key: cord-0758629-oj314x9t
authors: Farkash, Inbal; Feferman, Tali; Cohen-Saban, Noy; Avraham, Yahel; Morgenstern, David; Mayuni, Grace; Barth, Natasha; Lustig, Yaniv; Miller, Liron; Shouval, Dror S.; Biber, Asaf; Kirgner, Ilya; Levin, Yishai; Dahan, Rony
title: Anti-SARS-CoV-2 antibodies elicited by COVID-19mRNA vaccine exhibit a unique glycosylation pattern
date: 2021-11-24
journal: Cell Rep
DOI: 10.1016/j.celrep.2021.110114
sha: a8edf994672d5da50f8f691c99802ba36ea9986d
doc_id: 758629
cord_uid: oj314x9t

Messenger RNA-based vaccines against COVID-19 induce a robust anti-SARS-CoV-2 antibody response with potent viral neutralization activity. Antibody effector functions is determined by its constant region subclasses as well as by its glycosylation patterns, but their role in vaccine efficacy is unclear. Moreover, whether vaccination induces antibodies similar to those in patients with COVID-19 remains unknown. We analyze BNT162b2 vaccine-induced IgG subclass distribution and Fc glycosylation patterns and their potential to drive effector function via Fc-gamma receptors and complement pathways. We identify unique and dynamic pro-inflammatory Fc compositions that are distinct from those in patients with COVID-19 and convalescences. Vaccine-induced anti-spike IgG is characterized by distinct Fab- and Fc-mediated functions between different age groups and in comparison to antibodies generated during natural viral infection. These data highlight the heterogeneity of Fc responses to SARS-CoV-2 infection and vaccination and suggest that they support long-lasting protection differently.

The outbreak of the COVID-19 pandemic, with its associated tremendous health and economic consequences, has sparked a worldwide effort to design vaccines using different platforms. Messenger RNA (mRNA)-based vaccines have demonstrated high efficacy in reducing infection, symptomatic disease and hospitalization caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (McDonald et al., 2021) . These vaccines elicit high titers of virus-neutralizing antibodies in the vast majority of vaccinated individuals (Goel et al., 2021; Wang et al., 2021) . This response is dominated by the immunoglobulin G (IgG) isotype, and serum levels of neutralizing IgG induced by the vaccine were correlated with its efficacy in preventing SARS-CoV-2 infection (Baden et al., 2020; Polack et al., 2020 ). Yet, the generated IgG Fc structures of the vaccine-induced antibodies, their contribution to vaccine efficacy, and whether they induce similar effector function to viral-induced immunoglobulin, remains unclear.

The response of IgG antibodies to infection and vaccination is elicited by two functional domains.

Whereas the variable antigen-binding fragment (Fab) domain confers their antigen binding specificity (Sela-Culang et al., 2013) , the constant crystallizable fragment (Fc) domain determines their effector function. The latter is achieved by engagement of this domain with Fcγ receptor (FcγR) pathways to activate innate and adaptive immune responses, including cross-presentation of antigens for activation of T cells, antibody-dependent cell-mediated phagocytosis (ADCP), antibody-dependent cellular cytotoxicity (ADCC), as well as complement-dependent cytotoxicity (CDC) Lu et al., 2018) . Fc function and, particularly, Fc-FcR interactions are important for the activity of neutralizing anti-SARS-CoV-2 monoclonal antibodies (mAbs), which have been developed to prevent or treat COVID-19 (Winkler et al., 2021; Yamin et al., 2021) . However, studies on COVID-19 vaccine response have mostly focused on Fab-mediated viral binding and neutralization and, consequently, Fc characteristics and function in vaccine-induced IgG remain poorly known. In addition to Fc-mediated elimination of infected cells and protective antiviral inflammation, modification of Fc structure and, particularly, of its glycan composition can affect IgG generation and the quality of Fabmediated neutralization, as was previously shown for seasonal influenza vaccination (Wang et al., 2015a) . It is unclear if such Fc glycan modifications occur in response to the novel mRNA-based anti-COVID-19 vaccine and if so, how they affect the response to the vaccine.

Another question that remains open is how the IgG Fc response elicited by the mRNA vaccine compares to the immune response occurring in individuals who were naturally infected with SARS-CoV-2. Patients with severe, but not mild, COVID-19 were reported to have a unique pro-inflammatory IgG signature during the early days post-infection, which was characterized by elevation in afucosylation of Fc glycans in IgG1 antibodies. This Fc modification resulted in increased IgG1 binding to FcRIIIA expressed on monocytes and macrophages and in subsequent release of inflammatory cytokines, which might have contributed to the development of pneumonia in these critically ill patients (Chakraborty et al., 2021a; Larsen et al., 2021) . Nevertheless, the posttranslational modifications occurring in IgG Fc domains of patients infected with SARS-CoV-2, convalescent and vaccinated individuals and the role of this domain in the immune response against this disease have yet to be fully elucidated.

To address these questions, we performed an in-depth analysis of longitudinal IgG Fc response to the mRNA vaccine. To that end, we established a cohort of SARS-CoV-2 naïve individuals aged 24-94 years who received two doses of SARS-CoV-2 mRNA BNT162b2 vaccine. We characterized Fc structure and function of anti-spike receptor binding domain (RBD) SARS-CoV-2 specific IgG generated by the two-dose vaccination regime over a 5 weeks period, and compared them to IgG from COVID-19 convalescences, mild and severe patients. The results provide new insights into the IgG response to SARS-CoV-2 mRNA vaccine and infection.

