Emerging SARS-CoV-2 variants of concern evade humoral immune responses from infection and vaccination

Emerging SARS-CoV-2 variants pose a threat to human immunity induced by natural infection and vaccination. We assessed the recognition of three variants of concern (B.1.1.7, B.1.351 and P.1) in cohorts of COVID-19 patients ranging in disease severity (n = 69) and recipients of the Pfizer/BioNTech vaccine (n = 50). Spike binding and neutralization against all three VOC was substantially reduced in the majority of samples, with the largest 4-7-fold reduction in neutralization being observed against B.1.351. While hospitalized COVID-19 patients and vaccinees maintained sufficient neutralizing titers against all three VOC, 39% of non-hospitalized patients did not neutralize B.1.351. Moreover, monoclonal neutralizing antibodies (NAbs) show sharp reductions in their binding kinetics and neutralizing potential to B.1.351 and P.1, but not to B.1.1.7. These data have implications for the degree to which pre-existing immunity can protect against subsequent infection with VOC and informs policy makers of susceptibility to globally circulating SARS-CoV-2 VOC.

at the time of assessment as well as a control wild-type (WT) S protein from the Wuhan Hu-1 virus (GenBank: MN908947.3) isolated in December 2019 (6,24,25). Overall, the antibody responses against 124 each S protein were heterogeneous and differed up to ~1200-fold between the strongest and weakest 125 responders. The recognition of the three VOC by convalescent patients was significantly reduced 126 compared to WT by an average of 2.4-fold, 3-fold and 4-fold for B.1.1.7, B.1.351 and P.1, respectively (p < 127 0.0001 for all, Fig. 1, B and C). Binding titers elicited by the mRNA vaccine were more homogeneous than 128 those elicited by natural infection and differed ~10-fold between responders, with all participants having 129 half-maximal binding titers (ED50s) exceeding 10 3 , except for one poor responder (Fig. 1B). Since the 130 variability of binding titers in convalescent sera is considerable, we examined whether this variability was 131 related to severity of disease (i.e. hospital admission, Fig. 1D). We observed a highly significant ~8-fold 132 difference in ED50s between non-hospitalized patients and hospitalized patients, which is in line with 133 previous reports of WT SARS-CoV-2 binding titers correlating with severity of disease (p < 0.0001) (26, 134 27). This difference in binding titers between non-hospitalized patients and hospitalized patients was 135 consistent for all three SARS-CoV-2 VOC studied (Fig. 1D). When comparing immune responses of 136 vaccine recipients with convalescent sera, we observed similar S protein binding titers between vaccinee 137 and hospitalized patients, which are an average ~4 to 11-fold higher compared to non-hospitalized 138 patients for all VOC (Fig. 1D). Taken together, these data indicate that vaccine recipients as well as 139 COVID-19 patients exhibit reduced binding to S proteins of the currently circulating VOC, with hospitalized 140 patients and vaccine recipients exhibiting higher binding titers overall compared to non-hospitalized 141 patients. 142

