Real-Time Monitoring of Adherence of Streptococcus Anginosus Group

+ MODEL RESMIC3092_proof ■ 3 August 2012 ■ 1/12

1 66 2 67 3 68 4 Research in Microbiology xx (2012) 1e12 69 5 www.elsevier.com/locate/resmic 70 6 71 7 72 8 73 9 Real-time monitoring of adherence of Streptococcus anginosus group 74 10 75 11 bacteria to extracellular matrix decorin and biglycan proteoglycans in 76 12 77 13 biofilm formation 78 14 79 15 80 16 Q3 Janine Landrygan-Bakri, Melanie J. Wilson, David W. Williams, Michael A.O. Lewis, 81 17 Rachel J. Waddington* 82 18 83 19 Tissue Engineering and Reparative Dentistry, School of Dentistry, Cardiff University, Heath Park, Cardiff CF14 4XY, UK 84 20 85 21 Received 2 May 2012; accepted 16 July 2012 86 22 87 23 88 24 89 25 90 Abstract 26 91 27 92 28 Members of the Streptococcus anginosus group (SAGs) are significant pathogens. However, their pathogenic mechanisms are incompletely 93 29 understood. This study investigates the adherence of SAGs to the matrix proteoglycans decorin and biglycan of soft gingival and alveolar bone. 94 30 Recombinant chondroitin 4-sulphate(C4S)-conjugated decorin and biglycan were synthesised using mammalian expression systems. 95 31 C4S-conjugated decorin/biglycan and dermatan sulphate (DS) decorin/biglycan were isolated from ovine alveolar bone and gingival connective 96 32 tissue, respectively. Using surface plasmon resonance, adherence of the SAGs S. anginosus, Streptococcus constellatus and Streptococcus 97 33 intermedius to immobilised proteoglycan was assessed as a function of real-time biofilm formation. All isolates adhered to gingival proteo- 98 34 glycan, 59% percent of isolates adhered to alveolar proteoglycans, 70% to recombinant decorin and 76% to recombinant biglycan. Higher 99 35 adherence was generally noted for S. constellatus and S. intermedius isolates. No differences in adherence were noted between commensal and 100 36 pathogenic strains to decorin or biglycan. DS demonstrated greater adherence compared to C4S. Removal of the GAG chain with chondroitinase 101 37 ABC resulted in no or minimal adherence for all isolates. These results suggest that SAGs bind to the extracellular matrix proteoglycans decorin 102 38 and biglycan, with interaction mediated by the conjugated glycosaminoglycan chain. 103 39 Ó 2012 Published by Elsevier Masson SAS on behalf of Institut Pasteur. 104 40 105 41 Keywords: Streptococcus anginosus group; Decorin; Biglycan; Dermatan sulphate; C4S; Surface plasmon resonance 106 42 107 43 108 44 109 45 110 46 111 47 Q1 1. Introduction gastrointestinal and genitourinary tract. Whilst present in low 112 48 numbers in dental plaque, SAGs are frequently isolated from 113 49 114 50 The Streptococcus anginosus group (SAG) comprises polymicrobial infections; they are predominant components of 115 51 microaerophilic bacteria which are generally regarded as a number of orofacial infections, including dento-alveolar and 116 52 members of the commensal flora of the body (Gossling, 1988; periodontal abscesses, and are frequently associated with 117 53 Whiley et al., 1992). However, SAG members are frequently failed root canal treatment (Jacobs et al., 2003; Ledezma- 118 54 isolated from a range of clinical sites, including liver and brain Rasillo et al., 2010; Lewis et al., 1986; Schuman and 119 55 abscesses, infective endocarditis and infections of the Turner, 1999; Siqueira et al., 2002; Okayama et al., 2005; 120 56 Whiley et al., 1992). It has been proposed that SAGs are 121 57 122 58 present early in the pathogenic process and may actually 123 * Corresponding author. Tel.: þ44 (0) 29 2074 2609. 59 initiate infection, thereafter preparing the environment for 124 E-mail addresses: [email protected] (J. Landrygan-Bakri), wilsonmj@ 60 cardiff.ac.uk (M.J. Wilson), [email protected] (D.W. Williams), subsequent colonisation by anaerobic species (Aderhold et al., 125 61 [email protected] (M.A.O. Lewis), [email protected] 1981; Gossling, 1988; Nagashima et al., 1999; Shinzato and 126 62 (R.J. Waddington). Saito, 1994). As microaerophilic bacteria, SAGs can 127 63 128 64 0923-2508/$ - see front matter Ó 2012 Published by Elsevier Masson SAS on behalf of Institut Pasteur. 129 65 http://dx.doi.org/10.1016/j.resmic.2012.07.006 130

Please cite this article in press as: Landrygan-Bakri, J., et al., Real-time monitoring of adherence of Streptococcus anginosus group bacteria to extracellular matrix decorin and biglycan proteoglycans in bioﬁlm formation, Research in Microbiology (2012), http://dx.doi.org/10.1016/j.resmic.2012.07.006 RESMIC3092_proof ■ 3 August 2012 ■ 2/12