To explore the involvement of the Fc domain in IgG response to COVID-19 vaccine, we collected blood samples from 131 individuals with no evidence of prior SARS-CoV-2 infection 2 weeks after the first vaccine dose with BNT162b2 ("pre-boost") and 2 weeks after the second vaccine dose ("postboost") ( Figure. 1A, Table 1 ). Pre-vaccine samples were available from 23 individuals in this cohort.

To characterize Fab-mediated IgG activity, we first analyzed serum titer of antibodies binding to the SARS-CoV-2 RBD, which correlates to and predicts the neutralization activity of the antibody (Wajnberg et al., 2020; Wu et al., 2020) . Two weeks after the first vaccine dose, 50 of the 131 participants (38.2%) were found to be positive for the presence of RBD-binding IgGs. One participant J o u r n a l P r e -p r o o f was also positive for anti-N SARS-CoV-2 protein IgG, and therefore was suspected as COVID -19 convalescence and excluded from the study. In line with previous reports (Sahin et al., 2020) , 2 weeks after the second vaccine dose, we observed a significant increase in IgG binding to RBD and 127 of the 130 individuals were RBD-positive ( Figure. 1B) . The three participants that did not develop postboost anti-RBD IgG response, presumably due to medical history associated with immune deficiencies, were excluded from further analysis (for more details, see Material and Methods). A strong correlation between pre-boost and post-boost anti-RBD IgG levels indicates that the early IgG response is predictive of the IgG levels induced by the two-dose vaccination ( Figure.1C ).

To determine the IgG Fc response to the vaccine, we first analyzed the subclass composition of serum anti-RBD IgGs (Figures. 1D, S1A). The initial, pre-boost IgG response was mediated by significant elevation of IgG1 and IgG3 levels. At post-boost, serum binding levels of all IgG subclasses were elevated, as IgG1 became the predominant subclass, followed by IgG3 and IgG2 and a negligible IgG4 response. Pre-boost anti-RBD levels of both IgG1 and IgG3 were significant predictors of post-boost anti-RBD titers (Figure. S1B). The overall dominance of IgG3 and IgG1 responses supports a strong Fc-mediated effector function. These two subclasses are characterized by high binding affinity to the activating FcγRs and the classic complement pathway, as compared to the lower affinity of IgG2 and IgG4 (Bruhns and Jönsson, 2015) . Moreover, the ratio between higher affinity (IgG1 + IgG3) and lower affinity (IgG2 + IgG4) subclasses in the anti-RBD response increased from the pre-boost to post-boost time points (Figures. 1E, S1A) . This indicates that the two-dose mRNA vaccine elicits an IgG subclass trajectory consistent with a pro-inflammatory Fc response.

The FcγR and complement binding properties of a given IgG are dictated by the combination of its IgG subclass and the linked post-translational Fc glycosylation (S et al., 2017) . We therefore analyzed Fc glycoform composition of vaccine-induced anti-SARS-CoV-2 IgGs. For that, we isolated SARS-CoV-2 specific IgGs from pre and post-boost serum samples of 39 individuals who were positive for anti-RBD IgG response at pre-boost examination, as well as from additional post-boost samples of 19 ageand sex-matched individuals that were negative at pre-boost (Figure. S1C). The IgG CH2 domain has a conserved glycan attached at the N297 position, which is composed of a core heptasaccharide structure that can be supplemented by additional saccharide units, namely fucose, N-acetylglucosamine We observed a significant increase in IgG1 fucosylation and sialylation and a decrease in bisecting GlcNAc modification ( Figure 1G ) from the pre-boost to the post-boost time points. Previously described post-translational modifications in response to influenza and tetanus vaccines include increased galactosylation (Selman et al., 2012; Wang et al., 2015b) . Therefore, the increased fucosylation and decreased bisection represent an antigen-specific Fc glycosylation kinetics that is unique to SARS-CoV-2 mRNA vaccination. Interestingly, increased sialyation seems to be a feature of both this mRNA-based vaccine and the mentioned tetanus and influenza vaccines. However, due to the temporally dynamic nature of the glycosylation modifications and the fact that different time points were analyzed (influenza, 0, 3 and 8 weeks post-vaccination; tetanus, 0, 2 and 4 weeks post vaccination), the results of this comparison should be interpreted with care. In addition, whereas the influenza vaccine may elicit a recall response to a previous viral infection, the response to tetanus vaccine in children is more likely naïve, similar to the response to SARS-CoV-2 vaccine. Fc sialyation was previously reported to play a role in affinity maturation of antigen-specific IgG1 generated by influenza vaccine (Wang et al., 2015b) . Interestingly, we observed that the elevation in IgG1 sialylation, but also in fucosylation and galactosylation, correlated with higher anti-SARS-CoV-2 IgG titers ( Figure S1F ). This result supports a role for these post-translation modifications in driving a robust antiviral IgG response upon SARS-CoV-2 mRNA vaccination. In addition, we detected changes in glycosylation patterns in IgG2 & IgG3 that were not reported for other vaccines in humans (Supp.

While the functional significance of IgG Fc glycosylation is mostly known for the IgG1 subclass, due to the robust IgG3 response induced by the mRNA vaccine, we analyzed Fc glycosylation of all IgG subclasses (Figure. S1D). Interestingly, the levels of vaccine-induced anti-SARS-CoV-2 IgG3 fucosylation decreased over time, as opposed to the increase in IgG1 fucosylation. This indicates a unique regulation process of fucosylation for each IgG subclass. In contrast, IgG3 sialylation increased in anti-SARS-CoV-2 antibodies over time and in comparison to total IgG, as observed in IgG1. The trends of the rest of the IgG3 glycan modifications, as well as of all glycan alterations in anti-RBD IgG2, which appeared only post-boost, were similar to what we observed in IgG1 but did not reach statistical significance in our cohort. To our knowledge, previous studies of other vaccines in humans did not analyze Fc glycosylation patterns in non-IgG1 subclasses; thus, structural comparison of IgG2/3

Fc between SARS-CoV-2 mRNA and other vaccinations is currently lacking.

