Summary
Background
Elevated SARS-CoV-2 IgG4 levels following repeated COVID-19 mRNA boosters may impact blood and mucosal antibody functions against Omicron variants and sarbecoviruses. This study characterised the rise of IgG4 in blood and saliva following four consecutive COVID-19 mRNA doses and examined how IgG4 modulated neutralising and non-neutralising responses.
Methods
Paired plasma and saliva samples were collected pre- and post-mRNA boosters from homologous mRNA vaccinees (2 primary mRNA + 2 mRNA booster) or adenovirus-vector primed vaccinees (2 Vaxzevria + 4 mRNA booster). The breadth and proportion of IgG4 responses towards Omicron variants and sarbecoviruses were measured using multiplex bead arrays. Epitope blocking and depletion assays assessed immune imprinting. Live virus assays established IgG4's contribution towards neutralisation.
Findings
Plasma and salivary IgG4 antibodies binding Omicron variants and sarbecoviruses increased post-booster across both cohorts. Although non-neutralising responses significantly improved post-booster, they negatively correlated with IgG4 against all variants tested. Immune imprinting biased IgG4 responses towards the ancestral receptor binding motif, limiting neutralising IgG4 antibodies towards Omicron variants. Increased proportions of class-switched IgG4 also corresponded with decreased concentrations of spike-specific IgG1, particularly against the RBD, dampening neutralisation.
Interpretation
Repeated COVID-19 mRNA boosters improved neutralising and non-neutralising activity across viral variants for both cohorts, though these responses were dampened by increasing IgG4 antibodies. This work emphasises the potential longitudinal effects of booster-induced IgG4 subclass switching.
Funding
This study was supported by the VC2 Research Fund and an NHMRC Investigator grant #2008092. Sample collection was supported by WHO Unity funds (2020/1085469-0) and WEHI Philanthropic Funds.
Keywords: COVID-19, mRNA booster, IgG4, Humoural immunity, Neutralisation, Mucosal antibodies
Research in context.
Evidence before this study
We have previously demonstrated that plasma and salivary IgG4 only rose in mRNA primary vaccinees and not adenovirus-vector primed vaccinees following their first mRNA booster. To evaluate existing evidence regarding salivary IgG4 expression following repeated mRNA boosters, we searched PubMed as of 11 June 2025, for publications studying IgG4 responses to “saliva” and “IgG4”, and “COVID-19” or “SARS-CoV-2”. No studies have investigated salivary IgG4 beyond the first mRNA booster. We also searched “IgG4” and “Vaxzevria”, “adenovirus” or “ChAdOx1” and did not find any studies assessing the levels of IgG4 in adenovirus-vector primed vaccinees who subsequently received multiple mRNA boosters. Finally, we searched “IgG4” and “sarbecovirus” and found no studies exploring the impact of IgG4 on cross-reactive humoural immunity towards sarbecoviruses. Furthermore, although three studies have shown that IgG4 negatively correlates with neutralisation, this has not been shown for Omicron variants such as BA.1 or XBB.1.5. No studies have also linked that increasing proportions of SARS-CoV-2 specific IgG4 is negatively associated with decreased concentrations of SARS-CoV-2 specific IgG1.
Added value of this study
This study demonstrates that the proportion of IgG4 in saliva further increases with repeated mRNA boosters, particularly in individuals who received primary mRNA vaccines. Furthermore, we found that repeated mRNA boosters (up to the fourth booster) could overcome adenovirus-vector (Vaxzevria) priming and induce elevated IgG4 which dampened neutralising and non-neutralising responses across viral variants. Cross-reactive IgG4 also modulated humoural responses towards sarbecoviruses from bats and pangolins. Booster-induced IgG4 responses were biased towards the ancestral receptor binding motif, limiting the contribution of IgG4 antibodies towards the neutralisation of Omicron variants BA.1 and XBB.1.5. Finally, individuals with increased proportions of class-switched IgG4 displayed decreased concentrations of spike-specific IgG1, particularly against the RBD, further contributing to weaker neutralising responses.
Implications of all the available evidence
Repeated mRNA boosters are effective in promoting SARS-CoV-2 immunity across both primary mRNA or adenovirus-vector (Vaxzevria) vaccinees despite increasing the proportions of IgG4 antibodies in their blood and saliva. Individuals with elevated class-switched IgG4 antibodies are associated with smaller improvements in neutralising and non-neutralising capacities against circulating viral variants. These findings emphasise the importance of understanding how mRNA vaccines and heterologous priming modulate IgG4 class switching to improve future vaccine designs and booster strategies.
Introduction
Periodic COVID-19 mRNA boosters have been recommended for the protection against severe disease and death following SARS-CoV-2 infection, particularly among vulnerable populations.1 These updated mRNA boosters improve waning humoural immunity by enhancing both neutralising and binding immunoglobulin G (IgG) antibody responses against emerging SARS-CoV-2 variants (Omicron BA.1, XBB.1.5, JN.1).2, 3, 4 Previous studies show that receiving up to three COVID-19 mRNA vaccinations caused uni-directional IgG subclass switching of spike (S) antibodies towards IgG4 (IgG3→IgG1→IgG2→IgG4).5, 6, 7, 8, 9, 10, 11, 12, 13, 14 In contrast to IgG3 and IgG1, IgG4 binds poorly to Fc gamma receptors (FcγR) present on NK cells (FcγRIIIa) or phagocytes (FcγRIIa) which diminishes Fc-mediated anti-viral activity, such as antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent cellular phagocytosis (ADCP) respectively.5, 6, 7 Furthermore, IgG4 spontaneously undergoes Fab-arm exchange in the body forming bispecific antibodies which could limit their binding strength due to their resulting monovalency.15,16
Elevated class-switched IgG4 after the third mRNA dose (first mRNA booster) was recently associated with a higher risk of breakthrough infection and shown to negatively correlate with neutralising titres against ancestral pseudovirus.17 However, it is not known if receiving additional mRNA boosters, as is currently recommended, will continue to exacerbate the proportion of class-switched IgG4 and further dampen both neutralising and non-neutralising antibody responses.18,19 Moreover, as most previous studies centered on ancestral responses, the impact of increasing IgG4 against emerging recently circulating SARS-CoV-2 variants and sarbecoviruses remains underexplored.19 Even less is understood about the epitopes targeted by class-switched IgG4, which would better inform future mRNA vaccine design.
In contrast to mRNA primary vaccinees, cohorts receiving adenoviral vector primary vaccines (Vaxzevria) did not have elevated IgG4 following their first mRNA booster.5,13,20 It has been suggested that initial immune priming—by prior COVID-19 infection or non-mRNA vaccination—could limit the magnitude of post-booster class-switched IgG4.13 However, it remains to be seen if repeated boosting with mRNA vaccines, as is currently recommended, would eventually overcome immune priming in these Vaxzevria-primed cohorts.
Interest in mucosal antibody responses against respiratory viral infections has been gaining traction in recent years. COVID-19 mRNA boosters have been demonstrated to elevate mucosal IgG antibodies similarly in tears, saliva and nasal fluid and impact antibody-mediated responses.3,4,6,21,22 While we have previously shown that IgG4 is increased in mucosal secretions (tears and saliva) following the third mRNA dose (first mRNA booster),6 the impact of subsequent mRNA boosters remains to be studied.
Here, by comparing paired plasma and saliva samples pre- and post-booster from primary mRNA vaccinees (mRNA cohort), we demonstrate that the proportions of class-switched IgG4 to total IgG in both compartments rose significantly by their fourth mRNA dose (second mRNA booster) and remained elevated for up to 12 weeks. We also show that IgG4 responses in primary adenovirus vector vaccinees (Vaxzevria cohort) grew incrementally following their second and fourth mRNA boosters/doses. Furthermore, increased proportions of IgG4 were observed against most circulating SARS-CoV-2 variants (BA.1, XBB.1.5, JN.1) and sarbecoviruses (bat, pangolin) screened.
While antibodies against all viral variants displayed enhanced FcγR binding post-booster across both cohorts, these responses also negatively correlated significantly with the proportion of IgG4. This is also in line with the negative correlation of SARS-CoV-2-specific IgG1 concentrations to the increased proportion of class-switched IgG4. Finally, we uncover that immune imprinting biased IgG4 antibodies towards the ancestral receptor-binding domain (RBD), particularly the receptor-binding motif (RBM) which binds the angiotensin-converting enzyme 2 (ACE2) receptor, thus limiting neutralising IgG4 antibodies towards Omicron variants.
Methods
Study participants
Paired plasma (EDTA; sodium heparin) and saliva (SalivaBio, Salimetrics) were collected from primary COVID-19 mRNA vaccinees pre- and 2 weeks post third mRNA dose (first mRNA booster) (n = 46), and pre-, 2, 4 and 12 weeks (n = 46) post fourth mRNA dose (second mRNA booster) as previously described across Victoria, Australia3,6 (November 2021–April 2023). Similarly, paired plasma and saliva were collected from primary COVID-19 vaxzevria vaccinees pre-, 2, 4, and 12 weeks post second mRNA booster/dose (n = 39), and pre- and 4 weeks (n = 20) post fourth mRNA booster/dose as previously described across Victoria, Australia3,4,20 (April 2022–April 2024). Exclusion criteria included prior COVID-19 infection or vaccination in the previous 3 months. Samples were stored at −80 °C until analysis. Total secretory IgA measurements were comparable across saliva samples (Figure S1a). Characteristics of the cohorts were self-reported by study participants and are described in Table S1.
Luminex bead-based multiplex assays
A custom SARS-CoV-2 bead array was designed using SARS-CoV-2 S Trimer and RBD (ancestral: #40589-V08H4, #40592-V08H; BA.1: #40589-V08H26, #40592-V08H121; XBB.1.5: #40589-V08H45, #40592-V08H146; JN.1: #40589-V08H59, #40592-V08H155; Sino Biological) (Figure S1b–c). The pan-sarbecovirus bead array was designed using S Trimers for a range of human (SARS-CoV-2: #SPN-C52H9-50ug; SARS: #SPN-S52H6-100ug; MERS: #SPN-M52H5-50ug; Acro Biosystems), bat (RaTG13: #SPN-B52H4-50ug, RsSHC014: #SPN-B52H8-50ug, HKU3: #SPN-B52H7-50ug; Acro Biosystems) and pangolin (GX-P5L: #SPN-P52H4-50ug, GD-1: #SPN-P52H3-50ug; Acro Biosystems) SARS-related coronaviruses. The human seasonal coronavirus panel was designed using S Trimers for endemic human coronaviruses (SARS-CoV-2: #40589-V08H4, HKU1: #40606-V08B; OC43: #40607-V08B; 229E: #40605-V08B; NL63: #40604-V08B; Sino Biological).
