Coronavac inactivated vaccine triggers durable, cross-reactive Fc-mediated phagocytosis activities

ABSTRACT

Although humoral responses elicited by infection or vaccine lost the ability to prevent transmission against Omicron, vaccine-induced antibodies may still contribute to disease attenuation through Fc-mediated effector functions. However, Fc effector function elicited by CoronaVac, as the most widely supplied inactivated vaccine globally, has not been characterized. For the first time, our study depicted Fc-mediated phagocytosis activity induced by CoronaVac, including antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent neutrophil phagocytosis (ADNP) activities, and further compared with that from convalescent individuals and CoronaVac recipients with subsequent breakthrough infections. We showed that 2-dose of CoronaVac effectively induced both ADCP and ADNP, but was substantially lower compared to infection, whereas the booster dose further augmented ADCP and ADNP responses, and remained detectable for 52 weeks. Among CoronaVac recipients, ADCP and ADNP responses also demonstrated cross-reactivity against Omicron subvariants, and breakthrough infection could enhance the phagocytic response. Meanwhile, serum samples from vaccinees, convalescent individuals with wildtype infection, BA.2 and BA.5 breakthrough infection demonstrated differential cross-reactive ADCP and ADNP responses against Omicron subvariants, suggesting the different subvariants of spike antigen exposure might alter the cross-reactivity of Fc effector function. Further, ADCP and ADNP responses were strongly correlated with Spike-specific IgG responses and neutralizing activities, indicating coordinated neutralization activity, ADCP and ADNP responses triggered by CoronaVac. Of note, the ADCP and ADNP responses were more durable and cross-reactive than corresponding Spike-specific IgG titers and neutralizing activities. Our study has important implications for optimal boosting vaccine strategies that may induce potent and broad Fc-mediated phagocytic activities.

KEYWORDS:

• COVID-19 vaccine
• Fc effector function
• SARS-CoV-2
• ADCP
• ADNP

Introduction

The worldwide vaccination campaign proved outstanding protection against the severity and fatality of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, not only for wildtype (WT) strain but also for emerging variants of concern (VOCs) [1]. The immune correlation analyses suggested a strong relationship between neutralizing antibody (NAb) concentrations and vaccine efficacy [2,3]. Real-world efficacy study showed that coronavirus disease 2019 (COVID-19) vaccines, including inactivated virus vaccines and mRNA vaccines, have comparable levels of protection efficacy [4], but different degrees of their neutralizing activities have been consistently reported. This raises the postulation that neutralizing antibodies might not be the key element responsible for immune protection against disease severity and fatality. Meanwhile, antibody binding titer might serve as a stronger surrogate of protection across vaccine platforms [5,6], even before the generation of neutralizing antibodies [7]. The above observations highlight a possible undefined role for alternative anti-viral mechanisms of antibody responses.

Indeed, beyond the neutralization, antibodies also act in a variety of alternative immunological protection manners to coordinate the immune system using Fc receptors (FcRs) [8]. In the light of many fatal pathogens, including influenza [9], anthrax [10], malaria [11], Ebola virus [12], and human immunodeficiency virus (HIV) [13], Fc-mediated effector functions have been linked with immune protection. Meanwhile, merging evidence showed that Fc effector activities elicited by vaccination or infection confer immune protection against SARS-CoV-2 in COVID-19 patient cohort studies [14,15]. Furthermore, intact effector functions of Spike protein-specific monoclonal antibodies are also required for optimal therapeutic efficacy [16,17,18]. mRNA vaccines, such as mRNA-1273 and BNT162b2, could induce robust humoral responses with Fc-mediated effector function [19]. However, whether the inactivated vaccine, the most widely administered COVID-19 vaccine globally, could elicit potent Fc-mediated effector functions, particularly across variants of concern remains elusive. Previously, we reported the dynamic, longitudinal antibody, B cell, and T cell responses following immunization of CoronaVac in a prospective cohort of SARS-CoV-2 naïve healthcare workers [20,21,22,23]. Here, we aimed to characterize the dynamic Fc-mediated effector function induced by CoronaVac, including antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent neutrophil phagocytosis (ADNP), against WT stain and Omicron subvariants. Additionally, ADCP and ADNP responses among CoronaVac recipients, convalescent individuals, and vaccinees with breakthrough infection were compared. Furthermore, the durability and cross-reactivity of ADCP, ADNP, binding antibody titer, and neutralizing antibody (NAb) responses were evaluated. Our study points to the potential importance of Fc effector function elicited by CoronaVac immunization, which might be responsible for the immune protection against the severity and fatality of COVID-19.

