The heterologous expression of most staphylococcal antigens was accomplished using Escherichia coli SCS1 cells equipped with a two-plasmid system purchased from Protagen AG (Germany). Sequences were derived from the S. aureus strain NCTC 8325 if not indicated otherwise. Five proteins were not suitable for expression in full length and were thus expressed as two separate domains (see Supplementary Table 1 for detailed information). A two-plasmid setup was used for all expressions. The helper plasmid pSE111 was selected via a kanamycin resistance cassette and contained genes coding for the Lac repressor protein and a rare tRNA for arginine. The expression plasmid pQE30NST was selected via an ampicillin resistance cassette and contained the sequence for the desired staphylococcal antigen joined to a N-terminal hexa-His-tag for purification via a multi-cloning site. The expression was controlled by the Lac repressor, allowing an induction using Isopropyl-β-D-1-thiogalactopyranoside (IPTG). The strains were cultivated in super broth supplemented with 100 μg/mL ampicillin and 30 μg/mL kanamycin at 37°C with orbital shaking at 250 rpm starting with an overnight pre-culture set-up as a dilution series followed by the main cultivation inoculated to an optical density at 540 nm (OD_540nm) of 0.1. When the culture reached an OD_540nm of 2.0, IPTG was added to a concentration of 1 mM. Two hours after induction, the cells were harvested by centrifugation. The bacterial pellet was washed with PBS, then flash frozen in liquid nitrogen, and stored at −80°C until cell disruption.

The expression of truncated, non-functional Protein A (Spa_trunc) was accomplished using E. coli BL21 (DE3) pLysS_SAC_spa_His, with the spa cassette inserted into pASK-IBA33+ (IBA GmbH, Germany) under the control of a tetracycline-inducible promoter with an ampicillin resistance cassette for selection. Cultivation was performed in super broth supplemented with 100 μg/mL ampicillin and 25 μg/mL chloramphenicol. The main culture was inoculated with growing bacteria from an overnight pre-culture to an OD_540nm of 0.1; target protein expression was induced at an OD_540nm of 0.5 by adding anhydrotetracycline to a concentration of 0.2 μg/mL. The cells were harvested 2 h after induction by centrifugation, washed with PBS, flash frozen in liquid nitrogen, and stored at −80°C.

Cell disruption was performed using glass beads for mechanical disruption within a FastPrep cell homogenizer (Thermo Savant, US). The harvested pellets were resuspended in lysis/equilibration/binding buffer and then shaken with the glass beads three times for 30 s with cooling on ice in between. After centrifugation of the glass beads the supernatant was collected. For purification of the His-tagged proteins Protino® 96 Ni-IDA purification plates were used according to the manufacturer's protocol (MACHEREY-NAGEL GmbH & Co. KG, Germany). The purity of the protein products was verified by one-dimensional sodium dodecyl sulfate polyacrylamide gel electrophoresis followed by Coomassie staining and additionally by mass spectrometric analyses of the eluates.

Human plasma samples were derived from the SHIP-TREND-0 cohort of the Study of Health in Pomerania (SHIP). In brief, a stratified random sample of adults living in the region of Western Pomerania was invited for a personal interview as well as medical examinations to assess their state of health. A full list of initial instruments of investigation can be found in Völzke et al. (38), followed by multiple subsequent initiatives like the description of the molecular epidemiology of nasal S. aureus isolates obtained from this cohort by Holtfreter et al. (39). A subset of 996 subjects was used for this study, for detailed characteristics see Table 1. The study protocol of the SHIP-TREND-0 cohort was approved by the local ethics committee of the University of Greifswald (registration no. BB39/08) with all participants giving informed written consent.

The serological assay to detect specific antibody titers directed against the staphylococcal antigens was performed according to Meyer et al. (40). All antigens were presented on the surface of magnetic fluorophore-coded microspheres (MagPlex®/Luminex®, US) in a 79-plex assay. To achieve a high quality of data, seven serial dilutions spanning the range from 1:50 to 1:204,800 were analyzed for each serum. Human IgG and IgA specific detection antibodies labeled with R-Phycoerythrin were purchased from Jackson ImmunoResearch Europe Ltd (Ely, UK; order numbers 109-116-098 and 109-115-011). To retrieve a single characteristic value per serum/antigen from the multiple data points of the dilution series, the xMAPr analysis tool (40) was used for curve fitting and calculation of the response value. To avoid missing values in the data set, an imputation was performed to include cases where curve fitting failed based on a loess fit over all measurement of a single dilution. All statistical testings and data visualizations were performed using R [v4.0.2; (41)] together with Rstudio [v1.3.1056; (42)] and the tidyverse package [v1.3.0; (43)]. Data analysis was conducted separately for IgG and IgA, binary variables like S. aureus carriage or biological sex were addressed using the Wilcoxon signed rank test, reporting p-values after Benjamini-Hochberg correction (44) for multiple testing with a significance level of 0.05, whereas continuous and discrete variables like BMI or age were analyzed using spline interpolations. The spline regression analysis was performed using R (v4.0.1) in combination with tidyverse (v1.3.0) and limma (v3.44.3) packages. The phenotypes age, BMI or serum glucose were used as predictor variables for a natural cubic spline regression with four degrees of freedom, followed by a linear modeling and an empirical Bayes moderate F-test for each antigen, for the detection of general changes over the predictor variable for each antigen. The resulting p-values were adjusted using the Benjamini–Hochberg correction (44) and the significance level was set to 0.05.

