Human antibody responses against non-covalently cell wall-bound Staphylococcus aureus proteins

Human antibody responses to pathogens, like Staphylococcus aureus, are important indicators for in vivo expression and immunogenicity of particular bacterial components. Accordingly, comparing the antibody responses to S. aureus components may serve to predict their potential applicability as antigens for vaccination. The present study was aimed at assessing immunoglobulin G (IgG) responses elicited by non-covalently cell surface-bound proteins of S. aureus, which thus far received relatively little attention. To this end, we applied plasma samples from patients with the genetic blistering disease epidermolysis bullosa (EB) and healthy S. aureus carriers. Of note, wounds of EB patients are highly colonized with S. aureus and accordingly these patients are more seriously exposed to staphylococcal antigens than healthy individuals. Ten non-covalently cell surface-bound proteins of S. aureus, namely Atl, Eap, Efb, EMP, IsaA, LukG, LukH, SA0710, Sle1 and SsaA2, were selected by bioinformatics and biochemical approaches. These antigens were recombinantly expressed, purified and tested for specific IgG responses using human plasma. We show that high exposure of EB patients to S. aureus is mirrored by elevated IgG levels against all tested non-covalently cell wall-bound staphylococcal antigens. This implies that these S. aureus cell surface proteins are prime targets for the human immune system.


Materials and Methods
Bacterial strains, plasmids and growth conditions. Strains and plasmids used in this study are listed in Table 1. L. lactis strains were grown at 30 °C in M17 broth (Oxoid Limited, Hampshire, UK) supplemented with 0.5% glucose (wt/vol) (GM17). When necessary the medium was supplemented with chloramphenicol (5 µg/ml) or erythromycin (5 µg/ml) for plasmid selection. S. aureus strains were grown at 37 °C, 250 rpm in Tryptone Soy Broth (TSB; Oxoid). E. coli strain MC1061 was grown at 37 °C, 250 rpm in Lysogeny broth (LB; Becton Dickinson, Breda, The Netherlands). When necessary, the medium was supplemented with ampicillin (100 µg/ml) for plasmid selection.

Isolation of S. aureus cell wall fragments.
Cell wall fragments (CWFs) from S. aureus were isolated as described previously 30 . In short, S. aureus Newman cells were collected by centrifugation, glass-beads were added (0.1 µm beads, Biospec Products, Bartlesville, USA), and cells were disrupted for 2 min in a Precellys 24 homogenizer (Bertin Technologies, Saint Quentin en Yvelines Cedex, France). The resulting CWFs were collected by centrifugation and boiled at 96 °C for 10 min in 4% sodium dodecyl sulphate. This step was repeated twice. CWFs were subsequently washed six times with Phosphate Buffered Saline (PBS) and stored at −20 °C until further use.

Identification of non-covalently cell surface-bound proteins.
Non-covalently cell wall-bound proteins of S. aureus were identified by using the amino acid sequences of known domains for non-covalent cell wall-binding (i.e. PROSITE PS51780, PS51782, PS51781, PS51109) in BLAST searches against the sequenced S. aureus Newman strain. The actual identification of expressed non-covalently cell wall-bound proteins was accomplished as schematically represented in Fig. 1A. Upon overnight culturing in TSB, S. aureus Newman cells were incubated with 2 M potassium thiocyanate (KSCN) for 5 min leading to the release of non-covalently cell wall-bound proteins. Liberated non-covalently cell wall-bound proteins in the soluble fraction were either TCA-precipitated or dialyzed against PBS using a 3,500 Molecular weight cut-off (MWCO) membrane (Spectrum laboratories Inc. USA) as described before 31 . Secreted proteins present in S. aureus Newman growth medium fractions were also collected, either by TCA precipitation or dialysis of the spent growth medium against PBS. Dialyzed non-covalently cell wall-bound proteins were added to prepared S. aureus CWFs and incubated for 5 min at 4 °C unless stated otherwise. CWFs containing non-covalently bound proteins were washed with PBS and non-covalently cell wall-bound proteins were released into the supernatant fraction by incubation with 2 M KSCN as described above. Released proteins were collected either by TCA precipitation or dialyzed overnight against demineralized water. Released proteins and cell pellets were separated by lithium dodecyl sulphate (LDS) -PAGE and the respective gels were stained with SimplyBlue SafeStain (Life Technologies, Grand Island, NY. USA). Protein bands were cut from the gels (Fig. 1B) and identified by Mass Spectrometry (MS) as described previously 32 .
