Peptidoglycan Contribution to the B Cell Superantigen Activity of Staphylococcal Protein A

The LysM domain is found in all kingdoms of life. While their function in mammals is not known, LysM domains of bacteria and their phage parasites are associated with enzymes that cleave or remodel peptidoglycan.

IMPORTANCE The LysM domain is found in all kingdoms of life. While their function in mammals is not known, LysM domains of bacteria and their phage parasites are associated with enzymes that cleave or remodel peptidoglycan. Plants recognize microbe-associated molecular patterns such as chitin via receptors endowed with LysM-containing ectodomains. In plants, such receptors play equally important roles in defense and symbiosis signaling. SpA of S. aureus carries a LysM domain that binds glycan strands of peptidoglycan to influence defined B cell responses that divert pathogen-specific adaptive immune responses.
KEYWORDS B cell receptor, LysM, protein A, Staphylococcus aureus, peptidoglycan, superantigens A ll clinical Staphylococcus aureus isolates elaborate staphylococcal protein A (SpA), a surface protein well known for its ability to bind host immunoglobulins (Igs) (1,2). SpA is comprised of five immunoglobulin binding domains (IgBDs) (3,4). Each IgBD assumes a triple-helical structure with a binding site for the Fcg domain of IgG (5). Interaction with Fcg prevents C1q binding and complement activation, effectively blocking a critical effector function of antibodies (6,7). Earlier work also demonstrated that S. aureus stimulates the proliferation of isolated human B cells in a manner that requires SpA (8,9). Purified SpA extracted from the bacterial envelope was shown to promote B cell proliferation in mice and human lymphocyte preparations in the presence of CD4 1 T helper cells (10)(11)(12). This mitogenic activity is facilitated by a second site of binding of the IgBDs to the conserved surface of Ig antigen-binding fragment (Fab), which is encoded by variable heavy 3 (V H 3) clan-related genes (13). SpA can cross-link the V H 3-IgM B cell receptors (BCRs) to trigger supraclonal proliferation of B cells (9,11). During infection of mice and humans, B cell superantigen activity leads to the expansion of V H 3 idiotype plasmablasts in blood and to the secretion of V H 3 idiotype antibodies that lack recognition of staphylococcal antigens (12,14). In this manner, SpA may contribute to the lifelong association of S. aureus with humans and the bacterium's ability to cause reiterative infections (15).
SpA is synthesized as a precursor in the bacterial cytoplasm and enters the secretory pathway via an N-terminal signal peptide (16). Once translocated across the plasma membrane, the signal peptide is cleaved by signal peptidase and sortase A (SrtA) cleaves the C-terminal LPXTG sorting signal after threonine (T) and transfers the processed polypeptide to the terminal glycine (G) residue of the peptidoglycan intermediate [C 55 -(PO 4 ) 2 -MurNAc(L-Ala-d-iGlu-L-Lys(NH 2 -Gly 5 )-D-Ala-D-Ala)-GlcNAc] known as lipid II (17,18). The SpA-lipid II intermediate is then incorporated into cell wall peptidoglycan and displayed on the bacterial surface (19). During cell division, staphylococci release a portion of their peptidoglycan and peptidoglycan-linked SpA into the extracellular milieu, owing to muralytic enzymes that degrade the cell wall in the vicinity of septal membranes (20)(21)(22). Released SpA is tethered to wall peptide with the C-terminal structure L-Ala-D-iGln-L-Lys(SpA-LPET-Gly 5 )-D-Ala-Gly 4 (23). Native SpA released from the staphylococcal cell wall, but not recombinant SpA purified from the cytoplasm of Escherichia coli, is sufficient to induce the expansion of V H 3 idiotype IgG and IgM in mice (12).
In this study, we asked whether SpA modification with peptidoglycan is necessary for the expansion of V H 3 clonal immunoglobulin. By engineering strains that carry the spa gene lacking defined domains of the protein, we observed that S. aureus spa DLPXTG , a variant that can no longer immobilize SpA in the envelope, fails to exhibit B cell superantigen activity in vivo and in vitro. Further, the predicted LysM domain of SpA, positioned between the envelope spanning region X (Xr) and the LPXTG sorting signal, also contributes to B cell superantigen activity by binding the glycan chains of peptidoglycan fragments. Thus, together with peptidoglycan linked to the LPXTG motif, the LysM domain activates immune cells to implement the B cell superantigen activity of staphylococcal protein A.

