A SARS-CoV-2 variant elicits an antibody response with a shifted immunodominance hierarchy

Many SARS-CoV-2 variants have mutations at key sites targeted by antibodies. However, it is unknown if antibodies elicited by infection with these variants target the same or different regions of the viral spike as antibodies elicited by earlier viral isolates. Here we compare the specificities of polyclonal antibodies produced by humans infected with early 2020 isolates versus the B.1.351 variant of concern (also known as Beta or 20H/501Y.V2), which contains mutations in multiple key spike epitopes. The serum neutralizing activity of antibodies elicited by infection with both early 2020 viruses and B.1.351 is heavily focused on the spike receptor-binding domain (RBD). However, within the RBD, B.1.351-elicited antibodies are more focused on the “class 3” epitope spanning sites 443 to 452, and neutralization by these antibodies is notably less affected by mutations at residue 484. Our results show that SARS-CoV-2 variants can elicit polyclonal antibodies with different immunodominance hierarchies.

None of the individuals had evidence of prior SARS-CoV-2 infection, so we presume these individuals experienced a primary B.1.351 infection.
To enable comparison of B.1.351-elicited antibodies to those elicited by infection with an early 2020 virus, we reexamined a set of convalescent plasma samples collected approximately 30 days post-symptom onset (mean 32, range 15-61 days) from 17 individuals with symptom onset on or prior to March 15, 2020 in Washington state, USA (Table 1) (27,28). At that time, most sequenced viral isolates in Washington state had spike sequences identical to Wuhan-Hu-1, although D614G viruses were also present at a low level (29,30). No other spike mutations were present at appreciable frequencies at that time.  (28). The neutralizing activity of the B.1.351 plasmas was at least as RBD-focused as the early 2020 virus plasmas, with most neutralizing activity of most plasmas from both cohorts a ributable to RBD-binding antibodies (Fig. 2B,C). There was a slight trend for the neutralizing activity of the B.1.351 plasmas to be more RBD-focused than the early 2020 plasmas, but the difference was not statistically significant (Fig. 2C). One caveat is that all neutralization assays were performed in 293T cells overexpressing ACE2, which tend to emphasize the effect of RBD-binding, ACE2-competitive antibodies more than assays performed on cells with lower levels of ACE2 expression (7,35,36).
Complete mapping of mutations in the B.1.351 RBD that reduce binding by polyclonal plasma antibodies elicited by B. 1

.351 infection
To determine how mutations within the RBD affect plasma antibody binding, we used a previously described deep mutational scanning approach. Briefly, this approach involves generating comprehensive mutant libraries of the RBD, displaying the mutant RBDs on the surface of yeast, and using fluorescence-activated cell sorting (FACS) and deep sequencing to quantify how mutations impact antibody binding (28,37).
Previously, we have performed such deep mutational scanning using the RBD from the Wuhan-Hu-1 isolate to map mutations that affect binding by polyclonal antibodies elicited by infection or vaccination that involves a RBD identical to that in Wuhan-Hu-1 (24,28,38).
However, for the current work we wanted to determine the specificity of antibodies elicited by  (25)). We used computational filters based on these measurements as well as a pre-sort of the library for RBDs that bind ACE2 with at least 1% the avidity of the unmutated B.1.351 RBD to filter spurious antibody-escape mutations that were highly deleterious or led to gross unfolding of the RBD. (all cells with the mutation in the plasma-escape bin) (data file S3). The escape fractions measured for independent biological replicate libraries were well-correlated ( fig. S3D), and in the sections below we report the average across the two replicate libraries. We represent the escape maps as logo plots, where the height of each le er is proportional to its escape fraction ( fig. S3A).

