The C‐type lectin receptor MGL senses N‐acetylgalactosamine on the unique Staphylococcus aureus ST395 wall teichoic acid

Abstract Staphylococcus aureus is a common skin commensal but is also associated with various skin and soft tissue pathologies. Upon invasion, S. aureus is detected by resident innate immune cells through pattern‐recognition receptors (PRRs), although a comprehensive understanding of the specific molecular interactions is lacking. Recently, we demonstrated that the PRR langerin (CD207) on epidermal Langerhans cells senses the conserved β‐1,4‐linked N‐acetylglucosamine (GlcNAc) modification on S. aureus wall teichoic acid (WTA), thereby increasing skin inflammation. Interestingly, the S. aureus ST395 lineage as well as certain species of coagulase‐negative staphylococci (CoNS) produce a structurally different WTA molecule, consisting of poly‐glycerolphosphate with α‐O‐N‐acetylgalactosamine (GalNAc) residues, which are attached by the glycosyltransferase TagN. Here, we demonstrate that S. aureus ST395 strains interact with the human Macrophage galactose‐type lectin (MGL; CD301) receptor, which is expressed by dendritic cells and macrophages in the dermis. MGL bound S. aureus ST395 in a tagN‐ and GalNAc‐dependent manner but did not interact with different tagN‐positive CoNS species. However, heterologous expression of Staphylococcus lugdunensis tagN in S. aureus conferred phage infection and MGL binding, confirming the role of this CoNS enzyme as GalNAc‐transferase. Functionally, the detection of GalNAc on S. aureus ST395 WTA by human monocyte‐derived dendritic cells significantly enhanced cytokine production. Together, our findings highlight differential recognition of S. aureus glycoprofiles by specific human innate receptors, which may affect downstream adaptive immune responses and pathogen clearance.


| INTRODUCTION
Staphylococcus aureus is a common member of the human microbiome and colonises up to 30% of the population, where it mostly resides in the nares and on the skin (Eriksen, Espersen, Rosdahl, & Jensen, 1995;Kluytmans, van Belkum, & Verbrugh, 1997;Wertheim et al., 2005). S. aureus is a leading cause of surgical site infections and skin infections as well as health care-associated pneumonias (Pozzi et al., 2017).
Treatment of infections is hampered by the continuous emergence of antimicrobial resistance, most prominently methicillin-resistant S.
aureus and vancomycin-resistant S. aureus (Weigel et al., 2003, Lakhundi & Zhang, 2018. Understanding the molecular mechanisms underlying different S. aureus infections will support the development of new treatment strategies including vaccines. Components of the bacterial cell envelope are critical for S. aureus host-pathogen interaction, both at the level of colonisation but also during systemic infection by evading host immune responses (Weidenmaier & Lee, 2016). One of the most abundant and exposed structures on the Gram-positive cell wall is wall teichoic acid (WTA).
WTA is a glycopolymer that is covalently bound to peptidoglycan.
C-type lectin receptors (CLRs) are a family of pattern-recognition receptors that are dedicated to sense both self and non-self glycan structures through their characteristic carbohydrate recognition domains (CRDs; Brown, Willment, & Whitehead, 2018). CLRs have a particular expression pattern on subsets of immune cells. We recently identified that the CLR langerin (CD207), which is exclusively expressed on Langerhans cells in the skin epidermis, interacts with S. aureus through WTA β-1,4-GlcNAc, which affects Langerhans cell responses and skin inflammation in mice . In contrast, S. aureus ST395 does not interact with langerin . However, both dermal dendritic cells (DCs) and dermal macrophages express the trimeric CLR macrophage galactose-type lectin (MGL; CD301), which recognises terminal GalNAc residues as a result of a Gln-Pro-Asp motif in its CRD (Tanaka et al., 2017).
GalNAc is incorporated into, among others, pathogen-produced lipooligosaccharides from Campylobacter jejuni and Neisseria gonorrhoeae (van Sorge et al., 2009;, and confers binding to MGL in a Ca 2+ -dependent manner, inducing uptake and cellular responses (van Liempt et al., 2007). We therefore hypothesised that S. aureus ST395 might also be recognised by MGL via α-GalNAc modifications on WTA and may impact downstream immune responses.  We have previously observed that langerin shows a certain level of species specificity, that is, mouse langerin does not interact with S. aureus (van Dalen et al., 2019). Therefore, we investigated FIGURE 1 Human and mouse macrophage galactose-type lectin (MGL) interact with Staphylococcus aureus ST395 strains in a tagN-dependent manner. (a) hMGL binding to different S. aureus ST395 lineage strains, USA300 wild-type (WT) and Newman WT detected by anti-hisTag-FITC antibody. Control represents S. aureus PS187 WT incubated with secondary detection antibody. (b and d) Interaction between (b) hMGL or (d) mMGL2 to S. aureus PS187 WT in the absence or presence of GalNAc (50 mM) or glucose (50 mM). (c and e) Binding of (c) hMGL or (e) mMGL2 to PS187 WT, GN1, GN1 + ptagN and two non-ST395 strains. Means of geometric mean fluorescence intensity ± standard error of mean from three independent experiments are shown. *p < .05, **p < .01, ***p < .005, ****p < .0001 interaction of PS187 with mouse homologue MGL2 (Singh et al., 2009). Like human MGL, mouse MGL2 interacted with PS187, could be blocked with GalNAc, and interaction was lost upon deletion of tagN ( Figure 1d,e), suggesting that the interaction is, at least partially, conserved across species.