RESULTS
S. aureus spa variants lacking defined C-terminal subdomains of protein A. The translational product of the S. aureus Newman spa gene encompasses an N-terminal signal peptide, five IgBDs of 58 to 61 residues (designated E, D, A, B, and C), the socalled region X (Xr) with 10 tandem repeats of the octapeptide EDNNKPGK, a predicted LysM domain, and a C-terminal LPXTG sorting signal (Fig. 1A). Earlier work demonstrated that staphylococcal secretion of SpA encompassing four or five IgBDs is essential for B cell superantigen activity (12). To investigate the contribution of the C-terminal domains of SpA to B cell superantigen activity, we generated S. aureus Newman variants with modified spa genes that lacked coding sequences for either Xr (spa DXr ), LysM (spa DLysM ), or the LPXTG sorting signal (spa DLPXTG ) (Fig. 1A). Production of the new SpA variants was assessed by immunoblotting extracts derived from cultures of bacteria fractionated into supernatant (S), cell wall (W), membrane (M), and cytoplasm (C). The S. aureus Newman Dspa strain, which does not produce SpA, was used as a control. Full-length SpA and SpA DXr and SpA LysM , lacking the C-terminal Xr and LysM domains, respectively, displayed similar fractionation profiles, with most of the proteins found in the cell wall (Fig. 1B). SpA is released from the bacterial peptidoglycan via the combined activities of the murein hydrolases AtlA, Sle, LytM, and LytN (23). Thus, SpA proteins were also found in the culture supernatant (Fig. 1B). SpA DLPXTG , which lacks peptidoglycan anchoring (LPET-COOH), was predominantly found in the culture supernatant (Fig. 1B). While similar between all mutants, total immune-reactive signals for SpA were reduced compared to those for wild-type bacteria. We attribute this difference to the reduced stability of proteins following secretion in the culture medium. In agreement with this notion, a representative immunoprecipitation experiment shows similar levels of production of SpA and SpA LysM (see Fig. S1 in the supplemental material). Immunoblotting with antibodies against the membrane protein sortase A (anti-SrtA), secreted coagulase (anti-Coa) and cytoplasmic ribosomal protein L6 (anti-L6) was used to validate the fractionation of cultures (Fig. 1B). To further visualize the surface display of SpA, microscopy images of bacteria were acquired using monoclonal antibody 3F6, which binds to the IgBDs of SpA, and Alexa Fluor 647-labeled goat anti-mouse IgG (red) (24). BODIPY-vancomycin (green) was used to visualize peptidoglycan. Cells of the wild type and spa DXr and spa DLysM variants displayed SpA on the surface. As expected, cells of the Dspa or spa DLPXTG mutant could not be labeled with 3F6 (Fig. 1C).
SpA variant-mediated B cell superantigen activity during mouse infection. Intravenous inoculation of mice with a sublethal dose of S. aureus Newman (1 Â 10 7 CFU) causes bacteremia and persistent abscess formation in all tissues of infected animals (25). During infection, V H 3 clan IgM and IgG antibodies are produced in a SpA- , and cytoplasm (C) and immunoblot analysis for the distribution of wild-type and mutant SpA. Antibodies against sortase, coagulase, and ribosomal protein L6 were used to control for successful fractionation. (C) BODIPY-vancomycin (green) staining of peptidoglycan and Alexa Fluor 647-labeled secondary antibody (red) detection of SpA monoclonal antibody 3F6 binding to S. aureus were used to visualize the surface display of SpA.
