Coronaviruses have been responsible for three pandemics this century with high mortality although they are usually responsible for only benign colds. The severe acute respiratory syndrome coronavirus (SARS-CoV) and the Middle East respiratory syndrome coronavirus (MERS-CoV) caused nosocomial outbreaks in 2002 and in 2012, respectively, due to a virus transfer from animals to humans (De Wit et al., 2016; Reis et al., 2021). However, in 2019, the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) emerged worldwide as the causative agent of the coronavirus disease 2019 (COVID-19), which can range from a mild disease course to pneumonia requiring hospitalization and to life-threatening multi-organ failure in the most severe cases. So far, around 431 million cases and almost 6 million deaths have been reported worldwide.

Immunoglobulins G (IgG) are the most abundant antibodies present in human blood with concentrations ranging from 7 to 18 mg/ml. Produced by mature plasma cells, they play a crucial role in the regulation of inflammation and immune reactions in humans and animals (Karsten et al., 2012; Hess et al., 2013; Bartsch et al., 2020). In humans, IgG is subdivided into four subclasses named according to their abundance: IgG₁, IgG₂, IgG₃, and IgG₄ (Schur, 1988). Although they share about 90% of amino acid homology, each subclass differs with respect to effector functions and binding affinities toward Fc gamma receptors (FcγRs) (Wang and Ravetch, 2019). Each IgG molecule consists of two parts: fragment antigen binding (Fab) that is sporadically glycosylated and fragment crystallizable (Fc), which is glycosylated at asparagine (Asn) 297 within each heavy chain. IgG glycosylation at Asn297 modulates its interactions with Fcγ receptors (FcγRs). In particular, absence of core-fucosylation was shown to tremendously increase the binding to FcγRIIIa and enhance antibody-dependent cellular cytotoxicity. This principle is used in the production of recombinant monoclonal antibodies by the pharmaceutical industry for anti-cancer therapy (Yu et al., 2017).

IgG glycosylation at Asn297 consists of complex-type biantennary N-glycans being highly fucosylated, slightly sialylated and having various degrees of galactosylation (Clerc et al., 2016). IgG glycosylation is age- as well as sex-dependent (Bakovic et al., 2013; Stambuk et al., 2020) and has been widely studied in health and disease. Fucosylation is relatively stable in adulthood and upon aging (Dall’olio et al., 2013; De Haan et al., 2016). On the opposite, sialylation and galactosylation decrease with age (Dall’olio et al., 2013) because their occurrence is related to estrogen regulation both in men and women (Ercan et al., 2017). Moreover, the latter glycosylation features are also gender-dependent: IgG galactosylation and sialylation are higher in pre-menopausal women as compared with men (Bakovic et al., 2013). After menopause, sialylation, and galactosylation levels are similar in men and women (Bakovic et al., 2013). Distinct IgG glycosylation features are also associated with a broad range of inflammatory pathologies. Precisely, IgG glycosylation is modulated during chronic as well as acute inflammation. In rheumatoid arthritis patients, decreased galactosylation was correlated with clinical parameters (Van De Geijn et al., 2009; Schwedler et al., 2018) and with decreased activity of galactosyltransferase in plasma B-cells (Axford et al., 1987). During bacterial and viral infections, decreased IgG galactosylation was observed for hepatitis B and tuberculosis (Pilkington et al., 1995; Ho et al., 2015; Irvine and Alter, 2020).

In COVID-19 patients, however, regulation of glycosylation has not been addressed in detail. Thus, in order to gain a better understanding of IgG glycosylation upon SARS-CoV-2 infection, we studied total IgG and anti-SARS-CoV-2 IgG glycosylation at Asn297 in a cohort of patients hospitalized in Berlin (Germany) during disease course.

In this study, we investigated the glycosylation profiles of total IgG₁, total IgG₂, and antigen-specific anti-S IgG₁ isolated from plasma samples of COVID-19 patients by means of MALDI-TOF-MS. The investigated cohorts consisted of 35 severe COVID-19 patients and 35 sex- and age-matched HC, whose demographics are presented in Table 1. A majority of COVID-19 patients (80%) was treated at intensive care unit (ICU) wards. The hospitalization duration varied from 4 to 57 days (median: 20 days) and a total of 33 patients survived the COVID-19 infection. For each COVID-19 patient, three plasma samples (referred to as first, middle, and last) were analyzed, each of which corresponded to a different time-point of patient’s hospital stay, namely the beginning, middle, and the end. The first anti-S IgG glycopeptide detection corresponded to hospitalization day 0–6 and, with few exceptions, matched with the first day when anti-S IgG could be unambiguously detected (≥1.1 IU). Altogether, the investigated material consisted of 140 plasma samples (HC: 35; COVID-19 patients: 105) resulting in 245 IgG glycopeptide samples (total IgG: 140; anti-S IgG: 105) being measured.

