key: cord-0326977-brtflzmw
authors: Newman, J.; Thakur, N.; Peacock, T. P.; Bialy, D.; Elreafey, A. M.; Bogaardt, C.; Horton, D. L.; Ho, S.; Kankeyan, T.; Carr, C.; Hoschler, K.; Barclay, W. S.; Amirthalingam, G.; Brown, K.; Charleston, B.; Bailey, D.
title: Neutralising antibody activity against SARS-CoV-2 variants, including Omicron, in an elderly cohort vaccinated with BNT162b2
date: 2021-12-24
journal: nan
DOI: 10.1101/2021.12.23.21268293
sha: 8a2ce92852dc0da4c30134b93b581f35098f4c03
doc_id: 326977
cord_uid: brtflzmw

SARS-CoV-2 variants threaten the effectiveness of tools we have developed to mitigate against serious COVID-19. This is especially true in clinically vulnerable sections of society including the elderly. Using sera from BNT162b2 (Pfizer-BioNTech) vaccinated individuals aged between 70 and 89 (vaccinated with two doses 3-weeks apart) we examined the neutralising antibody (nAb) response to wildtype SARS-CoV-2. Between 3 and 20-weeks post 2nd dose, nAb titres dropped 4.9-fold to a median titre of 21.3 (ND80) with 21.6% of individuals having no detectable nAbs at the later time point. Experiments examining the neutralisation of twenty-one different SARS-CoV-2 variant spike proteins confirmed a significant potential for antigenic escape, especially for the Omicron (BA.1), Beta (B.1.351), Delta (B.1.617.2), Theta (P.3), C.1.2 and B.1.638 variants. Interestingly, however, the recently-emerged sub-lineage AY.4.2 was more efficiently neutralised than parental Delta pseudotypes. Combining pseudotype neutralisation with specific receptor binding domain (RBD) ELISAs we confirmed that changes to position 484 in the spike RBD were predominantly responsible for SARS-CoV-2 nAb escape, although the effect of spike mutations is both combinatorial and additive. Lastly, using sera from the same individuals boosted with a 3rd dose of BNT162b2 we showed that high overall levels of neutralising antibody titre can provide significant levels of cross-protection against Omicron. These data provide evidence that SARS-CoV-2 neutralising antibodies wane over time and that antigenically variable SARS-CoV-2 variants are circulating, highlighting the importance of ongoing surveillance and booster programmes. Furthermore, they provide important data to inform risk assessment of new SARS-CoV-2 variants, such as Omicron, as they emerge.

The effects of the COVID-19 pandemic have, in some countries, been mitigated by the 41 implementation of highly efficacious vaccines, which have reduced hospitalisations and 42

deaths. In late 2020 and early 2021 this was widely achieved (in the UK, France and 43 elsewhere) through vaccination with BNT162b2 (Pfizer-BioNTech)a lipid nanoparticle-44 formulated, nucleoside-modified RNA vaccine encoding prefusion stabilized SARS-CoV-2 45 spike; or ChAdOx1 nCoV-19 (AZD1222, Oxford-AstraZeneca), an adenoviral-vectored 46 vaccine expressing wild-type (non-stabilised) spike [1, 2] . Initially in the UK, for BNT162b2, 47

two doses of this vaccine were administered 3-weeks apart, with many of the most clinically 48 vulnerable (within the nine priority groups established by the UK's Joint Committee on 49

Vaccination and Immunisation [JCVI] ) receiving their vaccines with this dosing interval. 50

However, in the UK, this schedule was quickly changed to 'up to 12-weeks', to maximise use 51 of limited supplies of these vaccines and to protect the largest possible number of people from 52 developing serious disease. This remained the strategy as vaccination was extended to the 53 priority groups further down JCVI's list (stratification based primarily on age), before being 54 opened up to all adults later in 2021, as well as children over 12. Third doses, as well as 55

boosters, are also now available to all adults [3] . To date, detailed information on vaccine 56 responses in elderly populations (>65) vaccinated with BNT162b2 3-weeks apart is lacking, in 57 particular data on neutralising antibody (nAb) titres over time, correlations between ELISA and 58 nAb titres, cross-protective nAb titres against SARS-CoV-2 variants, and the role of boosters 59 in enhancing this cross protection. These data are relevant to elderly cohorts in the UK, but 60 also internationally where the 3-week interval between doses is followed. Neutralising 61 antibodies in the elderly are of especial significance, as it is well established that vaccine 62 responses in this demographic are less robust [4, 5] , and this group may represent a 63

vulnerable cohort for SARS-CoV-2 variants. 64

Whereas the SARS-CoV-2 virus that caused the first global wave had little genetic, antigenic 65 or other phenotypic diversity, subsequent waves comprised extremely diverse SARS-CoV-2 66 'variants', defined by genetic, antigenic and phenotypic divergence from preceding strains [6] . 67

The most widespread or concerning of these were classified by the World Health Organisation 68

(WHO) and/or UKHSA as 'variants of concern (VOC)', 'variants of interest/variants under 69

investigation' (VOI, WHO; VUI, UKHSA) or 'variants under monitoring' VUM; these VOCs and 70

VUIs were subsequently given Greek letter identifiers [7] . The first described of these, the 71

Alpha variant (Pango lineage B.1.1.7), emerged in the UK around Autumn 2020 and showed 72

higher transmissibility than previous variants [8, 9] . The VOC Beta/B.1.351 was detected at a 73 similar time in South Africa, with the VOC Gamma/P.1 identified in Brazil not long after. All 74 three of these VOCs showed antigenic distance from the original strain, the spike protein of 75 which is used as the immunogen in BNT162b2 and ChAdOx1 nCoV-19 vaccines [10, 11] . 76