The LysM Domain of Staphylococcal Protein A ® dependent manner (26). We wondered whether mice infected with S. aureus spa DXr , spa DLysM , spa DLPXTG , and Dspa variants continue to expand V H 3 clonal antibodies. Because the variants produced less SpA protein, we first compared the abilities of the new strains to disseminate and replicate in mouse kidneys following intravenous infection. At day 5 postinfection, only animals infected with the Dspa variant displayed a significant reduction in bacterial loads in kidney tissues ( Fig. 2A). Next, cohorts of animals were infected with the same strains; blood was collected on days 5, 12, 19, and 26 postinfection. Serum samples were subjected to enzyme-linked immunosorbent assay (ELISA) with SpA KK , a protein A variant that binds V H 3 variant heavy chains but not Ig Fcg (12). Compared to mock-infected animals, mice infected with wild-type S. aureus exhibited a large expansion of V H 3 idiotype IgM in blood on day 5 postinfection (Fig. 2B). V H 3 idiotype IgM expansion abated by day 12 and was accompanied by a rise in V H 3 idiotype IgG sustained over 26 days (Fig. 2C). As expected, animals infected with the Dspa mutant did not display B cell superantigen activity ( Fig. 2B and C). Although slightly reduced compared to the case with the wild type, V H 3 antibody expansions were also observed in animals infected with the spa DXr variant. Infection of mice with S. aureus spa DLysM caused a modest expansion of V H 3 idiotype IgM on day 5, but unlike the case with the wild-type strain or the spa DXr variant, failed to activate V H 3 idiotype IgG expansion on day 12 ( Fig. 2B and C). Animals infected with the spa DLPXTG mutant did not display any B cell superantigen activity, corroborating the notion that peptidoglycan modification of naturally released SpA is key to its B cell superantigen activity ( Fig. 2B and C). The data suggest that the LysM domain also contributes to SpAinduced expansions of V H 3 idiotype immunoglobulins during infection.
Human B cell proliferation induced by S. aureus spa variants. Previous work reported that heat-killed S. aureus cells or staphylococcal extracts induce in vitro proliferation of isolated human B cells in a T cell-independent manner (8,27). To analyze the contribution of specific domains of SpA in this context, S. aureus Newman or its spa DXr , spa DLysM , spa DLPXTG , and Dspa variants were fixed with 0.5% formaldehyde, heat killed, washed in buffer, and added to B cells isolated from human blood mononuclear cell preparations using negative selection (Fig. S2A) or positive selection (Fig. S2B). Proliferation of carboxyfluorescein succinimidyl ester (CFSE)-stained CD19 1 cells was quantified by measuring dilution of fluorescence over 6 days (Fig. S2C). Compared to mock-treated lymphocytes, incubation of mixed populations of CD19 1 B cells/CD3 1 T cells or purified CD19 1 B cells with formalin-fixed S. aureus Newman at a ratio of 16 bacteria/cell caused proliferation of 16.93% 6 5.16% and 29.36% 6 2.94% of CD19 1 B cells, respectively ( Fig. 3A and B). In contrast, incubation of B cells with the Dspa mutant did not activate proliferation compared to that with phosphate-buffered saline (PBS) ( Fig. 3A and B). Incubation of CD19 1 B cells/CD3 1 T cells with S. aureus spa DXr caused proliferation of CD19 1 B cells in a manner similar to that of wild-type staphylococci (Fig. 3A). However, a small proliferation reduction was observed when S. aureus spa DXr was incubated with purified CD19 1 B cells (Fig. 3B). This difference may represent a small contribution of the Xr domain or reflect heterogeneity in blood donors. Incubation with formalin-fixed spa DLPXTG or Dspa bacteria significantly reduced CD19 1 B cell proliferation ( Fig. 3A and B). This is as expected since SpA DLPXTG is no longer retained in the envelope of bacteria and thus is washed away during the preparation of fixed cells. Interestingly, CD19 1 B cell proliferation was also significantly reduced upon incubation with spa DLysM fixed bacteria ( Fig. 3A and B). To ascertain that the phenotype was not the result of reduced display of SpA LysM , three new strains were generated whereby the wild-type spa, spa DXr , and spa DLysM genes were expressed under the control of the tetracycline-inducible promoter. The new strains, the itet-spa, itet-spa DLXr , and itet-spa DLysM variants, were grown to similar densities with adjusted concentrations of inducer to produce similar amounts of SpA proteins (Fig. 4A). Incubation of CD19 1 B cells/CD3 1 T cells with the itet-spa DLysM strain resulted in reduced expansion of B cells compared to that with itet-spa (isogenic wild-type) S. aureus (Fig. 4B), confirming that the LysM domain contributes to SpA activity.