The analytical workflow used in this study is presented in Figure 1. Inter-day repeatability of the sample preparation was verified by analyzing the same plasma sample in triplicate on three consecutive days. The results of repeatability testing are presented in Supplementary Figure 1. The mean coefficient of variation values were 3.11 for total IgG₁, 3.75 for total IgG₂, and 4.71 for anti-S IgG₁, indicating a very good repeatability of the applied method.

The representative MALDI-TOF mass spectra [M-H]^– of total and anti-S IgG glycopeptide fractions are presented in Figure 2. In the total IgG fraction, up to 28 glycopeptide signals were detected, of which 17 corresponded to IgG₁ and 11 to IgG₂ subclass (Figure 2A and Supplementary Table 1). Five IgG₂ structures (i.e., G1, G2, G1N, G2N, and G2S1) could not be unambiguously assigned due to the mass overlap with IgG₄ glycopeptides and hence were not included in the analysis. It should be noted that the anti-S IgG fraction contained solely IgG₁ glycopeptides (Figure 2B); this is why it is referred to as anti-S IgG₁ in the following parts of this work. As shown in Figure 2B for one representative COVID-19 patient, at the beginning of the disease (hospitalization day 6), the most abundant anti-S IgG₁ glycopeptides are the afucosylated G1 and G2 structures, whereas the fucosylated G0F and G1F glycopeptides become the most abundant structures later in the disease course (day 18 and day 31).

Aiming at performing statistical comparisons between clinical parameters and glycosylation features, the glycosylation profiles of total and anti-S IgGs were represented in the form of four glycosylation traits, namely fucosylation, galactosylation, sialylation, and bisection, calculated separately for each IgG subclass/type, as described in section “MALDI-TOF Measurements and Data Analysis.” The relative abundance and SD values of all total IgG₁, total IgG₂, and anti-S IgG₁ glycosylation traits and individual glycopeptide structures detected in HC and COVID-19 patients are presented in Supplementary Table 2.

As visible in Figure 3, COVID-19 was found to be associated with significant changes in total IgG₁, total IgG₂, and antigen-specific anti-S IgG₁ glycosylation. In general, all investigated IgG fractions presented similar patterns of COVID-19-related glycosylation alterations, marked by a gradual decrease in galactosylation and sialylation and a concomitant gradual increase in fucosylation (in the case of IgG₂, fucosylation was not determined). Bisection was the only glycosylation trait that showed a distinct profile of COVID-19-related alterations in total IgG₁/total IgG₂ and antigen-specific anti-S IgG₁ antibodies, in which it, respectively, decreased and increased in the course of the disease. IgG₂ was found to be marked by an overall lower galactosylation compared to both IgG₁ fractions, which seems to be a universal feature of IgG₂ glycosylation and is in line with previous reports (Bakovic et al., 2013; Wieczorek et al., 2020). Notably, the strongest COVID-19-related glycosylation alterations were recorded for antigen-specific anti-S IgG₁. Among all glycosylation traits, anti-S IgG₁ fucosylation and galactosylation had a particularly high variation in the course of the disease, with anti-S IgG₁ fucosylation (first: 83.3%; last: 95.9%) showing a 12% increase and anti-S IgG₁ galactosylation (first: 56.4%, last: 42.2%) a 14% decrease. Notably, as visible in Supplementary Figure 2, further stratification of COVID-19 cohort based on the length of hospitalization revealed that the above-described alterations prevail in patients with prolonged hospital stay (≥20 days).

Despite similar trends of COVID-19-related alterations, glycosylation profiles of total and antigen-specific anti-S IgG₁ antibodies were found to differ significantly, with the strongest discrepancy being observed at the beginning of hospitalization. Precisely, as visible in Figure 4, anti-S IgG₁ antibodies secreted early in the COVID-19 course were marked by significantly lower fucosylation and bisection and significantly higher galactosylation and sialylation as compared to total IgG₁ of COVID-19 patients. Notably, although anti-S IgG₁ glycosylation was observed to alter continuously in the disease course (Supplementary Figure 3), statistically significant differences between total and anti-S IgG₁ were as well detected at the end of hospitalization for the following glycosylation traits: galactosylation, sialylation, and bisection. It is also notable that the profile of total IgG₁ glycosylation was found to differ significantly between COVID-19 patients and HC (Figure 4). In particular, total IgG₁ glycopeptides of COVID-19 patients were marked by decreased abundance of bisecting GlcNAc and increased fucosylation both at the beginning and at the end of hospitalization. Contrarily, sialylation of total IgG₁ was unaltered in COVID-19 patients as compared to HC, whereas galactosylation was significantly decreased in COVID-19 patients at the end of hospitalization.