Expression of non-covalently cell wall-bound proteins in L. lactis. PCR was performed with the Pwo DNA polymerase (Roche Diagnostics, Woerden, The Netherlands), using chromosomal DNA of S. aureus NCTC 8325 as template. Primers listed in Table 2 were designed to amplify gene sequences without the region coding for the natural secretion signals and with 5′ end extensions for ligase-independent cloning (LIC) 33  Backbone Exchange was performed by mixing ~300 ng of pERL vector with ~300 ng of the pRE-USP harboring the gene of interest, both digested with SfiI (New England Biolabs, Ipswich, UK). Ligation was performed using T4 DNA Ligase (New England Biolabs) and the resulting vector was introduced into electrocompetent L. lactis PA1001 34 . For insertion of genes into plasmid pNG4210, primers (Table 2) were designed to amplify the respective sequences without the region coding for the natural secretion signals and with 5′ end extensions encoding BamHI (forward) or NotI (reverse) restriction sites. Briefly, digested PCR products and linearized plasmid were separated by agarose gel electrophoresis, and selected DNA fragments were gel-extracted and purified. Ligation of digested plasmid and PCR fragments was performed using T4 DNA ligase and the resulting plasmid was introduced into electrocompetent L. lactis PA1001 as described before 35 . For the expression of cloned genes, L. lactis cultures were induced in the exponential phase of growth (0.5 O.D. at 600 nm) by the addition of nisin (final concentration 3 ng/ml, Sigma-Aldrich, St. Luis, MO). After 4 h or overnight culturing, the cells were separated from the growth medium by centrifugation. Proteins in the nisin-induced growth medium fractions were precipitated with TCA (10% W/V) and resuspended in LDS gel loading buffer (Life Technologies). Cells in LDS sample buffer were disrupted with 0.1 µm glass beads in a Precellys 24 homogenizer. Both cellular and growth medium fractions were analyzed by LDS-PAGE (Life Technologies) and proteins were either visualized using protein staining with the SimplyBlue SafeStain (Life Technologies), or by blotting onto nitrocellulose membranes (Protan nitrocellulose transfer paper, Whatman, Germany) and subsequent immunodetection. For immunodetection, mouse anti-his tag primary antibodies (Life Technologies) and fluorescently labeled secondary antibodies (goat anti-mouse IRDye 800 CW, LI-COR Biosciences, Lincoln, NE. USA) were used. Antibody binding was visualized with an Odyssey infrared imaging system (LI-COR Biosciences).

Protein purification and activity measurements.
When expressed proteins remained cell-associated, they were liberated from the cells either with 2 M KSCN or 6 M urea, as required. Next, the protein-containing soluble fractions were separated from the cell fraction by centrifugation. Subsequently, his-tagged proteins were purified from the respective supernatant fractions using the HisLink Protein Purification resin (Promega Corporation, Madison, WI. USA), in the absence or presence of either 2 M KSCN or 6 M urea. The HisLink binding and washing buffer was composed of 0.1 M HEPES 7.5 pH, 0.5 M NaCl and 10 mM imidazole. The elution buffers were essentially the same, but contained 200 mM or 400 mM imidazole. The IsaA and FtsL proteins were purified as described previously 35,36 .