To examine how the dynamic changes in the IgG Fc structures translate to potential effector functions, we examined the FcγRs and complement component 1q (C1q) binding profiles of vaccine-induced IgG ( Figures 1H, S1E ). To determine pure Fc binding properties without the effect of the IgG titer, we corrected for the number of RBD-binding IgGs and characterized similar quantity of IgG samples. The effector function potency of a given IgG response is dictated by the balance between activating and inhibitory FcγR signaling it induces and is directly correlated with the activating/inhibitory (A/I) FcγR binding ratio (Anthony and Nimmerjahn, 2011; Nimmerjahn et al., 2015) . We therefore analyzed the A/I binding ratio of vaccine-induced anti-SARS-CoV-2 RBD IgG and identified significant increase from pre-to post-boost time points ( Figure 1H ). This kinetic change in FcγR binding intensity over time was not specific to any individual FcγR ( Figure S1E ), but seemed to be dominated by decreased binding to the inhibitory FcγRIIB post-boost. This highlights the complexity of IgG Fc-FcγR interactions, which can be dictated by the heterogeneous combination of subclasses and Fc glycoforms.

For instance, FcγRIIIA binding, which is expected to increase due to the robust post-boost increase in IgG1 and IgG3 response, can also be attenuated due to the relative decreased portion of afucosylated IgG1 and relative levels of IgG3 in the post-boost IgG pool. These opposite effects on IgG binding intensity to FcγRIIIA can neutralize each other, resulting in similar pre-boost vs. post-boost binding levels. Moreover, there was decreased binding to C1q in the post-boost time point. Overall, our data reveal unique kinetics of Fc structure generation over the 5 weeks following vaccination with BNT162b2. Moreover, these results suggest that the generated anti-RBD SARS-CoV-2 antibodies acquire increased capability to engage FcγR pathways, as a potential mechanism to eliminate infection in addition to their Fab-mediated neutralization activity.

Immune ageing is associated with a diminished ability to mount an effective response against pathogens (Ciabattini et al., 2018) , and age is a significant risk factor for severe disease and mortality from COVID-19 (Shahid et al., 2020; Williamson et al., 2020) . To determine how age affects IgG response to BNT162b2 vaccine, we compared the Fab and Fc responses, namely subclass trajectory, Fc glycosylation and engagement of immune receptors, in young vs. elderly individuals. In line with previous reports (Bates et al., 2021; Jalkanen et al., 2021) , we found a strong correlation between age and vaccine-induced anti-RBD titers ( Figure 2A ). Because individuals aged over 60 years are at increased risk for severe outcome from COVID-19 (Shahid et al., 2020) and were given priority for J o u r n a l P r e -p r o o f COVID-19 vaccination, we used this age as a cutoff in analyzing the vaccine response. Indeed, individuals aged over 60 years had lower anti-RBD SARS-CoV-2 titers in both pre-and post-boost time points, more so in the pre-boost time-point ( Figure 2B ). Whereas in individuals under 60 years old a combined IgG1-IgG3 response was observed 2 weeks after the first vaccine dose, the response in the older population was significantly dominated by IgG3 with negligible involvement of other subclasses, implying a delayed switch from IgG3 to IgG1 subclass ( Figures 2C, S3A ). However, in the post-boost time point, IgG1 became the predominant subclass in both age groups. At both time points, the individuals in the >60 years old group displayed decreased (IgG1+ IgG3)/(IgG2+IgG4) ratios as compared to the under 60 years individuals, due to lower IgG1 and IgG3 levels but similar IgG2 response ( Figures 2D, S3A ).

Ageing is associated with an inherent alteration in Fc glycosylation patterns (Chen et al., 2012) , which was observed when we compared total IgG glycosylation patterns between age groups ( Figure 2E ).

Decreased IgG1 sialyation and galactosylation and elevated IgG1 bisection were found in total IgG as well as anti-RBD IgG1 in the pre-boost examination among individuals who had an IgG1 response at that time point. At the post-boost time point, these typical age-dependent differences were attenuated and elevated IgG1 RBD fucosylation was the only significant Fc glycosylation pattern unique to the older population. IgG3 Fc glycosylation displayed similar trends where the initial IgG response resembled that of the inherent age-related differences, which was subsequently attenuated in the postboost time point. No age-dependent in differences IgG2 Fc glycosylation were observed at any time point. ( Figure S3B ). Given the differences in subclass trajectories and Fc glycosylation we observed in the older age group, we proceeded to compare binding profiles to FcγRs and C1q ( Figure 2F ). At pre-boost the elderly population displayed elevated binding to all the FcγRs and C1q. This finding could be explained by the predominance of IgG3 at this time point in the elderly (Vidarsson et al., 2014) . In the post-boost time point, decreased binding to FcγRIIb was the only significant difference, causing an overall increase in A/I ratio of RBD-specific IgGs in the elderly population. To determine whether this response was unique to the newly formed IgGs or an overall trend in elderly individuals, we compared the FcR binding capacity of total IgGs ( Figure S3C ) and found no baseline differences between the populations.