In the investigated COVID-19 cohort, the length of hospitalization was expectedly found to correlate with patients’ age (p = 0.038, Pearson r = 0.3552). Since, in line with literature data (Gudelj et al., 2018), IgG N-glycan composition in both healthy and COVID-19 patients was likewise observed to differ based on patients’ age (Supplementary Figure 4), we next tested whether this age-dependency is also reflected at the level of COVID-19-related glycosylation changes recorded in total IgG₁, total IgG₂, and anti-S IgG₁. Among all investigated antibody fractions, age-dependency of COVID-19-related glycosylation alterations could be detected only in anti-S IgG₁. The results of the correlation analyses are presented in Figure 5, in which the X axes represent patient age, whereas the Y axes represent the change (Δ) in the respective anti-S IgG₁ glycosylation trait recorded between the last and first hospitalization time-points. Notably, anti-S IgG₁ fucosylation was the only glycosylation trait, in which the difference between the final and the initial level was significantly correlated with patients’ age. Precisely, in the course of COVID-19 disease, younger patients exhibited significantly stronger alteration in anti-S IgG₁ fucosylation level as compared to older ones. Contrarily, in the case of galactosylation and sialylation, a more prominent change between the final and the initial level was observed in older COVID-19 patients, however, the respective correlations were statistically insignificant.

Since altered IgG glycosylation is a common feature of inflammatory conditions, we next tested whether COVID-19-related total and anti-S IgG₁ glycosylation alterations correlate with changes in CRP and plasma anti-S IgG concentration in the course of COVID-19. As compared to physiological CRP concentrations that range between 0.8 and 3.0 mg/L (Shine et al., 1981), CRP levels in COVID-19 patients were strongly increased, with the highest levels being expectedly recorded at the beginning of hospitalization (except for the two patients who did not survive COVID-19) (Supplementary Figure 5A). In the course of patients’ hospitalization, the CRP and anti-S IgG levels were observed to, respectively, decrease and increase in COVID-19 patients; however, for none of these parameters the difference between the final and the initial level was correlated with patients’ age (Supplementary Figures 5C,F). As visible in Table 2, Pearson’s correlation analyses revealed that COVID-19-related changes in total and anti-S IgG₁ glycosylation profiles are not correlated with changes in CRP levels. Contrarily, changes in anti-S IgG₁ galactosylation and sialylation were both found to significantly correlate with changes in plasma anti-S IgG concentration recorded between the last and the first time-point of patients’ hospital stay (Table 2).

An increasing number of evidences suggest that altered IgG glycosylation might be a factor contributing to disease severity in COVID-19. To further deepen the understanding of molecular signatures underlying the SARS-CoV-2 infection, in this work, we performed a longitudinal analysis of total and anti-S IgG Fc-glycosylation in a cohort of 35 hospitalized COVID-19 patients and 35 HC by means of MALDI-TOF-MS. To assure site- and subclass-specificity of determined glycosylation profiles, in this study, all analyses were performed at the level of tryptic glycopeptides.

Upon SARS-CoV-2 infection, virus-specific IgG antibodies are typically detected in blood within 7 days from symptom onset (Long et al., 2020; Dan et al., 2021; Pang et al., 2021; Semmler et al., 2021; Sterlin et al., 2021). In our study, the first detection of SARS-CoV-2-specific anti-S glycopeptides in plasma of affected patients occurred between the hospitalization day 0 and day 6. While this broad range might partly result from the differences in patients’ hospital admission time, it might as well reflect the inter-individual variability of the humoral immune response to SARS-CoV-2 infection.

