Pseudomonas aeruginosa biofilms display carbohydrate ligands for CD206 and CD209 that interfere with their receptor function

Bacterial biofilms represent a challenge to the healthcare system because of their resilience against antimicrobials and immune attack. Biofilms consist of bacterial aggregates embedded in an extracellular polymeric substance (EPS) composed of carbohydrate polymers, nucleic acids and proteins. Carbohydrates within P. aeruginosa biofilms include neutral and mannose-rich Psl, and cationic Pel composed of N-acetyl-galactosamine and N-acetyl-glucosamine. Here we show that P. aeruginosa biofilms display ligands for the C-type lectin receptors mannose receptor (MR, CD206) and Dendritic Cell-Specific Intercellular adhesion molecule-3-Grabbing Non-integrin (DC-SIGN, CD209). Binding of MR and DC-SIGN to P. aeruginosa biofilms is carbohydrate-and calcium-dependent and extends to biofilms formed by clinical isolates. Confocal analysis of P. aeruginosa biofilms shows abundant DC-SIGN ligands among bacteria aggregates while MR ligands concentrate into discrete clusters. DC-SIGN ligands are also detected in planktonic P. aeruginosa cultures and depend on the presence of the common polysaccharide antigen. Carbohydrates purified from P. aeruginosa biofilms are recognised by DC-SIGN and MR; both receptors preferentially bind the high molecular weight fraction (HMW; >132,000Da) with KDs in the nM range. HMW preparations contain 74.9-80.9% mannose, display α-mannan segments and alter the morphology of human dendritic cells without causing obvious changes in cytokine responses. Finally, HMW interferes with the endocytic activity of cell-associated MR and DC-SIGN. This work identifies MR and DC-SIGN as receptors for bacterial biofilms and highlights the potential for biofilm-associated carbohydrates as immunomodulators through engagement of C-type lectin receptors. Author Summary Selective engagement of pattern recognition receptors during infection guides the decision-making process during induction of immune responses. This work identifies mannose-rich carbohydrates within bacterial biofilms as novel molecular patterns associated with bacterial infections. P. aeruginosa biofilms and biofilm-derived carbohydrates bind two important lectin receptors, MR (CD206) and DC-SIGN (CD209), involved in recognition of self and immune evasion. Abundance of MR and DC-SIGN ligands in the context of P. aeruginosa biofilms could impact immune responses and promote chronic infection.


Introduction 71
Figure 1. P. aeruginosa biofilms display ligands for MR and DC-SIGN. P. aeruginosa biofilms generated in 96 well plates for 24 h were fixed and incubated with MR-CTLD4-7-Fc or DC-SIGN-Fc followed by anti-human Fc antibody conjugated to alkaline phosphatase. A. DC-SIGN and MR-CTLD4-7 bind to PAO1 biofilms and binding is selectively inhibited by mannose and fucose compared to galactose. Binding assays were performed in two NaCl concentrations (0.154 M and 1M) and for the 1M condition, incubation with Fc chimeric proteins was performed in the presence or absence of different concentrations of monosaccharides (1, 0.2 and 0.04 mM). Binding at 1M NaCl is comparable to that observed at 0.154 M (ns). Presence of monosaccharides significantly reduced protein binding. Mannose and fucose preferentially inhibited binding of MR-CTLD4-7 compared to galactose at 0.2 and 0.04 mM and binding of DC-SIGN at 1 and 0.2 mM. Mann: mannose, Fuc: fucose; Gal: galactose. N=3 in triplicate. One-way ANOVA with Tukey's multiple comparison test. B. DC-SIGN and MR-CTLD4-7 binding to PAO1 biofilms is Ca 2+ -dependent. Binding of MR-CTLD4-7 and DC-SIGN to PAO1 biofilms was tested in the presence and absence of 10 mM CaCl2. N=2 in triplicate. C. MR and DC-SIGN recognise biofilms formed by P. aeruginosa wound isolates. Binding of both lectins was significant in all instances except in the case of MR-CTLD4-7 binding to CW7. No binding of the control protein MR-CR-CR-FNII-CTLD1-3 (15) was observed. Two-way ANOVA with Tukey's multiple comparison test. Right panel: Biofilm formation by clinical isolates tested using crystal violet assay. N=3 in triplicate. Graphs show mean +/-SEM. ns: non-significant.   Figure S2 were incubated with DC-SIGN and MR-CTLD4-7 Fc-chimeric proteins followed by anti-human Fc-secondary antibody conjugated to Alexa 647 (magenta) and counterstained with DAPI (DNA, blue) and FM1-43FX (bacteria, green). Z-stacks were acquired for all samples using confocal microscopy (see videos S1 and S2). The top panels show slice 9 for DC-SIGN, and 12 for corresponding secondary Ab control (insets) and bottom panels slice 12 for MR-CTLD4-7 and its control (insets). The same settings for image acquisition and processing were maintained for test and control samples (Videos S3 and S4). Inset shows signal with biofilms incubated with secondary antibody only. No specific labelling was seen with MR-CR-FNII-CTLD1-3 Fc chimeric protein (video S5). Size bar 4 µm.

