Human B Cell Clonal Expansion and Convergent Antibody Responses to SARS-CoV-2

Materials Availability

Unique reagents generated in this study will be provided upon reasonable request.

Healthy Human Control Samples

The dataset containing healthy adult control immunoglobulin heavy chain (IGH) repertoires has been described previously (Nielsen et al., 2019). In summary, healthy adults with no signs or symptoms of acute illness or disease were recruited as volunteer blood donors at the Stanford Blood Center. Pathogen diagnostics were performed for cytomegalovirus, human immunodeficiency virus, hepatitis C and B virus, West Nile virus, human T cell leukemia virus, TPPA (Syphilis), and Trypanosoma cruzi. Volunteer age range was 17-87 with median and mean of 52 and 49, respectively.

IGH Library Prep and Sequencing

The AllPrep DNA/RNA kit (QIAGEN) was used to extract genomic DNA (gDNA) and total RNA from PBMCs. For each blood sample, six independent gDNA library PCRs were set up using up to100 ng template/library. Multiplexed primers to IGHJ and the IGHV FR1 or FR2 framework regions (3 FR1 and 3 FR2 libraries), per the BIOMED-2 design were used (van Dongen et al., 2003) with additional sequence representing the first part of the Illumina linkers (Table S3). In addition, for each sample, total RNA was reverse-transcribed to cDNA using Superscript III RT (Invitrogen) with random hexamer primers (Promega). Total RNA yield varied between patients and between 6 ng-100 ng was used for each of the isotype PCRs using IGHV FR1 primers based on the BIOMED-2 design (van Dongen et al., 2003) and isotype specific primers located in the first exon of the constant region for each isotype category (IgM, IgD, IgE, IgA, IgG). Primers contain additional sequence representing the first part of the Illumina linkers (Table S3). The different isotypes were amplified in separate reaction tubes. Eight-nucleotide barcode sequences were included in the primers to indicate sample (isotype and gDNA libraries) and replicate identity (gDNA libraries). Four randomized bases were included upstream of the barcodes on the IGHJ primer (gDNA libraries) and constant region primer (isotype libraries) for Illumina clustering (Table S3). PCR was carried out with AmpliTaq Gold (Applied Biosystems) following the manufacturer’s instructions, and used a program of: 95°C 7 min; 35 cycles of 94°C 30 s, 58°C 45 s, 72°C 60 s; and final extension at 72°C for 10 min. A second round of PCR using QIAGEN’s Multiplex PCR Kit was performed to complete the Illumina sequencing adapters at the 5-prime and 3-prime ends of amplicons; cycling conditions were: 95°C 15 min; 12 cycles of 95°C 30 s, 60°C 45 s, 72°C 60 s; and final extension at 72°C for 10 min. Products were subsequently pooled, gel purified (QIAGEN), and quantified with the Qubit fluorometer (Invitrogen). Samples were sequenced on the Illumina MiSeq using 600 cycle kits for isotype libraries (PE300) and 500 cycle kits for gDNA libraries (PE250).

IGH Sequence Analysis

Paired-end reads were merged using Fast Length Adjustment of Short reads software (Magoč and Salzberg, 2011), demultiplexed (100% barcode match), and primer trimmed. The IGHV, IGHD, and IGHJ gene segments and IGHV-IGHD (N1), and IGHD-IGHJ (N2) junctions were identified using the IgBLAST alignment program (Ye et al., 2013). Quality filtering of sequences included keeping only productive reads with a CDR-H3 region, and minimum IGHV-gene alignment score of 200. Sample cDNA or gDNA libraries with poor read coverage were excluded from further analysis (Table 1). For cDNA-templated IGH reads, isotypes and subclasses were called by exact matching to the constant region gene sequence upstream of the primer. Clonal identities within each subject were inferred using single-linkage clustering and the following definition: same IGHV and IGHJ usage (disregarding allele call), equal CDR-H3 length, and minimum 90% CDR-H3 nucleotide identity. A total of 1,259,882 clones (per sample, mean number of clones: 33,154; median number of clones: 18,503) were identified. A total of 24,888,790 IGH sequences amplified from cDNA were analyzed for the COVID-19 subjects (mean: 754,205 per sample; median: 650,812) and 68,831,446 sequences from healthy adult controls (mean: 603,785 per individual; median: 637,269). Each COVID-19 patient had on average 372,304 in-frame gDNA sequences and each adult control had an average of 8,402 in-frame gDNA sequences.