Materials and methods

Study cohort and sample collection

In our study, the COVID-19 vaccine cohort and three SARS-CoV-2 infection cohorts were included (Figure 1). In the COVID-19 vaccine cohort, thirty healthcare professionals receiving 3-dose CoronaVac (Sinovac Biotech, Beijing, China) were enrolled. The first and the second dose of CoronaVac were administered at 0.73 (interquartile range (IQR): 0.60, 0.90) months apart, and the third booster was given at 9.13 (IQR: 8.94, 9.44) months following the priming two-dose vaccination. To investigate the kinetics of ADCP and ADNP responses elicited by CoronaVac in the COVID-19 vaccine cohort, serum samples were collected at eight timepoints, including before the first dose (blood draw 1), 2 weeks following the first dose (blood draw 2), 2 weeks following the second dose (blood draw 3), 8 weeks following the second dose (blood draw 4), 36 weeks following the second dose (blood draw 5), 2 weeks following the third dose (blood draw 6), 8 weeks following the third dose (blood draw 7), and 52 weeks following the third dose (blood draw 8) (Figure 1A). The serum samples were collected from a registered prospective study (NCT04729374). The first infection cohort consisted of twelve participants infected with WT SARS-CoV-2 from January to February 2020 without previous vaccination of CoronaVac, and their demographic characteristics were provided in Figure 1C. Their serum samples were collected at 15.76 (IQR: 15.35, 16.09) months after COVID-19 disease onset. Additionally, we have two breakthrough infection cohorts. The cohort two had BA.2 breakthrough infection 3.93 (IQR: 3.90, 4.02) months after the immunization of 3-dose of CoronaVac, and serum samples were collected 2.37 (IQR: 2.27, 2.44) months after breakthrough infection. The cohort three had BA.5 breakthrough infection 12.43 (IQR: 12.37, 12.50) months after vaccination, and the samples were collected 0.47 (IQR: 0.33, 0.47) months after infection. These infection cohorts were screened daily for potential infection of SARS-CoV-2 by RT–PCR. If any individuals were confirmed with infection, the nasal swabs were sent for whole genome sequencing to identify the infected subvariants. All participants had written informed consent before any study procedures were undertaken. This protocol was approved by Nanjing Drum Tower Hospital Ethics Committee (2021-034-01).

Figure 1. The study design and the characteristics of participants in our cohort. (A) The study design of our vaccine cohort. (B) Three convalescent COVID-19 patient cohorts, including WT strain infection only (cohort 1), BA.2 (cohort 2), and BA.5 (cohort 3) breakthrough infection following 3-dose of CoronaVac. (C) The demographic and clinical characteristics of four study cohorts.

Proteins

Antigens used for ADCP and ADNP assay are as following: SARS-CoV-2 Spike protein and receptor-binding domain (RBD) protein from WT strain, Omicron BA.1, BA.2, BA.3, BA.4/5, BQ.1 and XBB.1. SARS-CoV-2 Spike protein of ancestral strain (cat# Z03481) and all RBD protein (cat# Z03483, cat# Z03728-100, cat# Z03740, cat# CP0007, cat# Z03745 for WT strain, Omicron BA.1, BA.2, BA.3, BA.4/5, respectively) were provided by Genscript (Jiangsu, China). BQ.1 Spike (cat# CG273) and XBB.1 Spike (cat# CG275) proteins were provided by Vazyme (Nanjing, China) Omicron BA.1 Spike ectodomain (GenBank: OL672836.1, residues 1-1205) was expressed as previously described [21]. The prefusion Omicron Spike ectodomain of BA.2, BA.3, BA.4/5 with proline substitutions at residues 983 and 984, a “GSAS” instead of “RRAR” at the furin cleavage site (residues 679-682), with a C terminal T4 fibritin trimerization motif, an HRV-3C protease cleavage site, a Twin-Strep-tag, and an 8×His-tag was cloned into vector pcDNA3.1 (Thermo Fisher Scientific, MA, USA) according to previous publications [24]. The protein was purified from Expi293 cells (Thermo Fisher Scientific, MA, USA) using affinity chromatography followed by Strep-Tactin resin (IBA, Göttingen, Germany, Cat 2-1201-500), detailed as described previously [25].

Coupling of fluorescent beads

To generate antigen-coupled beads, antigens were biotinylated with Sulfo-NHS-LC biotin (Thermo Fisher Scientific, MA, USA, cat# A39257) following the manufacturer's instructions, respectively. Unbound biotin was removed using a Zeba Spin desalting column (Thermo Fisher Scientific, MA, USA, cat# A44300). The biotinylated antigen was coupled in a 1:1 ratio to 1 μm size of yellow-green fluorescent NeutrAvidin (Invitrogen, MA, USA, cat# F8776) in the dark overnight at 4°C. Subsequently, coupled beads were washed with phosphate-buffered saline (PBS) (16,000 ×g, 15 min) and resuspended in PBS with 0.5% bovine serum albumin (BSA).

Cell cultures

THP-1 cells were obtained from American Type Culture Collection (ATCC) and maintained in complete R10 media [RPMI-1640 (Gibco, cat#: 11875119) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich, cat#: 12007C), 5% penicillin/streptomycin (50μg/ml; Gibco, cat#: 15140163), 5% l-glutamine (4 mM; Gibco, cat#: A2916801), and 5% Hepes buffer (pH 7.2) (50 mM; Gibco, cat#: 15630106)]. Cell culture densities were kept below 0.5 × 10⁶ cells/mL to maintain consistent assay performance.

To generate peripheral blood mononuclear cells (PBMCs), whole human blood with ethylenediamine tetraacetic acid (EDTA) was mixed at a 1:10 ratio with ammonium-chloride-potassium (ACK) lysis buffer (Gibco, cat#: A1049201) and incubated for 10 min at room temperature. The PBMCs were pelleted by centrifugation (500 ×g, 5 min) at room temperature and then washed with cold PBS. The WBCs were finally diluted in complete R10 media for ADNP assay.