| S. lugdunensis tagN encodes a GalNAc-transferase that produces a MGL ligand
Similar to S. aureus ST395 WTA, certain CoNS species express GroPtype WTA. In addition, several CoNS species express homologues of the tagN gene, suggesting that CoNS may decorate WTA in a similar fashion as S. aureus ST395 strains (Winstel et al., 2014). Indeed, complementation of PS187 GN1 with a tagN homologue from Staphylococcus carnosus restores GalNAc glycosylation and phage susceptibility (Winstel et al., 2014). Similarly, we were able to confer susceptibility to phage ϕ187 by complementing the GN1 mutant, for which no transductants were obtained, with tagN from Staphylococcus lugdunensis  Figure S1). These data suggest that S. lugdunensis tagN encodes a GalNAc transferase. However, it is likely not or only lowly expressed in S. lugdunensis in our culture conditions.

| S. aureus PS187 interacts with and activates human moDCs
MGL is expressed on a range of immune cells including human DCs and macrophages residing in skin and lymph nodes, blood CD1c + DCs, and immature moDCs (van Vliet, Gringhuis, Geijtenbeek, & van Kooyk, 2006, Schutz & Hackstein, 2014, Heger et al., 2018. To investigate the interaction of MGL with S. aureus ST395 strains in a more biologically relevant system, we used a cell-based assay with human immature moDCs. Fluorescein isothiocyanate (FITC)-labeled S. aureus PS187 WT bound readily and in a ratio-dependent manner to moDCs ( Figure 3a,b). Interestingly, binding was reduced for the tagN-deficient mutant and USA300 strains, which both do not express GalNAc on their surface ( Figure 3b). Binding to moDCs was restored to WT levels in the tagN-complemented strain ( Figure 3b). Complementary, we assessed the effect of different blocking agents, that is, ethylene glycol-bis(βaminoethyl ether)-N,N,N' ,N'-tetraacetic acid (EGTA), GalNAc, and glucose (as a control; Figure 3c). Binding of PS187 WT, but not of the GN1 mutant, was reduced upon coincubation of EGTA and GalNAc, but not glucose ( Figure 3c). These data demonstrate that the PS187-moDC interaction is partially preventable by addition of GalNAc or calcium scavenging, which is in line with a possible role for MGL.
Loss of interaction with MGL may affect immune activation of moDCs, such as expression of costimulatory molecules or cytokine production, resulting in different immunological responses. We therefore investigated moDCs maturation and cytokine production after stimulation with gamma-irradiated S. aureus PS187 WT, GN1, tagNcomplemented GN1, or USA300 WT for 16 hr. MoDCs upregulated maturation markers CD80, CD83, CD86, and CD40, indicating that all S. aureus strains activate moDCs (Figure 4a, Figure S2). We observed little effect on expression of HLA-DR except with PS187 WT ( Figure S2). However, there was no difference in the induction of moDC maturation by the different S. aureus strains (Figure 4a, Figure S2). We also analysed moDC cytokine production. S. aureus PS187 WT induced expression of IL-6, IL-12p70, IL23p19, IL-10, and TNFα, but not IL-4 when incubated with moDCs ( Figure 4b).
