key: cord-0260799-6cx82cr5
authors: Andreano, Emanuele; Paciello, Ida; Pierleoni, Giulio; Piccini, Giulia; Abbiento, Valentina; Antonelli, Giada; Pileri, Piero; Manganaro, Noemi; Pantano, Elisa; Maccari, Giuseppe; Marchese, Silvia; Donnici, Lorena; Benincasa, Linda; Giglioli, Ginevra; Leonardi, Margherita; De Santi, Concetta; Fabbiani, Massimiliano; Rancan, Ilaria; Tumbarello, Mario; Montagnani, Francesca; Sala, Claudia; Medini, Duccio; De Francesco, Raffaele; Montomoli, Emanuele; Rappuoli, Rino
title: COVID-19 mRNA third dose induces a unique hybrid immunity-like antibody response
date: 2022-05-10
journal: bioRxiv
DOI: 10.1101/2022.05.09.491201
sha: a5f512044dd6eb303908066038f2f45ec0b3aacb
doc_id: 260799
cord_uid: 6cx82cr5

The continuous evolution of SARS-CoV-2 generated highly mutated variants, like omicron BA.1 and BA.2, able to escape natural and vaccine-induced primary immunity1,2. The administration of a third dose of mRNA vaccines induces a secondary response with increased protection. We investigated, at single-cell level, the longitudinal evolution of the neutralizing antibody response in four donors after three mRNA doses3. A total of 4,100 spike protein specific memory B cells were single cell sorted and 350 neutralizing antibodies were identified. The third dose increased the antibody neutralization potency and breadth against all SARS-CoV-2 variants of concern as previously observed with hybrid immunity3. However, the B cell repertoire that stands behind the response is dramatically different. The increased neutralizing response was largely due to the expansion of B cell germlines poorly represented after two doses, and the reduction of germlines predominant after primary immunization such as IGHV3-53;IGHJ6-1 and IGHV3-66;IGHJ4-1. Divergently to hybrid immunity, cross-protection after a third dose was mainly guided by Class 1/2 antibodies encoded by IGHV1-58;IGHJ3-1 and IGHV1-69;IGHJ4-1 germlines. The IGHV2-5;IGHJ3-1 germline, which induced broadly cross-reactive Class 3 antibodies after infection or viral vector vaccination, was not induced by a third mRNA dose. Our data show that while neutralizing breadth and potency can be improved by different immunization regimens, each of them has a unique molecular signature which should be considered while designing novel vaccines and immunization strategies.

vaccines induces a secondary response with increased protection. We investigated, at single-cell level, the 23 longitudinal evolution of the neutralizing antibody response in four donors after three mRNA doses 3 . A total 24 of 4,100 spike protein specific memory B cells were single cell sorted and 350 neutralizing antibodies were 25 identified. The third dose increased the antibody neutralization potency and breadth against all SARS-CoV-2 26 variants of concern as previously observed with hybrid immunity 3 . However, the B cell repertoire that stands 27 behind the response is dramatically different. The increased neutralizing response was largely due to the 28 expansion of B cell germlines poorly represented after two doses, and the reduction of germlines 29 predominant after primary immunization such as IGHV3-53;IGHJ6-1 and IGHV3-66;IGHJ4-1. Divergently to 30 hybrid immunity, cross-protection after a third dose was mainly guided by Class 1/2 antibodies encoded by 31 IGHV1-58;IGHJ3-1 and IGHV1-69;IGHJ4-1 germlines. The IGHV2-5;IGHJ3-1 germline, which induced broadly 32 cross-reactive Class 3 antibodies after infection or viral vector vaccination, was not induced by a third mRNA 33 dose. Our data show that while neutralizing breadth and potency can be improved by different immunization 34 regimens, each of them has a unique molecular signature which should be considered while designing novel 35 vaccines and immunization strategies. 36

The emergence of SARS-CoV-2 variants of concern (VoCs) able to escape vaccine immunity elicited by the 38 spike (S) protein of the original virus isolated in Wuhan, China, have decreased the impact of vaccination 4,5 . 39