Analysis of ADCP and ADNP responses

Antibody-dependent cellular phagocytosis (ADCP) [26] by monocytes and antibody-dependent neutrophil phagocytosis (ADNP) [27] assay were performed as previously described with minor modifications. Briefly, to form immune complexes, antigen-coupled beads were incubated with 1:25 diluted serum samples for 2 h at 37°C, and then washed to remove unbound immunoglobulins. A blank control containing only beads and neutrophils or THP-1 was included as negative control. For ADCP, the immune complexes were incubated for 1 h with 25,000 THP-1 cells per well at a concentration of 2.5 × 10⁵ cells/ml in R10 media at 37°C, 5% CO2. For ADNP, PBMCs was resuspended at 2.5 × 10⁵ cells/ml in R10 media; 25,000 cells per well were added to each well and incubated with immune complexes for 1 h at 37°C, 5% CO2. After incubation, PBMCs were washed, and stained for CD66b+ (APC–conjugated anti-CD66b clone G10F5 (2 μg/ml); BioLegend, cat#: 17-0666-42) to identify neutrophils. Then, cells were fixed with 4% paraformaldehyde (PFA) (Alfa Aesar). Flow cytometry was performed to identify the percentage of cells that had phagocytosed beads as well as the geometric mean fluorescent intensity (gMFI) of beads that had been phagocytosed (bead+ THP-1 or neutrophils, defined as positive cells) (Figure S1). A minimum of 2000 cells were acquired to calculate the phagocytic score (phagoscore). Phagoscore  = % positive cells × geometric mean fluorescent intensity of positive cells/10,000. The corrected phagoscore was defined as the actual phagoscore derived from the sample minus phagoscore score derived from the blank control. If the phagoscore was calculated as below zero, it would be set as zero. Flow cytometry was performed with BD Aria III Flow Cytometer, and analysis was performed using FlowJo V10.7.1.

The linear regression model was used to determine the waning rate of ADCP and ADNP responses elicited by 2-dose CoronaVac using ADCP and ADNP responses at week 2, week 8, and week 36 following two primary doses. The linear regression model was used to show the waning rate of ADCP and ADNP responses elicited by 3-dose CoronaVac using ADCP and ADNP responses at week 2, week 8, and week 52 following the booster doses. The slope of the fitted line was referred to as the waning rate.

Correlation analyses

Spearman correlation analysis was used for correlation analysis between ADCP and ADNP, antibody Fc effector functions and Spike-specific IgG titer, and antibody Fc effector functions and neutralizing antibody titer of vaccinees. The correlation analyses used all the data we tested for ADCP and ADNP responses at eight different time points among CoronaVac vaccinees, including before the first dose, 2 weeks following the first dose, 2 weeks following the second dose, 8 weeks following the second dose, 36 weeks following the second dose, 2 weeks following the third dose, 8 weeks following the third dose, and 52 weeks following the third dose. The binding IgG titers and neutralizing activities (ID50) in this vaccine cohort were previously published and included for correlation analysis [21]. The correlation heatmap was prepared using ChiPlot (https://www.chiplot.online/correlation_heatmap.html).

Statistical analysis

Kruskal–Wallis test with correction by controlling the False Discovery Rate (FDR) for multiple comparisons was performed for the comparison of ADCP and ADNP responses elicited by CoronaVac at different time points and fold changes in ADCP or ADNP responses specific to each Omicron Spike subvariants relative to that specific to WT Spike. For multiple hypothesis testing, FDR correction was performed using the Benjamini-Hochberg procedure at the FDR < 0.05 significance threshold. Unpaired t test (normal distributed data) or Mann–Whitney U test (non-normal distributed data) was performed to compare ADCP and ADNP between vaccinees and SARS-CoV-2 infection cohorts. For comparison of ADCP and ADNP of antibodies specific to COVID-19 WT strain versus Omicron subvariants, paired t test (normal distributed data) or Wilcoxon matched-pairs signed rank test (non-normal distributed data) was performed for paired data. Data were analyzed using GraphPad Prism (version 9.0.1, La Jolla, California, USA). The statistical analyses were performed in a two-sided manner, and p value <0.05 was considered as statistically significant. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001. ns indicates no significant difference.

Results

Robust Fc-mediated phagocytosis activities were induced by three doses of CoronaVac vaccination

Given the emerging appreciation for the potential importance of humoral response beyond neutralization, we sought to profile antibody-mediated effector function induced by CoronaVac. Here, we characterized the anti-viral effector function of antibodies in the serum by evaluating antibody-mediated phagocytosis at four different time points, including baseline before the first dose, 2 weeks after the first dose, 2 weeks after the second dose and 2 weeks after the third dose (Figure 2). The first dose of CoronaVac did not elicit substantially increased ADCP and ADNP responses, but two-dose CoronaVac significantly induced WT Spike and RBD-specific ADCP activities (p = 0.0010 and p = 0.0197, respectively) and ADNP activities (p < 0.0001 and p < 0.0001, respectively), compared with the baseline. Moreover, the third dose of CoronaVac further augmented WT Spike-specific ADCP response by 3.04 folds (p = 0.0011) (Figure 2A), WT Spike-specific ADNP by 3.69 folds (p = 0.0029) (Figure 2C), and WT RBD-specific ADNP by 2.53 folds (p = 0.0023) (Figure 2D), compared with the corresponding peak humoral response post 2 doses.

Figure 2. Dynamic SARS-CoV-2 specific Fc-mediated phagocytosis responses 2 weeks following each time of CoronaVac immunization. Antibody-dependent cellular phagocytosis (ADCP) (A, B) and antibody-dependent neutrophil phagocytosis (ADNP) (C, D) specific to Spike protein and RBD protein of WT strain, as well as five circulating Omicron variants, including BA.1, BA.2, BA.3, BA.4 and BA.5, at baseline before the first dose (0w), week 2 post the first dose (1 dose 2 wks), week 2 post the second dose (2 dose 2 wks) and week 2 post the third dose (3 dose 2 wks) were analyzed. Statistics were calculated using Kruskal-Wallis test with correction by controlling the false discovery rate for multiple comparisons between timepoints. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001. ns indicates no significant difference.