2 J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12

131 proliferate in regions of low oxygen tension, further reducing that the binding to these cell surface receptors may result in 196 132 both the oxygen concentration and redox potential to favour host cell responses, including intracellular signalling, influ- 197 133 198 134 the growth of strict anaerobes. They have a marked propensity encing host cell survival, promotion of antimicrobial peptide 199 135 for haematogenous spread and thus infection involving SAGs synthesis and initiation of the inflammatory response (Baron 200 136 is often complicated by seeding of the infective process else- et al., 2009; Hoepelman and Tuomanen, 1992; Jockusch 201 137 where in the body (Gossling, 1988). Despite being almost et al., 1995; Wang et al., 2011), and thus they may be 202 138 universally susceptible to penicillin, infections involving significant in the establishment of infection. Although still in 203 139 SAGs are often difficult to treat (Green et al., 2001; Han and their infancy, these studies have led to suggested options for 204 140 Kerschner, 2001; Schuman and Turner, 1999). Although therapeutic strategies related to the elucidation of microbe/PG 205 141 206 142 a number of potential virulence factors have been investigated, interactions (Baron et al., 2009; Hoepelman and Tuomanen, 207 143 including platelet aggregation, fibrinogen and fibrin binding, 1992; Jockusch et al., 1995). Interactions with GAGs associ- 208 144 haemolytic and hydrolytic activity, the pathogenesis of SAGs ated with decorin and biglycan, CS and DS, have been 209 145 remains poorly understood (Jacobs et al., 2000; Kitada et al., demonstrated for species Streptococcus pyogenes (Frick et al., 210 146 1997; Shain et al., 1996; Willcox, 1995). 2003), Streptococcus uberis (Almeida et al., 1999a,b; Almeida 211 147 The extracellular matrix (ECM) of connective tissue et al., 2003) and Streptococcus pneumonia (Tonnaer et al., 212 148 provides a prime site for the adherence or attachment of 2006). However, no studies have investigated binding of 213 149 214 Streptococcus 150 bacteria in the establishment of infection. The principle species to the protein cores of these PGs. 215 151 elements may be considered as a collagen fibrous network Furthermore, no studies have investigated the interaction of 216 152 providing structural support, embedded in and interacting with these PGs with SAGs. 217 153 a non-collagenous matrix consisting of proteoglycans (PGs) Against this background, the aim of this study was to 218 154 and various glycoproteins. Across the breadth of connective explore the binding of various SAG of bacteria with the PGs 219 155 tissues, perhaps one of the most prominent PG families is that decorin and biglycan. Specifically, this study will assess the 220 156 of the small-leucine rich proteoglycans (SLRPs) decorin and attraction and binding strength of bacteria for PGs in real-time 221 157 222 158 biglycan. As for all SLRPs, decorin and biglycan possess using surface plasmon resonance (SPR), thus assessing 223 159 a common protein core domain structure, dominated by early events associated with bacterial biofilm formation and 224 160 a central domain consisting of 6e10 leucine-rich repeat colonization. This study will investigate SAG isolates recov- 225 161 sequences flanked by an N-terminal and C-terminal region ered from healthy body sites and clinical infections. These 226 162 with a conserved cysteine pattern (Iozzo, 1999). Close to the findings will be compared and interpreted in terms of the 227 163 N-terminal, decorin and biglycan are conjugated to one and differential binding of the individual strains with the CS- and 228 164 two glycosaminoglycan (GAG) chains respectively. These DS-conjugated forms of decorin and biglycan, present in 229 165 230 166 GAG chains represent long polymer disaccharide repeat mineralised and soft connective tissue. 231 167 sequences of an uronic acid and an N-acetyl hexosamine, 232 168 with molecular weight 20e40 kDa (Iozzo, 1999). Within 2. Materials and methods 233 169 mineralised matrices of bone and dentine, these GAG chains 234 170 are predominantly characterised as the glucuronic acid- 2.1. Recombinant proteoglycans 235 171 containing chondroitin sulphate (CS), whilst in soft connec- 236 172 tive tissues such as skin, ligament and tendon, the iduronic cDNAs for mouse decorin (mDCN-5) or biglycan (clone 3) 237 173 238 174 acid-containing dermatan sulphate (DS) predominates cDNAs were subcloned in the mammalian expression vector 239 175 (Waddington and Embery, 1991). In addition, PGs are also pcDNA3.1 myc/his (Invitrogen) followed by liposomal 240 176 present on the cell surface, with the most prominent members transfection into HeLa cells as previously described by Sugars 241 177 belonging to the syndecan family. These PGs are identified in et al. (2002). Transfected cells were cultured in Eagles 242 178 multiple forms on the majority of cell and tissue types, Minimum Essential Medium (EMEM) with Earles Balanced 243 179 although their expression is selectively regulated in a cell-, Salt Solution (EBSS), supplemented with 2 mM glutamine, 244 180 tissue- and development-dependent manner (Kim et al., 1994). 1% antibiotic/antimycotic solution and 2% foetal bovine 245 181 246 182 Structurally, these molecules consist of a core protein which serum (FBS) at 37 C, 5% CO2. Recombinant decorin or 247 183 crosses the cell membrane, and the extracellular domain biglycan was purified from the retained media using 1 ml 248 184 carries several heparin sulphate and/or chondroitin sulphate HiTrap nickel affinity chelating columns prepared as per 249 185 glycosaminoglycan chains. manufacturing instructions (GE Healthcare). His-tag 250 186 When considering ECM/microbial interactions in the recombinant proteins were eluted with 0.02 M sodium phos- 251 187 establishment of infection involving streptococcal species in phate, 0.5 M sodium chloride and 500 mM imidazole, pH 7.4, 252 188 general, much research in the literature has been restricted to dialysed exhaustively against double-distilled water (4 days 253 189 254 190 establishing binding of bacteria to heparin sulphate and containing the protease inhibitors 5 mM benzamidine-HCl, 255 191 heparin (Rostand and Esko, 1997), with interactions facilitated 1mMiodoaceticacid,5mMN-ethylmaleimide (Sigma 256 192 via the microbial surface cell recognition adhesion matrix Aldrich) and 1 day double-distilled water only) and recovered 257 193 molecules (MSCRAMMs) present on the bacterial surface by lyophilisation. 258 194 (Almeida et al., 1999a,b; Almeida et al., 2003; Egesten et al., Recombinant proteoglycans were further purified by anion 259 195 2011; Frick et al., 2003; Wang et al., 1993). There is evidence exchange using a 1 ml Resource Q column (GE Healthcare) in 260