From the start of 2021 many further variants, arose throughout the world, including B. southern Africa (specifically Botswana and South Africa), with isolations found across the 88 globe in the following days and weeks. Early data on increased transmissibility, community 89 displacement of Delta and evidence for significant antigenic variation support its immediate 90 classification as a VOC (Omicron) [17] [18] [19] [20] . 91

To examine antibody levels and T-cell responses following the extension to the COVID-19 92 vaccine schedule the UKHSA (formerly Public Health England, PHE) initiated a prospective 93

longitudinal audit of vaccinated adults (the CONSENSUS study). Within CONSENSUS, a 94 cohort of volunteers aged between 70 and 89 received the BNT162b2 vaccine 3-weeks apart. 95

In our study, using sera from 37 individuals in this cohort, we investigated the impact of waning 96

immunity on neutralisation of SARS-CoV-2, the correlation between ELISA, RBD-ELISA and 97

nAb titres, as well as the impact of various SARS-CoV-2 VOCs and VUMs/VUIs on viral 98

neutralisation. These data indicate that this clinically vulnerable priority group may be 99 especially susceptible to repeat infection with SARS-CoV-2 variants that are antigenically 100 distinct from the originally emerged strain; however, a 3 rd dose of BNT162b2 can mitigate 101 against this risk. This susceptibility is the result of low overall nAb titres and appears to be 102 mechanistically defined by specific changes to the SARS-CoV-2 spike RBD, in particular E484 103 changes. These data provide a key tool for risk assessing current and future SARS-CoV-2 104 variants, including Omicron/BA.1 or sub-lineages (BA.2). 105

Using a SARS-CoV-2 spike pseudotype-based micro-virus neutralisation assay (mVNT), we 107 determined neutralisation titres (ND80s) SARS-CoV-2 S ECLIA) and there was a strong correlation between mVNT and ELISA titres 126 in both age groups (70-79, Spearman r = 0.84, Figure 1C ; 80-89, Spearman r=0.91, Figure  127 1D ) and Beta ≤10 (IQR 10-10) ( Figure 2C ). For 140

Delta and Beta this equated to a drop in neutralising titre of 11.2-fold and 13.9-fold, 141

respectively. Neutralising titres to Delta at 20-weeks post-vaccination were slightly lower, 142 consistent with the waning antibody response evidenced in Figure 1 (Supplemental Figure 2A-143 B). However, there was evidence of affinity maturation in some individuals, especially to the 144

Beta VOC, although the median value remained unchanged (Supplemental Figure 2C -D).

VNT assays with replication competent SARS-CoV-2 WT D614 and the Beta VOC also 146 identified a reduction in neutralisation for Beta, with the calculated titres correlating well with 147 the mVNT pseudotype data (Supplemental Figure 3 ). Comparing the S ELISA (Roche Elecsys 148

anti-SARS-CoV-2 S ECLIA) with VOC ND80 titres we identified a strong correlation for D614G 149 (Spearman r = 0.86), Alpha (r = 0.83) and Delta (r = 0.86) neutralisation titres but a poor 150 correlation with Beta (r = 0.52) ( Figure 2D ). To investigate the mechanisms underpinning the 151 drop in neutralisation seen for specific VOCs we next performed a targeted ELISA with the 152 same sera and recombinant RBDs reflecting WT SARS-CoV-2, Alpha, Delta and Beta spike 153 sequences. In both the 70-79 and 80-89 age groups there was no significant difference in 154

ELISA titres between WT, Alpha and Delta RBDs ( Figure 2E -F); however, there was a 155 significant reduction in binding to the Beta RBD (70-79, 2.9-fold compared to WT; 80-89, 2.2-156 fold), partially correlating with the mVNT results. The S ELISA and RBD-ELISA data (both 157 based on antigens representing WT SARS-CoV-2 spike sequence) showed a strong 158 correlation (Spearman r = 0.93), indicative of good agreement between the two assays ( Figure  159 2G). However, the correlation between RBD-ELISA and ND80 titres was again poor for the 160

Beta VOC (Spearman r = 0.49), albeit relatively consistent for the WT (r = 0.83), Alpha (r = 161 0.80) and Delta (r = 0.86) RBD-ELISAs. 162

Using a smaller representative pool of sera from the same cohort (3-weeks post 2nd dose; 163 n=16 total; 70-79, n=11; 80-89 n=5), selected by ranking the neutralisation responses of the 164 whole cohort, we widened our analysis to fifteen other SARS-CoV-2 variants, including other 165

VUIs and VUMs, in particular AY.4.2 ( Figure 3A ). This timepoint was chosen as the titres were 166 higher at 3-rather than 20-weeks post 2 nd dose. Several variants showed a significant 167 reduction in neutralisation when compared to D614G (B.1); namely B. Figure 2B -C]). Interestingly, the 179 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in To provide better spatial representation of the antigenic relationships between the different 183 variants, we also performed antigenic cartographic analysis on the collated ND80 titres from 184 a sub-set of the sera (3 weeks post second-dose, n=14) tested against all available VOCs, 185

VUIs and VUMs in mVNTs. Antigenic cartography allows high resolution quantitative 186 comparison and visualisation of antigenic relationships. Antigenic distances between antigens 187 on the map are measured in antigenic units, with one antigenic unit (AU) being the equivalent 188 of a two-fold dilution in titre. The antigenic map of ND80 titres ( Figure 3C ) again highlights the 189 largest antigenic distance is between D614G and Beta ( clustering of sera means interpretation of the distances between the most divergent SARS-198