In addition, emerging variants of concern, especially Omicron subvariants, displayed an alarming immune evasion of neutralizing antibodies [21,28,29]. It remains elusive whether antibody-mediated Fc effector activities were cross-reactive with Omicron subvariants. Post-vaccine sera demonstrated detectable ADCP and ADNP activities specific to Omicron subvariants following the priming two-dose vaccination. When compared to Fc effector activities after two doses, the booster elicited a stronger ADCP response with a 1.65-fold (p = 0.0064) augment for BA.2 spike, a 2.22-fold (p = 0.0014) increase for BA.2 RBD, a 2.54-fold (p = 0.0006) enhancement for BA.4/5 RBD-specific ADCP (Figure 2B). Meanwhile, BA.1, BA.2, BA.3, BA.4/5 Spike-specific ADNP activity showed 2.23-fold (p = 0.0023), 1.72-fold (p = 0.0318), 2.30-fold (p = 0.0032) and 1.92 folds (p = 0.0425) increase after the booster dose (Figure 2C), while Omicron various RBD-specific ADNP activity exhibited a similar increment, ranging from 1.47-fold to 6.67-fold (Figure 2D).

Longitudinal 52-week follow-up of ADCP and ADNP activity induced by 3-dose of CoronaVac

To study the longitudinal antiviral Fc effector functions induced by CoronaVac, we tracked and characterized the ADCP and ADNP responses over time, including week 2 post the 2 doses (2 dose 2 wks), week 8 post 2 doses (2 dose 8 wks), week 36 post 2 doses (2 dose 36 wks), week 2 post 3 doses (3 dose 2 wks), week 8 post 3 doses (3 dose 8 wks) and week 52 post 3 doses (3 dose 52 wks) (Figure 3). To characterize the waning rates of ADCP and ADNP responses induced by CoronaVac, a linear regression analysis to calculate the waning rate of ADCP and ADNP responses following 2-dose and 3-dose CoronaVac. While ADCP and ADNP responses were detectable at week 2 following 2-dose CoronaVac vaccination, slightly declined at week 8, and further decreased at week 36 post 2-dose CoronaVac compared with week 2 post 2-dose (Figure 3). The third dose of CoronaVac elicited a robust recall of ADCP and ADNP responses across various spike or RBD antigens. ADCP and ADNP responses reached to the peak immunity 2 weeks following the third dose of CoronaVac, which were well maintained at a high level even at week 8. ADCP and ADNP activities were decreased at a large extent at week 52 after the vaccination of 3-dose CoronaVac (Figure 3). The waning rates of spike-specific ADCP and ADNP responses were 0.038 and 0.007 after two doses, respectively, whereas the waning rates of ADCP and ADNP responses were 0.11 and 0.028 after three doses, respectively (Figure S3, Table S1). Our data showed that the waning rate of 3-dose CoronaVac was even faster than that of 2-dose CoronaVac, possibly due to the relatively low magnitude of ADCP or ADNP response elicited by 2-dose CoronaVac. Overall, our data showed that suggesting 2-dose or 3-dose CoronaVac-induced Fc-mediated phagocytosis slowly declined.

Figure 3. Longitudinal SARS-CoV-2 specific Fc mediated phagocytic responses induced by 2-dose and 3-dose of CoronaVac. The ADCP (A, B) and ADNP (C, D) specific to Spike protein and RBD protein of WT strain, as well as five circulating Omicron variants, including BA.1, BA.2, BA.3, BA.4 and BA.5, were analyzed at six time points, including at week 2 post the second dose (2 dose 2 wks), at week 8 post the second dose (2 dose 8 wks), at week 36 post the second dose (2 dose 36 wks), at week 2 post the third dose (3 dose 2 wks), at week 8 post the third dose (3 dose 8 wks) and at week 52 post the third dose (3 dose 52 wks). Statistics were calculated using Kruskal-Wallis test with correction by controlling the false discovery rate between time points. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001. ns indicates no significant difference.

COVID-19 convalescent patients showed stronger ADCP and ADNP responses than 2-dose CoronaVac recipients

To compare the differences in Fc-mediated phagocytosis between CoronaVac recipients and COVID-19 convalescent patients, it is ideal to use samples collected at the same time point after antigen exposure. Since we only have the convalescent sera collected at 64 weeks after the infection of WT strain, therefore, the vaccinated sera samples at week 36 post the immunization of 2-dose CoronaVac were used for comparison. Our data showed that anti-viral ADCP and ADNP responses from vaccinees were pronounced lower compared to COVID-19 convalescent subjects. Specifically, lower magnitudes of ADCP response specified to WT spike (p < 0.0001), BA.1 RBD (p = 0.0281), BA.2 spike (p < 0.0001), BA.3 spike (p < 0.0001), and BA.4/5 spike (p < 0.0001) were observed in 2-dose CoronaVac recipients compared to that of convalescent individuals infected with WT strain (Figure 4A,B). A Similar trend of ADNP responses across WT strain and Omicron variants (p < 0.0001 for all) was also noted (Figure 4C-D). Considering the weakening characteristics of Fc effector with time, ADCP and ADNP responses at week 36 after infection in convalescent patients should be higher than that at 64 weeks. Therefore, our data suggested that COVID-19 convalescent patients showed stronger ADCP and ADNP responses than 2-dose CoronaVac recipients.

