Naturally enhanced neutralizing breadth to SARS-CoV-2 after one year

Over one year after its inception, the coronavirus disease-2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) remains difficult to control despite the availability of several excellent vaccines. Progress in controlling the pandemic is slowed by the emergence of variants that appear to be more transmissible and more resistant to antibodies 1,2 . Here we report on a cohort of 63 COVID-19-convalescent individuals assessed at 1.3, 6.2 and 12 months after infection, 41% of whom also received mRNA vaccines 3,4 . In the absence of vaccination antibody reactivity to the receptor binding domain (RBD) of SARS-CoV-2, neutralizing activity and the number of RBD-specific memory B cells remain relatively stable from 6 to 12 months. Vaccination increases all components of the humoral response, and as expected, results in serum neutralizing activities against variants of concern that are comparable to or greater than neutralizing activity against the original Wuhan Hu-1 achieved by vaccination of naïve individuals 2,5-8 . The mechanism underlying these broad-based responses involves ongoing antibody somatic mutation, memory B cell clonal turnover, and development of monoclonal antibodies that are exceptionally resistant to SARS-CoV-2 RBD mutations, including those found in variants of concern 4,9 . In addition, B cell clones expressing broad and potent antibodies are selectively retained in the repertoire over time and expand dramatically after vaccination. The data suggest that immunity in convalescent individuals will be very long lasting and that convalescent individuals who receive available mRNA vaccines will produce antibodies and memory B cells that should be protective against circulating SARS-CoV-2 variants.

To determine the neutralizing activity against circulating variants of concern/interest, we 92 performed neutralization assays on HIV-1 virus pseudotyped with the S protein of the following 93 SARS-CoV-2 variants of concern/interest: B.1.1.7, B.1.351, B.1.526 and P.1 1,14,15 . Twelve-94 months after infection neutralizing activity against the variants was generally lower than against 95 wild-type SARS-CoV-2 virus in the same assay with the greatest loss of activity against B.1.351 96

Memory B cells 104
The memory B cell compartment serves as an immune reservoir that contains a diverse collection 105 of antibodies 16,17 . Although antibodies to the N-terminal domain and other parts of S can also be 106 neutralizing, we limited our analysis to memory B cells that produce anti-RBD antibodies 107 because they are the most numerous and potent 18,19 . To enumerate RBD-specific memory B 108 cells, we performed flow cytometry using a biotin-labeled RBD 3 (Fig. 2a and Extended Data Fig.  109 3a and b). In the absence of vaccination, the number of RBD specific memory B cells present at 110 mutant RBDs were also enumerated by flow cytometry (Extended Data Fig. 3c). The number of 115 variant RBD cross-reactive B cells was directly proportional to but 1.6 to 3.2-fold lower than 116 wild-type RBD binding B cells (Fig. 2a). 117 118 The memory B cell compartment accumulates mutations and undergoes clonal evolution over the 119 initial 6 months after infection 4,9,21,22 . To determine whether the memory compartment continues 120 to evolve between 6 and 12 months, we obtained 1105 paired antibody heavy and light chain 121 sequences from 10 individuals that were also assayed at the earlier time points, 6 of which were 122 To determine whether the antibodies obtained from memory B cells 12 months after infection 155 bind to RBD we performed ELISAs (Fig.3a). 174 antibodies were tested by ELISA including: 1. 156 53 that were randomly selected from those that appeared only once and only after 1 year; 2. 91 expressing anti-RBD antibodies was efficient (Supplementary Tables 4 and 5). The geometric 161 mean ELISA half-maximal concentration (EC50) of the antibodies obtained after 12 months was 162 2.6 ng/ml, which was significantly lower than at 6 months irrespective of vaccination and 163 suggestive of an increase in affinity (Fig. 3a, Extended data Fig. 7 a and b

and Supplementary 164
Tables 4 and 5). Consistent with this observation there was an overall increase in the apparent 165 avidity of plasma antibodies between 1.3 and 12 months 3,4 (p<0.0001, Extended data Fig. 7c).  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. To determine whether the loss of non-neutralizing antibodies over time was due to preferential 184 loss of antibodies targeting specific epitopes, we performed BLI experiments in which a 185 preformed antibody-RBD complex was exposed to a second monoclonal targeting one of 3 186 classes of structurally defined epitopes 3,23 (see schematic in Fig. 4a). We assayed 60 randomly 187 selected antibodies with comparable neutralizing activity from the 1.3-and 12-month time 188 points. The 60 antibodies were evenly distributed between the 2 time points and between 189 neutralizers and non-neutralizers (Fig. 4). Antibody affinities for RBD were similar among 190 neutralizers and non-neutralizers obtained at the same time point (Fig. 4b, Extended Data Fig. 8). 191 When the two sets of un-related antibodies obtained from 1.3 and 12 months were compared, 192 they showed significantly increased affinity over time irrespective of their neutralizing activity 193 antibodies tested and therefore must bind to epitopes that do not overlap with the epitopes of 196 these classes of antibodies (Fig. 4c, and Extended Data Fig. 9). In contrast, all but 2 of the 30 197 neutralizers blocked class 1, or 2 antibodies whose target epitopes are structural components of 198 the RBD that interact with its cellular receptor, the angiotensin-converting enzyme 2 23,24 (ACE2) 199 ( Fig. 4c and Extended Data Fig. 9). In addition, whereas 9 of the 15 neutralizing antibodies 200 obtained after 1.3 months blocked both class 1 and 2 antibodies, only 1 of the 15 obtained after 201 12 months did so. In contrast to the earlier time point, 13 of 15 neutralizing antibodies obtained 202 after 12 months only interfered with C121, a class 2 antibody 3,23 ( Fig. 4c and Extended Data Fig.  203 9). We conclude that neutralizing antibodies are retained and non-neutralizing antibodies 204 targeting RBD surfaces that do not interact with ACE2 are removed from the repertoire over 205