Tetanus toxoid (#T3194-25ug; Sigma–Aldrich), influenza haemagglutinin (H1Cal2009: #11085-V08H; Sino Biological) and SIV gp120 (#40415-V08H; Sino Biological) were also added to arrays as positive and negative controls respectively. All antigens were covalently coupled to magnetic carboxylated beads (DiaSorin) using a two-step carbodiimide reaction and blocked with 0.1% BSA, before being stored in PBS 0.05% sodium azide until use.
To set up the multiplex assay, briefly, 20 μL of working bead mixture (700 beads per antigen-couple bead region; diluted in 0.1% BSA) and 20 μL diluted plasma (final dilutions: IgG4, 1:800–1:3200; total IgG, 1:800–1:12800; FcγR, 1:800) or 20 μL diluted saliva (final dilutions: IgG4, 1:12.5–1:25; total IgG 1:25–1:50; FcγR, 1:800) were added per well and incubated overnight at 4 °C on a shaker. For antibody quantification, anti-SARS-CoV-2 spike human IgG4 (Clone AM359b; # SPD-M402a-100ug; Acro Biosystems) was used to construct a standard curve against ancestral RBD and S Trimer (Figure S1d–f). Beads were then washed and incubated with the respective detectors.
Detection of IgG4 and total IgG were performed in a two-step manner. First, either biotin-conjugated mouse anti-human IgG4 (#3854-6-250; Mabtech; 1 μg/mL, 25 μL/well) or total IgG (#3850-6-250; RRID: AB_10666158; Mabtech; 1 μg/mL, 25 μL/well) respectively were added to the beads and incubated for 2 h at room temperature (RT) on a shaker. After washing, streptavidin R-PE (SAPE; #S866; Thermo Fisher Scientific; 1 μg/mL, 25 μL/well) were added to the beads and incubated for 2 h at room temperature (RT) on a shaker.
FcγR-binding was measured using recombinant soluble FcγR dimers (higher affinity polymorphisms FcγRIIIa-V158 and FcγRIIa-H131; lower affinity polymorphisms FcγRIIIa-F158 and FcγRIIa-R131; 1 μg/mL, 25 μL/well)6 which were first added to the beads, washed and followed by the addition of SAPE as above.
Finally, beads were washed and read on the Intelliflex. Assays were performed in duplicates.
RBD-depleted multiplex assays
To deplete ancestral RBD-specific antibodies from samples, we first diluted plasma (final dilutions: IgG4, 1:3200; total IgG, 1:12800) or saliva (final dilutions: IgG4, 1:25; total IgG, 1:50) in 0.1% BSA containing soluble ancestral RBD (#40592-V08H; final concentration 0.2 μg/mL; Sino Biological). The sample-RBD mixture was then incubated for 1 h at RT on a shaker. Next, the sample-RBD mixtures were added to wells with working bead mixtures containing soluble ancestral RBD (final concentration 0.2 μg/mL). As a negative control, beads coupled to ancestral Spike 2 (S2; #S2N-C52H5; Acro Biosystems) were also included into the bead array. Following which, the multiplex assays were conducted as described above. Successful depletion of RBD-specific antibodies was assessed by >90% loss in ancestral RBD responses while retaining ancestral S2 responses (<20% CV). Experiments were performed in duplicate.
IgG4 depletion for multiplex analysis
Depletion of IgG4 was only performed on plasma samples containing both at >100 ng/mL SARS-CoV-2 specific IgG4, and background levels of tetanus and influenza HA IgG4 (used as negative controls) to detect a discernible drop in IgG4-specific response following depletion. Selected plasma samples (50 μL) were first diluted 1:5 in sterile PBS (250 μL final volume; Gibco, Life Technologies). Next, Capture Select IgG4 (Hu) Affinity Matrix slurry (#2942902010; Thermo Fisher Scientific) was added directly to each diluted plasma samples (5 μL/tube). Samples were incubated 1 h at RT on a shaker. Samples were then transferred to pre-wet Pierce spin columns and centrifuged at 2000g for 1 min to remove the capture resins. Finally, IgG4-depleted plasma samples were assessed for a drop in IgG4 responses (>85% depletion) and unchanged tetanus and influenza total IgG responses (<20% CV) before being used in downstream multiplex and neutralisation assays (Figure S6A–B). In addition, mouse anti-human IgG1 (#3851-14-250; Mabtech; 1 μg/mL, 25 μL/well) was used to assess the rise in IgG1 binding following IgG4 depletion. Experiments were performed in duplicate.
RBD-blocking multiplex assays
To assess the epitopes targeted by class-switched IgG4, we utilised recombinant IgG1 WCSL 55, which had been previously mapped to the ancestral RBM.23 We also utilised recombinant nanobody WNb 36 which bound to the ancestral RBM, as well as nanobody WNb 15 which bound adjacent to the RBM.24 Recombinant IgG1 CR9114—targeting influenza haemagglutinin—was used as a negative control (<20% CV).23 Binding of WCSL 55 or both WNb 15, 36 to ancestral RBD was validated using biotin-conjugated mouse anti-human total IgG (1 μg/mL, 25 μL/well; Mabtech) or PE-conjugated rabbit anti-camelid VHH cocktail (#A02018-200; 1 μg/mL, 25 μL/well; Genscript). Blocking antibodies/nanobodies (10 μL; final dilution 500 ng/mL) were first added to the working bead mixture (20 μL) in the absence of sample and the beads were blocked for 1 h at RT on a shaker. Subsequently, diluted plasma samples (30 μL) (final dilutions: IgG4 1:3200; total IgG, 1:12800) were added to the wells and the multiplex assays were conducted as described above (final dilution 250 ng/mL of blocking antibody/nanobody). Experiments were performed in duplicate.
SARS-CoV-2 neutralisation assay
A replicating SARS-CoV-2 neutralisation assay was performed as previously described,25 against the ancestral variant with D614G mutation, Omicron BA.1 and Omicron XBB.1.5 isolates. Infectivity of virus stocks was then determined by titration on HAT-24 cells (a clone of transduced HEK293T cells stably expressing human ACE2 and TMPRSS2).26 Virus stocks were titrated in quintuplicate in three independent experiments to obtain mean 50% infectious dose (ID50) values.
To determine serum neutralisation activity, IgG4-depleted and non-depleted plasma samples were diluted 3-fold (1:50–1:109,350) and incubated with SARS-CoV-2 virus at a final concentration of 2 × ID50 at 37 °C for 1 h. Next, 40,000 freshly trypsinised HAT-24 cells in DMEM with 5% FCS were added and incubated at 37 °C. “Cells only” and “Virus + Cells” controls were included to represent 0% and 100% infectivity respectively. After 48 h, 10 μL of alamarBlue Cell Viability Reagent (ThermoFisher) was added into each well and incubated at 37 °C for 1 h. The reaction was then stopped with 1% SDS and read on a FLUOstar Omega plate reader. The relative fluorescent units (RFU) measured were used to calculate %neutralisation with the following formula: (“Sample”–“Virus + Cells”) ÷ (“Cells only”–“Virus + Cells”) × 100. IC50 values were determined using four-parameter non-linear regression in GraphPad Prism with curve fit constrained to have a minimum of 0% and a maximum of 100% neutralisation.
Paired IgG1 and IgG4 depletion for neutralisation assay
Paired IgG1 and IgG4 depletion were only performed on a subset of plasma samples 4 weeks post fourth mRNA dose that displayed >10% IgG4/Total IgG binding responses against the ancestral RBD and >1000 neutralising IC50 against the ancestral virus (n = 13). This allowed us to retain detectable neutralising titres in the depleted fractions following dilution. Briefly, plasma samples (50 μL) were first diluted 1:5 in sterile PBS (250 μL final volume; Gibco, Life Technologies). Capture Select IgG1 (Hu) Affinity Matrix slurry (60 μL/tube, #191303005; Thermo Fisher Scientific) or Capture Select IgG4 (Hu) Affinity Matrix slurry (5 μL/tube, #2942902010; Thermo Fisher Scientific) were then added respectively to each pair of diluted plasma samples. Samples were incubated 1 h at RT on a shaker. Samples were then transferred to pre-wet Pierce spin columns and centrifuged at 2000g for 1 min to remove the capture resins. Next, the paired IgG1-and IgG4-depleted plasma samples were assessed for a drop in IgG1 or IgG4 binding to ancestral RBD respectively using multiplex array (>95% depletion). Total IgG binding to ancestral RBD was also assessed across paired depleted plasma samples. Paired depleted plasma samples were further diluted in PBS as required to achieve similar levels of total IgG binding to ancestral RBD across each sample pair (<15% CV). Finally, paired depleted plasma samples were assessed in neutralisation assays.
Statistical analysis
Statistical analyses were performed with Graphpad Prism 10.4.2. Interpolation graphs for calculating antibody concentrations were performed using 4 PL curves fitted using least squares regression. A minimum cut-off of 10 ng/mL and 10 pg/mL were selected for plasma and saliva samples respectively due to the limits of detection of each assay. Interpolated concentrations across the respective dilutions were averaged while censoring saturating concentrations (plasma <20% CV; saliva <30% CV). Group analyses across timepoints utilised the Kruskal–Wallis test with Dunn's multiple comparisons test. Group analyses across variants used the Friedman test with Dunn's multiple comparisons test. Spearman's correlations were used to assess the relationships between FcγRs and antibodies. Paired analyses were performed using Wilcoxon matched-pairs signed rank test, while unpaired analyses utilised Mann–Whitney U test where applicable. Comparisons of paired FcγR responses across sarbecoviruses were performed using Wilcoxon matched-pairs signed rank test, with Holm-Šídák's multiple comparison test. No formal sample size determination or randomisation was performed in this study as samples were instead determined by availability across the 3 biobanks (Doherty, WEHI, VC2). Experiments were not blinded but were performed in duplicates and analysed by multiple researchers. Exclusion criteria are listed in the section “Study Participants”.