Figure 4. Fc mediated phagocytic responses induced by 2-dose of CoronaVac is lower than that in convalescent patients infected with WT SARS-CoV-2 without CoronaVac. The ADCP (A, B) and ADNP (C, D) specific to Spike protein and RBD protein of WT strain, as well as five circulating Omicron variants, including BA.1, BA.2, BA.3, BA.4 and BA.5, were compared between vaccinees at 36 weeks after 2 doses and convalescent patients infected with WT SARS-CoV-2 without CoronaVac at 15.76 (IQR: 15.35, 16.09) months after infection. When the data is normally distributed, unpaired t test is used. When the data is non-normal distribution, Mann-whitney U test is used for comparison between the two cohorts. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001. ns indicates no significant difference.

Omicron breakthrough infection enhanced Fc-mediated ADCP and ADNP responses via distinct epitopes

We then investigated whether Omicron breakthrough infection could enhance Fc-mediated effector functions, by comparing ADCP and ADNP responses among 3-dose CoronaVac vaccinees and BA.2 breakthrough individuals with previous 3-dose CoronaVac vaccination. On one hand, BA.2 breakthrough infection resulted in a significant increment of ADCP response specific to BA.2 Spike (1.62-fold, p = 0.0330), BA.3 Spike (4.81-fold, p < 0.0001), BA.4/5 Spike (1.50-fold, p < 0.0001), and BA.3 RBD (1.63-fold, p = 0.0097) (Figure 5A). Nevertheless, 3-dose CoronaVac vaccinees showed a comparable level of ADCP response as BA.2 breakthrough infection specific to the majority of RBD proteins except BA.3 (Figure 5B). On the other hand, BA.2 breakthrough infection leads to a remarkable augment for ADNP response specific to WT spike (3.47-fold, p < 0.0001), BA.1 spike (3.93-fold, p < 0.0001), BA.2 spike (2.89-fold, p < 0.0001), BA.3 spike (5.73-fold, p < 0.0001), and BA.4/5 spike (6.45-fold, p < 0.0001) (Figure 5C). Consistent with spike-specific ADNP activity, similar trends were also observed for ADNP activity specific to RBD from WT and Omicron subvariants (Figure 5D). Our data suggested that breakthrough infection might further enhance ADCP responses majorly through non-RBD region, while strengthen ADNP responses mostly depending on RBD region.

Figure 5. Fc effector Fc mediated phagocytic induced by 3-dose of CoronaVac is lower than that in BA.2 breakthrough infection among 3-dose of CoronaVac recipients. The ADCP (A, B) and ADNP (C, D) specific to Spike protein and RBD protein of WT strain, as well as five circulating Omicron variants, including BA.1, BA.2, BA.3, BA.4 and BA.5, were compared between vaccinees at week 8 after 3-dose of CoronaVac and convalescent patients with BA.2 breakthrough infection after receiving 3 doses of CoronaVac. When the data is normally distributed, unpaired t test is used. When the data is non-normal distribution, Mann-whitney U test is used for comparison between the two cohorts. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001. ns indicates no significant difference.

Cross-reactivity of antibody Fc-mediated phagocytosis among vaccinees and COVID-19 convalescent individuals

Due to substantial immune evasion for Omicron subvariants, we investigated whether ADCP and ADNP responses were also compromised against Omicron subvariants compared to WT strain. Four cohorts of vaccinees or COVID-19 convalescent subjects were used for analysis, including 3-dose CoronaVac recipients, WT COVID-19 convalescent subjects without any vaccination, and BA.2 or BA.5 convalescent subjects with the previous recipient of 3-dose CoronaVac. Overall, anti-viral ADCP and ADNP responses specific to Omicron subvariants in all four cohorts were lower compared to SARS-CoV-2 WT strain (Figure 6). Sera samples from CoronaVac recipients showed significantly lower ADCP response specific to BA.1 (2.74-fold, p < 0.0001), BA.2 (2.58-fold, p < 0.0001), BA.3 (5.03-fold, p < 0.0001), BA.4/5 (2.38-fold, p < 0.0001), BQ.1 (1.86-fold, p = 0.0034) and XBB.1 (6.29-fold, p = 0.0002) compared to WT spike (Figure 6A). Consistently, CoronaVac recipients have reduced ADNP response specific to BA.1 (1.95-fold, p = 0.0001), BA.2 (3.74-fold, p < 0.0001), BA.3 (3.46-fold, p < 0.0001), BA.4/5 (4.68-fold, p < 0.0001), BQ.1 (2.73-fold, p < 0.0001) and XBB.1 (5.18-fold, p = 0.0002) if compared to that specific to WT strain (Figure 6B). In the convalescent individuals with previous infection of WT strain, our data showed moderate reductions of ADCP and ADNP responses specific to Omicron subvariants compared to WT strain, ranging from 1.20-fold to 10.40-fold. Interestingly, BA.2 breakthrough infection only exhibited a mild reduction of Omicron-specific ADCP and ADNP responses, ranging from 1.79 folds to 5.94 folds. In the cohort of BA.5 breakthrough infections, compared to the WT-specific responses, slightly reduced ADCP responses specific to BA.1 (1.58-fold, p < 0.0001), BA.3 (2.39-fold, p < 0.0001), BA.4/5 (2.94-fold, p < 0.0001), BQ.1 (2.37-fold, p < 0.0001) and XBB.1 (2.75-fold, p < 0.0001) were observed, except that BA.2 specific ADCP response showed an 18.88-fold drop (p < 0.0001). A similar trend was observed in Omicron-specific ADNP response, with the reduction folds from 2.43 folds to 4.72 folds. Consistently, BA.2 specific ADNP response was remarkably reduced compared to WT strain (15.28-fold, p < 0.0001).