For each clone, the median somatic mutation frequency of reads was calculated. Mean mutation frequencies for all clonal lineages from a sample for each isotype were calculated from the median mutation frequency within each clone, and so represent the mean of the median values. Clones with < 1% mutation were defined as unmutated and clones with greater or equal to1% mutation were defined as being mutated. Subclass fractions were determined for each subject by dividing the number of clones for a given subclass by the total number of clones for that isotype category. Expanded clones within each sample were defined as clones that were present in two or more of the gDNA replicate libraries. Clonal expansion in the isotype data was inferred from the gDNA data. Analyses were conducted in R (Team, 2017) using base packages for statistical analysis and the ggplot2 package for graphics (Wickham, 2016).

To identify convergent rearranged IGH among patients with SARS-CoV-2 infection, heavy-chain sequences annotated with the same IGHV and IGHJ segment (not considering alleles) and the same CDR-H3 length were clustered based on 85% CDR-H3 amino acid sequence similarity using CD-HIT (Fu et al., 2012). To exclude IGH that are generally shared between humans and to enrich the SARS-CoV-2-specific IGH that are likely shared among the patients, clusters were selected as informative if (1) they contained at least five IGH sequences from each COVID-19 patient and were present in at least two subjects; (2) no IGH sequences from HHC samples (collected prior to the 2019 SARS-CoV-2 outbreak) were identified in the same convergent cluster. The same selection criteria were used to determine the convergent clusters between the COVID-19 samples and previously reported IGH sequences specific to SARS-CoV and SARS-CoV-2. Convergent IGH sequences between the deeply sequenced COVID-19 patients and 10x Genomics single B cell immune profiling on two COVID-19 patients were selected for mAb expression. To assess whether the number of convergent clones identified among COVID-19 patients was significantly greater than the number expected by chance in HHC repertoires, we randomly selected with replacement sets of 13 HHC individuals from the 114 available, determining the number of convergent clones detected in two, three, four or larger numbers of participants, and repeated this process 100 times to generate the distribution of convergent clone counts shown in Figure 3B.

Single-Cell Immunoglobulin Analysis

Single-cell immunoglobulin (Ig) libraries were prepared using the 10x Single Cell Immune Profiling Solution Kit (v1.1 Chemistry), according to the manufacturer’s instructions. Briefly, cells were washed once with PBS + 0.04% BSA. Following reverse transcription and cell barcoding in droplets, emulsions were broken and cDNA purified using Dynabeads MyOne SILANE (Thermo Fisher) followed by PCR amplification (98°C for 45 s; 15 cycles of 98°C for 20 s, 67°C for 30 s, 72°C for 1 min; 72°C for 1 min). For targeted Ig library construction, 2 μL of amplified cDNA was used for target enrichment by PCR (Human B cell primer sets 1 and 2: 98°C for 45 s; 6 and 8 cycles of 98°C for 20 s, 67°C for 30 s, 72°C for 1 min; 72°C for 1 min). Following Ig enrichment, up to 50 ng of enriched PCR product was fragmented and end-repaired, size selected with SPRIselect beads (Beckman Coulter), PCR amplified with sample indexing primers (98°C for 45 s; 9 cycles of 98°C for 20 s, 54°C for 30 s, 72°C for 20 s; 72°C for 1 min), and size selected with SPRIselect beads. Targeted single-cell Ig libraries were sequenced on an Illumina MiSeq to a minimum sequencing depth of 5,000 reads/cell using the read lengths 26bp Read1, 8bp i7 Index, 91bp Read2 and reads were aligned to the prebuilt GRCh38/Ensembl/10x reference dataset and consensus Ig annotation was performed using Cell Ranger (10x Genomics, version 3.1.0).