SARS-CoV-2 VOC are substantially less sensitive to serum NAbs 144
We next tested the neutralizing activity of convalescent and vaccinee sera (Fig. 2, table S1). To this end, 145 we generated lentiviral-based pseudoviruses of the currently widespread SARS-CoV-2 D614G (WT) 146 variant as well as B.1.1.7, B.1.351 and P.1. We detected substantial neutralizing activity against the WT 147 virus (half maximal neutralization titer, ID50 > 100) in 96% of convalescent patients irrespective of 148 hospitalization, and in all vaccine recipients, except the one poor responder ( Fig. 2A). Indeed, the overall 149 binding titers correlated well with the neutralization titers for all VOC (fig. S1C). Non-hospitalized patients 150 had the most heterogeneous responses, with WT neutralizing titers differing by up to ~150-fold, while titers 151 against WT were much more homogeneous in hospitalized patients (up to ~20-fold difference) and 152 vaccine recipients (up to ~12-fold difference) ( Fig. 2A). Consistent with the binding results, we observed a 153 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted June 1, 2021. ; https://doi.org/10.1101/2021.05.26.21257441 doi: medRxiv preprint groups, the difference was most apparent against the B.1.351 VOC, showing a reduction of ~4-, 7-and 5-155 fold in neutralizing titers for non-hospitalized patients, hospitalized patients and vaccine recipients, 156 respectively (p < 0.0001, Fig. 2, A and B). The data were corroborated in an authentic SARS-CoV-2 157 neutralization assay ( fig. S2, A-D). We did not observe an effect of medication, including the antiviral drug 158 remdesivir (rdv) and the anti-inflammatory drug dexamethasone (dexa) on neutralization activity in sera 159 from convalescent but previously hospitalized patients ( fig. S2E). 160 Some sera, most notably those from non-hospitalized patients, showed a complete loss of 161 neutralization against some VOC, whereas they did neutralize WT pseudovirus ( Fig. 2A). When the sera 162 that reached the limit of detection against one of the VOC (i.e., ID50 < 100) were excluded, the fold 163 To obtain more insight into the antibody specificities that were affected by the mutations in the VOC we 187 assessed a panel of NAbs with known specificities isolated from convalescent patients, some of which 188 have since been characterized structurally, and some of which have been evaluated in preclinical 189 protection models (25, 29, 30). Bio-layer interferometry (BLI) experiments showed that for the majority of 190 RBD-targeting NAbs that span the four known epitope clusters on the RBD (31) the binding to B.1.351 and 191 P.1 S protein was reduced substantially, while binding to the B.1.1.7 S protein was mostly similar to WT 192 binding. This is consistent with observations that the only RBD mutation in B.1.1.7, i.e. N501Y, has not 193 been associated with escape from antibodies and was probably selected for increased ACE2 binding (8,  (Fig. 3A). Similarly, COVA2-15, which is highly potent against WT,198 showed decreased binding kinetics to S proteins from B.1.351 and P.1 compared to WT and B.1.1.7 S 199 proteins (Fig. 3A). We also tested binding of RBD-targeting, SARS-CoV cross-NAbs COVA1-16 and 200 COVA2-02, which showed highly similar binding to all VOC S proteins tested, indicating that these NAbs in 201 particular target a conserved epitope that has not been influenced by the mutations present in these three 202 VOC. This is in line with previous findings of COVA1-16 able to recognize RBDs of pangolin and bat 203 origin, indicating that COVA1-16 recognizes an epitope that is highly conserved among sarbecoviruses 204 Finally, we sought to confirm these findings in pseudovirus neutralization assays (Fig. 3B, fig. S4). In this study, we show that COVID-19 convalescent patients as well as mRNA vaccine recipients sampled 240 at the expected peak of their immunity, showed a marked decrease in binding and neutralization potency 241 against two of three VOC currently circulating (Fig. 2, A  (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted June 1, 2021. ; https://doi.org/10.1101/2021.05.26.21257441 doi: medRxiv preprint infection in vaccine recipients are starting to emerge (33). 248 To date, most studies have focused on a specific VOC or on a specific set of samples (i.e. vaccine 249 recipients, convalescent patients or NAbs) (4, 10, 16, 20). Here, through systematic comparison of non-250 hospitalized patients, hospitalized patients and vaccine recipients in the context of multiple VOC, we are 251 able to assess the impact of specific sets of mutations. We observed that serum binding titers were 252 impacted most by the P.1 VOC, while for neutralizing titers against B.1.351 showed the largest reduction 253 in all three groups studied ( Fig. 1C and 2B). This irregularity might be explained by non-neutralizing Abs (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted June 1, 2021. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted June 1, 2021. ; https://doi.org/10.1101/2021.05.26.21257441 doi: medRxiv preprint against VOC, there are some limitations. In this study, we present a vaccinee population that consists of 309 primary health care workers that are generally of middle age, with only four (8%) participants being older 310 than 60 (Table 1). However, we found no correlation between age and titers in convalescent patients 311 against any of the VOC tested. Moreover, we have focused on the neutralization potential of immune sera 312 and did not examine antibody effector functions which have been implicated previously in the control of 313 SARS-CoV-2 infection (41). Similar to other studies, our convalescent and vaccinee sera samples were 314 collected four to six weeks after infection and four weeks after vaccination (i.e. at the peak of immunity), 315 respectively. While we present findings that may have implications for additional booster vaccines and the 316 monitoring of VOC, we did not examine the aspect of waning antibody titers and how those might 317 influence VOC binding and neutralization. Finally, we have merely examined an early line of defense 318 against infection with SARS-CoV-2 (i.e. serum antibody levels), while we expect memory B cells to play an 319 important part in re-infections with SARS-CoV-2 VOC. Indeed, swift re-activation of memory compartments 320 may lead to reduced transmissibility, a milder course of disease and a more potent immune response. 321 In conclusion, we observed a substantial reduction in binding and neutralization potency against for us and whether VOC, when not brought under control, will evolve further and continue to escape from 328 humoral immunity induced by infection or vaccination. 329 All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Study design 331
The sera of 69 SARS-CoV-2 infected adults were collected four to six weeks after symptom onset through 332 (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Multiplex protein microarray 364
HCoV-PMA slides were produced as described previously (23) . WT and VOC (B.1.1.7, B.1.351  (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Monoclonal antibodies 415
The NAbs used in this study were isolated from participants in the "COVID-19 Specific Antibodies" 416 buffer (PBS, 0.02% Tween-20, 0.1% BSA) for 300 s. After the biosensors were washed in a well containing running buffer to remove excess protein, the biosensors were dipped in a well containing 30 424 µg/mL NAb in running buffer for 120 s to measure association. Next, the biosensors were moved to a well 425 containing running buffer for 120 s to measure dissociation of the S protein-NAb complexes. hospitalized patients, hospitalized patients and vaccine recipients against WT pseudovirus and each of the 726 VOC pseudoviruses. ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; *, p < 0.05; ns, not significant. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.