Interestingly, at a cell-to-bacteria ratio of 1:2, cytokine production was significantly lower when strains did not produce GalNAcylated WTA, that is, PS187 GN1 and USA300 WT (Figure 4c). At higher ratios, this difference was robust for IL-6 and IL12p70 and trends remained for IL-10, IL23p19, and TNFα ( Figure 4c). Cytokine production by moDCs was restored to PS187 WT levels in cells stimulated with the tagNcomplemented strain ( Figure 4c). Overall, these data indicate that the production of select pro-inflammatory cytokines, that is, IL-6 and IL12p70, by moDCs is enhanced by recognition of the α-GalNAc modifications present on S. aureus PS187 WTA.
To determine whether differences in cytokine production are not just WTA GalNAc-dependent but also MGL-dependent, we attempted to block the interaction using a commercially available anti-MGL blocking antibody. These experiments are technically complicated by the presence of protein A and Sbi on the S. aureus surface, as these proteins bind IgG Fc, thereby possibly increasing DC interaction by binding to the blocking antibody. MoDC cytokine production in response to PS187 WT was not affected by the presence of either the blocking antibody or the isotype control antibody compared with bacteria alone ( Figure S3). We confirmed that incubation of the antibodies with moDCs by itself did also not significantly affect cytokine production ( Figure S4). Therefore, we are currently unable to prove that moDC cytokine production in response to S. aureus PS187 occurs through MGL.

| DISCUSSION
Here, we show the molecular interaction between WTA of S. aureus ST395 and MGL, an innate receptor of the CLR family. This interaction is dependent on α-GalNAc modifications of S. aureus WTA and contributes to increased cytokine production in MGL-expressing moDCs.
Although Winstel et al. showed the importance of S. aureus GalNAc glycosylation for phage infection (Winstel et al., 2014), there was no previous indication for interaction with human receptors. Because the ST395 lineage is present in nasal and blood culture isolates (Holtfreter et al., 2007), interaction with MGL may be biologically relevant in context of recognition and clearance by the immune system. Presence of the GalNAc-WTA epitope also affected DC cytokine production, especially increasing production of IL-6 and IL-12p70 across the tested range of bacteria-to-cell ratios. This is in contrast Toll-like receptor ligands to differentially affect IL-8, IL-10, and TNFα production (Heger et al., 2018;van Vliet et al., 2013). Because Toll-like receptor ligands differ between Gram-positive and Gram-negative bacteria, this may explain the different effects on DC cytokine production that we observe here. Alternatively, we can speculate that observed differences in DC cytokine production are not completely MGL-dependent. It cannot be excluded that additional receptors were triggered in the absence of WTA-GalNAc as a result of newly exposed structures on the S. aureus surface. Additionally, other receptors may be more important for induction of cytokines, which is also implied by the experiments using anti-MGL blocking antibodies, which did not affect cytokine production.   (Table S1) were grown either on Todd Hewitt (Oxoid) agar or in Todd Hewitt broth supplemented with chloramphenicol (Sigma-Aldrich) at a concentration 10 μg/ml when required. For all experiments, bacteria were grown overnight, subcultured the next day in fresh Todd Hewitt broth, and grown to exponential phase (optical density at 600 nm [OD 600 ] = 0.6) for use in experiments.