As observed with animal infection with live bacteria, the cell expansion of human lymphocytes upon incubation of fixed S. aureus Newman or its spa DXr variant resulted in robust IgM secretion; this was not observed with the spa DLPXTG and Dspa variants (Fig. 3C). S. aureus activation of IgM secretion by CD19 1 B cells did not require CD3 1 T Human B cell proliferation induced by purified SpA variants. To determine whether SpA alone is sufficient to induce proliferation of human CD19 1 B cell populations, full-length SpA and variants SpA DXr , SpA DLysM , and SpA DLPXTG were purified from the filtered medium of S. aureus cultures via affinity chromatography on IgG Sepharose (Fig. 5A). Recombinant SpA (rSpA) (Fig. 1A) starting with six histidyl residues instead of the signal sequence and ending with the sorting motif followed by the Strep-tag sequence (LPETGWSHPQFEK) was purified from the cytosol of E. coli (Fig. 5A). Incubation of mixed populations of CD19 1 B cells and CD3 1 T cells with SpA caused proliferation of CD19 1 B cells (32.46% 6 8.41%), while rSpA did not trigger B cell expansion (Fig. 5B). As a control, subjecting filtered, conditioned medium of centrifuged S. aureus Dspa cultures to affinity chromatography generated a sample that induced proliferation of 9.80% 6 5.37% of CD19 1 B cells in the mixed CD19 1 B cell/ CD3 1 T cell populations (Fig. 5B, Dspa mutant versus rSpA, P = 0.01). We surmise that the mitogenic activity associated with the Dspa sample may be due to lipoprotein or nucleic acid contamination and activation of Toll-like receptors (TLRs) (27). Incubation of CD19 1 B cell/CD3 1 T cell populations with nonmethylated CpG, a potent inducer of TLR9 signaling (27), induced the proliferation of 29.04% 6 5.68% CD19 1 B cells (Fig. 5B). SpA DXr induced CD19 1 B cell proliferation similar to that of SpA, while removal of the LysM domain (SpA DLysM ) and LPXTG sorting signal (SpA DLPXTG ) led to a gradually reduced ability to promote B cell proliferation (Fig. 5B). In fact, SpA DLPXTG activation was similar to that observed with Dspa (Fig. 5B). Purified SpA molecules were also added to purified (97 to 99% pure) CD19 1 B cells (Fig. 5C). Compared to CpG, SpA caused a modest CD19 1 B cell proliferation (4.14% 6 2.46% versus 19.95% 6 4.03% [Fig. 5C]). None of the other protein preparations (SpA DXr , SpA DLysM , SpA DLPXTG , rSpA, and supernatant of Dspa) induced the proliferation of purified CD19 1 B cells (Fig. 5C). To rule out a contribution from contaminating lipopeptides to CD19 1 B cell proliferation, SpA was also purified from the lgt mutant (28) (Fig. S3A). Lipoprotein diacylglycerol transferase (Lgt) catalyzes transfer of phosphatidylglycerol to the sulfhydryl moiety of a cysteine residue conserved in the signal peptides of lipoprotein precursors (29). This diacylglycerol modification modulates host immune responses (30) but is abrogated in the lgt mutant of S. aureus (28). Nonetheless, CD19 1 B cell expansion by SpA was independent of diacylglycerol modification of lipoprotein precursors catalyzed by Lgt (Fig. S3B). As observed earlier, mitogen-active SpA induced CD19 1 B cells to secrete IgM but only in the presence of CD3 1 T cells (Fig. 5D and E). Incubation with rSpA did not promote IgM secretion, whereas CpG-activated CD19 1 B cells secreted IgM regardless of CD3 1 T cells ( Fig. 5D and E). SpA DXr and SpA DLysM also activated IgM secretion in mixed CD19 1 B cell/CD3 1 T cell preparations, albeit to a lesser degree than wild-type SpA without significant difference (Fig. 5D). Similar lower levels of IgM secretion were observed when CD19 1 B cells/CD3 1 T cells were treated with SpA DLPXTG or with the Dspa control sample (Fig. 5D). Thus, SpA molecules are sufficient to promote proliferation of CD19 1 B cells and secretion of V H 3 idiotypic immunoglobulin in a CD4 1 T helper cell-dependent manner. However, both the LysM domain and LPXTG sorting motif are required for optimal B cell superantigen activity.