Ethics
The studies were approved by ethics committees at the Royal Melbourne Hospital (no. 2021/272), University of Melbourne (nos. 13793, 23497, 28318, 2056689), the Walter and Eliza Hall Institute (no. 20/08), Melbourne Health (no. RMH69108), and Austin Hospital (no. 75984/Austin-2021). All study participants provided written informed consent in accordance with the Declaration of Helsinki.
Role of funders
Funders had no role in study design, data collection, data analyses, interpretation or writing of this report
Results
IgG4 remains significantly elevated in the mRNA cohort after fourth mRNA dose
We consolidated paired plasma and saliva samples collected after mRNA boosters across 3 biobanks in Melbourne, Australia (Doherty, WEHI, VC2) (Table S1). All boosters were given more than 3 months after the most recent vaccination or COVID-19 infection. To evaluate the effects of repeated mRNA boosters on class-switched IgG4, we compared responses in the mRNA cohort following their third and fourth mRNA doses (first and second mRNA boosters) (Fig. 1a) using a SARS-CoV-2 multiplex array (Figure S1b–d)4,6
Fig. 1.
Increased proportion of ancestral spike IgG4 in plasma and saliva following repeated mRNA boosters. (a and b) Vaccination schedule and timeline of sample collection pre- and post-mRNA booster in cohorts who either received primary (a) mRNA vaccines (mRNA) or (b) adenoviral vector vaccines (Vaxzevria). (c–f) Proportions of ancestral (c, e) RBD- or (d, f) S Trimer-specific IgG4 (% IgG4/total IgG) antibodies in both plasma and saliva from the (c–d) mRNA cohort after their first (mRNA Dose 3; n = 46) and second (mRNA Dose 4; n = 46) mRNA boosters or the (e–f) Vaxzevria cohort after their second (mRNA Dose 2; n = 39) and fourth (mRNA Dose 4; n = 20) mRNA boosters. Numbers above list the %IgG4 present at each timepoint. Pie graphs describe the fraction of the cohort with detectable IgG4 responses above pre-pandemic baseline (also see Figure S2). Statistical significance was calculated within each group against their earliest listed timepoint using the 2-tailed Kruskal–Wallis test with Dunn's correction for multiple comparisons. Where significant, p values were reported (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).
In line with previous observations,5,6 ancestral spike-specific (RBD, S Trimer) plasma IgG4 rose over 10-fold (Kruskal–Wallis corrected; p < 0.01) 2 weeks after the mRNA cohort received their third mRNA dose (Figure S2a–b) and decayed after 12 weeks, with 82% of the cohort retaining class-switched IgG4 above background (pie chart; Fig. 1c and d). Saliva trended similarly to plasma although only IgG4 responses to ancestral S Trimer were significant 2 weeks post third mRNA dose (Kruskal–Wallis corrected; p < 0.01), and just 33% of the cohort maintained elevated IgG4 to both ancestral RBD and S Trimer after 12 weeks (Fig. 1c and d; Figure S2c–d).
By contrast, 93% and 70% of the mRNA cohort had detectable IgG4 levels to ancestral spike (RBD, S Trimer) in their plasma and saliva respectively just prior to their fourth mRNA dose (Fig. 1c and d; Figure S2a–b). This rise in IgG4 could be driven by the breakthrough Omicron infections that occurred between the third and fourth mRNA doses (41% of the cohort; Table S1).6 Class-switched IgG4 targeting ancestral spike continued to rise after the fourth mRNA dose and remained elevated over baseline in both plasma and saliva in over 90% of the mRNA cohort (Kruskal–Wallis corrected; p < 0.0001) (Fig. 1c and d; Figure S2a–b).
Ancestral spike-specific IgG4 in both compartments (blood, saliva) trended similarly to that observed for total IgG (Figure S2a–d). However, when we calculated the percentage of ancestral spike-specific IgG4 to total IgG, we noticed that IgG4 formed a larger proportion of all ancestral spike-specific IgG in both plasma and saliva by the fourth mRNA dose (2W mRNA Dose 4; RBD, S Trimer; Plasma: 9%, 7%; Saliva: 14%, 12%; Kruskal–Wallis corrected; p < 0.001), as compared to the third mRNA dose (2W mRNA Dose 3; RBD, S Trimer; Plasma: 0%, 1%; Saliva: 1%, 1%). These percentages of IgG4 to total IgG in plasma and saliva remained elevated for up to 12 weeks post fourth mRNA dose (Fig. 1c and d).
Smaller proportion of IgG4 observed with Vaxzevria cohort even after fourth mRNA dose
Given that we previously observed minimal class-switched IgG4 in the Vaxzevria cohort following their first mRNA booster/dose,20 we decided to investigate the effects after their second and fourth mRNA boosters/dose (Fig. 1b; Figure S2e–f). These booster schedules were selected to line up with our mRNA cohort which received two mRNA boosters (ie. their fourth mRNA dose in total).
Comparing antibody responses before the second booster, concentrations of ancestral S Trimer-specific plasma IgG4 for the Vaxzevria cohort (second mRNA dose) were 100-fold lower than that for the mRNA cohort (fourth mRNA dose) (median [IQR]; 27 [13–137] ng/mL vs 3248 [627–8051] ng/mL) (Figure S2b, f). Ancestral spike-specific (RBD, S Trimer) IgG4 antibodies in plasma and saliva rose significantly in the Vaxzevria cohort after the second mRNA dose and peaked at 4 weeks (Kruskal–Wallis corrected; p < 0.0001) (Figure S2e–h). Although these responses decayed by 12 weeks, over 75% of individuals at this timepoint still retained elevated plasma and salivary IgG4 levels over background (Kruskal–Wallis corrected; p < 0.05) (Fig. 1e and f; Figure S2e–h).
By the start of their fourth mRNA dose, over 90% of the Vaxzevria cohort had detectable ancestral S Trimer-specific IgG4 above background in their plasma and saliva (Kruskal–Wallis corrected; p < 0.05) (Figure S2e–h). IgG4 concentrations rose significantly 4 weeks after their fourth mRNA dose (Kruskal–Wallis corrected; p < 0.0001) despite remaining lower than that observed with the mRNA cohort 4 weeks post fourth mRNA dose (S Trimer; Vaxzevria vs mRNA; median [IQR]; 1186 [388–2920] vs 8848 [2930–21029] ng/mL) (Figure S2e–h). The reduced rates of IgG4 class-switching by the Vaxzevria cohort were further highlighted by the lower percentages of ancestral spike-specific IgG4 to total IgG present in plasma and saliva respectively even after their fourth mRNA dose (4W mRNA Dose 4; RBD, S Trimer; Plasma: 2%, 3%; Saliva: 7%, 9%; Kruskal–Wallis corrected; p < 0.05) (Fig. 1e and f).
Proportion of IgG4 to total IgG remain elevated across circulating SARS-CoV-2 variants
Antibodies targeting circulating viral variants play key roles in modulating breakthrough infections.2,6,17 Recently, Martín Pérez et al. demonstrated that elevated plasma IgG4 against ancestral or BQ.1.1 RBD was associated with a heightened risk of symptomatic breakthrough infections.17 To understand if IgG4 remains elevated against recent Omicron variants, we investigated RBD-specific IgG4 in both plasma and saliva against the ancestral virus and key Omicron variants BA.1, XBB.1.5 and JN.1 that were present from 2022 to 2024 and incorporated into updated vaccines (mRNA: Booster 2, mRNA Dose 4; Vaxzevria: Booster 4, mRNA Dose 4) (Fig. 2a–c; Figure S3a).
Fig. 2.
Spike IgG4 antibodies were influenced by ancestral imprinting. (a–c) Concentrations (ng/mL; pg/mL) and proportions (% IgG4/total IgG) of (a–b) plasma and (c) salivary IgG4 antibodies targeting ancestral and Omicron variants (BA.1, XBB.1.5, JN.1) RBDs in samples collected from the (a, c) mRNA (4W mRNA Dose 4; n = 24) or (b) Vaxzevria cohorts (4W mRNA Dose 4; n = 20). (d–f) Concentrations (ng/mL; pg/mL) and proportions (% IgG4/total IgG) of (d–e) plasma and (f) salivary IgG4 antibodies targeting ancestral and Omicron variants (BA.1, XBB.1.5, JN.1) S Trimers in samples collected from the (d, f) mRNA (4W mRNA Dose 4; n = 24) or (e) Vaxzevria cohorts (4W mRNA Dose 4; n = 20). Numbers listed atop show the % IgG4 present at each timepoint, while numbers listed under triangles describe fold decrease in variant-specific IgG4 as compared to that for ancestral. (g–i) Proportion of IgG4 and total IgG binding to ancestral and Omicron S Trimers in (g–h) plasma and (i) saliva samples from the (g, i) mRNA (4W mRNA Dose 4; n = 21) or (H) Vaxzevria cohorts (4W mRNA Dose 4; n = 17) which are not cross-reactive to ancestral RBD in successfully depleted samples. Pre-pandemic baselines are illustrated by dotted lines. Statistical significance was calculated within each group against the ancestral response using the 2-tailed Friedman test with Dunn's correction for multiple comparisons. Where significant, p values were reported (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).
Plasma IgG4 targeting BA.1 and XBB.1.5 RBDs were detectable above background at the respective timepoints, although these responses were markedly lower than that detected for ancestral RBD (mRNA, Vaxzevria; BA.1: 46-, 14-fold less; Friedman corrected; p < 0.05; XBB.1.5: 137-, 32-fold less; Friedman corrected; p < 0.0001) (Fig. 2a and b). These ancestrally biased RBD-specific IgG4 responses trended similarly with that observed with total IgG (mRNA, Vaxzevria; BA.1: 27-, 27-fold less; Friedman corrected; p < 0.05; XBB.1.5: 80-, 49-fold less; Friedman corrected; p < 0.0001) (Figure S3b–c). As such, the overall percentage of RBD IgG4 to total IgG across variants for the mRNA cohort remained comparable to the ancestral, though a modest increase was detected for the Vaxzevria cohort (ancestral vs BA.1; 2% vs 4%; Friedman corrected; p < 0.01) (ancestral vs XBB.1.5; 2% vs 3%; Friedman corrected; p < 0.001) (Fig. 2a and b).