As a consequence, the COVID-19 pandemic continues to impact the health, the economy and the freedom 40 of people worldwide in spite of the several billion doses of vaccines already deployed. This scenario raises 41 new important questions about the use of the existing vaccines and the immunity they can provide to tackle 42 current and future variants. Therefore, it became of utmost importance to understand the nature and the 43 quality of the immune response elicited by the different vaccines used worldwide. In our previous study we 44 analyzed at single cell level the immune response induced by two doses of the BNT162b2 mRNA vaccine in 45 naïve people and in people that had been previously infected by the SARS-CoV-2 virus 3 . In this study we 46 analyzed at single cell level the longitudinal B cell and neutralizing antibody response of the same naïve 47 people after a third immunization. We found that, while the overall immune response after a third dose is 48 similar to that observed in the hybrid immunity of vaccinated people previously infected by the virus, the B 49 cell repertoire that stands behind the response is dramatically different. 50

To evaluate the longitudinal evolution of the neutralizing antibody response, four seronegative donors that 54 participated in our previous study after two doses of BNT162b2 mRNA vaccine 3 , were re-enrolled after 55 receiving a third dose. None of the subjects were exposed to SARS-CoV-2 infection between the second and 56 third vaccination dose. Data after the third dose (seronegative 3 rd dose; SN3) described in this study were 57 compared to those obtained from the same subjects after the second dose (seronegative 2 nd dose; SN2) and 58 to those of subjects with hybrid immunity (seropositive 2 nd dose; SP2) previously described 3 . Three subjects 59 received the BNT162b2 (VAC-001, VAC-002 and VAC-008) vaccine, while one subject (VAC-010) received the 60 mRNA-1273 vaccine. Blood collection occurred at an average of 58 days post third vaccination dose. Subject 61 details are summarized in Extended Data Table 1 . The frequency of CD19 + CD27 + IgD -IgMmemory B cell 62 6 in Class 3 and 4 nAbs compared to SN2 and a similar distribution to SP2. When we compared the GM-IC100 of 115 classes of nAbs in the SN3 with the SN2 and SP2 groups, we observed similar neutralization potency when 116 tested against the SARS-CoV-2 virus originally isolated in Wuhan, China (Extended Data Fig. 5) . Following, we 117 aimed to understand which classes of RBD (n = 154) and NTD (n = 43) targeting nAbs in SN3 were mainly 118 responsible for the cross-protection against the SARS-CoV-2 VoCs. Overall, we observed that Class 1/2 nAbs 119 are the most abundant family of antibodies against all tested variants, followed by Class 3, Not-competing, 120 NTD and Class 4 nAbs (Fig. 2b-c) . Finally, we compared the functional antibody response between SN3 and 121 SP2 against the highly mutated omicron BA.1 and BA.2 viruses. Interestingly, we observed that a third mRNA 122 booster dose increases neutralization potency and evasion resistance of Class 1/2 nAbs against both omicron 123 BA.1 and BA.2 compared to SP2. The opposite trend was observed for Class 3 nAbs, which had a lower 124 frequency of cross-protection in SN3 compared to SP2 (Fig. 2d-e) . 125 126 Antibody gene repertoire 127