B.1.351-elicited antibodies focus on different epitopes than early 2020 convalescent samples
We examined the sites and epitopes to which mutations had the greatest effect on antibody binding. We use the Barnes, et al. (23) antibody epitope classification scheme, in which there are antibody classes 1 through 4 (Fig. 3A). The class 1, 2, and 3 antibodies are often potently neutralizing, while the class 4 antibodies are usually less potently neutralizing in vitro (31-33, 39, 40). Relative to Wuhan-Hu-1, B.1.351 contains mutations in or proximal to the class 1, 2, and 3 epitopes (K417N, E484K, and N501Y, respectively) (Fig. 3A), although the N501Y mutation has li le effect on polyclonal convalescent antibody binding or neutralization for Wuhan-Hu-1-like viruses (7,8,41). viruses in Washington state, USA (28). These 11 samples are a subset of the 17 whose RBD-targeting neutralizing activity is described above (Fig 2B,C). Specifically, binding of the early 2020 plasmas were most affected by mutations to the class 1 and 2 epitopes, with mutations to sites 456, 486, and 484 having some of the largest effects on binding to the RBD ( Fig. 4, fig. S4, data file S4), although mutation to site 456 have li le effect on neutralization in vitro reflecting the common hyperfocusing of neutralizing antibody responses (28,38). While the B.1.351 plasmas were also strongly affected by mutations to the class 2 epitope and site 484, mutations to the class 1 epitope had li le effect. Moreover, while both groups of plasmas are affected by class 3 epitope mutations, the relative importance of class 3 mutations is greater for the B.1.351 plasmas (Fig. 4A,B). to perform neutralization assays with the lowest-potency sample). We also chose four early 2020 samples with substantial RBD-focused neutralizing activity and with antibody-binding escape maps representative of the early 2020 cohort as a whole ( fig. S4) to the reduction caused by removing all RBD-binding antibodies from the plasmas (Fig. 5).
Therefore, the neutralizing activity of early 2020 plasmas is often highly focused on site 484, as has been described previously (2,7,8,28,38,(41)(42)(43)(44) plasmas than for early 2020 plasmas, consistent with the deep mutational scanning escape maps. The G446V mutation to the class 3 epitope had a slightly larger, but still modest, effect on neutralization for the B.1.351 plasmas than for most of the early 2020 plasmas (Fig. 5, fig. S5).
No tested mutation, nor the 417-484-501 triple mutant, reduced neutralization by the B.1.351 plasmas as much as removing all RBD-binding antibodies (Fig. 5), a result in stark contrast to that observed for the early 2020 plasmas.

Discussion:
We found that a SARS-CoV-2 variant induces antibody responses with different immunodominance hierarchies than early SARS-CoV-2 viral isolates. Changes in immunodominance hierarchies over time and asymmetric antigenic drift have also been observed for influenza virus (14)(15)(16)45 (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. target an epitope containing site 484 (24,42,50,51), and are derived from common antibody germline genes (e.g., IGHV3-53/66, IGHV3-30, IGHV1-2 (23, 52,53 Our study has several limitations. The cohorts of individuals infected with early 2020 and B.1.351 viruses are small, and are geographically and temporally distinct. Nevertheless, the two cohorts are relatively well-matched with respect to age, sex, and days-post symptom onset of sample collection ( Table 1) and assays were performed under comparable conditions. Our deep mutational scanning measured binding to yeast-displayed RBD, which may not capture all relevant features of full-length spike in the context of virus. Finally, our neutralization assays used pseudotyped lentiviral particles and ACE2-overexpressing cells, and some recent works suggest that the relative importance of different spike epitopes for neutralization can depend on the viral system and target cell line used (7,35,36,60).
Although the B.1.351 variant has now been displaced by the Delta variant, our results illustrate the need to understand immunity elicited by different SARS-CoV-2 variants. As population immunity due to infection or vaccination increases, preexisting immunity is becoming an increasingly important driver of SARS-CoV-2 evolution (61), as has shown to be the case for seasonal coronaviruses (62,63). Moreover, as individuals begin to accumulate more complex SARS-CoV-2 immune histories due to multiple infections and/or vaccinations, the effects of immune imprinting or original antigenic sin (64,65) may start to interact with the variant-specific immunodominance hierarchies we have described to create increasingly diverse antibody specificities in the human population.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint

Description of cohort and ethics statement
Samples were collected from participants enrolled in a prospective cohort study approved by the Biomedical Research Ethics Commi ee (BREC) at the University of KwaZulu-Natal (reference BREC/00001275/2020). Wri en informed consent was obtained from each participant. The mean age was 54 years (median 53; range 26-78 years). Four were males and 5 were females. All participants had symptomatic SARS-CoV-2 infection and a positive SARS-CoV-2 qPCR from a swab of the upper respiratory tract, and all participants required hospitalization. All 9 participants were HIV-negative. Early-2020 convalescent plasma samples were previously described (27,28) and collected as part Five cases were hospitalized, 2 were asymptomatic, and the remainder were symptomatic non-hospitalized. The neutralization activity of plasma samples before and after depletion of RBD-binding antibodies in Fig. 2 and RBD binding-escape maps in fig. S4 were previously reported (28), but neutralization assays for all 30-days post-symptom onset plasmas in Fig. 5, fig. S5 were newly performed in this study. The neutralization assays on the 100-day early 2020 samples in fig. S5 were previously reported (38). This work was approved by the University of Washington Institutional Review Board.

Plasma separation from whole blood
Plasma was separated from EDTA-anticoagulated blood by centrifugation at 500 rcf for 10 min and stored at −80 °C. Aliquots of plasma samples were heat-inactivated at 56 °C for 30 min and clarified by centrifugation at 10,000 rcf for 5 min, after which the clear middle layer was used for experiments.
Inactivated plasma was stored in single-use aliquots to prevent freeze-thaw cycles. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made This sequence has 5' and 3' flanking sequences that are unmutated in the variant libraries (lower case).

Construction of B.1.351 RBD yeast-displayed DMS library
The uppercase portion is the RBD coding sequence, amino acids N331-T531 (Wuhan-Hu-1 spike numbering). The libraries were designed to contain all 19 amino acids at each site in the RBD, without stop codons, with no more than one amino-acid mutation per variant. The variant gene fragments were PCR-amplified with these primers: 5'-tctgcaggctagtggtggag-3' and 5'-agatcggaagagcgtcgtgtagggaaagagtgtagatctcggtggtcgccgtatca aa ctc agga cga caca c-3'.
(primer-binding regions underlined in the sequence above). A second round of PCR was performed using the same forward primer (5'-tctgcaggctagtggtggag-3') and the reverse primer 5'-ccagtgaa gtaatacgactcactatagggcgaa ggagctcgcggccgcnnnnnnnnnnnnnnnnagatcggaagagcgtcgtgtag-3' to append the Nx16 barcodes and add the overlapping sequences to clone into the recipient vector backbone as described in (25,66).
Failed positions in the Twist-delivered library (sites 362, 501, and 524 in Wuhan-Hu-1 numbering) were mutagenized in-house using a PCR-based method with NNS degenerate primers and cloned into the unmutated wildtype backbone plasmid using NEB HiFi assembly, exactly as described in (66). These were then PCR-amplified using the same 5'-tctgcaggctagtggtggag-3' and 5'-ccagtgaa gtaatacgactcactatagggcgaa ggagctcgcggccgcnnnnnnnnnnnnnnnnagatcggaagagcgtcgtgtag-3' primers to pool with the barcoded Twist library gene fragments.
The barcoded variant gene fragments were cloned in bulk into the NotI/SacI-digested unmutated wildtype plasmid, as described in (25,66). The Genbank plasmid map for the fully assembled, barcoded Colonies from bo lenecked transformation plates were scraped and plasmid purified. Plasmid libraries . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint (10 µg plasmid per replicate library) were transformed into the AWY101 yeast strain (67) according to the protocol of Gie and Schiestl (68).

PacBio sequencing to link variant mutations and barcodes
As described by Starr et al. (25), PacBio sequencing was used to generate long sequence reads spanning the Nx16 barcode and RBD coding sequence. PacBio sequencing amplicons were prepared from library plasmid pools via NotI digestion, gel purification, and Ampure XP bead clean-up. Sample-specific barcodes and SMRTbells were ligated using the HiFi Express v2 kit. The multiplexed libraries were