Noncovalent peptidoglycan modification of released SpA. Cell wall hydrolases release LPXTG anchored SpA from peptidoglycan with the C-terminal structure L-Ala-D-iGln-L-Lys(SpA-LPET-Gly 5 )-D-Ala-Gly 4 (23). Earlier findings, and data presented here, suggest that this covalent modification is critical for the B cell superantigen activity of SpA (12). However, the LPXTG wall peptide modification does not account for the contribution of the predicted LysM domain. LysM domains have been described for proteins from bacteria, plants, and eukaryotes as independent folding units; in bacteria and plants, the LysM domain binds oligosaccharides comprised of GlcNAc and other amino sugar residues (31). To investigate the molecular basis for LysM domain-mediated B cell superantigen activity, purified SpA molecules released into the conditioned medium of cultures from S. aureus Newman and its spa DLysM variants were cospotted with matrix and analyzed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS), focusing on signals with m/z 700 to 4,000 in the positive ion mode (Table 1)  . Several other ion signals, predicted as peptidoglycan fragments with glycan chains of 2 or 3 MurNAc-GlcNAc repeats and attached wall peptides, were also detected in wild-type SpA samples (Table 1). Importantly, none of the ion signals listed in Table 1 was observed for the SpA DLysM sample. Taken together, these experiments suggest that released SpA copurifies with noncovalently bound peptidoglycan comprised of glycan chains that are 3 to 6 amino sugar residues in length. Copurification with glycan chains was not observed for the SpA DLysM variant.
Recombinant rSpA modified with staphylococcal peptidoglycan displays mitogenic activity. Previous studies demonstrated that a construct containing only the five IgBDs of SpA (5-IgBD) can bind V H 3 idiotypic immunoglobulins and mediate cross-linking of BCRs on isolated human CD19 1 B cells but cannot induce B cell expansion (12,27,33). When B cells were costimulated with the TLR2 agonist Pam3CSK4 or with the T cell cytokine interleukin 2 (IL-2), the 5-IgBD polypeptide promoted B cell proliferation but failed to promoted Ig production (27). rSpA used in this study was purified from the cytosol of E. coli and is not processed by leader peptidase or sortase A (Fig. 1A). Like the 5-IgBD construct (27), rSpA cannot induce B cell expansion, immunoglobulin gene expression, or IgM secretion (Fig. 4BD). rSpA with its intact LysM and sortase cleavage site (LPET/G) is an ideal tool to evaluate the contribution of peptidoglycan ligands. Purified peptidoglycan was treated with mutanolysin, and solubilized fragments were separated by reversed-phase high-performance liquid chromatography (rpHPLC) (Fig. 7A). Fractions eluting with increasing retention times were pooled and subjected to MALDI-TOF MS ( Fig. 7A (Fig. 7B, right). rSpA was incubated with mutanolysin-treated peptidoglycan or purified dimer and trimer and sortase A. Following the removal of excess ligands and enzyme, rSpA preparations were dialyzed against buffer control and added to mixed populations of CD19 1 B cells/CD3 1 T cells. As expected, mocktreated rSpA resulted in negligible CD19 1 B cell proliferation compared to that with buffer (0.80% 6 0.468% versus 0.69% 6 0.192%) (Fig. 7C). Incubation with rSpA exposed to mutanolysin-treated peptidoglycan, either the total peptidoglycan digest or pooled fractions containing peptidoglycan trimer, stimulated a small but significant proliferation of B cells (4.13% 6 0.834% and 2.39% 6 0.547%, respectively). Thus, peptidoglycan modification of rSpA results in a gain of mitogenic activity.

DISCUSSION
During cell division, S. aureus forms septal membranes and synthesizes cross wall peptidoglycan with attached SpA at midcell (34,35). Once cross wall assembly is completed, muralytic enzymes split the peptidoglycan layer and separate two spherical cells (21,(36)(37)(38)(39). Muralytic enzyme cleavage at cross wall peptidoglycan also provides for the surface display of cell wall-anchored SpA and for the release of peptidoglycanlinked SpA into the extracellular medium (21,23). Early work demonstrated that S. aureus stimulates the proliferation of isolated human B cells in a SpA-dependent manner (8,9). SpA purified from S. aureus cultures promotes B cell proliferation in mice and in human lymphocyte preparations. B cell proliferation requires direct interaction of SpA with B cell receptors as well as CD4 1 T helper cells (10)(11)(12). Recombinant SpA (rSpA) purified from E. coli does not promote B cell proliferation, which prompted a model whereby the peptidoglycan-modified LPXTG motif must be critical for the B cell superantigen activity of SpA. In this study, we validated this model by generating an S. aureus strain producing SpA lacking its C-terminal cell wall tethering motif, SpA DLPXTG . We completed our analysis by generating strains that produce SpA lacking the C-terminal Xr and LysM domain, SpA DXr and SpA DLysM , respectively. When used as a mitogen, SpA DLysM displayed reduced activity on human B cells compared to those with SpA and SpA DXr . We propose that this loss of activity correlates with LysM-mediated binding of peptidoglycan glycan chains. Further, some B cell superantigen activity could be restored following incubation of E. coli purified rSpA with sortase A and muralytic enzyme-treated peptidoglycan.