CoV-2 variants on the antigenic map (e.g P.3 to Beta) is less reliable than their distances to 199 D614G, demonstrated by the confidence coordination areas for their positions (Supplemental 200 Figure 5A ). In addition, to identify the amino acid changes within the RBD responsible for the 201 significant drop in neutralisation in this cohort we repeated the WT, Alpha, Delta and Beta weeks post 2 nd dose this was 92% and 86%, respectively. However, at 4 weeks following the 217 3 rd dose all volunteers now, significantly, had detectable titres against Omicron ( Figure 4B -C). 218

In the 70-79 age group, when compared to D614G (median ND80 ≥530.7; IQR 318.3-883.4), 219

there was a 53.1-fold reduction (median ND80 ≤10; IQR 10-10) in neutralisation of Omicron 220 at 3 weeks post vaccination, but only an 8.0-fold (D614G; median ND80 ≥3070.9; IQR 1498.4-221

3798.1 vs Omicron; median ND80 384.3; IQR 189.7-735.7) drop 4 weeks after the 3 rd dose 222

( Figure 4B ). In the older age group (ages 80-89) the median titres for D614G were 1354.1 223 (IQR 953.3-2103.9) after the 3 rd dose, whilst for Omicron this was 80.8 (IQR 59.5-239.2), 224 Figure 4C ), a 16.8-fold reduction. Correlating the S ELISA (Roche S1) values for these 225 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101/2021.12.23.21268293 doi: medRxiv preprint samples with their Omicron ND80 titres further illustrates the improved neutralisation titres 226

following a 3 rd dose of vaccination and also identified an improved correlation between ND80 227

and ELISA (20 weeks post 2 nd dose, Spearman r = 0.53; 4 weeks post 3 rd dose, Spearman 228 r=0.82 Figure 4D -E). 229

At a global level many elderly populations have been protected from COVID-19 by the 231 implementation of mass vaccination. However, lower overall Ab responses and concerns of 232 waning immunity in elderly vaccinees have highlighted the potential impact of antigenically 233 distinct variants on controlling this disease. Using sera from an elderly cohort double-234 vaccinated with BNT162b2 we demonstrated significantly reduced neutralisation of the Delta 235

and Beta VOCs, which is compounded by waning immunity (Figures 1-2 comparison of the polyclonal B-cell response in various age-groups; however, Greaney et al., 254 demonstrated the response is skewed to a single class of antibodies which target an epitope 255

in the RBD encompassing position 484, which is substituted in Beta (E484K) [29] . One 256 limitation of our study is that for some VOCs, e.g. Beta, almost all the sera dropped below the 257 limit of detection of our assay (ND80 of 10), providing less granularity, complicating 258

comparisons with other studies, and obfuscating correlations with more sensitive Ab-detection 259 assays such as the RBD-ELISA. 260

Elsewhere, our data on Omicron neutralisation following 2 doses of BNT162b2 fits with 261 recently reported findings. Dejnirattisai et al., Cameroni et al., showed drops in neutralisation of >30-fold [18-20, 30] . However, it is clear from our data, and 263

that of [20] that boosting with a 3 rd dose of BNT162b2 generates a much 264

higher overall titre of neutralising antibodies to D614G (median ND80 ≥3070.9 in the 70-79 265 age group, Figure 4B ) and that these high titres enhance cross-protection against Omicron 266

and mitigate against the significant drops in neutralisation seen after two doses, highlighting 267

the important role of boosters in providing robust long-term immunity to SARS-CoV-2. 268

A smaller pool of antigenicity data is available on the wide range of other variants (VUIs/VUMs) 269

that have emerged since the beginning of the pandemic. Our data highlighted several 270 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in 2.5AU from D614G ( Figure 3B ). As discussed above these differences might be attributable 284

to laboratory or cohort-specific factors. Nevertheless, the main strength of our approach has 285 been to use a single set of sera to compare the relative neutralisation of VOCs and VUIs/VUMs 286 ( Figure 3 ). 287

As summarised by Harvey et al., [37] our molecular understanding of SARS-CoV-2 variant 288 spike mutations and immune escape is now relatively well-established, permitting the 289 mechanistic contextualisation of these neutralisation data. The most pertinent question raised 290 by our data set is, 'why are Omicron and Beta so bad?'. From a variant monitoring perspective, 291

the answers to this question can be used as a framework to understand other SARS-CoV-2 292 variants, old or new. Since Omicron has so many changes in its spike protein, it is difficult to 293 draw any meaningful conclusions without doing targeted mutagenesis. However, a number of 294 the other variants share enough similarity to Beta that useful comparisons and conclusions 295

can be drawn by correlating individual mutations to varied antigenicity. Compared to variants 296 such as Omicron or B.1.638, the spike mutation profile of Beta is relatively simple, with only 9 297 changes ( Figure 3A ). Focusing on the RBD, the amino acid changes K417N and E484K are 298

well associated with antigenic escape from monoclonal antibodies and the polyclonal B-cell 299

response [22, 23, 29] . Changes at position 417 affect class 1 Ab binding while 484 300 modifications affect the epitope bound by class 2 Abs, which dominate the polyclonal response 301

to the RBD [37] . The importance of these particular RBD-changes to Ab-binding was 302 confirmed by our RBD-ELISA ( Fig.2E -F and Fig.3D ). Of the single amino acid mutated RBDs 303

we assessed, only those containing K417N (1.2 fold) or E484K/D (E/K, 1.6-fold; E/D, 1.4-fold) 304

significantly reduced Ab-binding, when compared to WT. The E484D change, although not 305 present in any variant we analysed, has previously been implicated in escape from monoclonal 306 nAbs [37] . Combining these changes with N501Y in the Beta RBD appeared to be additive 307 and led to a concomitant 2.2-fold reduction in Ab-binding, indicating that the loss of Beta VOC 308 neutralisation we, and others [22] , have observed ( Figure 2B -C) is due to a reduction in class 309 1 and 2 nAb binding. Interestingly, the magnitude of reduction for the complete Delta RBD 310