Determining the effects of mutations on RBD expression and ACE2 binding to filter the library for functional variants
The effects of each mutation on RBD expression on the surface of yeast and on ACE2 binding were measured essentially as described previously for the Wuhan-Hu-1 RBD (25). Specifically, each biological replicate library was grown overnight at 30°C in 45mL SD-CAA media (6.7g/L Yeast Nitrogen Base, . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint Cells were processed on a BD FACSAria II and sorted into four bins from low to high RBD expression (measured by myc-FITC staining) or ACE2 binding (measured by streptavidin-PE fluorescence). The RBD expression sort bins were set such that bin 1 would capture 99% of unstained cells, and the remaining 3 bins divide the remainder of each mutant RBD library into equal tertiles. For The effects of each mutation on RBD expression and ACE2 binding were determined as described in (25). RBD mutant expression and ACE2 binding scores were calculated according to the equations in (25). For ACE2 binding, a score of -1.0 corresponds to a 10-fold loss in affinity (K d ) compared to the wildtype RBD. For RBD expression, a score of -1.0 corresponds to a 10-fold reduction in mean RBD-myc-FITC fluorescence intensity. These measurements were used to computationally filter library variants that were highly deleterious for RBD expression or ACE2 binding and would likely represent spurious antibody-escape mutations (see below for details). The ACE2 binding and RBD expression scores for the single amino-acid mutations in the B.1.351 RBD are available at h ps://github.com/jbloomlab/SARS-CoV-2-RBD_B.1.351/blob/main/data/final_variant_scores.csv.
As previously described, prior to performing the antibody-escape experiments, the yeast libraries were pre-sorted for RBD expression and binding to dimeric ACE2 (ACROBiosystems AC2-H82E6) to eliminate RBD variants that are completely misfolded or non-functional, such as those lacking modest ACE2 binding affinity (37). Specifically, unmutated B.

Depleting plasma of nonspecific yeast-binding antibodies prior to antibody-escape experiments
Prior to the yeast-display deep mutational scanning, plasma samples were twice-depleted of nonspecific yeast-binding antibodies. AWY101 yeast containing a negative control (containing an empty vector pETcon plasmid) were grown overnight at 30°C in galactose-containing media. Then, up to 50 microliters of plasma samples were incubated, rotating, with 40 OD units of the yeast for 2 hours at room temperature in a total volume of 1mL. The yeast cells were pelleted by centrifugation, and the supernatant was transferred to an additional 40 OD units of yeast cells, and the incubation was repeated overnight at 4°C. Before beginning the plasma-escape mapping experiments, the negative control yeast were pelleted by centrifugation and the supernatant (containing serum antibodies but not negative control yeast or yeast-binding antibodies) was used in plasma-escape mapping.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

.351-convalescent individuals
Plasma mapping experiments were performed in biological duplicate using the independent mutant RBD libraries, similarly to as previously described for monoclonal antibodies (37) and polyclonal plasma samples (28). Mutant yeast libraries induced to express RBD were washed and incubated with plasma at a range of dilutions for 1 hour at room temperature with gentle agitation. For each plasma, we chose a sub-saturating dilution such that the amount of fluorescent signal due to plasma antibody binding to RBD was approximately equal across samples. The exact dilution used for each plasma is given in fig. S3. HiSeq 2500 with 50 bp single-end reads. To minimize noise from inadequate sequencing coverage, we ensured that each antibody-escape sample had at least 2.5x as many post-filtering sequencing counts as FACS-selected cells, and reference populations had at least 2.5e7 post-filtering sequencing counts.