Figure 6. Cross-reactivity of antibody Fc-mediated phagocytic functions among vaccinees and COVID-19 convalescent individuals. The ADCP (A) and ADNP (B) responses specific to WT strain and Omicron subvariants in different cohorts, including vaccinees with 3 doses of CoronaVac, convalescent patients infected with WT strain without CoronaVac, BA.2 and BA.5 breakthrough infection after receiving 3 doses of CoronaVac. The average fold decrease was shown at the top of each bar. When the data is normally distributed, paired t test is used to compare the Fc effector functions between ancestral strain and Omicron subvariants. When the data is non-normal distribution, Wilcoxon matched-pair signed rank test is used. (C) Fold changes of ADCP and ADNP responses specific to Omicron subvariants Spike relative to that specific to WT strain. The dotted line represents no fold difference, while lines indicate the median. Statistical differences in fold changes between different cohorts were calculated using the Kruskal-Wallis test with correction by controlling the False Discovery Rate. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001. ns indicates no significant difference.

To directly compare the cross-reactivity of ADCP and ADNP responses in four different cohorts, the fold differences of Fc-mediated phagocytosis for Omicron variants versus WT strain were compared (Figure 6C). Compared to convalescent individuals only infected with WT strain, CoronaVac recipients had greater fold differences of ADCP responses specific to BA.2, BA.3, BA.4/5 versus WT spike (p = 0.0019, p < 0.0001, p < 0.0001, respectively), but smaller fold differences of ADNP responses specific to BA.2 and BA.4/5 versus WT spike (p < 0.0001 and p = 0.0381, respectively). Furthermore, BA.5 breakthrough infection had less affected ADCP responses against BA.1 and XBB.1 versus WT spike (p = 0.0261 and p = 0.0071, respectively), compared to CoronaVac recipients. Similarly, BA.2 breakthrough infection cohort demonstrated more cross-reactive ADNP responses specific to BA.1 and BA.3 spike (p = 0.0269 and p = 0.0244, respectively), compared to CoronaVac recipients. BA.5 breakthrough infection also induced more cross-reactive ADNP responses specific to BA.1, BA.3, BA.4/5, and XBB.1 spike, but without statistical differences, compared to CoronaVac recipients. Our data suggested distinct cross-reactive profiles for ADCP and ADNP responses induced by different histories and sequences of SARS-CoV-2 antigen exposure.

Correlation analyses of ADCP and ADNP responses induced by CoronaVac

To identify the potential relationship of ADCP and ADNP responses among CoronaVac recipients, Spearman correlation analysis was used for correlation analysis between ADCP and ADNP, antibody Fc effector functions and Spike-specific IgG titer, and antibody Fc effector functions and neutralizing antibody titer of vaccinees. Our data revealed strong positive correlations between ADCP and ADNP responses specific to WT spike (r = 0.77, p < 0.0001), BA.1 spike (r = 0.59, p < 0.0001), BA.2 spike (r = 0.86, p < 0.0001), BA.3 spike (r = 0.67, p < 0.0001), and BA.4/5 spike (r = 0.70, p < 0.0001) (Figure 7, Figure S4A). Consistent with the previous finding that the antibody levels were correlated with Fc-mediated effector function [15], the spike-specific IgG levels were positively correlated with the corresponding ADCP responses specific to WT Spike (r = 0.77, p < 0.0001), BA.1 Spike (r = 0.48, p = 0.0008), BA.2 Spike (r = 0.57, p < 0.0001), BA.4/5 Spike (r = 0.35, p = 0.0165), but not BA.3 Spike (r = 0.11, p = 0.4510) (Figure 7, Figure S4B). Besides, spike-specific IgG titer was positively associated with the corresponding ADNP responses specific to WT Spike (r = 0.67, p < 0.0001), BA.1 Spike (r = 0.56, p = 0.0001), BA.2 Spike (r = 0.68, p < 0.0001), BA.3 Spike (r = 0.65, p < 0.0001), BA.4/5 Spike (r = 0.55, p < 0.0001), respectively (Figure 7, Figure S4C). Since neutralizing activity is also a critical arm of humoral responses, we analyzed the potential correlation between serum neutralizing activity and Fc effector functions, using previously reported neutralization activity from the same cohort [15]. NAb titers against WT strain were strongly correlated with the ADCP responses specific to WT Spike (r = 0.69, p < 0.0001), BA.2 Spike (r = 0.5772, p = 0.0004), but not BA.1 Spike (r = 0.0097, p = 0.9567), BA.2 Spike (r = 0.58, p = 0.0004), BA.4/5 Spike (r = 0.22, p = 0.2118) (Figure 7, Figure S4D). Besides, the NAb titer against WT strain was positively associated with the relative ADNP responses specific to WT Spike (r = 0.69, p < 0.0001), BA.1 Spike (r = 0.65, p < 0.0001), BA.2 Spike (r = 0.62, p < 0.0001), BA.3 Spike (r = 0.76, p < 0.0001), BA.4/5 Spike (r = 0.53, p = 0.0014), respectively (Figure 7, Figure S4E). Our data highlighted a coordinated evolution of Fc-mediated phagocytic activity and neutralizing activity across diverse antigens elicited by CoronaVac vaccination.

