the absence of mAb. All experiments were performed in triplicates.
Cryo-EM model building and structure refinement
The structure of the ancestral spike protein trimer in 1-up conformation PDB: 7A2572 was used as a starting model for model building. The mAb 23 model was predicted using AlphaFold73. The model was mutated and rebuilt manually to reflect the used spike sequence and structure. Structure refinement and manual model building were performed using COOT v0.9.8.9174and PHENIX v1.2075, respectively, in interspersed cycles with secondary structure, Ramachandran, rotamers and bond geometry restraints. Structure figures were generated with UCSF ChimeraX v1.676 and PyMol v.2.5.777. Please see table Supplementary Data 8 for refinement and validation results.