In contrast, the differences in plasma IgG4 concentrations against the S trimer of the ancestral strain relative to variants were much smaller (mRNA, Vaxzevria; BA.1: 3-, 2-fold less; Friedman corrected; p < 0.0001; XBB.1.5: 3-, 3-fold less; Friedman corrected; p < 0.0001) (Fig. 2d and e), and in line with that observed for total IgG (Figure S3b–c). Proportions of IgG4 to total IgG were comparable between BA.1 and ancestral S Trimers. On the other hand, we did detect increased percentages of IgG4 against XBB.1.5 (15%; Friedman corrected; p < 0.01) and JN.1 S Trimer (15%; Friedman corrected; p < 0.001) as compared to the ancestral (9%) for the mRNA cohort. A modest increase was also detected for JN.1 S Trimer with the Vaxzevria cohort (ancestral vs JN.1; 3% vs 4%; Friedman corrected; p < 0.01) (Fig. 2d and e).
Complementary observations to plasma were made for salivary IgG4 from both cohorts, although RBD-specific responses to the Omicron variants were markedly weaker in saliva (Fig. 2c and f; Figure S3a). This is in concordance with salivary IgG antibodies being transudates from blood, albeit at lower concentrations (Figure S3b–e). Taken together, our data show that elevated proportions of IgG4 are also detected across Omicron variants including JN.1.
mRNA booster-induced IgG4 trend similarly to total IgG
mRNA vaccines leave immunological imprints which bias RBD-specific antibodies and persist after secondary exposures.27 Over 40% of the mRNA cohort and 90% of the Vaxzevria cohort had Omicron breakthrough infections by their latest booster (mRNA: Booster 2, mRNA Dose 4; Vaxzevria: Booster 4, mRNA Dose 4) (Table S1). Furthermore, the Vaxzevria cohort received the monovalent XBB.1.5 as their fourth mRNA booster/dose (Fig. 1b). To understand if ancestral RBD-imprinting would still influence IgG responses in both cohorts following their respective boosters, we depleted antibodies targeting the ancestral RBD from our samples (Figure S4a–f).
With the mRNA cohort, we found that 90% of IgG4 antibodies targeting BA.1 or XBB.1.5 RBD respectively were still cross-reactive to the ancestral RBD 4 weeks post fourth mRNA dose (Figure S4b). The Vaxzevria cohort showed a slight departure from ancestral imprinting with 83% and 77% drops in IgG4 targeting BA.1 or XBB.1.5 RBDs respectively following 4 weeks post fourth mRNA dose (Figure S4f). We also determined that RBD-specific IgG4 formed over 50% of plasma and salivary responses targeting the ancestral trimer following the fourth mRNA dose (Fig. 2g–i). In contrast, less than 20% of plasma IgG4 targeting JN.1 trimer were still cross-reactive to the ancestral RBD (Friedman corrected; p < 0.0001; Fig. 2g–i). Importantly, IgG4 responses trended similarly to total IgG responses in both cohorts (Fig. 2g–i), suggesting that this observation is likely IgG subclass-independent. Similar trends towards the ancestral RBD were also observed across both cohorts following prior mRNA boosters (mRNA: 2W Booster 1, mRNA Dose 3; Vaxzevria: 2W Booster 2, mRNA Dose 2) (Figure S4g–l), highlighting the challenges in inducing antibodies against novel RBD epitopes of emerging SARS-CoV-2 variants.
Rising proportions of SARS-CoV-2 IgG4 negatively correlate with IgG1 across variants
COVID-19 mRNA boosters are effective at elevating total IgG directed against the SARS-CoV-2 spike protein (Figure S2a–h). In concordance, S trimer-specific IgG1—which usually forms the bulk of S Trimer-specific total IgG responses—is elevated across both the mRNA and Vaxzevria cohorts following their respective mRNA boosters (Figure S5a–b). However, increased class-switching from IgG1 to IgG4 could gradually limit the overall concentration of spike-specific IgG1 antibodies available post-booster. Longitudinal studies on allergen immunotherapy have documented that repeated periodic vaccination over the course of 2–3 years caused initial surges in IgG1 responses which were gradually overtaken by the sustained rise in class-switched IgG4.28,29
Here, we explored if individuals with higher concentrations of spike-specific IgG4, would have a lowered expression of spike-specific IgG1, by correlating the binding concentrations (ng/mL) of S Trimer-specific plasma IgG1 against IgG4 in both cohorts following their respective mRNA boosters (Fig. 3a–d, Table S2). For the mRNA cohort, correlations between their IgG1 and IgG4 binding concentrations grew increasingly negative across all variant S Trimers at 4 weeks after their fourth mRNA dose (r = −0.53 to −0.66; Spearman; p < 0.01) (Fig. 3a and c) and remained significant for up to 12 weeks (Spearman; p < 0.05). On the other hand, S Trimer-specific IgG1 responses from the Vaxzevria cohort only weakly associated with IgG4 after their second mRNA dose (Fig. 3b and d). These negative correlations, however, grew stronger by the fourth mRNA dose and remained significant 4 weeks after their fourth mRNA dose (r = −0.62 to −0.71; Spearman; p < 0.01) (Fig. 3b and d). When we accounted for the proportion of spike-specific IgG4 present in individuals (% IgG4/total IgG), these negative associations with IgG1 grew stronger across both cohorts especially at 4 weeks after their respective mRNA boosters (mRNA: r = −0.79 to −0.83; Spearman; p < 0.0001; Vaxzevria: r = −0.83 to −0.87; Spearman; p < 0.0001) (Figure S5c–f; Table S2).
Fig. 3.
IgG4 responses negatively correlated with IgG1 following repeated mRNA boosters. (a–b) Spearman's correlations of IgG1 and IgG4 antibodies (ng/mL) targeting ancestral and Omicron variants (BA.1, XBB.1.5, JN.1) S Trimer in plasma samples collected from the (a) mRNA (4W mRNA Dose 4; n = 24) or (b) Vaxzevria cohort (4W mRNA Dose 4, n = 20). (c–d) Heat map summarises Spearman's correlations between plasma IgG1 and IgG4 antibodies targeting ancestral and Omicron variants S Trimer from the (c) mRNA (Pre-, 4W, 12W mRNA Dose 4, n = 44) or (d) Vaxzevria cohort (4W, 12W mRNA Dose 2, n = 37) (Pre-, 4W mRNA Dose 4, n = 20) (also see Table S2). Where significant, p values were reported (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).
Increased proportions of class-switched IgG4 in plasma have previously correlated with increased avidity.5,8 To address if these observed negative correlations were driven by competitive binding of IgG4 against IgG1 antibodies in our assay, we depleted total IgG4 from a subset our plasma samples (mRNA: 4W mRNA Dose 4; Vaxzevria: 4W mRNA Dose 4) with clearly detectable levels (>100 ng/mL) of spike-specific IgG4 using capture resins (>94% depleted; Wilcoxon; p < 0.0001; Figure S6a–c). The removal of IgG4 enhanced IgG1 binding to the S Trimer across all variants (13–22% increase; Wilcoxon; p < 0.0001) (Figure S6d). However, the negative correlations of IgG1 binding concentrations to %IgG4 remained significant even after IgG4 antibodies were successfully depleted (Figure S6e), suggesting that the weakened IgG1 responses observed were not solely impacted by competitive binding of increasing IgG4 antibodies.
Increasing IgG4 to Omicron variants negatively associates with Fc receptor binding
Contrary to IgG1 and IgG3, IgG4 binds poorer to FcγRIIa and FcγRIIIa (traditionally associated with ADCP and ADCC respectively) and plays a crucial role in modulating excessive activation by IgG-mediated Fc binding.5, 6, 7 We and others have previously shown that increasing IgG4 after the third mRNA dose negatively correlated with antibody-mediated Fc receptor binding to ancestral spike, particularly FcγRIIIa.5, 6, 7 Here, we expand upon this observation and explore the impact of repeated mRNA boosters particularly focussing on circulating Omicron variants BA.1, XBB.1.5 and JN.1.
Four weeks after their fourth mRNA dose, plasma and salivary antibodies from the mRNA cohort displayed significantly improved engagement (Kruskal–Wallis corrected; p < 0.0001) across both FcγRIIaH and FcγRIIIaV—the higher affinity Fc polymorphisms—against both ancestral and Omicron S Trimers (BA.1, XBB.1.5 and JN.1) (Fig. 4a and b; Figure S7a–b). Importantly, these responses remained elevated above pre-booster levels for up to 12 weeks in both plasma and saliva (Kruskal–Wallis corrected; p < 0.01).
Fig. 4.
Elevated IgG4 negatively associates with Fc receptor binding across Omicron variants. (a–b) FcγRIIIaV binding (median fluorescence intensity; MFI) across ancestral and Omicron variant S Trimers in (a) plasma and (b) saliva samples from the mRNA cohort pre- and post-booster 2 (mRNA Dose 4; n = 44). (c–d) Heat map summarises Spearman's correlations of both FcγRIIaH and FcγRIIIaV respectively against % IgG4 antibodies (% IgG4/total IgG) targeting ancestral and Omicron variants S Trimer in (c) plasma and (d) saliva samples pre- and post-booster 2 from the mRNA cohort (mRNA Dose 4; n = 44) (also see Table S2). (e–f) FcγRIIIaV binding (MFI) across ancestral and Omicron variant S Trimers in (e) plasma and (f) saliva samples from the Vaxzevria cohort pre- and post-booster 4 (mRNA Dose 4; n = 20). (g–h) Heat map summarises Spearman's correlations of both FcγRIIaH and FcγRIIIaV respectively against % IgG4 antibodies targeting ancestral and Omicron variants S Trimer in (g) plasma and (h) saliva samples from the Vaxzevria cohort post-booster 2 (mRNA Dose 2; n = 27), and pre- and post-booster 4 (mRNA Dose 4; n = 20) (also see Table S2). (i–k) Paired FcγRIIIaV responses against ancestral and XBB.1.5 S Trimers before and after successful depletion of IgG4 antibodies in a subset of individuals across both (i) or the respective (j) mRNA (4W mRNA Dose 4; n = 21) and (k) Vaxzevria cohorts (4W mRNA Dose 4; n = 10). Pre-pandemic baselines are illustrated by dotted lines. (a–b) Statistical significance was calculated within each group against the pre-booster timepoint using the 2-tailed Kruskal–Wallis test with Dunn's correction for multiple comparisons. (e-f, i-k) Statistical significance was calculated using Wilcoxon matched-pairs signed rank test. Where significant, p values were reported (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).