(1.1-fold) or Delta-specific single amino acid RBD substitutions was lower (L452R, 1.1-fold) or 311 the same as K417N (T478K, 1.2-fold) and non-significant in our studies ( Figure 3D ). This might 312 explain the more moderate reductions in neutralisation seen with this VOC (Figure 2B -C) or 313 point to another region of the Delta spike being responsible for antigenic variance. Further, 314

our RBD-ELISA data on E484Q, which only moderately reduced Ab-binding (1.1-fold) might 315 explain the relatively minor effects of B.1.617.3 on neutralisation (1.6-fold reduction, relative 316 to D614G, Figure 3B ). 317 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in neutralisation. However, the addition of N501Y (which is also present in Beta, Mu, Theta and 323 C.1.2) seemed to enhance escape from neutralisation, e.g., P.3/Theta (7.2-fold, 3.5AU).

Although there is some evidence that N501Y modifies antigenicity [38] , the presence of this 325 mutation in Alpha does not appear to affect its neutralisation ( Figure 2B-C, [39] ). Moreover, 326 this substitution has most frequently been mechanistically linked to an increase in affinity for 327

ACE2 [40] . In the context of neutralisation of E484-mutated variants it may be that the 328 increased affinity of N501Y for ACE2 has a compound effect on class 2 antibodies trying to 329 bind spike, as these antibodies' affinities for spike have already been reduced by, for example, 330

the E484K substitution. We recently showed that the RaTG13 spike is efficiently neutralised 331

by SARS-CoV-2-specific sera, which was surprising given the large number of RBD 332

substitutions between these two spike proteins [41] . However, the affinity of RaTG13 spike for 333 human ACE2 is markedly lower than SARS- recurrently deleted region 4 (RDR4), which corresponds to exposed loops on the surface of 342 the spike NTD. RDR4 deletions have previously been associated with the loss of monoclonal 343

antibody binding (4A8) [43] . Likewise, L18F also sits within the NTD supersite and its 344 modification may also affect antibody binding. 345

To summarise, the significant antigenic escape properties of Beta and Omicron are likely the 346 result of a combination of changes to the spike protein, which act in synergy and in an additive 347 manner to avoid Ab-recognition, with the other variants we tested lacking some or all of these 348

features. These hypotheses could be tested by introduction of K417N into C. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101/2021.12.23.21268293 doi: medRxiv preprint changes in the spike NTD, Y145H and A222V. Interestingly, despite these additional changes, 365

pseudotypes bearing the AY.4.2 spike as well as Delta + A222V alone were more efficiently 366 neutralised by sera from our BNT162b2-cohort ( Figure 3 ) when compared to Delta (albeit not 367 significantly; Wilcoxon matched-pairs signed rank test). These data are similar to those 368 reported by Lassauniere et al., who also examined BNT162b2 mRNA vaccine-elicited sera 369

[47]. A222V is the second most common substitution in SARS-CoV-2 Spike (after D614G), yet 370

to date had no established phenotypic impact on spike function [37] . It remains unclear why 371

A222V increases the neutralisation of Delta; however, this could relate to increased 372 accessibility of the RBD to nAbs. It remains to be seen how the prevalence of AY.4.2 will be 373 affected by the introduction of Omicron to the UK; however, epidemiologists are predicting that 374

Delta and its sub-lineages will soon be replaced by this new variant [13] . 375

Another pertinent question that our data set raises is whether spike-based ELISA titres 376 correlate well with virus (or pseudovirus) neutralisation titres. From a practical perspective, if 377

ELISAs can be used to establish an immune threshold that in turn reliably correlates with a 378 certain level of nAbs, then that would be advantageous for clinical management of This is especially true if a threshold can easily be established for each variant as and when 380 they emerge. Unfortunately, whilst our data on a BNT162b2-vaccinated elderly cohort 381 identified a reliable correlation between WT (D614 or D614G), Alpha and Delta, it was 382 disrupted by more antigenically distinct variants like Beta ( Figure 2 ) and Omicron ( Figure 4D ).

Whilst the RBD-ELISA with a Beta RBD was useful for mechanistically underpinning our 384 understanding of 'why Beta is so bad', and the antigenic relationships were largely consistent 385 between mVNT and ELISA using antigenic maps, the results did not improve the correlation 386 with neutralisation titres, although a better correlation exists with higher titre sera. However, a 387 major limitation of our findings is that our mVNT was clearly less sensitive than the ELISA or 388

RBD-ELISA, with many Beta mVNTs falling below the limit of detection (<10 ND80). A younger 389 cohort, one sampled from vaccinees who had the extended interval between 1 st and 2 nd dose 390

[48], or people who receive a 3 rd dose ( Figure 4 ) who collectively have higher overall titres, 391 might represent a better benchmark for understanding and testing this correlation in more 392 detail. 393 A younger cohort, including recipients of other vaccines, could also improve the resolution of 394 the antigenic maps. Antigenic cartography offers great potential as a robust and repeatable 395 way to measure antigenic changes among variants, and test hypotheses on the effect of 396

individual and combinations of spike protein mutations [49, 50] . However, using sera with 397 similar neutralisation profiles and from volunteers with unknown history of exposure to other 398 viruses limits the resolution of the maps [51]. Using sera from other species with a known 399 exposure history has also demonstrated its utility in rigorous comparison of antigenic 400 distances, overcoming these issues [52] . 401