J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12 3

261 326 order to remove co-eluting a2HS-glycoprotein (Sugars et al., (GE Healthcare), integrated into a fast performance anion 262 2002). Recombinant proteoglycans were applied to the exchange chromatography system (GE Healthcare) and eluted 327 263 328 column in 7 M urea, 0.05 M triseHCl buffer, pH 6.8 (1 mg/ with 7 M urea, 0.05 M sodium acetate buffer, pH 6.8, 2 ml/ 264 e 329 265 ml) and eluted using an expanded 0 1 M sodium chloride min. Bound material was selectively eluted with a linear 330 266 gradient. Pooled fractions were obtained according to peaks 0e1 M sodium chloride gradient over 240 ml. Eluting frac- 331 267 within the 280 nm profile, exhaustively dialysed against tions were pooled according to the protein peaks identified in 332 268 double-distilled water (with protease inhibitors as above), the 280 nm profile, dialysed against double-distilled water 333 269 lyophilised and re-dissolved in double-distilled water to (as above) and recovered by lyophilisation. Samples were 334 270 20 mg/ml, which was adjusted following determination of dissolved to 20 mg/ml in double-distilled water, with the 335 271 336 protein concentration using the bicinchoninic acid (BCA) concentration adjusted following determination of the 272 m 337 273 assay kit (Pierce Endogen Ltd., Cheshire, UK). 1 l of each of concentration using the BCA protein assay (Pierce Endogen 338 274 the pooled fraction samples was applied to a nitrocellulose Ltd., Cheshire, UK), and proteoglycan-rich fractions were 339 275 membrane (HybondÔ C, Amersham Pharmacia Biotech) identified by immunoreactivity with anti-chondroitin sulphate 340 276 which was then blocked in 10% non-fat milk (MarvelÔ), in antibody using the dot blot assay described above. 341 277 TBS (10 mM triseHCl, 0.15 M sodium chloride, pH 7.4) for 342 278 30 min. Proteoglycan-rich fractions were identified by incu- 2.3. Chondroitinase ABC digestion 343 279 344 280 bation with mouse monoclonal anti-chondroitin sulphate, 345 281 CS-56 (Sigma Aldrich; 1:200 dilution in 3% non-fat milk in Samples of recombinant decorin, biglycan or proteoglycan 346 282 TBS) for 1 h. Subsequent to washing with TBS, membranes from gingival or alveolar bone were mixed with an equal 347 283 were incubated with goat anti-mouse IgM conjugated to volume of 0.2 M triseHCl buffer, 0.06 M sodium acetate, pH 348 284 alkaline phosphatase (Sigma Aldrich; 1:10,000 dilution) for 8 containing 2 units/ml chondroitinase ABC (protease-free, 349 285 1 h, washed with TBS and immunoreactivity was visualised Seikagku Corporation, Tokyo) and incubated at 37 C for 350 286 using NBT/BCIP chromogenic substrate (Promega). 1e2 h. Core proteins were separated from chondroitinase 351 287 352 288 ABC and GAG fragments using gel filtration chromatography 353 289 2.2. Isolation of proteoglycans from gingiva and on a pre-packed Sephadex 75 HR 10/30 column (Amersham 354 290 alveolar bone Pharmacia Biotech.), eluted with 4 M urea, 0.05 M TriseHCl, 355 291 pH 8, containing 0.35 M sodium acetate, at a flow rate of 356 292 Gingiva was dissected 1e2 mm below the gingival-tooth 1 ml/min 280 nm absorbance was monitored and protein 357 293 358 junction around the upper and lower dentition of 9-month- fractions were pooled, dialysed against ddH2O