We then interrogated the functional antibody repertoire. Initially, we analyzed all immunoglobulin heavy 128 chain sequences retrieved from the three different groups (SN2 n = 58; SN3 n = 288; SP2 n = 278), and their 129 respective V-J gene rearrangements (IGHV;IGHJ) 3 . Interestingly, SN2 and SN3 share only 20.2% percent 130 (23/114) of SN3 IGHV;IGHJ rearrangements, while SN3 and SP2 share up to 45.6% (52/114). In addition, we 131 observed that the frequency of antibodies encoded by predominant rearrangements induced by 2 132 vaccination doses (i.e. IGHV3-30;IGHJ6-1, IGHV3-33;IGHJ4-1, IGHV3-53;IGHJ6-1 and IGHV3-66;IGHJ4-1) 3,6,9,10 133 were reduced after a third booster dose, while we observed an expansion of the antibody germlines IGHV1-134 58;IGHJ3-1 and IGHV1-69;IGHJ4-1 which were previously found to be predominant in SP2 3,11 ( Fig. 3a;  135 Extended Data Fig. 6 ). These latter germlines previously showed high level of cross-neutralization activity 136 against SARS-CoV-2 VoCs 12-14 . In addition, we observed in one donor (VAC-001) an important expansion of 137 the germline IGHV1-46;IGHJ6-1 after receiving a 3 rd booster dose ( Fig. 3a; Extended Data Fig. 6) . Conversely 138 to what found in SP2 3 , we did not observe the expansion of the Class 3 targeting antibody germline IGHV2-139 5;IGHJ4-1, which so far has been observed only in previously infected vaccinees or subjects immunized with 7 adenoviral vectors 3,15,16 . Following, we aimed to identify expanded clonal families within the same four 141 donors after a second and third vaccination dose. Sequences (SN2 n = 43; SN3 n = 288) were clustered by 142 binning the clones to their inferred germlines (centroids) and according to 80% nucleotide sequence in the 143 heavy complementary determining region 3 (CDRH3). Clusters were defined as antibody families including at 144 least five or more members as previously described 17 . Of the 331 sequences, 226 (68.3%) were orphans (i.e. 145 did not cluster with other sequences), and only in six cases sequences from SN2 and SN3 were binned to the 146 same centroid (Fig. 3b) . Only five clusters were identified, three of which were composed by antibodies 147 belonging exclusively from the SN3 group. Of these clusters, two were formed by antibody sequences shared 148 between the SN2 and SN3 groups. The smallest cluster was composed by 11 antibody members encoded 149 from the IGHV1-58;IGHJ3-1 germline, while the biggest cluster, composed by 18 antibody members, were 150 encoded by the IGHV3-48/IGHV3-53/IGHV3-66 germlines (Fig. 3b) . Finally, we evaluated the V-gene mutation 151 levels, neutralization potency and breadth in nAbs encoded by reduced and expanded germlines following a 152 third vaccination dose. Our data showed that IGHV3-53;IGHJ6-1 and IGHV3-66;IGHJ4-1 germlines, expanded 153 in SN2 and reduced in SN3, have a similar level of V gene somatic mutations, are poorly cross-reactive 154 especially against omicron BA.1 and BA.2, and show medium neutralization potency (IC100 mainly between 155 100 and 1,000 ng ml -1 ). Differently, the antibody germlines IGHV1-58;IGHJ3-1 and IGHV1-69;IGHJ4-1, mainly 156 expanded in SN3 and poorly represented in SN2, show between 5 and 15-fold higher V gene somatic 157 mutation levels, higher neutralization potency and breadth against all SARS-CoV-2 VoCs (Fig. 3c-f) . 158

In agreement with other longitudinal studies 9,10,18 , we found that a third dose of mRNA vaccine induces an 161 immune response similar to the hybrid immunity observed in people vaccinated after SARS-CoV-2 infection. 162

This antibody response is characterized by a small increase in S protein binding antibodies, a strong increase 163 in neutralizing potency and a considerable increase in antibodies able to cross-neutralize emerging variants, 164 including omicron BA.1 and BA.2. The increased potency and breadth are due to a significant expansion of S 165 protein specific MBCs, which is even higher than that observed in subjects with hybrid immunity, and by a 166 8 strong increase of V gene somatic mutations. What is new in our study is that the increased potency and 167 breadth observed after a third booster dose was due mostly to Class 1/2 nAbs, while Class 3 antibodies had 168 a lower frequency and breadth compared to subjects with hybrid immunity. In addition, we found that a third 169 mRNA dose did not induce a strong response against the distantly related SARS-CoV-1, suggesting that 170 additional doses of homologous vaccines against SARS-CoV-2 will focus the antibody response against this 171 virus instead of broadening cross-protection to other coronaviruses. Another unique observation of this 172 study is that the increased neutralization potency and breadth is not due to a linear evolution of the B cells 173 producing nAbs after two vaccine doses but is due mostly to the expansion of new B cells which were not 174 detected after primary immunization. Indeed, the secondary response induced by a third vaccine dose did 175 not derive from expanded B cell clones, but it was dominated by singlets that constituted 68% of the entire 176