| WTA isolation and analysis by polyacrylamide gel electrophoresis (PAGE)
WTA was isolated as previously described (Winstel et al., 2013).
Briefly, overnight culture of S. aureus PS187 was grown in BM (0.5 % w/v yeast extract; 1% w/v Soy peptone; 0.5% NaCl; 0.1 % K2HPO3) supplemented with 0.25% w/v glucose was harvested by centrifugation and washed using ammonium acetate buffer (AAB, 20 mM, pH 4.8). Bacterial cells were opened using a Euler cell mill (2.5 mL AAB/4.5 glass beads/1 g cell pellet). The obtained lysate was digested overnight with RNAse and DNAse at 37°C, subsequently treated by ultrasonification, and incubated with 2% sodium dodecyl sulfate (SDS) for 1 hr at 60°C. Purified peptidoglycan was washed extensively with AAB. WTA was released by 5% tri chloroacetic acid (TCA) treatment for 4 hr at 60°C. The supernatant was neutralised using NaOH and dialyzed against ddH2O.

| Isolation of human monocytes and differentiation to immature DCs
Buffy coats from healthy anonymous donors were purchased from Sanquin Amsterdam and obtained according to the good clinical practice in accordance with the declaration of Helsinki. Donors have given their written consent to the study. Peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats using Ficoll-Paque PLUS (GE Healthcare) density gradient and monocytes were obtained as described in Sallusto and Lanzavecchia (1994). Briefly, harvested PBMCs were washed twice with RPMI 1640 (Lonza) supplemented with 5% foetal bovine serum (FBS, Biowest). Monocytes were further isolated from the PBMC fraction using density gradient of 60%, 47.5 %, and 34 % Percoll (Sigma-Aldrich) in RPMI 1640 + 10% FBS. Harvested monocytes were washed three times with RPMI 1640 + 5% FBS and incubated at the concentration 0.5 × 10 6 cells/ml with differentiation medium consisting of RPMI 1640 supplemented with 10% HyClone FBS (GE Healthcare), 800 IU/ml GM-CSF (Bio Connect), 250 IU/ml IL-4 (Thermo Fisher Scientific), 100 IU/ml penicillin-streptomycin, and 2.4 mM L-glutamine for 5 to 7 days to obtain immature DCs.

| Binding of FITC-labeled bacteria to moDCs
To perform bacteria binding assays, S. aureus strains were labelled with FITC (Sigma-Aldrich). Five milligrams of bacterial culture in exponential phase were pelleted and resuspended in cold PBS with 0.1% BSA. Bacteria were incubated with 0.5 mg/ml FITC for 30 min on ice protected from light. Bacteria were washed three times with cold PBS + 0.1% BSA supplemented with 1% ammonia and resuspended in TSM + 0.1% BSA at OD 600 of 0.4. Immature moDCs were harvested by centrifugation and resuspended in TSM + 0.1% BSA (1 × 10 6 cells/ml). Cells were incubated with bacteria at 1:2, 1:5, 1:10, and 1:20 cell to-bacteria ratios in a 96-well round bottom plate for 30 min in 4°C protected from light.
For blocking, cells were preincubated for 15 min at room temperature with 1 mM EGTA (Brunschwig Chemie), 50 mM GalNAc (Fluka, Sigma-Aldrich), or 50 mM glucose (Merck). Next, cells were incubated with bacteria at 1:10 cell-to-bacteria ratio for 30 min at 4°C, protected from light. Samples were washed with TSM + 1% BSA, fixed using 1% formaldehyde in PBS, and analysed using flow cytometry. Microscopy pictures were prepared using 1:50 cell to bacteria ratio suspensions.
Cells were attached to the glass slides using a Shandon Cytospin 3 centrifuge. Cellular membranes were stained using WGA-Alexa Fluor 647 (Thermo Fisher Scientific), cell nucleus with DAPI (Sigma-Aldrich).

| Statistical analysis
Data obtained from flow cytometry was analysed using FlowJo 10 (FlowJo LLC). Statistical analysis of data was performed using GraphPad Prism 7.02 (GraphPad Software). One-way analysis of variance followed by Dunnett's or Tukey's test or two-way analysis of variance followed by Tukey's test were performed. Only significant differences between samples (p < .05) were indicated on graphs.