In summary, our data highlight the propensity of certain SARS-CoV-2 variants to partially avoid 403

vaccine-derived immunity. The BNT162b2-vaccinated elderly cohort we investigated showed 404 evidence of waning immunity at 20-weeks post-vaccination, which could potentially 405 exacerbate this escape. However, high titres could be rescued with a 3 rd dose, which provided 406 cross-protective immunity against Omicron. Developing a better understanding of how these 407 titres relate to a well-defined correlate of immunity will be an important step in understanding 408 the wider implications of this data on the management of COVID-19 and whether the risk of 409 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101/2021.12.23.21268293 doi: medRxiv preprint breakthrough infections, hospitalisation and deaths is increased, either by waning immunity or 410 new variants. Interestingly, three of the vaccinated volunteers in our study were SARS-CoV-2 411 N Ab positive (Elecsys® Anti-SARS-CoV-2 N, Roche), indicative of previously resolved 412

infections. The nAb and ELISA titres from two of these individuals were the highest in our 413 cohort, supporting observations that infection and vaccination provide a robust Ab response 414

to . Clearly, we can achieve similar titres through boosting ( Figure 4) 

Healthy participants were recruited as part of the "COVID-19 vaccine responses after 451 extended immunisation schedules" (CONSENSUS) study [21] . Participants aged 70-90 years 452

in January 2021 were recruited through North London primary care networks. Sera samples 453 used in this study were taken at 3 and 20 weeks after 2 doses of Pfizer/BioNTech BNT162b2 454 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in 

Mutant SARS-CoV-2 expression plasmids were generated by site-directed mutagenesis, by 469

using the QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent) or were 470 synthesised by Geneart (Thermo Fisher). All SARS-CoV-2 spike expression plasmids were 471

based on a codon optimised Wuhan-Hu-1 reference sequence (Genbank ID NC_045512.2) 472

[57], with the additional substitutions K1255*STOP (also referred to as Δ19 mutation or 473 cytoplasmic tail truncation). A list of the SARS-CoV-2 spike variants and their associated 474 mutations can be found in Supplemental Table 1 . Some substitutions here differ from the 475 lineage defining sequence for the named variant (Pango); these were included as they were 476

substitutions that were highly sampled in submitted sequences and predicted to be a plausible 477

worst case antigenic escape for that lineage. 478

Lentiviral-based SARS-CoV-2 pseudotyped viruses were generated in HEK293T cells 480 incubated at 37°C, 5% CO2 as previously described [58] . Briefly, cells were transfected with 481 900 ng of SARS-CoV-2 spike (see Supplemental Table 1 ), 600 ng p8.91 (encoding for HIV-1 482 gag-pol), 600 ng CSFLW (lentivirus backbone expressing a firefly luciferase reporter gene) 483

with PEI (1 µg/mL) transfection reagent. Supernatants containing pseudotyped SARS-CoV-2 484

were harvested and pooled at 48 and 72 hours post transfection, centrifuged at 1,300 x g for 485 10 minutes at 4 °C to remove cellular debris and stored at -80 °C. SARS-CoV-2 486 pseudoparticles were titrated on HEK293T cells stably expressing human ACE2 S-CoV-2 pps 487 and infectivity assessed by measuring luciferase luminescence after the addition of Bright-Glo 488 luciferase reagent (Promega) and read on a GloMax Multi+ Detection System (Promega). 489

Sera was diluted 1:5 in serum-free media in a 96-well plate in triplicate and titrated 3-fold. A 491 fixed titred volume of SARS-CoV-2 pseudoparticles was added at a dilution equivalent to 10 5 -492 10 6 signal luciferase units in 50 µL DMEM-10% and incubated with sera for 1 hour at 37 °C, 493 5% CO2 (giving a final sera dilution of 1:10). Target cells expressing human ACE2 were then 494 added at a density of 2 x 10 4 in 100 µL and incubated at 37 °C, 5% CO2 for 48 hours. Firefly 495 luciferase activity was then measured after the addition of BrightGlo luciferase reagent on a 496

Glomax-Multi + Detection System (Promega, Southampton, UK). Pseudotyped virus 497 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101/2021.12.23.21268293 doi: medRxiv preprint neutralisation titres were calculating by interpolating the point at which there was 80% 498 reduction in reduction in luciferase activity, relative to untreated controls, neutralisation dose 499 80% (ND80). 500

Antibody to the RBD of SARS-CoV-2 spike protein was measured using an in-house indirect 502

IgG RBD assay as described previously [59, 60] . Briefly, commercially synthesised 503 recombinant RBD subunit spike(Arg319-Phe541(V367F); SinoBiological Inc, Hong Kong, 504

PRC) with a C-terminal mouse Fc tag, was coated onto 96-well microtiter plates at 20 ng per 505 well at 4°C-8°C for a minimum of 16 hours. After washing and blocking, sera were analysed 506 at a dilution factor of 1 in 100 by serially diluting each serum sample starting at 1:100, (6-fold 507

with the highest dilution achieved 129600). The binding of IgG on the plate surface was 508 detected by using an anti-human IgG horseradish peroxidase antibody conjugate (Sigma-509

Aldrich, Poole, UK) and 3,3′,5,5′-Tetramethylbenzidine (Europa Bioproducts Ltd, Ely, UK). We 510 analysed samples in duplicate and evaluated optical density at 450 nm (OD450) Samples were 511 analysed in the presence of known positive controls (collected from individuals with confirmed 512 SARS-CoV2 infection) and a calibrator sample ("negative" added to four wells; collected prior 513

to the pandemic). Titres are expressed as serum fold-dilution required to achieve a T/N (test 514