T cell-independent proliferation of isolated human B cells has also been reported (27). However, such B cell expansion requires large numbers of staphylococci and Tolllike receptor 2 (TLR2) activation by bacterial lipoproteins (8,27). Such T cell-independent proliferation of human B cells may be restricted to tissues with significant pathogen burden, for example, the draining lymph nodes of deep-seated abscess lesions, where SpA cross-linking of B cell receptors may trigger staphylococcal uptake and TLR2 activation (40,41). In contrast, localized S. aureus infection with low pathogen burden triggers systemic B cell proliferation and secretion of V H 3 idiotypic immunoglobulin in humans (14). We propose that systemic modification of adaptive immune responses is associated with the release of SpA from the bacterial surface with two peptidoglycan modifications, wall peptides linked to the C-terminal LPXTG motif and glycan chains bound to the LysM domain. Released SpA molecules cross-link V H 3 idiotypic BCRs to activate B cell proliferation and immunoglobulin secretion; however, this cannot be optimally accomplished without CD4 1 T helper cells. Also required is receptor-interacting serine/threonine protein kinase 2 (RIPK2), presumably by implementing peptidoglycan signals via the nucleotide oligomerization domain 1 (NOD1) and 2 (NOD2) receptors (12,42,43). As NOD1 responds to muramyl peptide activation and NOD2 to wall peptides, it seems plausible that these two receptors may also be involved in perceiving peptidoglycan signals associated with SpA. Biotinylated peptidoglycan of S. aureus digested with mutanolysin has been shown to colocalize with NOD2 in primary mouse keratinocytes (44). It is unknown whether the mammalian immune system recognizes the ubiquitous polymeric glycan backbone of peptidoglycan via NOD2 or other receptors. Similarity searches show that humans and most mammals are endowed with the evolutionarily conserved proteins LysMD1, LysMD2, LysMD3, and LysMD4 with an N-terminal lysin motif (LysM). Only LysMD3 has been studied in mice. Mouse LysMD3 is a type II integral membrane protein that colocalizes with the Golgi apparatus (45). mLysMD3-deficient mice exhibit no obvious immune response deficiencies in a number of models of infection and inflammation (45). LysM domains have been best characterized for prokaryotes, bacteriophages, and plants. LysM domains are approximately 50 amino acids long, fold with a baab structure, and bind to GlcNAc-containing glycans, such as chitin, chitin-like compounds, and peptidoglycan (46). In bacteriophages and bacteria, LysM domains are mostly associated with catalytic protein modules for peptidoglycan assembly and degradation (22,46). In plants, membrane proteins with multiple LysM modules mediate immune responses to bacterial pathogens as well as symbiotic recognition of bacteria (47,48).
Future studies need to determine how B cells perceive the peptidoglycan signals of SpA, for example, following endocytosis of BCRs with cross-linked SpA, and to define the specific contributions of T cells to B cell proliferation and immunoglobulin secretion. In summary, this work provides insight into the mechanisms whereby localized infection with S. aureus and release of peptidoglycan-modified SpA trigger systemic B cell responses that divert adaptive immune responses to prevent the establishment of protective immunity.

MATERIALS AND METHODS
Bacterial strains and growth. S. aureus strains Newman and RN4220 and variants were grown in tryptic soy broth (TSB) or on tryptic soy agar (TSA) at 37°C. Escherichia coli strains DH5a and BL21(DE3) were grown in Luria-Bertani broth (LB) or agar at 37°C. Ampicillin (100 mg/ml) and chloramphenicol (10 mg/ml) were used for the selection of pET-15b and pKOR1 derivatives in E. coli and S. aureus, respectively (49). Erythromycin (10 mg/ml) was used to select for ermB-marked mutations in S. aureus (50). Additional cloning procedures for the generation of spa strain variants are provided in the supplemental material, along with methods to examine the production and localization of SpA in the envelope of bacterial cells (Text S1; Table S1).