with protease 294 old Welsh Mule breed sheep (within 1e2 h of slaughter at the inhibitors (as described above) and lyophilised. Decorin or 359 295 360 296 local abattoir). Buccal plates were removed from both biglycan core protein fractions were identified by immuno- 361 297 maxillae and mandibles using a dental drill, molar and reactivity with the polyclonal antibody rabbit anti-mouse 362 298 premolar teeth were extracted and bone from the immediate decorin (LF-113; raised against C-terminus of decorin) or 363 299 areas around the sockets was removed using a surgical chisel. rabbit anti-mouse biglycan (LF-106; raised against 364 300 Proteoglycans were isolated from alveolar bone according to C-terminus of biglycan) (Fisher et al., 1995) using the above 365 301 the method of Waddington and Embery (1991). Briefly, soft described dot-blot analysis. 366 302 adherent tissue was removed from the bone pieces with 367 303 368 2.4. SDS PAGE and western blot analysis 304 collagenase dispase, dehydrated in ethanol washes, followed 369 305 by diethyl ether to remove lipids and then left at room 370 306 temperature for the residual solvent to evaporate. Bone was Samples of intact proteoglycans and their respective core 371 307 powdered at 20 C, demineralised with 10% EDTA (triso- proteins were mixed with an equal volume of 0.125 M 372 308 dium salt) solution, pH 7.45 and non-collagenous material was TriseHCl, pH 6.8, 20% glycerol, 4% SDS, 10% 2-b-mer- 373 309 extracted using 4 M guanidinium chloride, 0.05 M sodium capthoethanol and 0.004% bromophenol blue and examined 374 310 acetate, pH 5.9 and containing above named protease inhibi- by SDS-PAGE using the Phastsystem (GE Healthcare) as 375 311 376 312 tors. Solubilised non-collagenous proteins were recovered by previously described by Waddington et al. (1993). SDS-6H 377 313 centrifugation at 1800 g, 15 min, dialysed against double- molecular weight markers (Sigma Aldrich) were included on 378 314 distilled water and lyophilised. Gingival tissue was washed each gel. Following separation, gels were stained using the 379 315 in phosphate-buffered saline (PBS) then sequentially washed Silver Stain Kit (GE Healthcare). For western blot analysis, 380 316 twice in ethanol, followed by diethyl ether for 10 min and left separated gels were electroblotted onto a nitrocellulose 381 317 at room temperature for the solvent to evaporate. Non- membrane (HybondÔ C, Amersham Pharmacia Biotech) with 382 318 collagenous gingival tissue components were extracted into transfer buffer, 25 mM Tris HCl buffer in 20% methanol, 383 319 384 4 M guanidinium chloride extraction buffer and recovered as containing 192 mM glycine, pH 8.3. Membranes were blocked 320 Ô e 385 321 for treatment of alveolar bone tissue samples. with 10% non-fat milk (Marvel ), in TBS (10 mM tris HCl, 386 322 Non-collagenous protein extracts from gingival or alveolar 0.15 M sodium chloride, pH 7.4) for 30 min and then incu- 387 323 bone were dissolved in 7 M urea, 0.05 M sodium acetate bated with primary polyclonal antibody rabbit anti-mouse 388 324 buffer, pH 6.8, at a concentration of 10 mg/ml 1 ml samples decorin (LF-113) or rabbit anti-mouse biglycan (both diluted 389 325 were applied to a HiLoad 16/10 Q-Sepharose column 1:200 in 3% milk, TBS) for 1 h. Immunoreactivity was 390