OD to negative OD) of 5 (T/N = 5 serves as cut-off for positive samples) by xy interpolation 515 from the RBD data series (dilution, x versus OD450, y). Samples which were below the cut-516 off in the initial dilution (ie negative), were expressed as <100. 517

Sera samples were tested by commercial ELISA as previously described [21] . Nucleoprotein 519 (N) antibodies were measured as a marker of previous SARS-CoV-2 infection (Anti-SARS-520

CoV-2 total antibody assay, Roche Diagnostics, Basel, Switzerland) and spike (S) protein 521 antibodies were measured as an indication of infection or vaccination (Elecsys Anti-SARS-522

CoV-2 S total antibody assay, Roche Diagnostics: positive ≥ 0.8 arbitrary units (au)/mL to 523 assess vaccine response). 524

Sera were serially diluted 1:2 in media containing 1% foetal calf serum (FCS) and incubated 526 with 112.5 plaque forming units (PFU) of SARS CoV-2 (hCoV-19/England/02/2020, 527 EPI_ISL_407073, or Beta (B.1.351), kindly provided by Public Health England) for one hour 528

at 37 °C, 5% CO2. 75 PFU of neutralisation mixture in a total of 200µl were then added to 96-529 well plates containing a ~80-90% confluent monolayer of Vero-E6-TMPRSS2 (a gift from 530

Stuart Neil, KCL, London) cells were incubated for 6 days at 37 °C, 5% CO2 in quadruplicate 531 per serum sample. Inoculum was then removed, and cells were fixed with formalin for 30 532 minutes before staining with 0.1% toluidine blue in PBS. ND80 titres were calculated using a 533

Spearman and Karber formula. Thawed virus aliquots used for VNT were back-titrated 1:10 534 by TCID50 on VERO-E6-TMPRSS2 cells to confirm the titre at time of use. 535

Antigenic maps were made with antigenic cartography techniques described previously [50] , 537

implemented in the R package Racmacs (version 1.1.12; https://acorg.github.io/Racmacs/). In 538

brief, titres were first converted into serum-antigen target distances, and sera and antigens 539

were then positioned on a map in a way that minimised the difference (residual sum of 540 squares) between target distances and corresponding map distances, using multidimensional 541

scaling. The target distance for each serum-antigen pair was calculated by subtracting the 542 logarithm (log2) of the titre, from the highest log2(titre) for the serum against any antigen; thus, 543 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 higher reciprocal titres resulted in shorter target distances. Multidimensional scaling was 544 carried out with 1000 random restart optimisations, to avoid local optima and increase the 545 likelihood of finding the best fit to the measured titres. The resulting maps were ordered 546 according to total error, and compared for self-consistency; the figures and descriptions in this 547 manuscript pertain to the maps with the lowest total error. Antigenic distances were measured 548 from the lowest error antigenic map. 549

The antigenic map of the ND80 titres was made based on titres from a subset of 14 sera: two 550 sera were removed before mapping, as their titres were consistently below or only marginally 551 above the detection limit, and they did therefore not contain valuable information for 552 cartography. ND80 titres from different experiments were merged without normalisation 553

procedures. For each serum, a single overall log2(titre) to D614G and B.1.617.2/Delta was 554 calculated by taking the mean of the log2(titre) values for these antigens across experiments. 555

Such average log2(titre) values were excluded (replaced with NA/'missing value') in case the 556 standard deviation of log2(titre) values for the antigen was equal to or exceeded 1. 557

To determine the optimal number of dimensions for representing the data, prediction 558 experiments were performed: antigenic maps were made while omitting a random 10% of 559 titres; the excluded titres were predicted according to their relative positions in the map; and 560 the predicted titres were then compared to the actual titres (on a log scale). Antigenic maps 561

were made in two, three, four and five dimensions, using 1000 optimizations; for each 562 dimension, 100 prediction tests were performed. The mean root mean square error (RMSE) 563 associated with the prediction of ND80 titres was 0.94 (SD: 0.15) for detectable titres and 1.32 564

(SD: 0.68) for non-detectable titres (i.e., below the limit of detection, <10), for two dimensions; 565 0.97 (SD: 0.14) and 1.37 (SD: 0.61) for three dimensions; 0.96 (SD: 0.14) and 1.34 (SD: 0.60) 566

for four dimensions; and 0.95 (SD: 0.14) and 1.35 (SD: 0.58) for five dimensions. The mean 567

RMSE associated with the prediction of RBD-ELISA titres was 0.55 (SD: 0.16) for detectable 568 titres and 1.27 (SD: 0.13) for non-detectable titres (i.e., below the limit of detection, <100), for 569 two dimensions; 0.56 (SD: 0.15) and 1.26 (SD: 0.10) for three dimensions; 0.56 (SD: 0.14) 570 and 1.25 (SD: 0.10) for four dimensions; and 0.56 (SD: 0.15) and 1.25 (SD: 0.11) for five 571 dimensions. Overall, for each dataset, the mean RMSE was similar across dimensions, and 572

in each case the mean RMSE for prediction of detectable titres corresponded to less than a 573 twofold dilution (one antigenic unit). Therefore, we considered two-dimensional maps 574 sufficient for representing the SARS-CoV-2 antigenic data. 575 576 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in collected from the same individuals at 3-(n=37 total) and 20-weeks (n=35 total) post 2 nd dose, 584

with a vaccination interval of 3 weeks between 1 st and 2 nd doses. Titres are expressed as 585 serum fold-dilution required to achieve 80% virus neutralisation, with the titre (ND80) 586 calculated by xy interpolation from the mVNT data series (dilution, x versus luciferase activity, 587 RLU, y). Statistical comparison of ND80 titres at 3 and 20 weeks was performed using a 588

Wilcoxon matched-pairs signed rank test (*; <0.05; ***, <0.001). Fold changes in median ND80 589 between 3 and 20 weeks are indicated. The detection limit of the assay is indicated with a 590 dotted line. The correlation between ND80 and S ELISA (Roche S) titres recorded from each 591 volunteer, was examined at 3 (n=37) (C) and 20 (n=35) (D) weeks post 2 nd dose, with statistical 592 analysis of the matrix performed using a nonparametric Spearman correlation (r). 593 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in Titres are expressed as serum fold-dilution required to achieve 80% virus neutralisation, with 602 the titre (ND80) calculated by xy interpolation from the mVNT data series (dilution, x versus 603 luciferase activity, RLU, y). Statistical comparison of ND80 titres against the D614G reference 604 was performed using a Wilcoxon matched-pairs signed rank test (***, <0.001; ****, <0.0001).