In our study, COVID-19 was found to be associated with significant changes in total and antigen-specific IgG glycosylation. In particular, anti-S IgG₁ produced in the early stage of the disease was found to be marked by strongly decreased core-fucosylation, which is in line with previously reported data (Hoepel et al., 2021; Larsen et al., 2021). Of note, this particular feature of SARS-CoV-2-specific antibodies has important functional consequences. Precisely, although afucosylation of the IgG Fc portion does not influence the binding toward viral particles, it enhances by several folds the binding affinity toward FcγRIIIa receptors on the surface of innate immune cells (Wang et al., 2017; Chakraborty et al., 2021). These Fc-FcγR interactions are crucial for Fc-mediated effector functions such as antibody-dependent cellular cytotoxity and phagocytosis, which next to viral neutralization are the primarily mechanisms contributing to anti-viral host protection (Taylor et al., 2021). Data reported by Bye et al. (2021) indicate, however, that low fucosylation of anti-SARS-CoV-2 Spike IgG might be a double-edged sword; while it facilitates the recovery process by potentiating anti-viral immune responses, it might contribute to COVID-19 patients’ death by enhancing pathogenic platelet activation and thrombosis (Bye et al., 2021). In line with this data, immune complexes engaging afucosylated IgG molecules were shown to stimulate the expression of pro-inflammatory cytokines (e.g., IL-6, TNF, and IL-1β) in macrophages and natural killer cells, generating a prothrombotic environment (Chakraborty et al., 2021; Larsen et al., 2021). In addition, high titers and low fucosylation of anti-S IgG₁ were recently shown to promote inflammation by alveolar macrophages (Hoepel et al., 2021).

Consistent with data of Larsen et al. (2021), in this work, the initially low fucosylation of antigen-specific IgG₁ was found to increase continuously over time, eventually approaching the levels observed in total IgG₁. It seems plausible that this timely restricted production of highly potent afucosylated anti-viral antibodies helps preventing excessive and potentially harmful immune activation. Contrary to the above described findings, in the study of Chakraborty et al. (2021), afucosylation levels were shown to be stable over time. This discrepancy might be caused by the fact that, in the latter study, the investigated samples were not collected at the onset of anti-S IgG₁ expression.

Interestingly, some reports indicate that low fucosylation of antigen-specific IgG antibodies might as well play a role in other acute viral infections. For instance, a longitudinal study of dengue-infected patients showed that, at the early stage of the disease, IgG glycosylation profile is marked by high afucosylation, which decreases in the course of the disease the way we have measured here for anti-S IgG (Wang et al., 2017). Contrarily, antibodies against internal viral proteins such as nucleocapsid in COVID-19 or the parvovirus B19 were not afucosylated but rather highly fucosylated (Larsen et al., 2021). These seemingly contradictory data suggest that the biological function of IgG fucosylation might differ depending on the nature of the infectious agent, but this has not been investigated in details so far.

Besides profiling the COVID-19-related IgG glycosylation changes, the aim of our study was to determine whether observed alterations are associated with patients age and whether they correlate with changes in CRP and anti-SARS-CoV-2 IgG plasma concentrations. Notably, for the vast majority of conducted analyses, statistically significant trends were observed exclusively in antigen-specific anti-S IgG₁, which further confirms that the glycosylation of bulk and anti-viral IgG in COVID-19 might be distinctly regulated. In this work, anti-S IgG₁ fucosylation was the only glycosylation trait whose change was correlated with patients’ age during hospitalization. Precisely, in the course of the disease, anti-S IgG₁ antibodies of younger COVID-19 patients displayed a more prominent alteration in the fucosylation level as compared to older patients. In the light of what has been written above, it seems plausible that this dynamic timely restricted transition from highly pro-inflammatory afucosylated phenotype at seroconversion to less pro-inflammatory fucosylated anti-S IgG₁ phenotype observed later in the disease course comprises an innate regulatory mechanism that allows younger individuals to mount a more potent virus-specific immune response that is limited to the early phase of the disease. Following this understanding, a weaker change in anti-S IgG₁ fucosylation observed in older COVID-19 patients could contribute to compromised or less balanced anti-viral response. In line with this rationale, the quality of the humoral response was shown to decline with age, which was linked to diminished potential of aged B-cells to undergo somatic hypermutations (Frasca et al., 2017). Correspondingly, the quantity and glycosylation profile of anti-SARS-CoV-2 antibodies elicited in response to COVID-19 mRNA vaccination were shown to differ in younger and older individuals (Farkash et al., 2021). In line with our data, anti-SARS-CoV-2 antibodies produced after the first and the second vaccination showed higher variation with respect to the fucosylation level in younger as opposed to older individuals (Farkash et al., 2021). Interestingly, in our study, no correlation was observed between the change in anti-S IgG₁ fucosylation and the change in anti-SARS-CoV-2 antibody titer in the course of hospitalization, implying that the above-described age-dependent transformation of IgG fucosylation in COVID-19 patients is independent of anti-S IgG abundancy in blood. Nevertheless, considering that the vast majority of the investigated cohort was represented by severe COVID-19 patients, who presented overall high anti-Spike IgG titers, the latter observation necessitates validation in a larger and more diversified cohort.