time. 206
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. To determine whether there was an increase in neutralization breadth over time, the neutralizing 208 activity of the 60 antibodies was assayed against a panel of RBD mutants covering residues 209 associated with circulating variants of concern: R346S, K417N, N440K, A475V, E484K and 210 N501Y ( Fig. 4d and  The increase in breadth and overall potency of memory B cell antibodies could be due to shifts in 215 the repertoire, clonal evolution, or both. To determine whether changes in specific clones are 216 associated with increases in affinity and breadth, we measured the relative affinity and 217 neutralizing breadth of matched pairs of antibodies expressed by expanded clones of B cells that 218 were maintained in the repertoire over the entire observation period 3,4 . SARS-CoV-2 neutralizing 219 activity of the antibodies present at 1.3 or 12 months was not significantly correlated with 220 affinity at either time point when each time point is considered independently (Fig. 4e). 221 However, there was a significant increase in overall affinity over time including in the 4 pairs of 222 antibodies with no measurable neutralizing activity ( Fig. 4f and Supplementary Table 7).     was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.07.443175 doi: bioRxiv preprint 341 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  test. f, BLI affinity measurements for same n=36 paired 1.3-and 12-month antibodies as in e. 360 Statistical significance was determined using two-tailed Wilcoxon test. g, IC50 values for n=30 361 paired neutralizing antibodies isolated at indicated timepoints against indicated mutant SARS-362 CoV-2 pseudoviruses. Antibodies are divided into groups i-iii, based on neutralizing activity: (i) 363 potent clonal pairs that do not improve over time, (ii) clonal pairs that show increased activity over 364 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  The E484K and K417N/E484K/N501Y (KEN) substitution, as well as the deletions/substitutions 463 corresponding to variants of concern were incorporated into a spike protein that also includes the 464 R683G substitution, which disrupts the furin cleaveage site and increases particle infectivity. 465 Neutralizing activity against mutant pseudoviruses were compared to a wildtype SARS-CoV-2 466 spike sequence (NC_045512), carrying R683G where appropriate.   was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Antibody sequencing, cloning and expression 532
Antibodies were identified and sequenced as described previously 3 . In brief, RNA from single 533 cells was reverse-transcribed (SuperScript III Reverse Transcriptase, Invitrogen, 18080-044) and 534 the cDNA stored at −20 °C or used for subsequent amplification of the variable IGH, IGL and 535 IGK genes by nested PCR and Sanger sequencing. Sequence analysis was performed using 536 MacVector. Amplicons from the first PCR reaction were used as templates for sequence-and 537 ligation-independent cloning into antibody expression vectors. Recombinant monoclonal 538 antibodies were produced and purified as previously described 3 . 539 540

Biolayer interferometry 541
Biolayer interferometry assays were performed as previously described 3 . Briefly, we used the The plasma SARS-CoV-2 antibody avidity assay were performed as previously described 47 . 562 563

Computational analyses of antibody sequences 564
Antibody sequences were trimmed based on quality and annotated using Igblastn v.1.14. with 565 IMGT domain delineation system. Annotation was performed systematically using Change-O 566 toolkit v.0.4.540 48 . Heavy and light chains derived from the same cell were paired, and 567 clonotypes were assigned based on their V and J genes using in-house R and Perl scripts (Fig.  568 2d). All scripts and the data used to process antibody sequences are publicly available on GitHub 569 (https://github.com/stratust/igpipeline). 570

571
The frequency distributions of human V genes in anti-SARS-CoV-2 antibodies from this study 572 was compared to 131,284,220 IgH and IgL sequences generated by 49 and downloaded from 573 cAb-Rep 50 , a database of human shared BCR clonotypes available at https://cab-574 rep.c2b2.columbia.edu/. Based on the 91 distinct V genes that make up the 6902 analyzed 575 sequences from Ig repertoire of the 10 participants present in this study, we selected the IgH and 576 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.07.443175 doi: bioRxiv preprint IgL sequences from the database that are partially coded by the same V genes and counted them 577 according to the constant region. The frequencies shown in (Extended data Fig. 4) are relative to 578 the source and isotype analyzed. We used the two-sided binomial test to check whether the 579 number of sequences belonging to a specific IgHV or IgLV gene in the repertoire is different was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.07.443175 doi: bioRxiv preprint symptom-free (-) 12 months post-infection. Statistical significance was determined using the 661 two-tailed Mann-Whitney test in a, b and d, and using the two-sided Fisher's exact test in c. 662 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  unequal variance were used to compare the frequency distributions., significant differences are 715 denoted with stars (* p < 0.05, ** p < 0.01, *** p < 0.001, **** = p < 0.0001). 716 717 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.  of RBD-binding memory B cells from ten convalescent individuals, a, phylogenetic tree graph 737 shows clones from convalescent non-vaccinees, b, same as a, but from convalescent vaccinees. 738 Numbers refer to mutations compared to the preceding vertical node. Colors indicate timepoint; 739 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.07.443175 doi: bioRxiv preprint grey, orange and red represent 1.3, 6 and 12 months respectively, black dots indicate inferred 740 nodes, and size is proportional to sequence copy number; GL = germline sequence. 741 742 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 2, 2021.  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.