However, correlations between plasma %IgG4 and Fc receptor binding to Omicron S Trimer also grew significantly negative in the mRNA cohort at 4 weeks post fourth mRNA dose (Fig. 4c; Figure S7c; Table S2) (p < 0.0001; FcγRIIaH and FcγRIIIaV; across all variants). This association weakened after 12 weeks, particularly for XBB.1.5 and JN.1 to both FcγRIIaH (Spearman; p < 0.001) and FcγRIIaR (the weaker polymorphism; Spearman; p < 0.01) respectively receptors (Fig. 4c; Figure S7d; Table S2). In contrast, the correlation of plasma %IgG4 to FcγRIIIaV and FcγRIIIaF (weaker polymorphism) still remained strongly negative (r = −0.81 to −0.80; r = −0.85 to −0.80; Spearman; p < 0.0001) (Fig. 4c; Figure S7d; Table S2), highlighting the differential effects of IgG4 against the different FcγRs.7
While salivary antibodies for the mRNA cohort displayed similar trends to plasma 4 weeks after the fourth mRNA dose, the negative associations of %IgG4 to FcγRIIaH and FcγRIIIaV in saliva were weaker than that observed in plasma, particularly for JN.1 (saliva vs plasma; FcγRIIaH, r = −0.74 vs −0.34; FcγRIIIaV r = −0.81 vs −0.65) (Fig. 4d). This could be in line with the concentration-dependent changes in IgG4 activity, in which IgG4 instead compensates for Fc receptor binding, particularly FcγRIIa, in environments with low IgG1 concentrations,7 although the limits of detection of our assays could also contribute to the weaker associations observed.
Plasma and salivary antibodies from the Vaxzevria cohort also displayed significantly improved engagement to Omicron variants (Wilcoxon; p < 0.05) across both FcγRIIaH and FcγRIIIaV following their fourth mRNA dose (Fig. 4e and f; Figure S7e–f). However, plasma %IgG4 correlated negatively with both FcγRIIaH and FcγRIIIaV binding respectively 4 weeks after the fourth mRNA dose (r = −0.73 to −0.88; r = −0.78 to −0.85; Spearman; p < 0.001) (Fig. 4g; Figure S7g; Table S2). Similar negative trends were also observed for both FcγRIIaR and FcγRIIIaF binding (weaker polymorphism) (Figure S7h; Table S2). The association of salivary %IgG4 to Fc receptor binding on the other hand remained weak despite the fourth mRNA dose (Fig. 4h), highlighting the potential effects of different antibody concentrations present across both systemic and mucosal compartments.
To better evaluate the direct impact of IgG4 on the observed Fc responses, we depleted total IgG4 from a subset our plasma samples from both cohorts (Figure S6a–c). The removal of IgG4 rescued S Trimer specific antibody-medicated engagement to FcγRIIIaV in plasma, particularly for the mRNA cohort (Ancestral: 20% increase; XBB.1.5: 15% increase; Wilcoxon; p < 0.0001) (Fig. 4i–k, Figure S6f). However, changes to FcγRIIaH responses were more variable, corroborating with the differential effects of IgG4 on the respective FcγRs (Figure S6g).
Limited IgG4 back-boosting to seasonal HCoVs
Back-boosting of antibodies targeting seasonal human coronaviruses (HCoVs), particularly OC43 and HKU1, post-mRNA vaccination has been attributed to cross-reactivity to the conserved S2 region30,31 (Fig. 5a). Here, we investigated if IgG4 antibodies induced post-mRNA booster were cross-reactive to seasonal HCoVs (Figure S8a). While cross-reactive IgG4 against seasonal HCoVs, particularly HKU1, were marginally detectable 4 weeks post fourth mRNA dose in the mRNA cohort, they were limited to only 1% or less of the total IgG4 generated against ancestral SARS-CoV-2 (Friedman corrected; p < 0.001) (Fig. 5b and c). This is in line with previous observations that cross-reactive antibodies to seasonal HCoVs were mainly elicited by primary COVID-19 vaccines in adults.32
Fig. 5.
Cross-reactive IgG4 bind sarbecoviruses and impact Fc-engagement. (a) Phylogenetic tree of seasonal human coronaviruses (HCoVs) and SARS-CoV-2. (b) Plasma IgG4 concentrations (ng/mL) from the mRNA cohort against HCoVs and SARS-CoV-2 S Trimers. (c) HCoV S Trimer-specific plasma IgG4 concentrations from the mRNA cohort expressed as a percentage of their SARS-CoV-2 plasma IgG4 response. (d) Phylogenetic tree of sarbecoviruses and MERS. (e–f) Sarbecovirus S Trimer-specific plasma IgG4 concentrations from the (e) mRNA and (f) Vaxzevria cohorts expressed as a percentage of their respective SARS-CoV-2 plasma IgG4 response. (g–h) Proportions S Trimer-specific plasma IgG4 (% IgG4/total IgG) antibodies across sarbecoviruses from the (g) mRNA and (h) Vaxzevria cohorts. (i–j) FcγRIIIaV binding (median fluorescence intensity; MFI) across sarbecoviruses pre- and 4 weeks post-their respective boosters for the (i) mRNA and (j) Vaxzevria cohorts. (k–l) Heat map summarises Spearman's correlations of both FcγRIIaH and FcγRIIIaV respectively against % IgG4 antibodies binding sarbecoviruses in plasma 4 weeks post-their respective boosters for the (k) mRNA and (l) Vaxzevria cohorts (also see Table S2). (m) % IgG4 binding to sarbecovirus S Trimers in plasma from the mRNA (4W mRNA Dose 4; n = 23) which are not cross-reactive to ancestral RBD in successfully depleted samples. Pre-pandemic baselines are illustrated by dotted lines. (b-c, e, g, i, k) Plasma samples from the mRNA cohort (4W mRNA Dose 4; n = 24). (f, h, j, l) Plasma samples from the Vaxzevria cohort (4W mRNA Dose 4; n = 20). (b-c, e-h, m) Statistical significance was calculated within each group against the SARS-CoV-2 response using the 2-tailed Friedman test with Dunn's correction for multiple comparisons. (i–j) Statistical significance was calculated using Wilcoxon matched-pairs signed rank test with Holm-Šídák test for multiple comparisons. Where significant, p values were reported (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).
Cross-reactive IgG4 negatively impact Fc engagement to sarbecoviruses
SARS-related beta coronaviruses (sarbecoviruses) have been highlighted by the WHO as priority pathogens with outbreak potential.33 These viruses usually circulate in non-human mammal species, such as bats (RaTG13, RsSHC014, HKU3) and pangolins (GX-P5L, GD-1), but are known to make cross-species jumps to humans (SARS, SARS-CoV-2)34 (Fig. 5d). Antibodies generated by COVID-19 vaccines have been demonstrated to be cross-reactive across a range of sarbecoviruses and even induce neutralising activity.35,36 To understand the impact of cross-reactive IgG4 on sarbecoviruses, we interrogated post booster plasma samples from both cohorts (mRNA: 4W mRNA Dose 4; Vaxzevria: 4W mRNA Dose 4) (Figure S8b).
Over 50% of ancestral S Trimer specific IgG4 were cross-reactive to SARS-CoV-2-related sarbecoviruses RaTG13, GX-P5L and GD-1 (90–97% identity to ancestral S Trimer) across both cohorts (Fig. 5e and f). Lower levels of IgG4 cross-reactivity (19–41% cross-reactive IgG4; Friedman corrected; p < 0.0001) remained clearly detectable for SARS and SARS-related sarbecoviruses (RsSHC014, HKU3) with reduced identity scores (>70% identity to ancestral S Trimer). Only the antigenically distant MERS (35% identity to ancestral Trimer) showed barely detectable levels of cross-reactive IgG4 post-booster (1%; Friedman corrected; p < 0.0001).
As before with SARS-CoV-2 variants, IgG4 cross-reactivity across sarbecoviruses trended similarly to that for total IgG (Figure S8c, e). Consequently, the proportion of IgG4 to total IgG antibodies remained fairly comparable between SARS-CoV-2 (mRNA: 9%; Vaxzevria: 4%) and SARS-CoV-2-related sarbecoviruses (median % IgG4 across variants; mRNA: 10–14%; Vaxzevria: 4–7%) within each respective cohorts, though significant increases were detected across both cohorts for both SARS-related sarbecoviruses RsSHC014 and HKU3 (mRNA: 13–18%; Vaxzevria: 10–11%; Friedman corrected; p < 0.001) (Fig. 5g and h).
Antibody-mediated FcγRIIaH and FcγRIIIaV engagement also showed significant increases across all sarbecoviruses tested across both cohorts (Wilcoxon corrected; p < 0.01) (Fig. 5i and j; Figure S8d, f), despite negatively correlating with the respective proportions of IgG4 responses (% IgG4/total IgG) (Fig. 5k, l; Table S2). Salivary responses trended similarly to plasma across both cohorts (Figure S9a–j; Table S2). Altogether, our findings suggest that cross-reactive antibodies post-mRNA boosters could offer broad Fc-mediated responses against sarbecoviruses, though these might be dampened by rising levels of cross-reactive IgG4.
Majority of ancestrally imprinted IgG targets the receptor-binding motif (RBM)
SARS-CoV-2-related sarbecoviruses RaTG13, GX-P5L and GD-1 share high sequence similarity to the ancestral RBD (87–97%) (Fig. 5d). As such, we wondered if ancestral imprinting could be contributing to the bulk of cross-reactive IgG observed post-booster in the mRNA cohort (Fig. 5e). Utilising ancestral RBD-depleted plasma from the mRNA cohort as before, we observed that only GD-1, and not RaTG13 or GX-P5L, showed similar losses to IgG4 and total IgG binding as ancestral SARS-CoV-2 (Friedman corrected; p < 0.001) (Fig. 5m; Figure S8g). Delving into their RBD sequences revealed that despite sharing several mutation sites with RaTG13 and GX-P5L across the RBD, GD-1 only had a single mutation in the RBM (shaded in green; Figure S8h) which facilitates binding to the ACE2 receptor.