4 J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12

391 visualised following incubation with goat anti-rabbit IgG on Fastidious Anaerobic Agar (FAA) (Lab MÔ International 456 392 conjugated to horse radish peroxidase (diluted 1:1000 in 3% Diagnostic Group plc, Bury, UK), supplemented with 5% v/v 457 393 458 394 milk, TBS) and detection using the ECL Plus western blotting defibrinated sheep blood (TCS Bioscience Ltd., Buckingham, 459 395 detection reagent kit (GE Healthcare). UK) and purity was confirmed regularly. Bacterial cells were 460 396 inoculated into 10 ml of brain heart infusion (BHI) broth 461 397 2.5. Identification of GAG constituents (Oxoid Ltd., Basingstoke, UK) and subsequently incubated 462 398 overnight in an anaerobic cabinet (10% v/v CO2, 20% v/v H2, 463 399 464 Samples of intact proteoglycans were digested with an 70% v/v N2), at 37 C (Don Whitley Scientific Ltd., Shipley, 400 equal volume of 10 mg/ml type XIV protease (Sigma Aldrich) UK) to the stationary growth phase (determined from growth 465 401 466 in 0.2 M triseHCl and 10 mM calcium chloride pH 7.5 and curves experiments). Bacterial cells were harvested by 402 467 403 incubated at 55 C for 20 h. Samples were subsequently centrifugation at 3000 g for 10 min and washed twice in 468 404 recovered by lyophilisation, reconstituted with ddH2O to the filtered, degassed HBS-EP buffer (0.01 M Hepes, 0.15 M 469 405 equivalent of the original starting volume of the PG sample NaCl, 0.003 mM EDTA, pH 7.4, containing 0.005% surfactant 470 406 and examined by cellulose acetate electrophoresis as previ- P20). The cells were sonicated (60 W, 10 s) to disrupt the 471 407 ously described by Waddington et al. (2004). GAGs were streptococcal chains and re-suspended in HBS-EP buffer to an 472 408 identified as judged by their electrophoretic mobility against OD ¼ 0.5 (approximately 1 108 cfu/ml, determined 473 409 (550 nm) 474 410 the standards hyaluronan (HA), heparin sulphate (HS), der- by serial dilutions, plated onto FAA plates and colony counts 475 411 matan sulphate (DS), chondroitin 4-sulphate (C4S) and following overnight incubation). 476 412 chondroitin 6 sulphate (C6S) (Sigma Aldrich) separated on the 477 413 same membrane. 2.7. Detection of GAG depolymerase activity 478 414 479 415 2.6. Bacterial source and preparation Depolymerisation of GAG was assessed using the method 480 416 of (Tipler and Embery, 1985). Briefly, 4 ml BHI broth (Gibco) 481 417 482 418 Isolates of SAG and sources are detailed in Table 1. The supplemented with either chondroitin-4-sulphate (Whale 483 419 isolates had previously been characterised and assigned to cartilage, type A, Sigma Aldrich) or hyaluronic acid (bovine 484 420 Streptococcus constellatus, S. anginosus or Streptococcus umbilical cord, grade 1, Sigma Aldrich), final concentration 485 421 intermedius using a phenotypic identification scheme (Bartie 400 mg/ml, was inoculated with 1 ml of bacterial suspension 486 422 et al., 2000; Whiley et al., 1990). Bacteria were subcultured (harvested at mid-log phase of growth) and incubated under 487 423 anaerobic conditions for 7 days. After incubation, samples 488 424 were centrifuged at 3000 g for 10 min, 20 ml of each super- 489 425 490 Table 1 426 natant was added to 4 ml of distilled water and 232 nm 491 Identity and source of SAG isolates used in this study. GAG depolymerase 427 absorbance (due to unsaturated bond formation following the 492 activity indicated where detected. 428 elimination reaction of HA and CS) recorded against non- 493 429 Species Reference Clinical source GAG depolymerase inoculated broth (negative controls). 494 430 activity detectable 495 431 S. constellatus 350/96 Dento-alveolar 2.8. Surface plasmon resonance analysis 496 432 abscess 497 433 S. constellatus F436 Lung abscess 498 434 S. constellatus 322/95 Dento-alveolar Positive for chondroitinase The interactions between mixed SLRP preparations from 499 435 abscess and hyaluronidase activity ovine tissues and recombinant decorin and biglycan with the 500 436 S. constellatus 48C Tongue SAG isolates were investigated by real-time biomolecular 501 437 S. constellatus 34C Tongue Positive for chondroitinase interaction analysis (BIA) using surface plasmon resonance 502 438 and hyaluronidase activity (SPR) technology on a BIAcoreÒ 3000 system (BIAcore, 503 439 S. intermedius HW13 Dento-alveolar 504 abscess Uppsala, Sweden) using a C1 sensorchip (BIAcore). The C1 440 505 S. intermedius 127/95 Dento-alveolar Positive for chondroitinase chip has a carboxymethylated surface with no dextran matrix 441 506 abscess and hyaluronidase activity 442 and is useful when studying interactions where the analyte is 507 S. intermedius 447/96 Dento-alveolar 443 large, such as bacteria. Recombinant proteoglycans or tissue- 508 abscess 444 extracted proteoglycans (ligand) were dissolved in 10 mM 509 S. intermedius 11C Tongue m 445 S. intermedius 30C Tongue Positive for chondroitinase phosphate buffer, pH 7.1 (150 g/ml; Biacore), whilst 510 446 and hyaluronidase activity deglycosylated protein cores were dissolved in 10 mM 511 447 S. intermedius 84C Plaque sodium acetate buffer (150 mg/ml; BIAcore). At a flow rate of 512 448 S. anginosus 43586/96 High vaginal 10 ml/min, the docked chip was first treated with BIA nor- 513 449 514 swab malisation solution (70% w/w glycerol in water) (BIAcore) 450 S. anginosus 39/2/14A Unknown 515 451 S. anginosus 670/95 Dento-alveolar to create a standardised total reflection and then 516 452 abscess equilibrated with HBS-EP buffer. Proteoglycan ligands were 517 453 S. anginosus 16C Plaque immobilised on the sensor chip surface of flow cell 2 using 518 454 S. anginosus 19C Plaque amino coupling by first treating the surface (to produce 519 455 S. anginosus 43C Tongue active ester groups) by injecting 70 ml of 400 mM 1-ethyl-3- 520