repertoire. In addition, the germlines IGHV3-53;IGHJ6-1/IGHV3-66;IGHJ4-1 which dominated the neutralizing 177 response in donors infected with the original Wuhan virus and in subjects immunized with two doses of 178 mRNA vaccines 3,6,19-21 , decreased in frequency and did not improve in potency or cross-neutralization after a 179 third dose. Conversely, the antibody germlines IGHV1-58;IGHJ3-1 and IGHV1-69;IGHJ4-1 became largely 180 responsible for the improved potency and cross-neutralization observed after a third dose. Interestingly, the 181 IGHV1-58 germline predominant in this group was previously shown to recognize a "supersite" on the S 182 protein surface and to be expanded following beta or omicron breakthrough infection in vaccinated 183 individuals 12,14,22 . A unique observation of this work is that the highly cross-reactive germline IGHV2-5;IGHJ4-184 1, found in subjects with hybrid immunity and in people vaccinated with viral vectors 3,16 , is absent after a 185 third mRNA vaccine, suggesting that the induction of this germline may require endogenous production of 

Peripheral blood mononuclear cells (PBMCs) and single cell sorting strategy were performed as previously 294 described 3,6 . Briefly, PBMC were isolated from heparin-treated whole blood by density gradient 295 centrifugation (Ficoll-Paque™ PREMIUM, Sigma-Aldrich) and stained with Live/Dead Fixable Aqua 296 (Invitrogen; Thermo Scientific) diluted 1:500. After 20 min incubation cells were saturated with 20% normal 297 rabbit serum (Life technologies) for 20 min at 4°C and then stained with SARS-CoV-2 S-protein labeled with 298

Strep-Tactin®XT DY-488 (iba-lifesciences cat# 2-1562-050) for 30 min at 4°C. After incubation the following 299 staining mix was used CD19 V421 (BD cat# 562440, 1:320), IgM PerCP-Cy5.5 (BD cat# 561285, 1:50), CD27 PE 300 (BD cat# 340425, 1:30), IgD-A700 (BD cat# 561302, 1:15), CD3 PE-Cy7 (BioLegend cat# 300420, 1:100), CD14 301 PE-Cy7 (BioLegend cat# 301814, 1:320), CD56 PE-Cy7 (BioLegend cat# 318318, 1:80) and cells were incubated 302 at 4°C for additional 30 min. Stained MBCs were single cell-sorted with a BD FACS Aria III (BD Biosciences) 303 into 384-well plates containing 3T3-CD40L feeder cells, IL-2 and IL-21 and incubated for 14 days as previously 304 described 24 . 305 306 ELISA assay with SARS-CoV-2 and SARS-CoV-1 S protein prefusion trimer 307 mAbs and plasma binding specificity against the S-protein trimer was detected by ELISA as previously 308 described 3 . Briefly, 384-well plates (microplate clear, Greiner Bio-one) were coated with 3 µg/mL of 309

Structural basis for potent antibody neutralization of SARS-CoV-2 variants including 221

Recall of pre-existing cross-reactive B cell memory following Omicron breakthrough 223 infection. bioRxiv

mRNA vaccination in octogenarians 15 and 20 months after recovery from 225

COVID-19 elicits robust immune and antibody responses that include Omicron

Broadly neutralizing antibodies against Omicron-included SARS-CoV-2 variants induced 227 by vaccination

Systematic analysis of human antibody response to ebolavirus glycoprotein reveals 229 high prevalence of neutralizing public clonotypes. bioRxiv