Previous literature has shown that a high frequency of serum antibodies targeted the RBM following mRNA vaccination and subsequent breakthrough infections.37 To assess if class-switched IgG4 was targeting the RBM, we performed competitive blocking experiments, using recombinant IgG1 monoclonal antibody (mAb) WCSL 55, which had been previously mapped to the ancestral RBM23 (Fig. 6a; Figure S10a). The addition of WCSL 55 IgG1 mAb blocked over 50% of IgG4 binding to the ancestral RBD (Wilcoxon; p < 0.0001) following boosting in both the mRNA (4W mRNA Dose 4) and Vaxzevria cohorts (4W mRNA Dose 4) (Fig. 6b and c). In contrast, IgG4 binding remained unchanged when similar concentrations of the influenza HA mAb CR9114 were added as controls (Figure S10b).
Fig. 6.
Majority of ancestral-RBD imprinted IgG4 targets the receptor-binding motif and IgG4 contributes to ancestral neutralisation. (a) Diagram representing the binding sites for monoclonal WCSL 55 and nanobodies WNb 36 and WNb 15 on the RBD. (b–c) Plasma IgG4 binding concentrations to ancestral RBD from a subset of individuals with clearly detectable IgG4 responses from the (b) mRNA (4W mRNA Dose 4; n = 23) and (c) Vaxzevria cohorts (4W mRNA Dose 4; n = 18) respectively, in the presence/absence of blocking monoclonal/nanobodies WCSL 55, WNb 36 and WNb 15. (d) Paired plasma neutralisation IC50 against ancestral, BA.1 and XBB.1.5 live viruses before and after successful depletion of IgG4 antibodies in a subset of individuals across both the mRNA (pink; 4W mRNA Dose 4; n = 21) and Vaxzevria cohorts (purple; 4W mRNA Dose 4; n = 10). (e) Spearman's correlations of plasma neutralisation IC50 and the proportion of RBD IgG4 antibodies (% IgG4/total IgG) across ancestral, BA.1 and XBB.1.5 for both the mRNA (4W mRNA Dose 4; n = 24) and Vaxzevria cohorts (4W mRNA Dose 4; n = 20). (f) Spearman's correlations RBD IgG1 concentrations (ng/mL) and RBD %IgG4 for ancestral, BA.1 and XBB.1.5 across both the mRNA (4W mRNA Dose 4; n = 24) and Vaxzevria cohorts (4W mRNA Dose 4; n = 20). (g) Spearman's correlations of plasma neutralisation IC50 and the concentration of RBD IgG1 antibodies across ancestral, BA.1 and XBB.1.5 for both the mRNA (4W mRNA Dose 4; n = 24) and Vaxzevria cohorts (4W mRNA Dose 4; n = 20). Pre-pandemic baselines and neutralisation assay threshold (IC50 = 50) are illustrated by dotted lines. (b–d) Statistical significance was calculated using Wilcoxon matched-pairs signed rank test. Where significant, p values were reported (∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001).
To further validate that ancestrally-imprinted IgG4 was indeed targeting the RBM, we repeated the competitive blocking experiments using nanobody WNb 36 which bound to the RBM, and for comparison we used nanobody WNb 15 which bound adjacent to the RBM24 (Fig. 6a; Figure S10a). WNb36 was successful in blocking a majority of IgG4 (mRNA: 71%; Vaxzevria: 43%; Wilcoxon; p < 0.0001) (Fig. 6b and c) antibodies from binding to the ancestral RBD, while WNb15 only blocked a smaller proportion of IgG4 (mRNA: 4%; Vaxzevria: 21%; Wilcoxon; p < 0.01) (Fig. 6b and c). Similar trends were also observed with total IgG binding to the ancestral RBD, with WNb36 blocking a larger proportion of total IgG than WNb 15 (mRNA, Vaxzevria; WNb36: 77%, 71%; Wilcoxon; p < 0.0001; WNb15: 28%, 26%; Wilcoxon; p < 0.0001) (Figure S10c–d), corroborating that ancestral imprinting towards the RBM is likely IgG subclass-independent.
Increasing proportions of IgG4 impact neutralisation against Omicron variants
The RBM is a mutational hotspot particularly for emerging Omicron variants (Figure S10e) and has been implicated for the loss of neutralising activity directed at newer variants.38 Given that most booster-induced plasma IgG is imprinted to the ancestral RBD (Fig. 2a and b), particularly the RBM (Fig. 6b and c), we hypothesised that IgG4 could contribute less towards the neutralisation of Omicron variants. To investigate this, we performed live virus neutralisation assays (ancestral, BA.1, XBB.1.5) using untreated and IgG4-depleted plasma from a subset of the mRNA (4W mRNA Dose 4) and Vaxzevria (4W mRNA Dose 4) cohorts with clearly detectable levels of spike-IgG4 and neutralising activity (Figure S6b-c; Figure S11a).
As expected, while the loss of IgG4 resulted in a significant decrease in neutralising activity against the ancestral virus (−22%; Wilcoxon; p < 0.0001), its effects were less pronounced for both BA.1 (−5%; Wilcoxon; p < 0.05) and XBB.1.5 (+4%; Wilcoxon; ns) respectively (Fig. 6d; Figure S11b-c). Understandably, as the Vaxzevria cohort received more mRNA boosters, particularly the later XBB.1.5 monovalent boosters, they exhibited stronger neutralising responses to XBB.1.5 than the mRNA cohort (Fig. 6d and e; Figure S11a). Despite the modest age gap between the mRNA and Vaxzevria cohorts (Table S1) due to vaccine availability and the prioritisation of the elderly during vaccine/booster rollouts, age did not influence neutralisation across our cohorts (Figure S11d–e).
Interestingly, when we explored the association of neutralisation (IC50) against proportion of IgG4 (% IgG4/total IgG) in untreated plasma, we observed strong negative correlations from our cohorts across all strains; ancestral (mRNA: r = −0.45; Spearman; p < 0.05; Vaxzevria: r = −0.77; Spearman; p < 0.0001), Omicron variants BA.1 (mRNA: r = −0.46; Spearman; p < 0.05; Vaxzevria: r = −0.58; Spearman p < 0.01) and XBB.1.5 (Vaxzevria: r = −0.77; Spearman; p < 0.001) (Fig. 6e). We had previously observed that the concentration of S Trimer specific IgG1 negatively correlated with increasing % IgG4 post booster (Figure S6c–f). Here, we found that RBD-specific IgG1 concentrations also negatively associated with % IgG4 across variants in both the mRNA (r = −0.75 to −0.81; Spearman; p < 0.0001) and Vaxzevria cohorts (r = −0.72 to −0.78; Spearman; p < 0.001) (Fig. 6f; Figure S11f–g). Furthermore, RBD-specific IgG1 concentrations positively correlated with neutralisation across variants (Spearman; p < 0.05) (Fig. 6g), suggesting that neutralisation is primarily linked to the available concentrations of IgG1.
To explore the impact of IgG subclasses on neutralisation, we performed IgG1 or IgG4 depletions (Figure S12a–b) on a small subset of plasma samples collected 4 weeks post fourth mRNA dose across both cohorts which displayed enhanced neutralising activity (>1000 neutralising IC50) and a larger proportion of class-switched IgG4 (>10%) towards the ancestral virus/RBD. Depleted fractions were diluted so that each pair would show similar levels of total IgG binding to the ancestral RBD (Figure S12c), before being utilised in an ancestral live virus neutralisation assay. IgG4 depletion (i.e. samples enriched in IgG1) had 55% (Wilcoxon; p < 0.01; Figure S12d) more neutralising activity then the IgG1 depleted fraction (enriched in IgG4), despite displaying similar total IgG binding to the ancestral RBD. While this observation is supportive of our hypothesis that SARS-CoV-2 specific IgG4 antibodies may have poorer neutralising capacity than IgG1 antibodies, we acknowledge that this exploratory experiment is limited by the small sample size.
Discussion
mRNA vaccines have immense potential and versatility to rapidly pivot and address a wide range of diseases and emerging pathogens. Understanding the influence of elevated IgG4 antibodies following repeated COVID-19 mRNA vaccines/boosters is crucial for informing future vaccination strategies and improving vaccine designs to maximise the protective capacity of mRNA vaccines. Here, we demonstrate that the proportions of class-switched plasma and salivary IgG4 (% IgG4/total IgG) rose significantly in both our mRNA and Vaxzevria cohorts after 4 exposures each to the COVID-19 mRNA vaccine/booster. It has been suggested that priming from initial vaccine exposure could shape downstream class-switching to IgG4.13 While the Vaxzevria cohort indeed had a smaller overall proportion of IgG4 as compared to the mRNA cohort, gradual sustained growth in IgG4 responses was still observed up to the fourth mRNA booster/dose, suggesting that repeated mRNA boosters could overcome initial vaccine priming.
Spike-specific IgG4 antibodies trended similarly to total IgG responses across Omicron variants in both plasma and saliva, resulting in comparable proportions of increased IgG4 (% IgG4/total IgG) across variants, including JN.1. Both of our cohorts were strongly primed by vaccination against the ancestral spike, which skewed their total IgG and by extension IgG4 responses. Ancestral imprinting—particularly against the mutational hotspot RBM—could limit cross-reactive IgG antibodies capable of targeting the RBDs from emerging variants.37 Indeed, the majority of IgG antibodies targeting JN.1 across both cohorts were non-RBD, which is in line with literature showing a declining efficiency of boosting neutralising responses against emerging variants with each subsequent boosters.2 This bias towards non-RBD antibodies places a heavier emphasis on non-neutralising responses, which are negatively impacted by rising IgG4. Future studies should explore if repeated Omicron monovalent mRNA boosters would influence the specificity of class-switched IgG4.
In contrast, populations who received the inactivated ancestral COVID-19 vaccine (Coronavac) first followed by repeated Omicron breakthrough infections have been shown to override ancestral imprinting.38,39 Particularly, antibodies targeting regions of the RBM in these populations were dominantly Omicron-specific. It remains to be seen if these populations could go on to develop Omicron-specific IgG4 responses, and how that would impact humoural immunity against future Omicron variants. Similarly, while children have been shown to develop ancestral spike-specific IgG4 following their second COVID-19 ancestral mRNA vaccine,14 little is known if children whose first exposure were Omicron antigens would respond differently.