Overall, these results reveal distinct Fab and Fc functional properties of anti-RBD SARS-CoV-2-J o u r n a l P r e -p r o o f elicited IgG among vaccinated adults aged >60 years. Despite their reduced Fab-mediated anti-RBD titer and neutralization activity (Bates et al., 2021) , the effectiveness of mRNA vaccination in this age group seems to be similar to the effectiveness in younger adult populations (Baden et al., 2020; Polack et al., 2020; Tenforde, 2021) . Our data suggest that this could be explained by increased pre-boost Fc engagement of complement and post-boost activation of FcγR pathways, which may represent distinct levels of IgG-mediated protective mechanism of older adults. However, this aspect needs to be further characterized by functional studies, such as ADCP/ADCC assays.

There are known differences between the sexes in the frequency and intensity of inflammation and immune-related diseases. Furthermore, male patients with COVID-19 tend to develop a more severe disease outcome (Takahashi et al., 2020) and display lower anti-RBD titers compared to female patients (Peckham et al., 2020) . Hence, we proceeded to examine gender-related IgG response to the vaccine.

We did not find significant differences in anti-RBD IgG titers, IgG subclass distribution or Fc glycosylation patterns, or in Fc engagement of downstream receptors ( Figure. S4 ).

To compare IgG trajectories between vaccination-induced response and natural SARS-CoV-2 infection, we analyzed the responses in our vaccinated cohort as well as those of convalescent, mild and severe patients with COVID-19 at similar time points, as measured from either symptom onset or diagnosis (Table 2, Figure 3A ). At the 2-week time point, patients with severe COVID-19 had higher titers of RBD-binding IgGs as compared to both vaccinated individuals and patients with mild COVID-19. Differences between severe patients and vaccinated individuals were observed in levels of all IgG subclasses, whereas differences between severe and mild patients were mainly in IgG1 and IgG3 levels ( Figure 3B ). Unlike in vaccinated individuals, in patients with both mild and severe COVID-19 an initial IgG2 response was evident at the early time point and was then elevated. At the 2-week time point, the (IgG1+ IgG3)/(IgG2+ IgG4) ratio was higher in the severe patients as compared to both mild patients and vaccinated individuals. However, following the second vaccine dose, this ratio increased significantly over patients with both severe and mild COVID-19 at an equivalent time point from disease onset. Altogether, these data indicate that the anti-RBD IgG titer and Fab-mediated neutralization potential following the two vaccine doses are higher than in COVID-19 convalescences J o u r n a l P r e -p r o o f and similar to patients suffering from a severe disease, while the vaccine-induced subclass composition is enriched in pro-inflammatory IgGs, as compared to both convalescent and severe patients.

Next, we compared glycosylation patterns of RBD-specific and total IgGs in the vaccinated vs. patient cohort, at 5 weeks from diagnosis. In RBD-specific IgGs of patients with COVID-19, IgG1 Fc fucosylation was decreased as compared to vaccinated individuals, a trend that was correlated with disease severity ( Figure 3C ). This gradient between mild and severe patients is consistent with previous reports on the association of increased afucosylated anti-SARS-CoV-2 IgG early after infection and severe disease outcome (Chakraborty et al., 2020; Larsen et al., 2021) . Similar trends distinguishing between vaccination and disease severities were observed in anti-RBD IgG1 sialyation and galactosylation. Together with distinct Fc glycan profiles of IgG2 and IgG3/4 ( Figure S5A ), these unique subclass-specific Fc glycosylation changes support the notion of different regulatory process in response to vaccine vs. acute infection.

We then compared the binding capacity of anti-RBD IgGs of vaccinated and patients with COVID-19

to FcγRs and C1q. At 2 weeks, patients with mild COVID-19 displayed attenuated binding of RBDspecific IgGs compared to vaccinated individuals. High variability and a small cohort size prevented us from determining whether at that time point, severe COVID-19 patients had increased potential to engage effector function, as would be expected given their subclass composition and enrichment of afucosylated IgG1. Post-boost vaccine-elicited RBD-specific IgGs were characterized by a trend toward increased engagement of effector function relative to COVID-19 patients, as indicated by both C1q binding capacities and FcγR A/I ratio.

While the mRNA-based vaccine elicits a response only against the viral spike protein, the immune response during SARS-CoV-2 infection is broader, targeting multiple viral antigens (Chakraborty et al., 2021b) . To evaluate IgG response of non-RBD binding antibodies in the sera of patients with COVID-19, we analyzed Fc structure and function of their total IgG. At 5 weeks post-diagnosis, there was a similar gradient of Fc glycan modifications in the total IgG from severe patients through to vaccinated individuals, as observed in RBD-specific IgGs. For instance, decreased galactosylation was observed in all IgG subclasses of severe, but not in mild/convalescent patients, as compared with vaccinated individuals ( Figure S5B ). Moreover, total IgGs of severe patients displayed elevated binding capabilities to FcγRs and C1q as compared to mild patients at all time points. As opposed to J o u r n a l P r e -p r o o f anti-RBD IgGs, total IgGs of the vaccinated cohort displayed lower binding capacities compared to patients, as well as lower A/I ratios. This indicates an increased systemic inflammatory state and a more comprehensive IgG response against a multitude of SARS-CoV-2 epitopes in patients with COVID-19, as opposed to the specific anti-spike response that is generated by the mRNA-based vaccine.

Overall, these results indicate that 2 weeks after the first vaccine dose or disease onset, patients with COVID-19 had higher anti-RBD IgG responses, increased binding to both activating and inhibitory 

Lastly, we examined whether severe side effects to the vaccine were predictors of IgG response. For binding to activating FcRs after the second dose. In addition to the increased antibody neutralization activity that is generated from vaccine prime to boost, we report here an increased FcR engagement potential of the Fc domain of generated antibodies. Upon post-vaccine exposure to SARS-CoV-2, these Fc structures in the generated antibody-virus immune complexes (ICs) can affect the phagocytic activity of these complexes. Thus, even when neutralizing IgG levels are highly effective, the overall FcR engagement potential will dictate the intensity of IC clearance and of the sequential antiviral immune responses, including cross-presentation to prime and activate T cell response and elimination of infected host cells.