J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12 5

521 (3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC)/ 3. Results 586 522 70 ml 100 mM N-hydroxysuccinimide (NHS) solutions. After 587 523 588 m m 3.1. Characterisation of proteoglycan ligands 524 activation, 35 l of 150 g/ml of ligand in appropriate buffer 589 525 was applied to the surface. The system was washed through 590 526 with HBS-EP buffer to remove any loosely associated ligand. Recombinant proteoglycans were synthesised by trans- 591 527 Finally, residual NHS esters were inactivated with 70 mlof fected HeLa cells and purified from the culture media by 592 528 1 M ethanolamine hydrochloride, pH8.5. Flow cell 1 was used nickel affinity chromatography. Previous studies demonstrated 593 529 594 as an in-line reference for flow cell 2, where the blank surface that recombinant proteoglycans copurify with a2HS glyco- 530 was exposed to amine coupling in the absence of ligand. protein which is present within the serum of the culture media 595 531 596 Thirty-five ml of fibronectin 10 mM sodium acetate buffer (Sugars et al., 2002). Recombinant decorin and biglycan were 532 m 597 533 (150 g/ml; Biacore) coupled to the surface served as a posi- therefore further purified by anion exchange chromatography 598 534 tive control. Bacterial cells within the HBS-EP buffer, pH 7.4 using a resource Q column. A substantial amount of a2HS 599 535 were injected over the various ligand-prepared surfaces at glycoprotein was observed to elute from the column prior to 600 536 20 ml/min for 125 s at 25 C, followed by a 5 min dissociation application of the sodium chloride gradient and recombinant 601 537 phase (HBS-EP buffer only). The bacteria/ligand interaction decorin or biglycan eluted at 0.3e0.35 M sodium chloride. 602 538 produces an increase in the SPR signal, measured as The elution profiles were consistent with those previously 603 539 604 540 a response unit (RU). Residual bound bacteria was washed off published as part of detailed descriptions for the purification 605 541 with 100 mM NaOH, pH 9.5 for 1 min and the chip surface re- protocols of these proteoglycans (Sugars et al., 2002). Silver- 606 542 equilibrated with HBS-EP buffer for 5 min between the stained SDS-PAGE gels of the purified recombinant proteo- 607 543 application of bacterial test isolates. Each isolate was analysed glycan are shown in Fig. 1A. For recombinant biglycan, 608 544 in triplicate in a randomised order. protein bands were prominent at approximately 98, 66 and 609 545 45 kDa, whilst for recombinant decorin, bands were prominent 610 546 2.9. Statistical analysis at 66 and 45 kDa. Digestion of either recombinant decorin or 611 547 612 548 biglycan with chondroitinase ABC to remove the GAG chains 613 549 The RU data obtained was assessed by analysis of variance yielded a single band at 45 kDa, consistent with the known 614 550 (ANOVA) using SPSS 12. The data was evaluated both in molecular weight for the respective core proteins. Western blot 615 551 terms of differences between strains for a specific substrate analysis confirmed the immunoreactivity of the 45 kDa band 616 552 and differences for substrates within a particular strain. with antibodies against decorin or biglycan, respectively 617 553 Differences were considered significant at p < 0.05 and data (Fig. 1B). Both recombinant biglycan and decorin were 618 554 expressed as the mean value of triplicate assays standard identified as C4S-rich, as demonstrated by their immunore- 619 555 620 556 deviation. activity with monoclonal antibody CS56 (Fig. 1C). Previous 621 557 622 558 623 559 624 560 625 561 626 562 627 563 628 564 629 565 630 566 631 567 632 568 633 569 634 570 635 571 636 572 637 573 638 574 639 575 640 576 641 577 642 578 643 579 644 580 645 581 646 582 647 583 648 Fig. 1. Characterisation of recombinant proteoglycans, decorin and biglycan. Proteoglycans were synthesised within a mammalian expression system and purified 584 by anion exchange chromatography. Core proteins relating to either decorin or biglycan were detected by SDS PAGE (A) or western blot analysis (B). The presence 649 585 of chondroitin sulphate chains was confirmed by immunoreactivity to the monoclonal antibody CS56. DCN: decorin; BGN: biglycan. 650