Analysis of deep sequencing data to compute each mutation's escape fraction
Escape fractions were computed as described in (37), with minor modifications as noted below. We used the dms_variants package (h ps://jbloomlab.github.io/dms_variants/, version 0.8.10) to process Illumina sequences into counts of each barcoded RBD variant in each pre-selection and antibody-escape population. For each plasma selection, we computed the escape fraction for each barcoded variant using the deep sequencing counts for each variant in the original and plasma-escape populations and the total fraction of the library that escaped antibody binding via the formula provided in (37). These escape . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint fractions represent the estimated fraction of cells expressing that specific variant that falls in the escape bin, such that a value of 0 means the variant is always bound by plasma and a value of 1 means that it always escapes plasma binding.
We then applied a computational filter to remove variants with >1 amino-acid mutation, low sequencing counts, or highly deleterious mutations that might cause antibody escape simply by leading to poor expression of properly folded RBD on the yeast cell surface (25,37). Specifically, we removed variants that had ACE2 binding scores < −3.0 or expression scores < −1.0, after calculating mutation-level deep mutational scanning scores for this library as in (25). An ACE2 binding score threshold of -3.0 retained 99.4% and an RBD expression score threshold of -1.0 retained 93.8% of all RBD mutations observed >=50x in GISAID as of Aug. 1, 2021 (fig. S2C).
We also removed all mutations where the wildtype residue was a cysteine. There were 2,014 out of the possible 3,653 mutations to non-disulfide bond residues in the RBD that passed these computational filters.
The reported antibody-escape scores throughout the paper are the average across the libraries; these scores are also in data file S3. Correlations in final single-mutant escape scores are shown in fig.

S3D.
For plo ing and analyses that required identifying RBD sites of strong escape, we considered a site to mediate strong escape if the total escape (sum of mutation-level escape fractions) for that site exceeded the median across sites by >5-fold, and was at least 5% of the maximum for any site. For experiments involving D614G spike, we used spike-pseudotyped lentiviral particles that were generated essentially as described in (70), using a codon-optimized SARS-CoV-2 spike from Wuhan-Hu-1 strain that contains a 21-amino-acid deletion at the end of the cytoplasmic tail (27) and the D614G . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint mutation that is now predominant in human SARS-CoV-2 (30). The plasmid encoding this spike, HDM_Spikedelta21_D614G, is available from Addgene (#158762) and BEI Resources (NR-53765), and the full sequence is at (h ps://www.addgene.org/158762). Point mutations were introduced into the RBD of this plasmid via site-directed mutagenesis.

Titering of pseudotyped lentiviral particles
Titers of spike-pseudotyped lentiviral particles were determined as described in (70)   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint

Depletion of RBD-binding antibodies from polyclonal sera
Two rounds of sequential depletion of RBD-binding antibodies were performed for vaccine-elicited sera.
Magnetic beads conjugated to the SARS-CoV-2 B.1.351 RBD (ACROBiosystems, MBS-K032) were prepared according to the manufacturer's protocol. Beads were resuspended in ultrapure water at 1 mg beads/mL and a magnet was used to wash the beads 3 times in phosphate-buffered saline (PBS) with 0.05% bovine serum albumin (BSA). Beads were then resuspended in PBS with 0.05% BSA at 1 mg beads per mL. Beads (manufacturer-reported binding capacity of 10-40 µg/mL anti-RBD antibodies) were incubated with human plasma at a 2:1 ratio beads:plasma, rotating overnight at 4°C or for 2 hours at room temperature. A magnet (MagnaRack Magnetic Separation Rack, Thermo Fisher Scientific, CS15000) was used to separate antibodies that bind RBD from the supernatant, and the supernatant (the post-RBD antibody depletion sample) was removed. A mock depletion (pre-depletion sample) was performed by adding an equivalent volume of PBS + 0.05% BSA and rotating overnight at 4°C or for 2 hours at room temperature. Up to three rounds of depletions were performed to ensure full depletion of RBD-binding antibodies. For the neutralization assays on these plasmas depleted of RBD-binding antibodies, the reported plasma dilution is corrected for the dilution incurred by the depletion process. Note that these assays were performed in 293T cells over-expressing human ACE2, which may underestimate contributions of non-RBD-binding antibodies to viral neutralization (7,35,60).