Fold changes in median ND80, compared to D614G are indicated. The detection limit of the 606 assay is indicated with a dotted line. The correlation between ND80 and S ELISA (Roche S) 607

titres recorded for each volunteer (n=37), was then examined (D) for each VOC, with statistical 608 analysis of the matrix performed using a nonparametric Spearman correlation (r Dunn's multiple comparisons of column means (**, <0.005; ****, <0.0001). Fold changes in 616 median titre, compared to WT are indicated. The detection limit of the assay is indicated with 617 a dotted line. The correlation between RBD-ELISA and S ELISA titres (Roche S) (G) or ND80 618 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101/2021.12.23.21268293 doi: medRxiv preprint titres for each VOC (H) recorded from each volunteer (n=37) was then examined, with 619 statistical analysis of the matrix performed using a nonparametric Spearman correlation (r). taken from an individual who tested positive for SARS-CoV-2 Nucleoprotein by ELISA, 629

indicative of previous infection. Titres are expressed as serum fold-dilution required to achieve 630 80% virus neutralisation, with the titre (ND80) calculated by xy interpolation from the mVNT 631 data series (dilution, x versus luciferase activity, RLU, y). Statistical comparison of ND80 titres 632 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 against the D614G reference was performed using a Wilcoxon matched-pairs signed rank test 633 (*, <0.05; **, <0.005, ***, <0.001; ****, <0.0001). The separate graphs represent experiments 634 performed on different days, with each assay including the D614G pseudotype as a reference. 635

Fold changes in median ND80, compared to D614G are indicated. The detection limit of the 636 assay is indicated with a dotted line. (C) Two-dimensional antigenic map of variants, based 637 on the ND80 titres in (B). Multidimensional scaling was used to position the sera and variants 638

to best fit target distances derived from the titres. The map is the lowest error solution of 1000 639

optimisations. Variants are represented by solid circles, sera by open squares (with the 640

Nucleoprotein-ELISA-positive serum in red). Two sera were not used in mapping because of 641 titres consistently below the detection limit. The spacing between grid lines represents one 642 antigenic unit, equivalent to a two-fold dilution in ND80 titres. Neutralisation titres calculated using pseudotypes bearing the SARS-CoV-2 D614G or 657

Omicron spike and sera from a cohort of BNT162b2 vaccinated individuals (n=19), recruited 658

as part of the UK CONSENSUS trial, aged 70-79 (n=12, solid circle symbols) (B) or 80-89 659

(n=7, open circle symbols) (C). Sera samples were taken from the same volunteers at the 660 indicated times post 2nd or 3rd doses of vaccination. Symbols in red represent samples taken 661 from individuals who tested positive for SARS-CoV-2 Nucleoprotein by ELISA, indicative of 662 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted December 24, 2021. ; https://doi.org/10.1101/2021.12.23.21268293 doi: medRxiv preprint previous infection. Titres are expressed as serum fold-dilution required to achieve 80% virus 663 neutralisation, with the titre (ND80) calculated by xy interpolation from the mVNT data series 664 (dilution, x versus luciferase activity, RLU, y). Fold changes in median ND80 between D614G 665

and Omicron are indicated. The detection limit of the assay is indicated with a dotted line. The 666

correlation between Omicron ND80 and S ELISA (Roche S) titres recorded from each 667 volunteer, was examined at 20 weeks post 2nd dose (D) and 4 weeks post 3rd dose (both 668 n=19) (E), with statistical analysis of the matrix performed using a nonparametric Spearman 669 correlation (r). 670

Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine

Safety and immunogenicity of the ChAdOx1 nCoV-19 vaccine 675 against SARS-CoV-2: a preliminary report of a phase 1/2, single-blind, randomised 676 controlled trial

Coronavirus (COVID-19) 678 vaccination information for public health professionals

Age-related immune response heterogeneity to SARS-CoV-2 681 vaccine BNT162b2

Age-Dependent Reduction in Neutralization against Alpha and 683 Beta Variants of BNT162b2 SARS-CoV-2 Vaccine-Induced Immunity. Microbiol 684 Spectr

SARS-CoV-2 one year on: evidence for ongoing viral adaptation

SARS-CoV-2 Variants of Interest and Concern naming scheme 688 conducive for global discourse

Estimated transmissibility and impact of SARS-CoV-2 lineage 690 B.1.1.7 in England

Preliminary genomic characterisation of an emergent SARS-CoV-692 2 lineage in the UK defined by a novel set of spike mutations. 2020: virological.org

Detection of a SARS-CoV-2 variant of concern in South Africa. 694 Nature

Genomics and epidemiology of the P.1 SARS-CoV-2 lineage in 696

Emergence and phenotypic characterization of C.1.2, a globally 698 detected lineage that rapidly accumulated mutations of concern. medRxiv

Technical briefing 31, in SARS-CoV-2 variants of concern 701 and variants under investigation in England. 2021, UK HSA