6 J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12

651 analysis had indicated that recombinant decorin and biglycan by protease digestion, the GAG chains were analysed by 716 652 are synthesised with C4S chains, representing 76% of the total cellulose acetate. Electrophoretic mobility of the released 717 653 718 654 GAG content. In addition, chondroitin sulphate chains with GAGs compared to commercial GAG standards confirmed that 719 655 reduced sulphation were also identified (approx 24% total decorin and biglycan isolated from alveolar bone were 720 656 GAG) (Sugars et al., 2002). substituted by C4S. Analysis of gingival proteoglycans on 721 657 Non-collagenous proteins isolated from ovine gingiva or silver-stained SDS PAGE gels identified two strong bands at 722 658 alveolar bone were fractionated using a two-stage anion approximately 45 kDa. In addition a sharp band was observed 723 659 exchange chromatography procedure. 280 nm absorbance at approximately 50 kDa and protein staining materials were 724 660 profiles were similar to those previous reported for isolation of apparent at the top of the gel (molecular weight 220 kDa), 725 661 726 662 proteoglycans from connective tissues (Waddington et al., which were lost following removal of the GAG with chon- 727 663 1993, 2003) and have proven successful in purifying proteo- droitinase ABC (Fig. 2A). Western blot analysis again 728 664 glycans from other glycoprotein matrix components. For confirmed the 45 kDa band to be core proteins decorin and 729 665 extracts from gingival and alveolar bone, a broad biglycan (Fig. 2B). Cellulose acetate electrophoresis of the 730 666 proteoglycan-rich fraction eluted from Q Sepharose with GAG chains (core protein digested with protease) indicated 731 667 sodium chloride concentrations within the range of that these proteoglycans were substituted by C4S and DS. 732 668 0.3e0.4 M. These proteoglycan-rich fractions were further 733 669 734 670 purified with Resource Q, eluting with 0.35 M sodium chlo- 3.2. Interaction of SAG with proteoglycans and 735 671 ride. Silver-stained SDS PAGE analysis for alveolar bone fibronectin 736 672 proteoglycan indicated a protein band at approximately 737 673 45 kDa with strong staining for protein material also identified Fig. 3 shows typical sensograms obtained following injec- 738 674 within the molecular weight range of 220e55 kDa, repre- tion of various SAG isolates over immobilised gingival 739 675 sentative of proteoglycan substituted by a GAG chain of proteoglycans (Fig. 3A) or fibronectin (Fig. 3B). For all 740 676 heterogeneous length (Fig. 2A). This was confirmed by the ligands, sensograms recorded specific interactions of the SAG 741 677 742 678 appearance of a single 45 kDa band after digestion of the GAG isolates with the proteoglycan ligand. No interaction events 743 679 chains with chondroitinase ABC. Western blot analysis were reported for the flow of SAG isolates over the control cell 744 680 confirmed this 45 kDa band to be the core proteins of decorin containing no ligand, indicating that, in these conditions, the 745 681 and biglycan (Fig. 2B). Following removal of the protein core bacteria were unable to bind to the surface of the chip and did 746 682 747 683 748 684 749 685 750 686 751 687 752 688 753 689 754 690 755 691 756 692 757 693 758 694 759 695 760 696 761 697 762 698 763 699 764 700 765 701 766 702 767 703 768 704 769 705 770 706 771 707 772 708 773 709 774 710 775 711 776 712 777 Fig. 2. Characterisation of the proteoglycans decorin and biglycan from alveolar bone and gingival connective tissues. Non-collagenous proteins were extracted 713 778 from demineralised tissues with 4 M guanidinium chloride and proteoglycans were purified by anion exchange chromatography. Proteoglycan-rich fractions 714 containing both decorin and biglycan were characterised by SDS PAGE (A) and western blot analysis. The GAG component conjugated to the proteoglycans was 779 715 identified by cellulose acetate electrophoresis (C). DCN: decorin; BGN: biglycan. 780