We previously reported that S. aureus carriers mount a strain-specific antibody response against the superantigens produced by their colonizing strain with limited cross-reactivity to other superantigens (22). To re-evaluate this observation in our cohort, we took advantage of the genetic information that was available for a subset of 240 colonizing S. aureus isolates from this study cohort, which comprised the spa type, clonal complex, and presence of genes coding for the superantigens TSST-1 and the staphylococcal enterotoxins (39). To test for strain-specific antibody responses we compared anti-superantigen IgG and IgA profiles in individuals colonized with strains harboring or lacking the corresponding superantigen genes.

Carriers of tst-encoding S. aureus isolates (n = 48) showed a significantly higher IgG and IgA response against TSST-1 and also SEP than carriers of tst-negative isolates (n = 192) (Supplementary Figure 20). Similarly, individuals colonized with S. aureus strains harboring the pathogenicity island (SaPI1)-encoded superantigen genes sek and/or seq (n = 7 for sek/seq and n = 2 for seq alone) showed higher antibody responses for SEK, SEQ, and Ear, with Ssl1 additionally for individuals with seq-positive strains.

Surprisingly, there was no correlation between the occurrence of the sep gene and the SEP-specific antibody titers. This might be explained by a discrepancy between the sep allele amplified with our PCR and that encoding the recombinant SEP protein used for the measurement of antibody binding. The primer sequence for the genetic analysis was derived from strain S. aureus N315 (45), while the sep sequence of strain T0131 was used for recombinant protein expression. The two SEP-variants have only 77% sequence identity. Comparing the antibody titers of carriers colonized with strains equipped with sep to those without the gene the former showed significantly increased levels of antibodies against Ssl6 (SAS0389) and BOF837_902400132 for IgG and IgA, and additionally SplA, SplB, and SplC for IgA only.

This study provides a comprehensive overview on the S. aureus-specific antibody repertoire in a large population-based study, SHIP-TREND-0. Using a dilution series approach rather than single-point measurements allowed us to precisely quantify anti-staphylococcal antibodies in both low and high titer sera and to detect subtle differences [for details on the used method see (40)]. We mainly focused on extracellular antigens of S. aureus including both secreted and surface-associated proteins, since these are accessible to the humoral immune system when facing live bacterial cells and hence induce a strong antibody response (25). Antigens were derived from S. aureus NCTC8325 only, but there are numerous allelic variants in the species S. aureus. Studies of superantigens illustrate that the antibody response to S. aureus virulence factors has a strain-specific component (22, 46). However, superantigens are an extreme example of antigenic variability in S. aureus that exceeds allelic variation. Most antigens are more conserved than superantigens, and allelic variation usually affects only a small portion of an antigens. Thus, conserved antigens and epitopes likely dominate the (polyclonal) antibody response to S. aureus and therefore, the analysis of antigens from a selected, well-established strain, while not perfect, provides a reliable estimate of the immunodominance of S. aureus antigens.

Using available meta data we could pinpoint multiple factors correlating with the strength of the anti-staphylococcal antibody response in naturally exposed humans. Comparing our results to earlier studies, we corroborate the large inter-individual differences in the anti-staphylococcal antibody profiles that cannot be tracked down to single factors (36). Nevertheless, we report several factors such as smoking habits that have not been directly linked to reduced antibody levels until now, or BMI and serum glucose showing only small but distinct differences. Combining our antibody data with the genetic background information on the colonizing S. aureus strains we could show the impact of single staphylococcal antigens on the antibody response in individuals whose immune system is confronted with these proteins on a regular basis.

Both serum IgG and IgA titers against S. aureus were highly heterogeneous. Anti-staphylococcal IgG and IgA were detected in every single individual, albeit in highly variable quantity and composition. In this respect, our data confirm and extend previous studies (e g., 36) because we analyzed a large set of antigens (n = 79) and many individuals (n = 996). The highly individual antibody repertoire likely reflects the history of encounters with S. aureus. Indeed, we observed that carriers mount a specific response against the superantigens of their colonizing strains (namely TSST-1, SEK, and SEQ), which is in line with our previous observations (22). However, genetic variation between strains can make it hard to track down such correlations, as shown for SEP.