Rebinding of isolated proteins to S. aureus cells. Overnight growth cultures of S. aureus Newman
ΔspaΔsbi were resuspended to 1 optical density measured at 600 nm in 800 µl of PBS (pH 7) or 50 mM sodium acetate (pH 5) and incubated with 1-3 µg of histidine-tagged fusion proteins for 10 min. After incubation, cell pellets and supernatants were processed, and localization of tagged proteins was assessed by LDS-PAGE and Western blotting as described above.
Enzyme-linked immunosorbent assay (ELISA). Plasma    The EB patients included six patients with junctional EB (EB01, EB02, EB09, EB15, EB53, EB60), one patient with EB simplex (EB11), and one patient with dystrophic EB (EB51) 19 . ELISA plates were coated overnight at 4 °C with histidine-tagged fusion proteins (100ng/well) diluted in carbonate coating buffer (50 mM sodium carbonate, pH 9.6) 6 . Subsequently, the plates were blocked with PBS containing 5% skim milk. Patient and healthy control plasma samples were processed as previously described 19 . Serial dilutions of plasma (1000-to 2,000,000-fold) were prepared in PBS-Tween 20/5% skim milk. Specific anti-human IgG secondary antibodies coupled to horseradish peroxidase (dilution 1:8,000; Southern Biotechnology, Birmingham, AL) were used according to the manufacturer's recommendations. Horseradish peroxidase activity was quantified by measuring the hydrolysis of the substrate (O-Phenylenediamine, Sigma-Aldrich) at OD 492 in a plate reader (Biotek Powerwave XS2, USA). The raw ELISA data are provided in Supplementary Information Table 1. Titers were calculated in arbitrary units (AU) through extrapolation using linear regression for data points from known dilutions giving OD 492 values between 0.1 and 1.0. All calculated R 2 linear regression values (Pearson product moment correlation coefficient) were above 0.98. IgG titers in plasma samples of EB patients were averaged and normalized by adjusting the averaged IgG titers in the control plasma samples of healthy volunteers to a single arbitrary level (AU = 10) and adjusting accordingly the averages for the EB patient samples. In brief, obtained EB and control averages were multiplied by a numeric factor that resulted in the average of all controls equal to 10 AU. After plotting normalized values, the differences in the average IgG levels measured for plasma samples from EB patients and healthy control individuals were compared for the different analyzed proteins. Statistical significance of the ELISA data was tested with a Mann-Whitney U test for two independent samples (Supplementary Information

Results and Discussion
Selection of non-covalently cell wall-bound proteins. To pinpoint a panel of non-covalently cell wallbound proteins of S. aureus, we performed an extensive bioinformatics analysis using the genome sequence of S. aureus strain Newman and, in addition, we surveyed the results of our previous analysis of the cell surface proteome of this S. aureus strain 29 . The results are summarized in Table 3. Specifically, the bioinformatics approach identified several SERAMS, in particular the extracellular adhesive protein Eap, the extracellular matrix protein Emp, the extracellular fibrinogen-binding protein Efb, and coagulase 37 42 . Except for the amidase from phage phiNM2 and the transmembrane protein EbpS, all other identified proteins carried a predicted signal peptide for export from the cytoplasm (indicated as S in Table 3). Of note, Eap, Atl, IsaA, SsaA2, Sle1, LukG and LukH have also previously been identified as non-covalently cell wall-bound proteins 16,29,43,44 . Further, Sle1 was shown to be localized in the vicinity of the S. aureus cross-wall 43,45 , while Atl was found to be preferably bound to the septal region 44 .
To detect the actually produced non-covalently cell wall-bound proteins of S. aureus Newman and to ensure that the identified proteins do indeed behave as non-covalently cell wall-bound proteins, we extracted these proteins with the chaotrope KSCN from the staphylococcal cells, reattached them to isolated S. aureus CWFs, and re-extracted the proteins with KSCN (schematically represented in Fig. 1A). The proteins thus obtained were separated by LDS-PAGE. Eight dominant bands were detected, excised from the gel and identified by MS (Fig. 1B). The identifiers and characteristics of the identified proteins are summarized in Table 3. In this respect, it is noteworthy that our unpublished proteomics data indicate that only about 16% of the KSCN-extractable proteins are specifically cell wall-bound proteins.