In line with the results of Hoepel et al. (2021) and Larsen et al. (2021), anti-S IgG₁ antibodies investigated in the present study were marked by high galactosylation and sialylation as compared to their corresponding levels in total IgG₁. Particularly, this glycosylation profile of anti-SARS-CoV-2 antibodies was specific to the early stage of the disease, as the disparity between total and antigen-specific IgG galactosylation/sialylation diminished continuously toward the end of hospitalization. Notably, this profile was shown to be associated with previous natural infection and recent immunization (Wang et al., 2015; Sonneveld et al., 2017; Hoepel et al., 2021). Interestingly, contrary to the trends observed for fucosylation, COVID-19-related decrease in anti-S IgG₁ galactosylation and sialylation showed no correlation with patients’ age; instead, they were found to negatively correlate with changes in anti-S IgG plasma concentration. Considering that, along with previous reports (Hoepel et al., 2021; Larsen et al., 2021), anti-S IgG₁ concentration and anti-S IgG₁ galactosylation/sialylation levels were, respectively, observed to increase and decrease in the course of the disease, the above findings imply that a strong increase in anti-S IgG concentration, reported to occur in severe COVID-19 patients (Larsen et al., 2021; Madariaga et al., 2021), might be accompanied with less prominent decrease in anti-S IgG₁ galactosylation and sialylation. Notably, the persistence of highly galactosylated/sialylated anti-S IgG₁ antibodies that are otherwise limited to acute, early stage of the disease, could potentially contribute to the severity of the disease. This observation is advocated by the fact that platelet-mediated thrombosis that contributes to increased mortality in critically ill COVID-19 patients was shown to require both low levels of fucosylation and high levels of galactosylation in the anti-S IgG Fc domain (Bye et al., 2021).

Expectedly, inflammation accompanying SARS-CoV-2 infection was reflected by strong CRP elevation in COVID-19 patients at the beginning of hospitalization. Afterward, in the COVID-19 patients who survived the disease, CRP levels were found to gradually decrease in the disease course, with the biggest change being observed during the first half of hospitalization. In our study, the difference in CRP level showed no correlation with changes observed in IgG glycosylation traits.

It should be noted that the present study suffers from some limitations. First and foremost, the analyses were conducted on a relatively small number of samples, with COVID-19 cohort consisting predominantly of patients having severe disease symptoms. Therefore, it would be meaningful to validate these findings in a larger and more diversified cohort. Additionally, anti-S IgG fractions investigated in our study contained exclusively IgG₁ glycopeptides, as antibodies of this subclass dominate the immune response directed toward S1 subunits of the viral Spike protein (Yates et al., 2021). It would be of interest to determine whether reported trends are as well observed in IgG₃, whose titers were recently reported to increase upon SARS-CoV-2 infection, particularly in response to SARS-CoV-2 Spike S2 subunit (Luo et al., 2021; Yates et al., 2021).

In conclusion, the results presented in this study confirm previous findings showing that anti-S IgG₁ antibodies produced in COVID-19 patients are marked by differential glycosylation profiles, which normalize gradually in the course of the disease. Using a German cohort, we were able to show that COVID-19-related glycosylation alterations that occur in antigen-specific anti-S antibodies are to some extend dependent on patient’s age and anti-S IgG quantity. Further studies are needed to determine whether these observed trends are specific to anti-Spike S1 subunit IgG₁ antibodies and whether they are likewise detected in COVID-19 patients suffering from less severe disease symptoms.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Charité – Universitätsmedizin Berlin, Campus Virchow-Klinikum, Germany (no. EA2/095/20) and by the Ärztekammer Berlin, Germany (Eth-23/20). All experiments for EA2/095/20 were performed in accordance with relevant guidelines and regulations. Additional written informed consent was taken for Eth-23/20.

VB contributed to the conception and design of the study. KK, BH, JR, and GH coordinated the collection of samples and database. CS and MG performed the experiments and data analysis. VB, MG, and CS wrote the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

KK is partly contractually provided to Labor Berlin – Charité Vivantes GmbH. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.