Using our highly sensitive multiplex approach, we were able to detect antibodies against all tested antigens. This emphasizes the ubiquitous nature of S. aureus. Likely, every individual encounters numerous S. aureus strains with very diverse antigen repertoires throughout their lifetime. This infers that even persistent S. aureus carriers are exposed to different S. aureus isolates, either by consecutive colonization with different isolates over the years, by co-colonization with multiple strains, or by infection with exogenous S. aureus strains (17, 46). In addition, antibody binding could reflect cross-reactivity to closely related antigens such as superantigens (47) and pore-forming toxins.

Even though anti-S. aureus antibodies obviously do not eliminate S. aureus from the nasal mucosa and do not reliably protect from infection, there is growing evidence that a strong immune memory, as reflected by high antibody titers, protects from severe S. aureus disease (23–25). Our data clearly show that around 6% of the tested individuals were low responders with antibody response values in the lower third of the binding range for more than 80% of the antigens. Based on our current knowledge, these individuals would be prone to a severe disease course in case of S. aureus infection. Tailored preventive or therapeutic measures, e g., decolonization, intravenous immunoglobulins (IVIG), and monoclonal antibodies could improve the outcome of such vulnerable patients in the future (48, 49).

The protective potential of anti-staphylococcal antibodies has been most convincingly shown for the superantigen-driven toxic shock syndrome (TSS). Anti-TSST-1 antibodies are highly prevalent in the Caucasian population (about 90%). Only people without detectable anti-TSST-1 antibodies are at risk of TSS, and symptoms can be treated by IVIG containing neutralizing antibodies (50, 51). Using the multiplex approach with seven serial dilutions of serum samples, we detected antibodies against TSST-1 in all samples (response values spanning from 7.5^*10² to 1.3^*10⁷), pointing toward a higher assay sensitivity as compared to the ELISA used in earlier studies (51). This suggests that very low anti-TSST-1 antibody responses might not confer protection.

The quality of the antibody response to S. aureus depends on the nature of the encounter with the bacteria. Antibody responses in the mucosa are governed by IgA, with local mucosal titers being reflected in the systemic availability of specific monomeric IgA (28, 36), while invasive infections are controlled by IgG. This dichotomy is also reflected by our results. The ranking of antigens based on the median antibody response was similar for only 28 antigens, suggesting that different gene expression profiles are executed during mucosal encounters vs. systemic infection. Similarly, others also observed limited overlap in the anti-staphylococcal IgG and IgA response in healthy carriers and non-carriers (30, 36). CHIPS ranked first for both IgG and IgA as previously reported (36), implying a high immunogenicity of this immune evasion factor. Moreover, the overlap between high IgG vs. IgA responders was limited, suggesting that these immune response modes are indeed independent.

In this study, we used serum IgA as a surrogate marker for mucosal IgA production, because nasal secretions were not available. This approach is supported by growing evidence for a clonal relationship of serum IgA and mucosal IgA responses. For example, proteomic analyses of a celiac disease cohort identical antigen-binding regions of gluten-specific serum IgA and mucosal IgA (secreted by gut-derived plasma blasts) (28). This suggests that serum and mucosal IgA originate from the same B cell clones but are produced by individual plasmablasts in different locations. Indeed, several studies demonstrate that B cells activated in the gut immune system give rise to plasma cells that reside in the lamina propria and produce dimeric IgA for transport across the epithelium as well as an equivalent population of plasma cells that migrate to the bone marrow and secrete monomeric IgA into the circulation (28, 52–54). Hence, serum monomeric IgA can serve as a proxy for the hardly accessible secretory dimeric IgA of mucosae (28). In line with this, it has already been shown by Verkaik et al. (36) that for staphylococcal antigens serum antibody titers of IgG and IgA can be correlated to those in the nasal secretion.

Carrying S. aureus as part of a healthy human microbiome has long been suspected to increase the amount of anti-staphylococcal antibodies due to repeated minor invasive episodes caused by the colonizing strain (36). With our data we support this notion. All antigen-specific antibody responses were stronger in carriers than in non-carriers, in most cases significantly. With ratios of 1.10–1.98 between the mean responses of the carrier group vs. the non-carrier group the direction of the trend is very clear. However, the effect size of carriage was minute compared to the variability of the individual antibody response to each antigen.