For our further studies on human antibody responses against non-covalently cell wall-bound proteins of S. aureus, we made a selection of 10 representative proteins. The inclusion criteria for these 10 proteins were identification by bioinformatics and/or biochemical analysis. Exclusion criteria were a lack of identification in previous biochemical or proteomics analyses 29 , the absence of a predicted signal peptide, the presence of an LPxTG motif for covalent cell wall binding, and known IgG-binding properties that would interfere with our further analyses. The domain structure of the selected proteins, highlighting domains potentially involved in cell wall binding, is represented in Fig. 2. It should be noted that Atl is synthesized in a pre-pro-form which, upon export, is cleaved into two moieties with an amidase domain (here termed Atl1) and a glucosamidase domain (here termed Atl2). Accordingly, the Atl2 moiety of Atl does not have its own signal peptide for export from the cytoplasm.
Cloning, production and isolation of non-covalently cell wall-bound proteins in L. lactis. Genes for the selected non-covalently cell wall-bound proteins of S. aureus were cloned into nisin-inducible expression vectors and introduced into L. lactis strain PA1001 for expression. In the case of Atl, the Atl1 and Atl2 moieties were expressed separately, each being secreted with the lactococcal Usp45 signal peptide. Of note, the L. lactis PA1001 strain lacks the genes for the major extracellular protease HtrA and the autolysin AcmA, which minimizes proteolysis and cell lysis, respectively 36,46 . Next, expression of the cloned genes was induced with nisin and the subcellular localization of the respective S. aureus proteins was determined. To this end, cells were separated from the growth medium by centrifugation and the respective fractions were analyzed by LDS-PAGE and Western blotting using anti-His 6 antibodies. As expected, all proteins were largely cell wall-bound (data not shown). By incubation of the cells with 2 M KSCN (Efb, Eap and Atl1) or 6 M urea (all seven other proteins), the expressed proteins were released, consistent with disruption of their non-covalent interactions with the cell wall. Upon centrifugation, the released proteins were purified from the resulting supernatant fractions using Ni-NTA agarose beads, and their potential to re-bind to cells of S. aureus Newman Δspa Δsbi was confirmed ( Fig. 3; data not shown for IsaA). Notably, Efb did not re-bind to S. aureus cells under the standard assay conditions (pH 7), but its binding to the cells could be demonstrated at a lowered pH of 5 (Fig. 3). Altogether, these findings imply that the purified proteins have retained their cell wall-binding capabilities. Of note, the cell wall binding of SA0710 had not been experimentally verified so far, despite the fact that this protein has a LysM domain.

Human IgG responses against non-covalently cell wall-bound proteins of S. aureus. To assess
whether EB patients and healthy volunteers mount immune responses against the selected non-covalently cell wall-bound proteins of S. aureus, we applied an ELISA approach. The membrane protein FtsL was included in this analysis as a control, because it is surface-exposed, but not bound to the cell wall 29 . As shown in Fig. 4A and Supplementary   investigated proteins. Importantly, the levels of IgGs against non-covalently cell wall-bound proteins in plasma samples from EB patients were, on average, about 10-fold higher than those from healthy carriers. In this respect, the largest differences were observed for Eap, Atl2 and IsaA, and the smallest for Sle1 and Emp. Only, the FtsL-specific IgG levels in plasma samples from EB patients and healthy volunteers did not differ significantly (Fig. 4A). In this respect, it should be noted that for many, but not all, S. aureus antigens elevated IgG levels were previously observed in plasma from EB patients 19,28 . For example, no significant differences in IgG levels of EB patients and healthy volunteers were previously observed for the ClfA, ClfB and IsdH proteins bound to the staphylococcal cell wall. We further inspected the overall IgG responses to all non-covalently cell wall-bound proteins per plasma sample, excluding FtsL. As shown in Fig. 4B, the normalized average IgG levels against the eleven remaining S. aureus antigens were higher for the eight plasma samples from EB patients than for the six healthy volunteers. Further, the carriage of multiple S. aureus strains by EB patients correlated with higher normalized average IgG levels (Fig. 4B). Thus, the highest normalized average IgG levels were observed for patients EB01 and EB51 who, respectively, were previously shown to carry 7 and 4 different S. aureus types at 3 time points of sampling 19,28 .