DC-SIGN binds to planktonic P. aeruginosa 143
Further analysis of DC-SIGN and MR binding to P. aeruginosa biofilms using ELISA-based 144 assays identified binding of DC-SIGN to biofilms generated by the Psl-deficient mutant ΔwspF 145 Δpsl (16)( Table 1); this mutant is not expected to contain mannose-rich carbohydrates. The 146 ΔwspF background confers constitutive high levels of cyclic-di-GMP, overproduces Psl and 147 promotes biofilm formation; a phenotype that resembles that of small rough colony variants 148 found during chronic infection (17). MR-CTLD4-7 displays reduced binding to ΔwspF Δpsl 149 biofilms indicating that the psl operon is likely responsible for the generation of MR-CTLD4-7 150 ligands ( Figure 3A). We investigated the possibility of DC-SIGN interacting with planktonic 151 cells, which could account for binding to biofilms in the absence of Psl. DC-SIGN binds 152 planktonic P. aeruginosa cultures and binding was independent of Psl and/or Pel ( Figure 3B).  Binding of MR-CTLD4-7 but not DC-SIGN to PAO1 biofilms depends on the presence of the psl operon. Biofilms formed by wspF pel (Psl+/Pel-) or wspF psl (Psl-/Pel+) were generated in 96 well plates for 24 h, fixed and incubated with MR-CTLD4-7-Fc or DC-SIGN-Fc followed by anti-human Fc antibody conjugated to alkaline phosphatase. Two-way ANOVA with Dunnett's multiple comparison test. N=3 in triplicate. Right panel. Biofilms formation was confirmed using crystal violet assay. B. Planktonic cultures of P. aeruginosa PAO1 and different mutants were collected, fixed and used to coat wells of MaxiSorp plates. Wells were incubated with MR-CTLD4-7-Fc or DC-SIGN-Fc followed by anti-human Fc antibody conjugated to alkaline phosphatase. DC-SIGN, but not MR, bind planktonic bacteria and binding is independent from the presence of Psl and/or Pel. N=4 in triplicate. C. DC-SIGN binding to planktonic bacteria is dependent on presence of CPA LPS which is absent in the rmd and wbpL mutants. N=3 in triplicate. Man-PAA and Fuc-PAA refer to commercial mannose and fucose polymers. Right panel: Adherence of planktonic cells to the wells was confirmed by ELISA using an antibody against P. aeruginosa (Anti-PA). N=3 in triplicate. Graphs show mean +/-SEM. 162