Animal experiments. Animal experiments were conducted in accordance with institutional and federal guidelines following experimental protocol review and approval and supervision by the Institutional Biosafety Committee (IBC) and the Institutional Animal Care and Use Committee (IACUC) at the University of Chicago. Overnight cultures of bacteria were diluted 1:100 into fresh TSB and grown at 37°C with rotation to an A 600 of 0.4. Staphylococci were sedimented by centrifugation at 13,000 Â g for 5 min, washed, and suspended in PBS to an A 600 of 0.4. Staphylococcal challenge doses (0.1-ml suspensions of bacteria in PBS) were quantified by spreading of sample aliquots on TSA plates, incubation, and enumeration of CFU. Cohorts (n = 10) of 6-week-old BALB/c mice were anesthetized via intraperitoneal injection of anesthetic (64.2 mg/ml of ketamine and 4.9 mg/ml of xylazine per kg of body weight). Anesthetized mice were infected by injection of 0.1 ml of 1 Â 10 8 CFU S. aureus into the periorbital venous sinus of the right eye. Mice were anesthetized and bled with heparin-coated capillary tubes from periorbital venous sinuses on days 5, 12, and 19 or by terminal cardiac puncture (day 26). Coagulated blood specimens were centrifuged at 10,000 Â g; serum supernatant was retrieved for measurements of V H 3 clonal IgG and IgM contents using ELISA (see Text S1 for details on ELISA). To ensure reproducibility, mouse experiments were repeated.
Preparation of test articles used in B cell proliferation assays. Details for B cell isolation and proliferation assays are provided in the supplemental material (Text S1). To examine the effect of killed bacteria on B cell proliferation, overnight cultures of S. aureus were diluted 1:100 into fresh TSB and grown at 37°C to an A 600 of 0.4. For itet-spa and itet-spa DLysM strains, 80 ng/ml and 400 ng/ml of anhydrotetracycline were added to cultures, respectively. Staphylococci were centrifuged, washed, and normalized to a load of 4 Â 10 8 CFU/ml. Bacteria were killed by addition of 0.5% formaldehyde and incubation at room temperature for 3 h, followed by heat treatment at 80°C for 3 min (9). Killed staphylococci were centrifuged for 5 min at 13,000 Â g, the supernatant was discarded, and bacteria were suspended in the same volume of PBS. To examine the effect of purified SpA on B cell proliferation, first the sbi::erm allele was crossed in each background, yielding S. aureus sbi::ermB (isogenic wild type lacking Sbi), spa DXr sbi::ermB, spa DLysM sbi::ermB, spa DLPXTG sbi::ermB, and Dspa sbi::ermB variants, which were grown overnight in TSB and diluted 1:100 into 8 liters of fresh TSB. Cultures were grown with rotation at 37°C for 3 h and centrifuged at 8,000 Â g for 10 min. Culture supernatants were filtered through 0.22-mm membranes (Millipore) and subjected to affinity chromatography over 2 ml of IgG Sepharose 6 Fast Flow (GE Healthcare). Bound proteins were eluted with 0.1 M glycine-HCl (pH 3.0), followed by immediate pH neutralization with 1 M Tris-HCl (pH 8.0). Eluates were concentrated in a vacuum centrifuge concentrator, dialyzed against PBS, and used at a concentration of 80 nM in B cell proliferation assays. Protein concentrations were determined with the Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific).
The LysM Domain of Staphylococcal Protein A ® Purification of peptidoglycan and recombinant proteins and binding experiments. Peptidoglycan was prepared as described previously (51), suspended in water, normalized to an A 600 of 10, and stored at 220°C. Recombinant proteins were produced using E. coli BL21(DE3) from crude lysate generated as described previously (26,51,52). Additional details are provided in the supplemental material (Text S1).
Statistical analysis. Statistical analysis was performed with GraphPad Prism (GraphPad Software, Inc., La Jolla, CA; version 7.04). Statistically significant differences were calculated by using statistical methods as indicated. P values of #0.05 were considered significant.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.05 MB.