In allergen immunotherapy, induction of class-switched IgG4 following repeated allergen exposures dampen IgE-mediated inflammatory responses.28 Strobl et al. have shown that the first mRNA booster induced a surge in RBD-specific IgE responses, which in turn correlated significantly with elevated IgG4 antibodies.40 Mice models demonstrated that mRNA vaccines were more potent than Vaxzevria at inducing S protein in draining lymph nodes for up to 30 h post-vaccination,41 while studies on immunosuppressive drugs have highlighted the importance of IL-4, IL-13 and TNF in RBD-specific IgG4 class switching.42 At present, it remains unclear if the induction of IgG4 antibodies is driven by mRNA vaccine formulations, dosage or elevated expression of immunogens. Future studies could utilise updated low-dose mRNA vaccines such as mNexspike, as well as bivalent mRNA COVID-19 influenza combination vaccines to explore the impact of dosage and if mRNA-induced IgG4 responses would extend beyond SARS-CoV-2 antigens.
Co-administration of COVID-19 mRNA boosters and the annual influenza vaccine has been adopted to coincide with seasonal outbreaks and boost anti-viral immunity.4,43 Current literature shows that concurrent administration does not result in the rise of influenza-specific IgG4 by 6 months, suggesting that the induction of IgG4 could be formulation or immunogen-specific.43 However, as we have observed with our Vaxzevria cohort, repeated exposures to the mRNA booster can eventually lead to a gradual sustained rise in IgG4 over time. As co-administration becomes more commonplace yearly, particularly with vulnerable populations, follow-up studies should be done to explore if influenza-specific IgG4 antibodies remain undetectable.
Reports of elevated spike IgG4 associating with decreased neutralising activity,17,44 with one study suggesting that IgG4 may be associated with increased risk of breakthrough infections, though further research with larger cohorts are needed to confirm this. Here, we also observe that increased proportions of IgG4 negatively correlated with neutralisation of ancestral, BA.1 and XBB.1.5 live viruses. Interestingly, depletion of IgG4 antibodies suggested that IgG4 played a more modest role in the direct neutralisation of Omicron variants. Class-switching to IgG4 is thought to increase antibody avidity, which in turn gradually replaces the binding activity of IgG1.5,28,29 In this way, class-switching from IgG1 to IgG4 is thought to retain neutralising activity, while dampening non-neutralising responses to reduce inflammation and allergies. However, Garcia et al. have suggested that mRNA vaccines could promote antibody class switching without somatic hypermutations, limiting affinity maturation.45 Furthermore, most IgG4 antibodies spontaneously undergo Fab arm exchange in vivo, and the resulting monovalency could greatly limit their binding capability and in turn neutralising activity against Omicron variants.15,16 While the gradual accumulation of IgG4 in allergen immunotherapy is also usually against a conserved target, SARS-CoV-2 is constantly mutating. Although we do observe an accumulation of IgG4 antibodies targeting ancestral spike—in line with ancestral imprinting—future research should explore if IgG4 Fab arm exchange also exacerbates poorer binding towards variant RBDs.
Sarbecoviruses have shown their potential for global outbreaks.33 The current sustained population immunity against SARS-CoV-2 could afford some protection against zoonotic spillovers.35,36 Research on rapidly evolving SARS-CoV-2 variants have repeatedly highlighted that neutralising responses are usually limited against emerging escape mutants.38,46 Here, we demonstrate that cross-reactive antibodies against a range of sarbecoviruses are capable of engaging Fc receptors, which could offer non-neutralising immunity.
While the strength of our study lies in the paired plasma and mucosal samples across timepoints, the moderate size of our cohorts may limit the interpretation of some observations. Future studies should continue to investigate the impact of subsequent mRNA boosters with larger cohorts and confirm if heterologous priming would indeed limit the overall proportion of class-switched IgG4. Surrogate Fc-dimer binding assays utilised in our study have been previously shown to correlate strongly with cell-based antibody-dependent functional assays,47,48 and allow us to study Fc-binding responses in both plasma and saliva. Our data suggests that class-switching from IgG1 to IgG4 may dampen neutralising activity, however future work should confirm its implications on emerging viral variants.
In summary, we find that despite increasing proportions of spike-specific IgG4 observed with repeated mRNA vaccination, the overall antibody binding to FcγR continues to improve across Omicron variants and sarbecoviruses, even after the fourth mRNA dose. Although the reduced availability of IgG1 antibodies—arising from IgG4 class-switching post-booster—dampened live virus neutralisation, updated mRNA boosters still improved neutralising activity towards Omicron variants. Future research exploring the immunological mechanisms and longitudinal effects of mRNA booster-primed IgG4 sub-class switching will be critical for vaccine design strategies aimed towards maximising the protective potential of mRNA vaccines.
Contributors
Conceptualisation: K.J.S.; Methodology: K.J.S., W.S.L., and A.L.E.; Formal analysis: K.J.S., W.S.L., A.L.E, and A.W.C.; Investigation: K.J.S., W.S.L., and A.L.E.; Resources: K.J.S., L.C.A, J.A., H.E.K., E.M.E, N.K.W, J.J.C., J.M.T., P.P., W.H.T., L.E.D., A.K.W., J.A.J., S.J.K. and A.W.C.; Writing—Original draft: K.J.S.; Writing—Review & editing: K.J.S., W.S.L., A.L.E, L.C.A, J.A., H.E.K., E.M.E, N.K.W, J.J.C., J.M.T., P.P., W.H.T., L.E.D., A.K.W., J.A.J., S.J.K. and A.W.C.; Supervision: K.J.S., and A.W.C.; Funding acquisition: K.J.S., W.S.L., S.J.K. and A.W.C. Verified underlying data: K.J.S., and A.W.C. All authors have read and approved the final version of the manuscript.
Data sharing statement
All data reported in this study are available from the lead contact Amy Chung (awchung@unimelb.edu.au) upon request.
Declaration of interests
The authors declare that they have no conflicts of interest. Funders had no role in the design, analysis or reporting of this study.
Acknowledgements
We thank all the cohort participants for generously providing their samples. We thank Thakshila Amarasena, Kathleen Wragg, Penny Konstandopoulus, Grace Gare, Julie Nguyen, Kirsty Field (University of Melbourne). The authors are grateful to Bruce Wines and Mark Hogarth (Burnet Institute) for sharing their soluble Fc receptor dimers. This research has been conducted using samples and data from Victorian Critical Vaccinees Collection (VC2). We also acknowledge VC2 investigators and sites, and the support of the Victorian Government. This study was supported by the VC2 Research Fund (K.J.S.) and an NHMRC Investigator grant #2008092 (A.W.C). Sample collection was supported by WHO Unity funds (2020/1085469-0), and WEHI Philanthropic funds. K.J.S, W.S.L, W.H.T. (2016908), L.E.D., A.K.W. J.A.J. and S.J.K. are supported by NHMRC Investigator grants. J.A.J. is supported by the Charles and Sylvia Viertel Foundation.
Footnotes
Supplementary data related to this article can be found at https://doi.org/10.1016/j.ebiom.2025.106087.
Appendix A. Supplementary data
References
- 1.Australian Government Department of Health and Aged Care COVID-19 vaccine advice and recommendations 2025. https://www.health.gov.au/our-work/covid-19-vaccines/getting-your-vaccination Available from:
- 2.Cromer D., Reynaldi A., Mitchell A., et al. Predicting COVID-19 booster immunogenicity against future SARS-CoV-2 variants and the benefits of vaccine updates. Nat Commun. 2024;15(1):8395. doi: 10.1038/s41467-024-52194-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lee W.S., Audsley J., Trieu M.C., et al. Randomized controlled trial reveals no benefit to a 3-month delay in COVID-19 mRNA booster vaccine. J Clin Invest. 2024;134(17) doi: 10.1172/JCI181244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lee W.S., Selva K.J., Audsley J., et al. Randomized trial of same- versus opposite-arm coadministration of inactivated influenza and SARS-CoV-2 mRNA vaccines. JCI Insight. 2025;10(4) doi: 10.1172/jci.insight.187075. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Irrgang P., Gerling J., Kocher K., et al. Class switch toward noninflammatory, spike-specific IgG4 antibodies after repeated SARS-CoV-2 mRNA vaccination. Sci Immunol. 2023;8(79) doi: 10.1126/sciimmunol.ade2798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Selva K.J., Ramanathan P., Haycroft E.R., et al. Preexisting immunity restricts mucosal antibody recognition of SARS-CoV-2 and Fc profiles during breakthrough infections. JCI Insight. 2023;8(18) doi: 10.1172/jci.insight.172470. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Aurelia L.C., Purcell R.A., Theisen R.M., et al. Increased SARS-CoV-2 IgG4 has variable consequences dependent upon Fc function, Fc receptor polymorphism, and viral variant. Sci Adv. 2025;11(9) doi: 10.1126/sciadv.ads1482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gelderloos A.T., Verheul M.K., Middelhof I., et al. Repeated COVID-19 mRNA vaccination results in IgG4 class switching and decreased NK cell activation by S1-specific antibodies in older adults. Immun Ageing. 2024;21(1):63. doi: 10.1186/s12979-024-00466-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kalkeri R., Zhu M., Cloney-Clark S., et al. Altered IgG4 antibody response to repeated mRNA versus recombinant protein SARS-CoV-2 vaccines. J Infect. 2024;88(3) doi: 10.1016/j.jinf.2024.106119. [DOI] [PubMed] [Google Scholar]