We observed age differences in vaccine-induced IgG Fab and Fc structures and receptor engagement properties. The vaccine-primed IgG subclass trajectory we identified was characterized by an initial IgG3 response, mainly in individuals over 60 years old, whereas in younger individuals, IgG1 response was also generated at the pre-boost time point. The delay in IgG3 to IgG1 switch, with associated increased binding capacities to FcγRs and C1q that we observed in older participants could be related to immune ageing, which could also be relevant in the context of inflammatory-mediated acute illnesses such as COVID-19. Moreover, reduced RBD binding and neutralization potential, but increased C1q binding pre-boost and A/I FcγR binding ratio post-boost characterized the vaccine-generated antibodies in older adults. These age-dependent differences in IgG characteristics may result in activation of distinct mechanisms of vaccine efficacy. Interestingly, slight reduction in vaccine efficacy for asymptomatic infection, but not for symptomatic and severe COVID-19, was reported in older adults (Haas et al., 2021; Jalkanen et al., 2021) . While the reduced neutralization activity in older adults could result in increased asymptomatic infection, increased Fc engagement may enhance SARS-CoV-2 clearance and compensate for the reduced IgG titer, resulting in similar levels of protection from symptomatic infection as in younger individuals with higher neutralizing IgG levels.

While the A/I FcγRs binding ratio was elevated from vaccine prime to boost, this kinetic change was not significant for any individual FcγR and seemed to be dominated by decreased binding to the (Vidarsson et al., 2014) . In our cohort, the ratios of IgG1 and IgG3 to IgG2 and IgG4 in patients with severe COVID-19

were significantly higher than in mild patients and vaccinated individuals at the early time point.

However, following the vaccine boost, these ratios in the vaccinated surpassed those of the severe patients, supporting the notion of vaccine-induced increased A/I FcγR engagement by IgG.

In summary, we identified unique and dynamic modifications in the IgG scaffold during antibody response to the BNT162b2 vaccine. These age-but not sex-dependent alterations in IgG subclasses, Fc glycosylation, and binding to immune receptors were distinct from those we detected in COVID-19 convalescents and patients and from previously reported responses to other types of vaccines. We suggest that alongside the serum levels and neutralization activity of the antibodies, these Fc features may affect the quantity, quality and mode of protection achieved by COVID-19 mRNA vaccine.

This study focuses on the structure of IgGs produced against the RBD domain of the SARS-CoV-2 virus. We did not address the humoral response mediated by non-IgG antibody isotypes. Moreover, we did not address the response by non-RBD binding IgGs, such as those that bind other epitopes of the spike protein (in the vaccinated cohort) or other proteins of the virus (in infected individuals).

Additional limitations arise from the methods used in this study. While we tried to provide a comprehensive characterization of the Fc glycoforms in all IgG subclasses, the mass spectrometrybased method used to analyze IgG Fc glycans cannot distinguish quantitatively between glycans attached to IgG3 and IgG4 subclasses. Therefore, glycan analysis results refer to these two subclasses together. However, when we assessed the RBD-IgG subclass response with ELISA, we did not detect a significant IgG4 response. Thus, we assumed negligible IgG4 levels in our samples and, throughout the text, we refer to the glycosylation data of IgG3/4 as IgG3 glycosylations. When we assessed antibody responses by ELISA, we could not distinguish between quantities of IgGs produced as opposed to higher binding affinities of IgGs. Another limitation was the relatively small cohort of patients, especially in the early time point, which reduced the power of the statistical analysis and our ability to draw conclusions from it. Lastly, while we were able to provide structural information on the produced IgGs and determine their Fc receptor binding capabilities, given the low affinity of monomeric IgGs to FcRs and the low sensitivity of the assay, it is possible that the effects were underestimated. A functional assay such as ADCC/ADCP might add another aspect to the importance of the presented findings. Figure S1 ). Data are presented as scatter plots indicating individual measurements (dots); black line represents the mean; error bars represent Standard Deviation (SD). Unless otherwise mentioned, unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Non-parametric Spearman's correlation was used, n=127.

B -Pre-and post-boost anti-RBD IgG titers when using a cutoff of 60 years of age (age≤ 60, n=60; age >60, n=67). Dotted line depicts threshold for positivity. C Pre-and post-boost anti-RBD IgG subclass distributions by age. Fold change from baseline anti-RBD IgG levels were determined as described above. (age≤ 60, n=60; age >60, n=67). D -Age dependent pre-and post-boost (IgG1+IgG3)/(IgG2+IgG4) ratios of anti-RBD IgG (age≤ 60, n=60; age >60, n=67). E -Agedependent Fc glycosylation patterns of IgG1 among vaccinated individuals. Levels are compared for the total IgGs produced at the pre-boost time point (age≤ 60, n=27; age>60, n=32), for individuals who had an IgG1 response at pre-boost (age≤ 60, n=9; age>60, n=3) and for individuals who had an IgG1 response at post-boost (age≤ 60, n=24; age>60, n=15). F -Age-dependent binding activity of RBD-J o u r n a l P r e -p r o o f specific IgGs to FcγRs and C1q. Binding to each receptor was determined at pre-boost (age≤ 60, n=22; age>60, n=15) and post-boost (age≤ 60, n=32, age>60, n=27). Ratios between RBD-specific IgG binding to activating inhibitory receptors were determined as described above. See also Figure S3 . Data are presented as scatter plots indicating individual measurements (dots); black line represents the mean; error bars represent SD. Unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

The circular bar plots depict the mean percentile of each feature ranging from 0-1.

vaccine dose. Anti-RBD total IgG and subclass composition were determined for the indicated serum samples. Symptom severity was determined by the participants and was assigned as no/mild side effects and severe side effects groups. No side effects, n=114; severe side effect, n=13. See also Figure   S6 . Data are presented as scatter plots indicating individual measurements (dots); black line represents J o u r n a l P r e -p r o o f the mean; error bars represent SD. Unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. 