Men had on average higher anti-S. aureus antibody levels than women, both IgG and IgA. This is unusual because generally women produce higher antibody levels than men. Klein and Flanagan reviewed multiple features of immunity in humans influenced by biological sex (55). Combining the results of many studies, they concluded that regardless of age, females tend to show stronger antibody responses than males, with higher basal levels of immunoglobulin and increased B cell numbers (55). The more pronounced antibody response to S. aureus antigens in men could be explained by their more extensive exposure to the bacteria. Using culture-based detection methods, men tended to have higher colonization rates than females (12). DNA sequencing, however, revealed overall higher colonization rates (53%) that were equal for men and women (13). This discrepancy can be explained by a 10–100-fold higher absolute abundance of S. aureus in men and its influence on culture outcomes. This higher bacterial load is probably the main reason for the consistently higher anti-staphylococcal antibody levels in men. In our study cohort we reported a culture-based carriage rate of 27.0 vs. 24.3% for males and females, respectively, but according to Liu et al. (13) we might have failed to detect carriers with low bacterial loads, especially among female study participants (39).

Only the IgG responses consistently decreased with age. Regarding IgA, there was no clear trend; antibody binding to two thirds of the antigens decreased with age, the remainder showed the opposite trend. The decline in specific IgG titers could reflect a gradual decline of the adaptive immune functions with age, called immunosenescence. It diminishes the host's ability to clear pathogens and also to develop long-term memory to vaccination. This results in a steady increase of the incidence of infections with age in the elderly. In the humoral immune system alterations in the B cell repertoire and subcompartment distribution, as well as defects in B lymphopoiesis, cell development and homeostasis are observed. These lead to reduced pathogen-specific antibodies in the elderly [reviewed in (56, 57)]. Since age-dependent analyses of the B cell response and antibody repertoires against S. aureus have not been performed yet, it remains to be shown which features of immune aging are responsible for the decline in S. aureus-specific antibody levels. S. aureus colonization rates also decrease with increasing age in humans, which is counterintuitive considering the decreasing antibody levels. A noteworthy exception from this general trend are two matching candidate antigens namely LukF-PV and LukS-PV where antibody titres increased with age. The genes for the pore-forming toxin Luk-PV are very rare in colonization strains in our cohort; they are carried by only 0.1% of the nasal isolates. Probably age groups face different strains because of their different lifestyle; e g., contacts at work or in health care facilities. Long-term observational studies correlating the changes in the antibody repertoire with challenges by various strains with distinct sets of mobile genetic elements are required to assess the impact of such factors.

Besides sex and age, we included additional lifestyle factors in our study, revealing even more pieces of influence on the anti-staphylococcal antibody repertoire in the puzzle of individual variability. Obesity, chronic diseases of the skin (e g., atopic eczema), but also smoking habits have been shown to alter colonization rates (7, 11, 14). With our data we could show that BMI, elevated serum glucose, and regular smoking impact on the antibody levels against staphylococcal antigens, while medically diagnosed allergies did not play a role.

Elevated BMI and serum glucose levels were inversely correlated with antibody responses against most of the included antigens. This may be due to the impaired function of the adaptive immunity in obese people already described in the context of vaccination against hepatitis and influenza [reviewed in (58)]. We found the opposite trend in our IgA measurements: antibodies to most antigens increased with increasing BMI or serum glucose. This falls in line with the observation of a higher S. aureus colonization rate in obese individuals and patients with diabetes than in healthy individuals (59). The higher challenge to the immune system of the nasopharyngeal mucosa could trigger more IgA secretion.

Smoking has been described to alter the nasal microbiome and decrease the S. aureus colonization rate in humans (14). Interestingly smoking was associated with lower specific IgG levels, while for IgA no clear trend could be observed. Higher anti-Ssl9 and lower anti-FnbA levels in current smokers were sticking out. Obviously, there must be other factors besides colonization that influence the systemic (IgG) and local (IgA) immune response to S. aureus.

Our analyses have implications for vaccine development against S. aureus. The selection of vaccine targets needs to consider their relevance in different staphylococcal diseases, prevalence in clinical isolates but also the pre-existing immunity in the population and the quality of the immune response that is generated during infection. Judging by the broad range of the median antibody response values observed in this healthy cohort that is spanning several orders of magnitude, selecting vaccine targets with a low pre-existing response may improve vaccine efficacy. Notably, past and current vaccine candidates strongly differ in their natural immunogenicity, with rather low antibody levels against ClfA (passive vaccine Tefibazumab) (60), and intermediate to high antibody levels against pore-forming toxins (Hla, HlgBC, LukED, LukSF, LukS-PV) (34, 61). However, the huge interindividual variation in antibody titers that cover several orders of magnitude suggests that a considerable proportion of individuals lacks protective antibody levels even against the highly immunogenic bacterial proteins. These would benefit from vaccination. Besides active vaccination, passive vaccination approaches might help to overcome this problem in the future: monoclonal antibody cocktails could be individualized by factoring in the virulence factor repertoire of the invasive strain as well as the antibody profile of the patient.