IGH Sequence Analysis for mAb Expression

IGH sequences from single cells with paired productive heavy and light chains were searched against COVID-19 patient bulk IGH repertoires to identify convergent sequences according to the following criteria: utilization of the same IGHV and IGHJ genes; same CDR-H3 lengths; and CDR-H3 amino acid sequences that were within a Hamming distance cutoff of 15% of the length of the CDR-H3. Two native heavy and light chain pairs, designated mAb2A and mAb4A, which were found in convergent clusters characterized by low- to mid-SHM frequencies and included at least one class-switched member, were selected for cloning and expression. Additionally, four IGHs were selected from clones that clustered with SARS-CoV-2 RBD-specific antibody BD-494. Full-length light chain sequence for the BD-494 was not available so to produce mAbs for CoV.1-CoV.4, the four IGH were paired with a germline-reverted light chain that utilizes IGKV1-9, IGKJ3, and the published CDR-L3 sequence for BF-494, ‘QQLNSYPFT’.

mAb Cloning

Heavy and light chain sequences were synthesized by Integrated DNA Technologies as gBlocks encoding full-length heavy and light chain V(D)J regions. gBlocks were resuspended at 50 ng/μL and amplified with AmpliTaq Gold (Applied Biosystems) following the manufacturer’s instructions, using a program of: 95°C 7 min; 30 cycles of 94°C for 30 s, 55°C for 45 s, 72°C for 60 s; and final extension at 72°C for 10 min. Products were gel purified (QIAGEN) and cloned as in-frame fusions to human IgG1, IgK or IgL constant regions into the pTT5 vector (Durocher et al., 2002) (National Research Council (NRC), Canada) using Gibson Assembly Master Mix (New England Biolabs) for 45 min at 50°C. Assembled constructs were verified by Sanger sequencing.

mAb Expression and Purification

Constructs were transiently transfected in HEK293-EBNA1-6E cells (NRC) or Expi293F cells (Thermo Fisher) at a density of 1.2-1.6 million cells/mL or 3 million cells/mL, respectively, using 25 kDa linear polyethylenimine (PEI) at a 3:1 PEI:DNA ratio in OptiMEM reduced serum medium (GIBCO), with a heavy chain: light chain ratio of 1:1. HEK293-EBNA1-6E cells were maintained in Freestyle 293 Expression Medium (GIBCO) and were supplemented with 0.5% tryptone 24-36 h after transfection. Expi293F cells were maintained in Expi293 Expression Medium (Thermo Fisher). Cell supernatants were harvested after 96 h and filtered through 0.45-μM filters (Millipore). Antibodies were purified via HiTrap Protein A HP columns (GE Healthcare) run at a flow rate of 0.5-1 mL/min on Äkta Start protein purification system (GE Healthcare). Antibodies were eluted using 0.1M glycine pH 2, dialyzed with 3 changes of PBS pH 7.4 using Slide-A-Lyzer-G2 10K dialysis cassettes (Thermo Fisher), and concentrated using 30,000 kDa molecular weight cutoff polyethersulfone membrane spin columns (Pierce). Final concentrations of purified antibodies were quantified with Nanodrop 2000 (Thermo Fisher).

Author Contributions

C.A.B. and S.D.B. conceived the project. C.A.B., K.C.N., A.R., and A.J.R. recruited, enrolled, and consented patients and contributed to clinical sampling and processing and support of clinical sample collection. S.C.A.N., R.A.H., K.R., G.H.J., B.S., J.-Y.L., A.E.P., M.H., J.N., A.R.O.-C., K.E.Y., B.D., and B.A.P. performed the experiments. M.H., H.Y.C., A.T.S., K.C.N., T.S.J., P.S.K., and T.T.W. contributed to serological or 10× assays and analysis. S.C.A.N., F.Y., K.J.L.J., R.A.H., K.R., G.H.J., S.D.B., and B.A.P. contributed to data analysis. S.C.A.N., F.Y., K.J.L.J., R.A.H., K.R., K.E.Y., C.A.B., and S.D.B. wrote the manuscript. All authors edited the manuscript.