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Rony Dahan (rony.dahan@weizmann.ac.il).

This study did not generate new unique reagents.  This paper does not report original code  Any additional information required to reanalyze the data reported in this work paper is available from the Lead Contact upon request.

This study was designed to investigate the overall IgG response, IgG subclass distribution, subclassspecific Fc glycosylation patterns and effector functions of IgGs produced following administration of the SARS-CoV-2 mRNA-based BNT162b2 vaccine and to compare these to the response to natural infection with SARS-CoV-2. For this purpose, we analyzed plasma samples that had been collected independently from adults (aged ≥ 18 years) at four centers. Samples were obtained at the Sheba and Tel Aviv medical centers and at a motel designated for patients with mild COVID-19 in quarantine.

Blood from the vaccine cohort was collected at the Weizmann Institute of Science. Samples were coded and blinded and the relevant review boards in each institute (see below) approved this study.

The Pfizer-BioNTech COVID-19 vaccine, also known as tozinameran or BNT162b2, co-developed by BioNTech and Pfizer, is based on mRNA products that encode a genetically modified SARS-CoV-2 spike protein. The vaccination campaign in Israel started in December 2020.

Vaccine cohort individuals volunteered of them 127 were included in the study according to these criteria.

We performed screening for the presence of N-binding IgGs and found one volunteer with high binding titers, as well as high RBD-binding IgGs following the first vaccine. This volunteer was excluded from further analysis due to potential past infection with SARS-CoV-2. Of the remaining 130 volunteers, 127 were positive for the presence of anti-RBD antibodies at 5 weeks. Of the three that were negative, one had a history of chronic lymphocytic leukemia with known low immunoglobulin levels, one was on chronic replacement IVIG therapy following rituximab treatment for lymphoma 9 years earlier, and the third individual was an 86-year-old male with no known underlying medical condition. These individuals were excluded from further analysis.

We obtained blood samples from patients who had recovered from COVID-19 as well as de-identified samples from active patients at diverse clinical stages of the disease (asymptomatic, mild to severe).

The samples and relevant clinical data were obtained from Tel Aviv Medical Sheba Medical Center 

The hexahistidine-tagged SARS-CoV-2 RBD construct was a kindly provided by gift from F. Krammer RBD production, the secreted protein was collected from the supernatant and affinity purified on HisPur Ni-NT resin (ThermoFisher). For the production of N protein, cells were transfected as above, and the cell palette was frozen overnight at -20°C. Frozen pellets were thawed on ice for 15-20 minutes and re-suspended in lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 0.5% Nonident P 40

Substitute (IGEPAL Sigma) and EDTA-free protease inhibitors) at 4°C on dry ice for 20-25 minutes and further incubated on a rotator at 4°C for 30 minutes. Next, the lysate was centrifuged at 13,000 x g, 4°C for 15 minutes to pellet debris and incubated with the HisPur Ni-NT resin.

Polypropylene columns (Bio-Rad Laboratories) were loaded with the supernatant-resin mixtures and then washed with wash buffer (50 mM NaH2PO4H2O, 30 mM NaCl, 20 mM imidazole) four times.

Columns were then eluted using elution buffer, which contained a high (235 mM) concentration of imidazole (Sigma-Aldrich). Four fractions were collected from each column by incubating the resin in the column with 3 mL of elution buffer for each fraction. Eluate was collected directly in a 50 mL falcon tube placed on ice. Purified proteins were dialyzed against PBS and sterile filtered (0.22 μm).

Purity was assessed by SDS-PAGE and Coomassie staining .

5-10 mL of blood were collected from volunteers/patients into EDTA-containing tubes, (BD Biosciences, San Jose, CA) and then centrifuged at 1200 rpm for 10 minutes. Plasma was collected and kept for further analysis. Samples originating from patients underwent an additional heat inactivation step and was heat-inactivated for 30 minutes at 56°C to avoid potential viral spread.

Antibodies were purified from the obtained plasma by protein G Sepharose 4 Fast Flow beads (GE

Healthcare) according to the manufacturer's protocol. In short, 2 mL of beads were washed in PBS, resuspended with up to 15 mL of plasma and rotated overnight at 4°C. Next, the plasma-bead mix was transferred onto a column (Bio-Rad). Beads were then washed by several bead volumes until runthrough was clear of protein, as determined by a 260/280 ratio measured by a Nanodrop spectrophotometer. Elution of IgGs was obtained with Elution Buffer (ThermoFisher) into 10% 1 M Tris (pH 9.0) in tubes. Purified antibodies were concentrated, quantified and dialyzed in PBS and sterile filtered (0.22 μm) and kept at 4°C for further analysis. These IgGs were further analyzed via either ELISA and/or mass spectrometry (see below).

Pierce™ NHS-Activated Magnetic Beads (Thermo fisher) were used to purify antigen (RBD) specific IgG from total IgG. For column RBD immobilization, 10 µg of RBD were bound to 25 µL NHS magnetic bead slurry according to the manufacturer's protocol. Each column was incubated with up to 6 mg of total IgGs from each sample and subsequently eluted according to the manufacturer's protocol.