MR and DC-SIGN bind to carbohydrates produced by P. aeruginosa biofilms 163
To determine whether mannose-rich sugars from P. aeruginosa biofilms bound MR and DC-164 SIGN, carbohydrates from cultures of the Pel-deficient mutant ΔwspF Δpel (Table 1) were 165 purified as described (19). Two preparations generated independently, 1 and 2, were divided 166 into high (>45 kDa, HMW) and low molecular weight (<45 kDa, LMW) by gel filtration 167 chromatography based on protein standards (19). Gel permeation chromatography (GPC) 168 confirmed differences in size (15,370 Da for LMW-1 and 182,300 Da and 132,670 Da, for 169 HMW1 and HMW-2, respectively. LMW-2 was not investigated) ( Figure 4A). A substantial 170 amount of the material in all the samples (~33 -40% of the total mass) eluted with the included 171 volume. In our system, this means compounds with low MW, i.e. 1000 Da. Their nature is 172 unknown, but we propose that they could be carbohydrate breakdown products. No major 173 protein or DNA contamination were detected based on Silver ( Figure S3) and Coomassie 174 staining, protein quantification and spectrophotometry (data not shown). DC-SIGN and MR-175 CTLD4-7 bind to HMW and LMW preparations ( Figure 4B and C) but binding to HMW was 176 stronger. In contrast to their biofilm binding ability, MR-CTD4-7 and DC-SIGN bind similarly to 177 both HMW preparations. 178 Figure 4. Size analysis and binding to DC-SIGN and MR-CTLD4-7 of P. aeruginosa biofilm-associated carbohydrate. A. GPC analysis of HMW-1, LMW-1 and HMW-2 confirms successful fractionation into high and low MW forms. HMW-2 contained two peaks poorly resolved. This indicates that this sample is comprised of two components that are similar in MW and perhaps conformation. The MW for HMW-2 reflects the average MW for the entire sample. B. Lectin binding assays demonstrate binding of MR-CTLD4-7 and DC-SIGN to HMW-1, LMW-1, HMW-2 and LMW-2. Robust binding of HMW-1 and 2 was observed at 0.5 µg/ml while binding of LMW-1 and 2 at this concentration was substantially reduced. Dose-dependent binding of HMW-1 and HMW-2 to MR-CTLD4-7 and DC-SIGN occurs at lower doses ( Figure S4). Fc-chimeric proteins and anti-human Fc-secondary antibody conjugated to alkaline phosphatase were used. Graphs show mean ± SEM of 2 independent repeats done in duplicate. 179 Initial 1 H-NMR analysis indicated increased level of impurities in LMW-1 compared to HMW-1 180 and HMW-2 (data not shown), hence further work largely focused on HMW preparations. The 181 hydrolysed carbohydrate monomer compositions in weight % for HMW-1 is 74.9% mannose, 182 14.7% glucose, 7.4% galactose, and 3.0% rhamnose and for HMW-2 80.9% mannose, 11.0% 183 glucose, 2.3% galactose, and 5.7% rhamnose. The 1 H-NMR spectra of HMW-1 and HMW-2 184 are very similar ( Figure 5A) and show that, mannose, the major monomer present, arose from 185 mannan segments in the polymer ( Figure S5) (20). The mannose-rich composition of HMW 186 preparations agrees with previous findings (9) and is supported by its recognition by 187 Hippeastrum Hybrid Amaryllis (HHA) lectin ( Figure 5B) commonly used to detect the 188 mannose-rich carbohydrate Psl within P. aeruginosa biofilms (21). Binding of DC-SIGN and 189 MR to HMW-2 was further confirmed using bio-layer interferometry and purified full-length 190 human MR and biotinylated tetrameric DC-SIGN (22). Analysis of the binding kinetics revealed 191 that both receptors bound HMW-2 with KDs in the nM range ( Figure 5C). 192 Figure 5. Characterisation of high molecular weight biofilm carbohydrates. A. 1 H-NMR spectra from HMW-1 (top) and HMW-2 (bottom) demonstrate that they are very similar and contain primarily carbohydrates composed of α(1-6) linked mannose segments. B. HMW-2 is recognised by the mannose-specific lectin HHA in a lectin binding assay. HHA recognises both (1-3) and (1-6) α-linked mannose structures. LPS-free HMW-2 ( Figure S3) and HHA conjugated to alkaline phosphatase were used in these assays. Lewis x -PAA and Heat-Killed Candida albicans (HKC) were negative and positive controls, respectively. Graph shows Mean ± SEM of two independent repeats done in duplicate. C. HMW-2 binds rhDC-SIGN and rhMR. Tetrameric hDC-SIGN, biotinylated and immobilised on a streptavidin sensor and rhMR immobilised on a Ni-sensor were incubated with different HMW-2 concentrations. The table shows equilibrium dissociation constants for the receptor ligand interaction in μM (KD); receptor density on the biosensor surface (BMAX) and non-specific binding (KNON-