European Centre for Disease Prevention and Control, SARS-CoV-2 variants of 703 concern as of 13

SARS-CoV-2 B.1.617.2 Delta variant replication and immune 705 evasion

SARS-CoV-2 Spike Mutations, L452R, T478K, E484Q and P681R, 707 in the Second Wave of COVID-19 in Maharashtra, India. Microorganisms

SARS-CoV-2 Omicron has extensive but incomplete escape of Pfizer 710 BNT162b2 elicited neutralization and requires ACE2 for infection. medRxiv

Reduced neutralisation of SARS-COV-2 Omicron-B.1.1.529 713 variant by post-immunisation serum. medRxiv

Broadly neutralizing antibodies overcome SARS-CoV-2 Omicron 715 antigenic shift

mRNA-based COVID-19 vaccine boosters induce 717 neutralizing immunity against SARS-CoV-2 Omicron variant

Robust antibody responses in 70-80-year-olds 3 weeks after the 720 first or second doses of Pfizer/BioNTech COVID-19 vaccine

Potency of BNT162b2 and mRNA-1273 vaccine-induced neutralizing 723 antibodies against severe acute respiratory syndrome-CoV-2 variants of concern: A 724 systematic review of in vitro studies

Impact of the Delta variant on vaccine efficacy and response strategies

Expert Review of Vaccines

CoronaVac induces lower neutralising activity against variants 728 of concern than natural infection. The Lancet Infectious Diseases

SARS-CoV-2 variant B.1.617 is resistant to bamlanivimab and 731 evades antibodies induced by infection and vaccination

Reduced neutralization of SARS-CoV-2 B.1.617 by vaccine and 734 convalescent serum

Age-related immune response heterogeneity to SARS-CoV-2 736 vaccine BNT162b2

mRNA vaccination in people over 80 years of age induces strong 738 humoral immune responses against SARS-CoV-2 with cross neutralization of P.1 739 Brazilian variant. eLife

Mapping mutations to the SARS-CoV-2 RBD that escape binding 741 by different classes of antibodies

SARS-CoV-2 Omicron has extensive but incomplete escape of Pfizer 743 BNT162b2 elicited neutralization and requires ACE2 for infection

The continuous evolution of SARS-CoV-2 in South Africa: a new 745 lineage with rapid accumulation of mutations of concern and global detection

Characterization of the emerging B.1.621 variant of interest 748 of SARS-CoV-2

Genome sequencing and analysis of an emergent SARS-CoV-2 750 variant characterized by multiple spike protein mutations detected from the Central 751

Neutralization of Mu and C.1.2 SARS-CoV-2 Variants by Vaccine-753 elicited Antibodies in Individuals With and Without Previous History of Infection

Neutralization of the SARS-CoV-2 Mu Variant by Convalescent and 756 Vaccine Serum

The spike protein of SARS-CoV-2 variant A.30 is heavily mutated and 758 evades vaccine-induced antibodies with high efficiency

SARS-CoV-2 variants, spike mutations and immune escape. 761

Sensitivity of SARS-CoV-2 B.1.1.7 to mRNA vaccine-elicited 763 antibodies

Changes in symptomatology, reinfection, and transmissibility 765 associated with the SARS-CoV-2 variant B.1.1.7: an ecological study. Lancet Public 766 Health

The N501Y spike substitution enhances SARS-CoV-2 infection and 768 transmission

Pseudotyped Bat Coronavirus RaTG13 is efficiently neutralised by 770 convalescent sera from SARS-CoV-2 infected Patients. bioRxiv

Binding and molecular basis of the bat coronavirus RaTG13 virus to 773 ACE2 in humans and other species

Recurrent deletions in the SARS-CoV-2 spike glycoprotein drive 775 antibody escape

The furin cleavage site in the SARS-CoV-2 spike protein is 777 required for transmission in ferrets

SARS-CoV-2 sensing by RIG-I and MDA5 links epithelial infection 779 to macrophage inflammation

Neutralisation of SARS-CoV-2 Delta sub-lineage AY.4.2 and 783 B.1.617.2+E484K by BNT162b2 mRNA vaccine-elicited sera. medRxiv

Immunogenicity of standard and extended dosing intervals of 786 BNT162b2 mRNA vaccine

Substitutions near the receptor binding site determine major antigenic 788 change during influenza virus evolution

Mapping the antigenic and genetic evolution of influenza virus

Flavivirus-induced antibody cross-reactivity

Quantifying antigenic relationships among the lyssaviruses

mRNA vaccination boosts cross-variant neutralizing antibodies 796 elicited by SARS-CoV-2 infection

The ChAdOx1 vectored vaccine, AZD2816, induces strong 798 immunogenicity against SARS-CoV-2 Beta (B.1.351) and other variants of concern in 799 preclinical studies

Neutralizing antibody activity in convalescent sera from infection in 801 humans with SARS-CoV-2 and variants of concern

The furin cleavage site in the SARS-CoV-2 spike protein is 804 required for transmission in ferrets

Self-amplifying RNA SARS-CoV-2 lipid nanoparticle vaccine 806 candidate induces high neutralizing antibody titers in mice

Production of Recombinant Replication-defective Lentiviruses 809 Bearing the SARS-CoV or SARS-CoV-2 Attachment Spike Glycoprotein and Their 810 Application in Receptor Tropism and Neutralisation Assays. Bio-protocol

Seroprevalence of SARS-CoV-2 among Blood Donors and 813 Changes after Introduction of Public Health and Social Measures, London, UK. Emerg 814 Infect Dis

Antibodies to SARS-CoV-2 protect against re-infection during 816 outbreaks in care homes