RBD-specificity and purity of the eluted IgG was verified by RBD ELISA as described below. These RBD specific IgGs were further analyzed via either ELISA and/or mass spectrometry (see below).

Plasma samples were screened for binding to both RBD and N proteins as in , with small modifications. 96-well half-area ELISA microplates (Greiner Bio-One) were coated with 20 µl of RBD or N protein at 1 μg/ml/well in PBS and incubated at 37°C for 1 hour. All sequential steps were performed at room temperature. Plates were washed (PBS/Tween 0.05%) and 100 μl blocking solution (2% FCS/PBS) was added for 2 h. Then, samples from either vaccinated individuals or heat-inactivated plasma samples from patients with COVID-19 were diluted, first at 1:100 and then serially in blocking solution 1:4. 20 µl of the serially diluted samples were added to each well and incubated for 2 h. After washing, plates were incubated for 1 hour with horseradish 1:2500 peroxidase- in their anti-RBD/N titer were used as controls to standardize the reaction in all tested plates. Additional modified ELISA protocols, which are described in the following subsections, were all based on the above protocol.

To assess the serum distribution of RBD binding antibody subclasses, plate coating was done with RBD 2 µg/ml/well, and 2% FCS/PBS was used as blocking solution. The following mouse anti-human IgG secondary antibodies were used: IgG1 1:4000 (Southern biotech) and IgG3 1:500 (ThermoFisher).

Serum was initially diluted 1:25 and then serially diluted 1:2. For IgG2 1:500 (Southern biotech,) and

IgG4 1:500 (Southern biotech), serum was initially diluted 1:10 and then serially diluted 1:2.Trajectories of IgG subclass distribution over time ( Figures. 1, 2 , and .S3) are presented as fold change from baseline. To calculate the fold change, we determined for each subclass the pre-vaccine baseline mean level.

To determine the binding of total IgGs and anti-RBD IgGs to the different FcγRs, the human recombinant receptors FcγRIIA, IIB, and IIIA (Sino Biological) were immobilized to half area plates at 2 µg\ml in PBS. Blocking solution contained 10% BSA. Initial concentrations of 50 µg/ml of total IgG and 10 µg/ml of anti-RBD IgG and were serially diluted 1:2. HRP-conjugated anti-human IgG was used at 1:2500. For each FcR assay, the optimal concentration that fulfilled linearity was determined and these values were used for plotting.

C1q (Sigma) was immobilized to half area plates (5 µg/ml/well) and 5% BSA/PBS was used as blocking solution. Initial concentrations of 50 µg/ml of total IgG and 10 µg/ml of anti-RBD IgG were serially diluted 1:2. HRP-conjugated anti-human IgG was used at 1:2500. 

Data were searched against the IgG glycan-carrying peptide sequences allowing for deamidation and Fc glycan database using Byonic (v4.0.1.) search engine, and quantified using Skyline (v20.2.0.343).

The following peptide sequences were used for subclass determination: IgG1, EEQYNSTYR; IgG2, differentiate the readouts of the IgG3 and IgG4 peptides, these are shown together as IgG3/IgG4 in all the figures.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol Y, 2019) partner repository with the dataset identifier PXD02932.

Each feature in the polar plots is drawn as a wedge indicating the mean percentile of that group ranging Figure 1 . A -Anti-RBD IgG subclass distribution following the first and second vaccine doses (N=127). IgG subclasses were determined for each individual serum sample. These levels were then used to calculate IgG1+IgG3/IgG2+IgG4 ratios. B -Correlation of post-boost anti-RBD overall IgG titers to pre-boost IgG response, by subclass. Non-parametric Spearman's correlation was used. C -Isolation of RBD-specific IgGs from plasma samples from vaccinated individuals. Total IgGs and anti-RBD IgGs were prepared from plasma samples obtained after each vaccine dose as described in Methods. To determine RBD binding or its absence, total IgGs, unbound flow-through IgGs and eluted RBD-specific IgGs were subjected to an RBD binding ELISA assay. Purification of representative sample is shown. D -Fc glycosylation patterns in anti-RBD IgG2 and IgG3/4 and in total IgGs (as reference) from vaccinated individuals. Because there was no anti-RBD IgG2 response following the first vaccine dose, only total IgGs (n=59) and post-boost anti-RBD IgG2 glycosylations (n=39) were detected and shown. Detected pre-boost IgG3 glycosylations, n=28. E -FcγR and C1q binding of RBD-specific IgGs from vaccinated individuals, a (pre-boost, n=39; post-boost, n=59). F -RBD-specific IgG1 glycosylation patterns associated with overall IgG response, as determined by the non-parametric Spearman's correlation. Shown are glycosylation patterns that were in significant correlation with overall anti-RBD IgG response. Preboost IgG1 fucosylation, n=28; post-boost IgG1 glycosylations, n=39. Data are presented as scatter plots indicating individual measurements (dots); black line represents the mean; error bars represent SD. Unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

A serological assay to detect SARS-CoV-2 seroconversion in humans

The role of differential IgG glycosylation in the interaction of antibodies with FcγRs in vivo

Efficacy and Safety of the mRNA-1273 SARS-CoV

Age-Dependent Neutralization of SARS-CoV-2 and P.1 (2021). Variant by Vaccine Immune Serum Samples

Diversification of IgG effector functions

Mouse and human FcR effector functions

Proinflammatory IgG Fc structures in patients with severe COVID-19

Proinflammatory IgG Fc structures in patients with severe COVID-19

Divergent early antibody responses define COVID-19 disease trajectories. BioRxiv

Human IgG Fc-glycosylation profiling reveals associations with age, sex, female sex hormones and thyroid cancer