During S. aureus colonization and invasion, immune-competent individuals may rapidly mount antibody responses to a large panel of antigens. The antibody response profiles of individuals usually reflects the history of previous encounters with S. aureus 22 . Accordingly, they may change after every new encounter 21 , which could explain the strong variations in the profiles observed for different individuals 28,47,48 . Importantly, this was previously shown to even be the case upon controlled nasal inoculation of healthy human volunteers with S. aureus 49 .
In previous studies, we have reported that patients with EB develop wounds that are highly susceptible to S. aureus colonization 19,24-26 . Accordingly, it was shown that these patients mounted significantly higher immune responses against S. aureus antigens than healthy carriers 19,28 . In these studies, compared to healthy control individuals, elevated IgG levels were observed in plasmas of EB patients for three out of eleven tested cell wall-associated antigens (i.e. Efb, IsaA and IsdA), and for seven out of seventeen tested secreted antigens. Minor differences were observed for IgG responses against secreted superantigens 19 . In the present exploratory study, we show that EB patients carry significantly increased IgG levels against all eleven tested antigens that are non-covalently bound to the S. aureus cell wall, including Efb and IsaA. This novel observation highlights the strong immunogenicity of these antigens compared to other staphylococcal antigens that are covalently bound to the cell wall or secreted into the host environment. When IgG levels were compared for EB patients carrying multiple S. aureus types versus EB patients carrying only one S. aureus type, significant differences were observed for 4 out of 10 tested antigens, again including Efb and IsaA 19 . This general difference upon colonization with multiple S. aureus types was also observed in the present study, supporting the view that patients exposed to different S. aureus types are challenged with more staphylococcal antigens than patients carrying only one S. aureus type. Another novel finding is that EB patients showed a much larger variation in IgG levels against S. aureus antigens than healthy carrier controls. For example, the level of IgG against LukG in EB patient 02 was lower than the respective IgG levels in four of the healthy control individuals, whereas EB patient 53 showed the highest of all presently recorded IgG levels against LukG. Additionally, the comparison of plasma samples from EB patients and healthy control individuals revealed no significantly different IgG levels for other surface-exposed proteins of S. aureus, like FtsL. Thus, while it has been previously shown that EB patients have in general higher levels of antibodies against S. aureus antigens, our present study shows that (i) not all surface-exposed S. aureus proteins will elicit a similar antibody response in each EB patient, and that (ii) different EB patients similarly exposed to S. aureus may generate highly diverse patterns of antibody responses against particular S. aureus antigens. This relates most likely to their S. aureus contact history, which was previously shown to be variable over time 19,24,26 .
The species S. aureus is known to display high genomic plasticity. Although, the genes for some virulence factors are (almost) invariantly (Efb, Eap, Emp, IsaA) or frequently (Atl, LukG, LukH, SsaA2, Sle1) present in S. aureus isolates, their amino acid sequence identity and expression levels can show substantial inter-strain variations 13,17,21,29,[50][51][52] . This diversity could be a determinant for variations in the host immune responses to S. aureus. Our results with Eap and Efb, which present large inter-lineage amino acid sequence variation 51 , showed a highly variable immune response in IgG titers in EB patients compared to healthy carriers. Interestingly, previous studies on different (not EB-related) patient cohorts showed either lower levels of anti-Eap and Efb antibodies in patients 48,53,54 or, on the contrary, higher levels in infected patients than in healthy S. aureus carriers 23,47,55,56 . Nonetheless, consistently higher IgG titers against Atl 14 and IsaA 19 in sera of S. aureus-infected or colonized patients have been previously reported, which is in agreement with our own results.