Figure 7. Correlation analysis of ADCP, ADNP, IgG titer, and neutralization activity elicited by CoronaVac. Statistics were analyzed using Spearman correlation analysis. p < 0.05 was considered to be statistically significant.

ADCP and ADNP responses were more durable and cross-reactive than binding IgG responses and neutralizing activities induced by CoronaVac

To compare the persistence of ADCP responses, ADNP responses, and antibody responses induced by CoronaVac, we evaluated the waning rate of the above immune responses at week 2, week 8, and week 52 following 3-dose of CoronaVac (Figure 8A). The waning rates of WT-specific ADCP and ADNP responses among 3-dose CoronaVac recipients were 0.11 and 0.03, respectively, whereas the waning rate of binding IgG titer was 114.40, which was remarkably higher compared to that of ADCP and ADNP responses. Consistently, the waning rate of BA.1 specific ADCP and ADNP responses among 3-dose CoronaVac vaccinees were 0.03 and 0.03, respectively, whereas the waning rate of binding IgG titer was 41.16.

Figure 8. ADCP and ADNP responses were more durable and cross-reactive than IgG and NAb responses induced by CoronaVac. (A)The waning rate of ADCP, ADNP and IgG titer specific to WT strain and BA.1 Spike after 3 doses of CoronaVac. (B) The fold changes of NAbs, IgG titer, ADCP, and ADNP specific to BA.1 over WT Spike at week 2 following 3rd dose of CoronaVac. Statistical differences of fold changes among NAbs, IgG titer, ADCP, and ADNP specific to BA.1 over WT Spike were calculated using the Kruskal-Wallis test with correction by controlling the false discovery rate. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001. ns indicates no significant difference.

To examine the cross-reactivity among ADCP, ADNP, neutralizing responses, and IgG, we compared the fold changes of BA.1 versus WT Spike among serum samples from 13 CoronaVac recipients. Of note, the reduced fold change of NAb titer (7.26, p = 0.0001) and IgG titer (4.61, p = 0.0002) was significantly higher than that of ADCP responses (1.65, p = 0.0117) or ADNP responses (1.82, p = 0.0005) (Figure 8B). Altogether, our data suggested that the ADCP and ADNP responses were more cross-reactive than neutralizing activity and IgG titer (Figure 8B).

Discussions

Among the proposed non-neutralizing antibody immune mechanism of protection, beyond T cells [30], Fc receptor-mediated effector function has emerged as an important arm to the control of SARS-CoV-2 infection. Spike and RBD-specific phagocytosis have been linked to viral control in bronchoalveolar fluid [31]. Fatal COVID-19 patients did not fully develop a highly functional IgG response able to coordinate FcR binding and elicit innate immune effector function [15]. The above evidence highlighted the synergy of antibody antigen-binding domain (Fab) and constant domain (Fc) functions is required for early antiviral control and disease resolution.

Despite the previously published references, our work has further expanded our understanding of Fc effector functions of antibodies elicited by CoronaVac. For example, Mok et al. [32] only compared the plaque reduction neutralization (PRNT) and spike-binding antibody and T-cell responses between BNT162b2 and CoronaVac, but not Fc-mediated functions of SARS-CoV-2 specific antibody. Bartsch et al. [33] only analyzed the FcγR binding antibody profiles against SARS-CoV-2 VOCs in CoronaVac vaccinees, but not ADCP and ADNP responses. Tong et al. [34] analyzed the ADCP and ADNP in CoronaVac vaccinees, but they only included the samples following one dose and two doses of CoronaVac. In our current study, we included a longitudinal sample at different time points during the over vaccination of 3-dose CoronaVac and the ADCP and ADNP responses specific for emerging VOCs, including Omicron subvariants, which provide a timely and comprehensive insight of the mechanism of vaccine efficacy. Consistent with the Fc effector response elicited by mRNA vaccines [19], we showed that both ADCP and ADNP activities were effectively induced and further boosted with 3-fold increment, validating the necessity of the boosting dose of CoronaVac. Most ADCP and ADNP Fc effector responses reached peak immunity at week 2 following 2-dose and 3-dose CoronaVac vaccination, which did not substantially decline at week 8. ADCP responses in several individuals even increased from week 2 to week 8 post the booster dose of CoronaVac. This might be because some vaccinees had distinct kinetics of ADCP responses, which reached the peak ADCP responses at week 8 post 3-dose CoronaVac. In addition, compared with IgG titer specific to WT and BA.1 Spike, the corresponding ADCP and ADNP exhibited a slow waning rate. This suggests that ADCP and ADNP responses play a more important role in the long-term protection induced by CoronaVac. Overall, our data highlighted strong, long-lasting ADCP and ADNP responses induced by CoronaVac.

Omicron subvariants were able to evade the neutralizing activities elicited by vaccine or infection to a great extent [21]. A previous report has shown Fc effector function induced by Ad26. COV.2.S vaccine was well retained across D614G, Alpha, Beta, and Delta, but not as cross-reactive as those elicited by the Beta strain [35]. It is not known whether Fc effector function elicited by CoronaVac is compromised against Omicron strains. Consistent with our study, it was shown that two-dose of CoronaVac only showed limited cross-reactivity specific to Omicron for functional Fcγ-recpetor complexes, while the mRNA-vaccine BNT162b2 booster could broaden the Fc effector function [34]. Our data suggested that 3-dose CoronaVac could induce substantial ADCP and ADNP responses against Omicron subvariants. However, compromised ADCP and ADNP responses for Omicron subvariants were observed compared to WT spike. Specifically, convalescent sera from WT strain infected individuals and BA.5 breakthrough infected individuals showed greater than 10-fold drop of ADNP response against BA.2 than WT strain. Our data indicated ADNP responses were more resilient to Omicron sublineage recognition, especially for BA.2 strain.