Vaccination in the elderly: The challenge of immune changes with aging

Distinct antibody and memory B cell responses in SARS-CoV-2 naïve and recovered individuals after mRNA vaccination

Impact and effectiveness of mRNA BNT162b2 vaccine against SARS-CoV-2 infections and COVID-19 cases, hospitalisations, and deaths following a nationwide vaccination campaign in Israel: an observational study using national surveillance data

COVID-19 mRNA vaccine induced antibody responses against three SARS-CoV-2 variants

Afucosylated IgG characterizes enveloped viral responses and correlates with COVID-19 severity

Modulating IgG effector function by Fc glycan engineering

Beyond binding: antibody effector functions in infectious diseases

Comparative systematic review and meta-analysis of reactogenicity, immunogenicity and efficacy of vaccines against SARS-CoV-2. Npj Vaccines

FcR dependent mechanisms of cytotoxic, agonistic, and neutralizing antibody activities

Male sex identified by global COVID-19 meta-analysis as a risk factor for death and ITU admission

Signaling by Antibodies: Recent Progress

COVID-19 vaccine BNT162b1 elicits human antibody and TH1 T cell responses

The Structural Basis of Antibody-Antigen Recognition

The PRIDE database and related tools and resources in 2019: improving support for quantification data

Changes in antigen-specific IgG1 Fc Nglycosylation upon influenza and tetanus vaccination

COVID-19 and Older Adults: What We Know

Sex differences in immune responses that underlie COVID-19 disease outcomes

Effectiveness of Pfizer-BioNTech and Moderna Vaccines Against COVID

Among Hospitalized Adults Aged ≥65 Years -United States

SARS-CoV-2 mRNA vaccines induce persistent human germinal centre responses

IgG Subclasses and Allotypes: From Structure to Effector Functions

Robust neutralizing antibodies to SARS-CoV-2 infection persist for months

Anti-HA Glycoforms Drive B Cell Affinity Selection and Determine Influenza Vaccine Efficacy

Glycoforms Drive B Cell Affinity Selection and Determine Influenza Vaccine Efficacy

mRNA vaccine-elicited antibodies to SARS-CoV-2 and circulating variants

Factors associated with COVID-19-related death using OpenSAFELY

Human neutralizing antibodies against SARS-CoV-2 require intact Fc effector functions for optimal therapeutic protection

Neutralizing antibody responses to SARS-CoV-2 in a COVID-19 recovered patient cohort and their implications

Fc-engineered antibody therapeutics with improved efficacy against COVID-19

*** **

 Figure 1F . Detected attached Fc glycans by mass spectrometry. A list of glycoforms that were detected and quantified for each IgG subclass from total IgG preparation (A) or RBD-specific purified IgG (B) obtained from sera samples. A -Age-dependent anti-RBD IgG subclass distribution following the first and second vaccine doses. IgG subclasses were determined for each individual serum. These levels were then used to calculate IgG1+IgG3/IgG2+IgG4 ratios (preboost: age≤ 60, n=60; age>60, n=67; post-boost: age≤ 60, n=60; age>60, n=67). B -Age-dependent Fc glycosylation patterns in IgG2, IgG3/4 and total IgG from vaccinated individuals. Total IgGs: age≤ 60, n=32; age>60, n=27; pre-boost IgG3: age≤ 60, n=15; age>60, n=13; post-boost IgG2 and IgG3: age≤ 60, n=24; age>60, n=15. C -Binding of total IgGs to FcγRs and C1q. Binding to activating vs inhibitory FcγR ratio was as described above. Age≤ 60, n=32; age>60, n=27. Data are presented as scatter plots indicating individual measurements (dots); black line represents the mean; error bars represent SD. Unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. (n=56) and females (n=71). Values of each subclass were used to calculate IgG1+IgG3/IgG2+IgG4 ratios. C -Fc glycosylation patterns of total IgGs and post-boost RBD-specific IgG1, IgG2 and IgG3/4. Total IgG: male, n=29; female, n=30; postboost anti-RBD IgG: male, n=20; female, n=19. D -Binding of RBD-specific IgGs to FcγRs and C1q. Binding to each receptor was determined by ELISA at pre-boost (males n=20, females, n=19) and post-boost (males, n=29, females, n=30). Ratios between RBD-specific IgG binding to activating and inhibitory receptors were determined as described above. Data are presented as scatter plots indicating individual measurements (dots); black line represents the mean; error bars represent SD. Unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. -19 diagnosis (vaccinated, n=39; mild patients, n=8 ; severe patients, n=6. RBD-specific IgGs were isolated and Fc glycan structure was determined by mass spectrometry. Data are presented as violin plots with solid lines representing median and dotted lines upper and lower quartiles. B -Distinct patterns of subclass-specific Fc glycosylation, as determined by mass spectrometry, in total IgGs from vaccinated individuals and COVID-19 patients at 5 weeks from diagnosis (vaccinated, n=39; mild, n=5; severe, n=4). Data are presented as violin plots with solid lines representing median and dotted lines upper and lower quartiles. C -Binding of total IgGs to FcγR and C1q in severe COVID-19 patients, mild patients and vaccinated individuals (vaccinated, n=59; early mild, n=5; early severe, n=4; late mild, n=14; late severe n=12). Unless otherwise mentioned, data are presented as scatter plots indicating individual measurements (dots); black line represents the mean; error bars represent SD. Unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data are presented as scatter plots indicating individual measurements (dots); black line represents the mean; error bars represent SD. Unpaired two-sided Mann-Whitney U test was used to evaluate differences between groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.