- 10.Buhre J.S., Pongracz T., Kunsting I., et al. mRNA vaccines against SARS-CoV-2 induce comparably low long-term IgG Fc galactosylation and sialylation levels but increasing long-term IgG4 responses compared to an adenovirus-based vaccine. Front Immunol. 2022;13 doi: 10.3389/fimmu.2022.1020844. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Espino A.M., Armina-Rodriguez A., Alvarez L., et al. The Anti-SARS-CoV-2 IgG1 and IgG3 antibody isotypes with limited neutralizing capacity against Omicron elicited in a Latin population a switch toward IgG4 after multiple doses with the mRNA Pfizer-BioNTech vaccine. Viruses. 2024;16(2) doi: 10.3390/v16020187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Yoshimura M., Sakamoto A., Ozuru R., et al. The appearance of anti-spike receptor binding domain immunoglobulin G4 responses after repetitive immunization with messenger RNA-based COVID-19 vaccines. Int J Infect Dis. 2024;139:1–5. doi: 10.1016/j.ijid.2023.11.028. [DOI] [PubMed] [Google Scholar]
- 13.Kiszel P., Sik P., Miklos J., et al. Class switch towards spike protein-specific IgG4 antibodies after SARS-CoV-2 mRNA vaccination depends on prior infection history. Sci Rep. 2023;13(1) doi: 10.1038/s41598-023-40103-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kobbe R., Rau C., Schulze-Sturm U., et al. Delayed induction of noninflammatory SARS-CoV-2 spike-specific IgG4 antibodies detected 1 year after BNT162b2 vaccination in children. Pediatr Infect Dis J. 2024;43(12):1200–1203. doi: 10.1097/INF.0000000000004488. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Labrijn A.F., Buijsse A.O., van den Bremer E.T., et al. Therapeutic IgG4 antibodies engage in Fab-arm exchange with endogenous human IgG4 in vivo. Nat Biotechnol. 2009;27(8):767–771. doi: 10.1038/nbt.1553. [DOI] [PubMed] [Google Scholar]
- 16.Callaway H.M., Hastie K.M., Schendel S.L., et al. Bivalent intra-spike binding provides durability against emergent Omicron lineages: results from a global consortium. Cell Rep. 2023;42(1) doi: 10.1016/j.celrep.2023.112014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Martin Perez C., Ruiz-Rius S., Ramirez-Morros A., et al. Post-vaccination IgG4 and IgG2 class switch associates with increased risk of SARS-CoV-2 infections. J Infect. 2025;90(4) doi: 10.1016/j.jinf.2025.106473. [DOI] [PubMed] [Google Scholar]
- 18.Dionne K., Tauzin A., Bélanger É., et al. Impact of vaccine-elicited anti-spike IgG4 antibodies on Fc-effector functions against SARS-CoV-2. Viruses. 2025;17(5):666. doi: 10.3390/v17050666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tam J.C.H., Sibayan A.C., Seow J., et al. Spike-specific IgG4 generated post BNT162b2 mRNA vaccination is inhibitory when directly competing with functional IgG subclasses. Cell Rep. 2025;44(7) doi: 10.1016/j.celrep.2025.116000. [DOI] [PubMed] [Google Scholar]
- 20.Selva K.J., Ramanathan P., Haycroft E.R., et al. Mucosal antibody responses following Vaxzevria vaccination. Immunol Cell Biol. 2023;101(10):975–983. doi: 10.1111/imcb.12685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Declercq J., Gerlo S., Van Nevel S., et al. Repeated COVID-19 mRNA-based vaccination contributes to SARS-CoV-2 neutralizing antibody responses in the mucosa. Sci Transl Med. 2024;16(770) doi: 10.1126/scitranslmed.adn2364. [DOI] [PubMed] [Google Scholar]
- 22.Winklmeier S., Rubsamen H., Ozdemir C., et al. Intramuscular vaccination against SARS-CoV-2 transiently induces neutralizing IgG rather than IgA in the saliva. Front Immunol. 2024;15 doi: 10.3389/fimmu.2024.1330864. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wheatley A.K., Pymm P., Esterbauer R., et al. Landscape of human antibody recognition of the SARS-CoV-2 receptor binding domain. Cell Rep. 2021;37(2) doi: 10.1016/j.celrep.2021.109822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Pymm P., Adair A., Chan L.J., et al. Nanobody cocktails potently neutralize SARS-CoV-2 D614G N501Y variant and protect mice. Proc Natl Acad Sci U S A. 2021;118(19) doi: 10.1073/pnas.2101918118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Lee W.S., Tan H.X., Reynaldi A., et al. Durable reprogramming of neutralizing antibody responses following Omicron breakthrough infection. Sci Adv. 2023;9(29) doi: 10.1126/sciadv.adg5301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Tea F., Ospina S.A., Aggarwal A., et al. SARS-CoV-2 neutralizing antibodies: longevity, breadth, and evasion by emerging viral variants. PLoS Med. 2021;18(7) doi: 10.1371/journal.pmed.1003656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Voss W.N., Mallory M.A., Byrne P.O., et al. Hybrid immunity to SARS-CoV-2 arises from serological recall of IgG antibodies distinctly imprinted by infection or vaccination. Cell Rep Med. 2024;5(8) doi: 10.1016/j.xcrm.2024.101668. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Strobl M.R., Demir H., Sanchez Acosta G., et al. The role of IgG(1) and IgG(4) as dominant IgE-blocking antibodies shifts during allergen immunotherapy. J Allergy Clin Immunol. 2023;151(5):1371–1378.e5. doi: 10.1016/j.jaci.2023.01.005. [DOI] [PubMed] [Google Scholar]
- 29.Eckl-Dorna J., Weber M., Stanek V., et al. Two years of treatment with the recombinant grass pollen allergy vaccine BM32 induces a continuously increasing allergen-specific IgG(4) response. eBioMedicine. 2019;50:421–432. doi: 10.1016/j.ebiom.2019.11.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Singh G., Abbad A., Kleiner G., et al. The post-COVID-19 population has a high prevalence of cross-reactive antibodies to spikes from all Orthocoronavirinae genera. mBio. 2024;15(1) doi: 10.1128/mbio.02250-23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Amanat F., Thapa M., Lei T., et al. SARS-CoV-2 mRNA vaccination induces functionally diverse antibodies to NTD, RBD, and S2. Cell. 2021;184(15):3936–3948.e10. doi: 10.1016/j.cell.2021.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zhou F., Vahokoski J., Bergen C.-R.G., Langeland N., Cox R.J. Impact of ageing on homologous and human-coronavirus-reactive antibodies after SARS-CoV-2 vaccination or infection. NPJ Vaccines. 2024;9(1):37. doi: 10.1038/s41541-024-00817-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.World Health Organization Pathogens prioritization: a scientific framework for epidemic and pandemic research preparedness 2024. https://www.who.int/publications/m/item/pathogens-prioritization-a-scientific-framework-for-epidemic-and-pandemic-research-preparedness Available from:
- 34.Guillebaud J., Ou T.P., Hul V., et al. Study of coronavirus diversity in wildlife in Northern Cambodia suggests continuous circulation of SARS-CoV-2-related viruses in bats. Sci Rep. 2025;15(1) doi: 10.1038/s41598-025-92475-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Renner T.M., Stuible M., Cass B., et al. Reduced cross-protective potential of Omicron compared to ancestral SARS-CoV-2 spike vaccines against potentially zoonotic coronaviruses. Npj Viruses. 2024;2(1):58. doi: 10.1038/s44298-024-00067-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Tan C.W., Chia W.N., Young B.E., et al. Pan-Sarbecovirus neutralizing antibodies in BNT162b2-Immunized SARS-CoV-1 survivors. N Engl J Med. 2021;385(15):1401–1406. doi: 10.1056/NEJMoa2108453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Seow J., Jefferson G.C.E., Keegan M.D., Yau Y., Snell L.B., Doores K.J. Profiling serum immunodominance following SARS-CoV-2 primary and breakthrough infection reveals distinct variant-specific epitope usage and immune imprinting. PLoS Pathog. 2024;20(11) doi: 10.1371/journal.ppat.1012724. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Jian F., Wang J., Yisimayi A., et al. Evolving antibody response to SARS-CoV-2 antigenic shift from XBB to JN.1. Nature. 2025;637(8047):921–929. doi: 10.1038/s41586-024-08315-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Yisimayi A., Song W., Wang J., et al. Repeated Omicron exposures override ancestral SARS-CoV-2 immune imprinting. Nature. 2024;625(7993):148–156. doi: 10.1038/s41586-023-06753-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Portilho A.I., Silva V.O., Da Costa H.H.M., et al. An unexpected IgE anti-receptor binding domain response following natural infection and different types of SARS-CoV-2 vaccines. Sci Rep. 2024;14(1) doi: 10.1038/s41598-024-71047-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Fair-Makela R., Thoren P., Nasiaho J., et al. COVID-19 vaccine type controls stromal reprogramming in draining lymph nodes. Sci Immunol. 2025;10(110) doi: 10.1126/sciimmunol.adr6787. [DOI] [PubMed] [Google Scholar]
- 42.Valk A.M., Keijser J.B.D., van Dam K.P.J., et al. Suppressed IgG4 class switching in dupilumab- and TNF inhibitor-treated patients after mRNA vaccination. Allergy. 2024;79(7):1952–1961. doi: 10.1111/all.16089. [DOI] [PubMed] [Google Scholar]
- 43.Barouch S.E., Chicz T.M., Blanc R., et al. Concurrent administration of COVID-19 and Influenza vaccines enhances spike-specific antibody responses. Open Forum Infect Dis. 2024;11(4) doi: 10.1093/ofid/ofae144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Kalkeri R., Zhu M., Cloney-Clark S., et al. Anti-Spike IgG4 and Fc effector responses: the impact of SARS-CoV-2 vaccine platform–specific priming and immune imprinting. J Infect. 2025;91(2) doi: 10.1016/j.jinf.2025.106543. [DOI] [PubMed] [Google Scholar]
- 45.Garcia G.M., Ng J.C., Stewart A.T., et al. Multi-step antibody class switching in a primary human response is restricted after IGHG2 and dependent on B cell maturation stage. bioRxiv. 2025;2025 05.07.652638. [Google Scholar]
- 46.Liu J., Yu Y., Yang S., et al. Virological and antigenic characteristics of SARS-CoV-2 variants LF.7.2.1, NP.1, and LP.8.1. Lancet Infect Dis. 2025;25(3):e128–e130. doi: 10.1016/S1473-3099(25)00015-5. [DOI] [PubMed] [Google Scholar]
- 47.Lee W.S., Selva K.J., Davis S.K., et al. Decay of Fc-dependent antibody functions after mild to moderate COVID-19. Cell Rep Med. 2021;2(6) doi: 10.1016/j.xcrm.2021.100296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Selva K.J., van de Sandt C.E., Lemke M.M., et al. Systems serology detects functionally distinct coronavirus antibody features in children and elderly. Nat Commun. 2021;12(1):2037. doi: 10.1038/s41467-021-22236-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.