Measurement of plasma binding to RBD or spike by enzyme-linked immunosorbent assay (ELISA)
The IgG ELISAs for spike protein and RBD were conducted as previously described (71). Briefly, ELISA plates were coated with recombinant B.1.351 spike (purified and prepared as described in (71)) and RBD (ACROBiosystems, SPD-C52Hp) antigens described in at 2 µg/mL. Five 3-fold serial dilutions of sera beginning at 1:500 were performed in PBS with 0.1% Tween with 1% Carnation nonfat dry milk. Dilution series of the synthetic sera comprised of the anti-RBD antibody REGN10987 (72), which binds to both Wuhan-1-like RBD and B.1.351 RBD, and pooled pre-pandemic human serum from 2017-2018 (Gemini Biosciences; nos. 100-110, lot H86W03J; pooled from 75 donors) were performed such that the anti-spike antibody was present at a highest concentration of 0.25 µg/mL. REGN10987 was recombinantly produced by Genscript. The REGN10987 is the same as that used in (73). Pre-pandemic serum alone, without anti-RBD antibody depletion, was used as a negative control, averaged over 2 replicates. Secondary labeling was performed with goat anti-human IgG-Fc horseradish peroxidase (HRP) (1:3000, Bethyl Labs, A80-104P). Antibody binding was detected with TMB/E HRP substrate (Millipore Sigma, ES001) and 1 N HCl was used to stop the reaction. OD 450 was read on a Tecan infinite M1000Pro plate reader.

Data visualization
The static logo plot visualizations of the escape maps in the paper figures were created using the dmslogo package (h ps://jbloomlab.github.io/dmslogo, version 0.6.2) and in all cases the height of each le er indicates the escape fraction for that amino-acid mutation calculated as described above. For each sample, the y-axis is scaled to be the greatest of (a) the maximum site-wise escape metric observed for that sample, (b) 20x the median site-wise escape fraction observed across all sites for that plasma, or (c) an absolute value of 1.0 (to appropriately scale samples that are not noisy but for which no mutation has a strong effect on antibody binding). Sites K417, L452, S477, T478, E484, and N501 have been added to logo . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. plots due to their frequencies among circulating viruses. The code that generates these logo plot visualizations is available at h ps://github.com/jbloomlab/SARS-CoV-2-RBD_B.1.351/blob/main/results/summary/escape_profiles.md.
For the static structural visualizations in the paper figures, the RBD surface (PDB 6M0J) was colored by the site-wise escape metric at each site, with white indicating no escape and red scaled to be the same maximum used to scale the y-axis in the logo plot escape maps, determined as described above.
We created interactive structure-based visualizations of the escape maps using dms-view (74) that are available at h ps://jbloomlab.github.io/SARS-CoV-2-RBD_B.1.351/. The logo plots in these escape maps can be colored according to the deep mutational scanning measurements of how mutations affect ACE2 binding or RBD expression as described above.

Statistical Analysis
The percent of neutralizing activity of early-2020 and B.1.351-convalescent plasmas due to RBD-binding antibodies is plo ed with the plotnine python package, version 0.8.0  plasmas were performed with Wuhan-Hu-1 RBD proteins and D614G spike-pseudotyped lentiviruses. The data for the early 2020 viruses are reprinted from (28). Neutralization titers are in data file S1 and at . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  351 plasmas are in fig. S1, and the full curves for the early 2020 plasmas are shown in the supplement of (28).
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ;

B.1.351 versus early 2020 viruses. (A)
The total escape at each site is shown as a light gray line for each plasma in the early 2020 or B.1.351 cohorts. The thicker black line indicates the average for each cohort. Key antibody epitopes are highlighted, colored as in Fig. 2A. (B) The total escape at each site averaged across each cohort is mapped to the Wuhan-Hu-1 RBD surface (PDB 6M0J (76)), with sites colored from white to red, with white indicating no escape, and red being the site with the most escape. Interactive versions of logo plots and structural visualizations are at h ps://jbloomlab.github.io/SARS-CoV-2-RBD_B.1.351/. The early 2020 escape-mapping data in this figure were originally published in (28) and are reanalyzed here.
The full escape maps for the early 2020 samples are shown in Fig. S4 and the full escape maps for the B.1.351 samples are shown in Fig. 3.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint  Fig. 2; the underlying measurements for the early 2020 plasmas in Fig. 2 are shown in (28).
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Correlations between biological independent replicate library measurements of the effects of single mutations on ACE2 binding and RBD expression, measured as described in (25). See A key difference is that for the previously published Wuhan-Hu-1 experiments, dimeric rather than monomeric ACE2 was used (25).
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted October 13, 2021. ; https://doi.org/10.1101/2021.10.12.464114 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made