RT-PCR and expression vector cloning

Reconstructing a B-cell clonal lineage. I. Statistical inference of unobserved ancestors

Reconstructing a B-Cell Clonal Lineage. II. Mutation, Selection, and Affinity 459

streptavidin (Thermo Fisher) diluted in carbonate-bicarbonate buffer (E107, Bethyl laboratories) and 310 incubated at RT overnight. The next day, plates were incubated 1 h at RT with 3 µg/mL of SARS-CoV-2 or 311 SARS-CoV-1 S protein, and saturated with 50 µL/well of blocking buffer (phosphate-buffered saline, 1% BSA) 312for 1 h at 37°C. Following, 25 µL/well of mAbs or plasma samples, diluted 1:5 or 1:10 respectively in sample 313 buffer (phosphate-buffered saline, 1% BSA, 0.05% Tween-20), were added serially diluted step dilution 1:2 314 and then incubated at 1 h at 37°C. Finally, 25 µL/well of alkaline phosphatase-conjugated goat antihuman 315IgG and IgA (Southern Biotech) diluted 1:2000 in sample buffer were added. S protein binding was detected 316 using 25 µL/well of PNPP (p-nitrophenyl phosphate; Thermo Fisher) and the reaction was measured at a 317 wavelength of 405 nm by the Varioskan Lux Reader (Thermo Fisher Scientific). After each incubation step, 318 plates were washed three times with 100 µL/well of washing buffer (phosphate-buffered saline, 0.05% 319 Tween-20). Sample buffer was used as a blank and the threshold for sample positivity was set at 2-fold the 320 optical density (OD) of the blank. Technical duplicates were performed for mAbs and technical triplicates 321 were performed for sera samples. 322 323 ELISA assay with RBD, NTD and S2 subunits 324 mAbs identification and plasma screening of vaccinees against RBD, NTD or S2 SARS-CoV-2 protein were 325 performed by ELISA as previously described 3 . Briefly, 3 µg/mL of RBD, NTD or S2 SARS-CoV-2 protein diluted 326 in carbonate-bicarbonate buffer (E107, Bethyl laboratories) were coated in 384-well plates (microplate clear, 327Greiner Bio-one) and blocked with 50 µL/well of blocking buffer (phosphate-buffered saline, 1% BSA) for 1h 328 at 37°C. After washing, plates were incubated 1 h at 37 °C with mAbs diluted 1:5 in samples buffer 329 (phosphate-buffered saline, 1% BSA, 0.05% Tween-20) or with plasma at a starting dilution 1:10 and step 330 diluted 1:2 in sample buffer. Anti-Human IgG −Peroxidase antibody (Fab specific) produced in goat (Sigma) 331 37°C, 5% CO2. The mixture was then added to the wells of a 96-well plate containing a sub-confluent Vero E6 360 17 cell monolayer. Plates were incubated for 3-4 days at 37°C in a humidified environment with 5% CO2, then 361 examined for CPE by means of an inverted optical microscope by two independent operators. All nAbs were 362 tested a starting dilution of 1:5 and the IC100 evaluated based on their initial concentration while plasma 363 samples were tested starting from a 1:10 dilution. Both nAbs and plasma samples were then diluted step 1:2. To investigate the genetic similarity within and between lineages, a network map was built by representing 428 each clonal family with a centroid and connecting centroids sharing a similar sequence. The centroid 429 sequence was computed with Cloanalyst to represent the average CDRH3 sequence for each clonal family, 430 and Hamming distance was calculated for each antibody CDRH3 sequence to represent the relationship 431 within the clonal family. Levenshtein distance was calculated between each centroid representative of each 432 clonal family to investigate the relationship between clonal families. Levenshtein distance was calculated 433 with the R package stringdistm v0.9.8 (https://cran.r-project.org/web/packages/stringdist/index.html) and 434 normalized between 0 and 1. A network graph was generated with the R package ggraph v2.