Our data revealed that breakthrough infection had a modest enhanced cross-reactivity profile across divergent antigens from Omicron sublineages. Intriguingly, ADCP and ADNP responses post BA.2 and BA.5 breakthrough infection did not show any substantially improved cross-reactivities against the matched infecting strains. This could because the initial immune background might also limit Fc effector activities induced by subsequent antigen stimulation, similar as the neutralization activity [36]. An alternative explanation is that Omicron antigen elicited Fc effector function was less cross-reactive to other VOCs [37]. Therefore, the sequence of the exposed Spikes might affect the extent of cross-reactivity.

Currently, there was not any well-established surrogate biomarker to represent effector function. A recent study shed light on the potential epitope of ADCP and ADNP, which demonstrated that RBD-specific antibody depletion did not have a major effect on Fc-mediated effector function in BNT162b2 and mRNA-1273 vaccine recipients. Conversely, the Fc-mediated effector function of convalescent plasma samples was largely affected by RBD-specific antibody depletion [19]. Therefore, there are differences in Fc programming induced by vaccination and infection in an epitope-specific manner. Compromised RBD binding IgG was accompanied by a loss of RBD-specific antibody Fcγ receptor (FcγR) binding, but Spike protein-specific antibodies exhibited persistent but reduced binding to RBD-specific FcγRs in samples from individuals who received the CoronaVac vaccine [33]. In our study, spike-specific ADCP responses were minimal affected by antibody responses specific to RBD region, while spike-specific ADNP responses mostly targeted at RBD region, suggesting that RBD-specific antibody might play a predominate role in ADNP response, but not ADCP responses. Given the neutralizing antibodies mostly depend on the RBD recognition, our current study and others showed that even with a reduction in the RBD-specific antibody abundance, vaccine-induced antibodies will continue to elicit robust Fc-mediated phagocytosis against neutralizing antibody-resistant VOCs including Omicron sublineages.

Our study also demonstrated that Spike-specific IgG titer and neutralizing antibody titer were strongly correlated with ADCP and ADNP responses against WT Spike and most Omicron subvariants. This suggested that humoral responses elicited by CoronaVac incorporating neutralization activities, ADCP and ADNP responses might be co-revolutionized, which might be partially associated with the immune protection elicited by vaccination. Such functional humoral trajectories were also observed in survived COVID-19 individuals, but not in fatal COVID-19 individuals [15]. Of note, Spike-specific IgG response and serum neutralizing activities were not correlated with BA.3 Spike-specific ADCP response. SARS-CoV-2 infection and vaccination induce antibodies with Fc-dependent effector function, of which neutralizing antibody was only a subset of immune protection. Meanwhile, in addition to IgG, the other antibody isotypes such as IgE and IgA also play a critical role in ADCP responses. Therefore, these might be result in a lack of correlation for BA.3 ADCP and IgG or neutralizing titer. Our data highlighted that ADCP and ADNP responses are also important parameters to evaluate the immunogenicity of vaccine and might serve as a surrogate immune biomarker against VOCs.

Our data have important implications for the future design of boosting vaccines, especially among individuals who have received the inactivated COVID-19 vaccines. Firstly, compared to infection, 2-dose CoronaVac vaccination only induced moderate levels of ADNP and ADCP responses, suggesting that Fc effector responses elicited by CoronaVac could be further expanded by an additional effective boosting strategy. Additionally, nuances in magnitude and breadth of Fc-mediated phagocytic responses from convalescent donors and vaccinees suggested that the exposure of spike from different genotypes, vaccine doses, and formulation differentially trigger the repertoire of induced antibodies, with implication for vaccination. It is speculated that the exposure of various SARS-CoV-2 antigens as well as mucosal exposure of SARS-CoV-2 antigen might lead to the humoral responses targeting the most highly conserved region of the SARS-CoV-2 spike antigen, which provides a qualitatively improved antibody response able to better leverage Fc-effector functions against conserved regions of the virus.

There are several limitations to our study. First, our study only studied ADCP and ADNP activities, while antibody-dependent NK cell degranulation and antibody-dependent complement deposition (ADCD) analysis were not included in our analysis. Additionally, whether the magnitude of the Fc effector phagocytosis responses could translate to real-world efficacy remains unclear. Future immune correlates analysis responsible for vaccine effectiveness will provide essential insights into the role of Fc effector phagocytosis responses during COVID-19 disease progression.

Collectively, our study provided a comprehensive analysis of Fc-mediated phagocytic responses elicited by CoronaVac. The vaccination of 3-dose CoronaVac was capable to induce robust and persistent ADCP and ADNP responses cross WT stain and divergent Omicron sublineage, which could be further improved by the vaccination strategy incorporating the spike immunogens associated with a more balanced and broad responses.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by the National Natural Science Foundation of China (92269118, 92269205), Scientific Research Project of Jiangsu Health Commission (M2022013), Clinical Trials from the Affiliated Drum Tower Hospital, Medical School of Nanjing University (2021-LCYJ-PY-10).