Effect of biofilm carbohydrate on human dendritic cells 194
Following on previous findings, we next explored the possibility of the mannose-rich HMW 195 biofilm carbohydrate preparations altering the phenotype of human moDCs (MR + , DC-SIGN + 196 cells, Figure S6). However, silver staining of HMW-1 and HMW-2 highlighted substantial 197 endotoxin contamination ( Figure S3A). Accordingly, HMW-1 and HMW-2 induced high levels 198 of TNF-α by moDCs that were reduced in the presence of polymyxin B. In addition, the pattern 199 of cytokines produced by moDCs in response to HMW-2 was indistinguishable from that of 200 purified endotoxin based on a cytokine microarray assay (Data not shown). LPS removal from 201 HMW-2 was achieved using an endotoxin removal column and confirmed using SDS-PAGE 202  aeruginosa and biofilms co-exist in the host, with planktonic bacteria primarily associated with acute infections and biofilms with chronic infections. Planktonic bacteria display traits associated with enhanced cytotoxicity (T3SS+) and ability to stimulate immune cells (Flagellin+) and can trigger multiple signalling pathways through engagement of pattern recognition receptors (Toll-like receptors are displayed as example). In this instance DC-SIGN, could modulate cellular activation and lead to upregulation of IL-10 production (as observed in other pathogens). Biofilms engage both DC-SIGN (through biofilm-associated carbohydrate and ligands in planktonic cells) and MR (through biofilm-associated carbohydrates). In addition, biofilm-associated bacterial cells could display reduced ability to cause cytotoxicity and stimulate pattern recognition receptors. In this instance MR and DC-SIGN could modulate an already altered cellular activation likely leading to further modulation of immunity away from pro-inflammatory Th1-dominated responses.  Binding of DC-SIGN and MR to P. aeruginosa biofilms was detected using ELISA-based 291 assays and confocal microscopy and there was good correlation between both assays with 292 DC-SIGN binding being more abundant and widely distributed than that of MR. Findings agree 293 with the broader binding specificity observed for DC-SIGN (recently reassessed in (30)). We 294 hypothesise that MR binds a subset of the DC-SIGN binding sites as suggested by the 295 clustering of the binding sites for both lectins. The confocal study also supports the 296 heterogeneity of mannose-rich structures within P. aeruginosa biofilms as both MR and DC-297 SIGN binding patterns differ from that of the HHA lectin, normally used for the detection of 298 mannose-rich biofilm-associated carbohydrates (21) (Compare Figure 2 and Figure S2). HHA 299 preferentially binds bacterial aggregates, which indicates preferential binding to mannose 300 structures associated to bacterial cells. It is possible that MR and/or DC-SIGN binding sites in 301 cell-associated carbohydrates are blocked through binding to the mannose-specific P. 302 aeruginosa lectin LecB that directly interacts with Psl (31). Our results agree with the existence 303 of distinct Psl epitopes (class I, II, and III) which can be targeted with different monoclonal 304 antibodies (mAbs) (32) and are differentially distributed within mature PAO1 biofilms (33). 305 The predicted carbohydrate structure for the HMW preparations used in this work do not 306 conform to that described for Psl, the mannose-rich neutral polysaccharide produced by PAO1 307 via the psl operon products. Byrd et al. described Psl as repeating pentameric units of D-308 mannose, L-rhamnose and D-glucose (9). In contrast, our preparation contains a small 309 proportion of galactose and, unlike the structure proposed for Psl, lacks mannose β anomers. 310 There is a high proportion of 1-6-linked-α-mannose with some 1-2 linkages, characteristics of 311 mannans. In C. albicans the structure of mannan varies depending on the culture conditions 312 (34) and it is highly feasible that differences in growth conditions, purification procedures, 313 including selection of HMW forms, and bacteria strain (WT vs wspF pel) could account for 314 these observations. In agreement with our findings Bates et al using the same purification In summary, this work demonstrates direct interaction between biofilm-associated 362 carbohydrates and immune C-type lectins and opens the possibility for these receptors to 363 contribute to the establishment of chronic infections. 364

Biofilm quantification assay 366
All strains (Table 1) were used as controls. Binding of HHA to purified carbohydrates was tested in a similar way 411 using alkaline phosphatase-conjugated HHA (20 µg/ml, LA-8008-1 EY laboratories). 412

Generation of human monocyte-derived dendritic cells 514
Human monocyte-derived dendritic cells (moDCs) were prepared from buffy coats (Blood 515 Transfusion Service, Sheffield, UK). Use of buffy coats was approved by the Faculty of 516 Medicine and Health Sciences Research Ethics Committee. PBMCs were isolated using 517 Histopaque-1077 (H8889, Sigma) and monocytes were isolated using human CD14