In the context of our present study, it is noteworthy that except for very severe cases, patients with EB appear to suffer infrequently from invasive staphylococcal disease, especially if one considers their high rates of colonization with S. aureus 28 . This has led to the proposal that the elevated levels of anti-staphylococcal IgGs could potentially be protective. Importantly, none of the patients who participated in this and our previous studies was treated for S. aureus bacteremia in the 5 years prior to donating the investigated plasma samples. Additional support for the idea that high anti-staphylococcal IgG levels in EB patients could be protective comes from studies with monoclonal antibodies against IsaA, showing protection against S. aureus infections in murine S. aureus infection models 6,7,57 . At present, it is not clear whether these findings for anti-IsaA antibodies can be extrapolated to other non-covalently cell wall-bound proteins. This idea is tempting in view of the present results, but it has to be noted that vaccination studies in murine models with an octa-valent vaccine, including Atl and IsaA, did not lead to protection against S. aureus challenges 18 . Yet, immune responses directed against other presently investigated antigens could be beneficial, not only by promoting opsonophagocytosis, but also by interfering with the biological activity of the different antigens as was recently shown for monoclonal antibodies against the staphylococcal complement inhibitor SCIN 58 . Clearly, potentially beneficial effects of using non-covalently cell surface-bound proteins as antigens for vaccination need to be thoroughly validated in immunization experiments. This is important, because strong immune responses against S. aureus antigens may not be (fully) protective. This is underscored by the observation that S. aureus carriers producing antibodies that neutralize superantigens can still develop sepsis, although they do have an improved prognosis 59 . Also, in animal experiments high antibody titers against S. aureus antigens, such as IsaA, are not necessarily protective against infections by this pathogen 18,60 . In this context, it is noteworthy that Hawkins et al. showed that natural exposure of humans to S. aureus elicited a strong antibody response against ClfA, but this response did not prevent the ClfA-mediated binding of S. aureus cells to fibrinogen, the natural ligand of ClfA 61 . In contrast, immunization with a ClfA-containing vaccine induced functional antibodies that prevented S. aureus from binding to fibrinogen. Another recent observation that encourages further research towards the development of an anti-staphylococcal vaccine was reported by Stentzel et al., who showed that immunoglobulin replacement therapy in STAT3 hyper-IgE syndrome patients with low S. aureus-specific serum IgG levels did not only increase the S. aureus-specific IgG levels, but also resulted in an attenuated clinical course of disease 62 .

Conclusion
In the present study, we assessed the immunogenicity of ten non-covalently cell surface-bound proteins of S. aureus, using plasma samples from patients with EB and healthy volunteers. Surface-exposed and secreted proteins of S. aureus have previously been studied as potential vaccine targets 9,10 . However, while most studies focused on covalently cell wall-bound proteins, the less-studied non-covalently cell wall-bound proteins could also be promising vaccine targets 16 . Therefore, we applied a combined bioinformatics and biochemical approach to select ten non-covalently cell wall-bound proteins of S. aureus for further analyses. These included Eap, Efb, EMP, IsaA, LukG, LukH, SA0710, Sle1 and SsaA2, as well as two separated domains of Atl. These potential antigens were expressed in L. lactis, purified and tested for antigenicity using human plasma samples. Remarkably, our present results show that the high exposure of EB patients to S. aureus was mirrored by significantly elevated IgG levels against all tested non-covalently cell wall-bound antigens. This is a novel finding, which suggests that these antigens on the cell surface of S. aureus are prime targets for the human immune system. In brief, obtained EB and control averages were multiplied by a numeric factor that resulted in the average of all controls equal to 10 AU. After plotting normalized values, the differences in the average IgG levels measured for plasma samples from EB patients and healthy control individuals were compared for the different analyzed proteins (isolates/tests). Per patient, the number of S. aureus types identified per number of sampling time points are indicated in parentheses 19,24,26 . Sa, S. aureus.