J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12 7

781 846 782 847 783 848 784 849 785 850 786 851 787 852 788 853 789 854 790 855 791 856 792 857 793 858 794 859 795 860 796 861 797 862 798 863 799 864 800 865 801 866 802 867 803 868 804 869 805 870 806 871 807 872 808 873 809 874 810 875 811 876 812 Fig. 3. Typical SPR sensograms obtained following injection of various SAG isolates (S. constellatus, F436, 322/95, 48C and S. anginosus 39/2/14A and 19C 877 813 shown) over immobilised gingival proteoglycans (A) or fibronectin (B). Proteoglycan was chemically immobilised to a C1 sensorchip (BIAcore) and SAG isolates 878 814 (1 108 cfu/ml) were injected over the immobilised proteoglycans and bacterial/ligand interaction measured as a response unit (RU). Fibronectin provided 879 815 a positive control. Adherence was taken as the entire process of attachment and accumulation of bacteria in the formation of biofilm on the surface and measured as 880 816 the difference in RU value from before injection of cells to the end of injection. 881 817 882 818 883 819 not form large aggregates which could be recorded by surface proteoglycan ligands. All strains demonstrated the ability to 884 820 resonance technology. Sensograms for bacterial/ligand inter- adhere to fibronectin, which was included as a positive control. 885 821 actions produced fluctuating profiles representing a series of For the S. anginosus commensal strain, 19C, 16C, 43C, S. 886 822 random association and dissociation events. This is likely to be constellatus commensal strain 34C, pathogenic strain F436 887 823 888 S. intermedius 824 due to the SAG isolate forming electrostatic interactions with and pathogenic strains 127/95, HW13 and 447/ 889 825 the proteoglycan ligand, which are readily broken by the shear 95, adherence to fibronectin was significantly higher compared 890 826 force generated by the eluting buffer flowing over the surface. to adherence to the respective proteoglycan ligands ( p < 0.01 891 827 This prevented calculation of affinity and dissociation for all comparisons). Gingival proteoglycans showed adher- 892 828 constants. In this study, the term adherence was therefore ence for all SAG strains, with significantly greater adherence 893 829 taken as the entire process of attachment and accumulation of to gingival proteoglycans compared with other proteoglycans 894 830 bacteria in formation of a biofilm on the surface, considering noted for S. anginosus 43C, 43586/96, 670/95, 39/2/14A, S. 895 831 896 both initial cell-to-surface adherence and subsequent cell-to- constellatus, 48C, F436, 322/95and S. intermedius, 84C447/95 832 < 897 833 cell adherence of bacterial cells (Eifukukoreeda et al., 1991). ( p 0.01 for all comparisons). Adherence of S. anginosus 39/ 898 834 Adherence was observed as the difference in RU value from 2/14A and S. constellatus 48C to gingival proteoglycans was 899 835 immediately before injection of cells (binding response at significantly higher compared to adherence to fibronectin 900 836 baseline) to the end of injection. Of note, the majority of ( p < 0.05 for both comparisons). 59% percent of strains, 901 837 strains that bound to the ligand surface to form a biolayer did which included both commensal and pathogenic isolates, 902 838 not dissociate immediately from the surface following the end demonstrated some ability to adhere to alveolar proteoglycans, 903 839 904 840 of injection, but required the application of 100 mM sodium 70% of strains bound to intact recombinant decorin and 76% 905 841 hydroxide to remove bound bacteria and return the RU to bound to recombinant biglycan. Almost all members of the S. 906 842 baseline. intermedius group were able to adhere to all proteoglycan 907 843 Mean RU values standard deviations from triplicate ligands examined, with all commensal strains showing strong 908 844 assays are presented in Table 2 and graphically in Fig. 4, adherence to recombinant decorin and biglycan and alveolar 909 845 illustrating trends in adherence patterns to the respective bone and gingival proteoglycans ( p < 0.01 when compared 910

8 J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12

911 Table 2 976 912 Mean response units (RUs) for strains of S. anginosus, S. constellatus and S. intermedius to recombinant and tissue-extracted proteoglycans, their respective core 977 913 proteins and fibronectin. Means determined from three independent assays (SD). Response units were determined as the increase in the SPR signal after 125 s of 978 914 injection of bacterial isolates compared with the SPR signal at start of injection. 979 915 Strain Recombinant decorin Recombinant biglycan Gingiva Alveolar bone Fibronectin 980 916 981 917 Intact Core Intact Core Intact Core protein Intact Core 982 918 protein protein protein 983 919 S. anginosus 984 920 19C NBD NBD NBD NBD 41.7 25.1 NBD 32.9 20.2 NBD 289 23.4 985 921 16C NBD NBD NBD NBD 95.8 45.8 NBD 4.7 1.4 NBD 2170.2 245.3 986 922 43C 9.2 4.9 NBD 9.6 2.9 NBD 39.9 3.4 NBD NBD NBD 32.8 15.2 987 923 43586/96 16.9 1.2 NBD 4.6 1.5 NBD 145.7 59.7 NBD NBD NBD 56.2 5.7 988 924 670/95 NBD NBD 9.0 5.5 NBD 206.2 79.0 NBD NBD NBD 45.8 26.3 989 925 39/2/14A NBD NBD NBD NBD 60.5 23.1 NBD 27.9 1.9 NBD 32.9 12.4 990 926 S. constellatus 991 48C 11.9 3.9 NBD NBD NBD 119.5 66.1 NBD NBD NBD 104 26.3 927 992 34C 38.4 21.2 NBD 28.5 11.8 NBD 54.8 25.8 NBD NBD NBD 82.6 5.2 928 993 350/96 19.4 5.8 NBD 40.6 13.7 NBD 204.4 135.9 NBD 26.6 13.4 NBD 75.1 14.6 929 994 F436 4.2 1.1 NBD 7.7 4.9 NBD 53.2 23.3 NBD 11.2 3.1 NBD 60.4 13.1 930 995 322/95 775.6 90.1 NBD 72.4 24.9 NBD 439.5 148.7 NBD 16.9 5.1 NBD 283 46.1 931 S. intermedius 996 932 11C 134.4 67.1 NBD 102.9 37.7 NBD 139.8 72.9 NBD 31.9 15.9 NBD 109.5 17.2 997 933 30C 136 17.2 NBD 1378.1 284.7 NBD 1532.2 114.0 74.3 58.9 112.4 50.1 NBD 1270.8 115 998 934 84C 130.5 40.2 NBD 266 114.7 NBD 1133.7 115.7 NBD 155 8.1* NBD 158.9 23.5 999 935 127/95 168.1 23.8 NBD 219.5 91.5 NBD 258.1 66.0 42.4 3.9 NBD NBD 452 48.2 1000 936 HW13 23.6 14.9 NBD 29.8 17.3 NBD 147.3 63.3 25.0 5 64.0 13.1 NBD 251 32.1 1001 937 447/95 NBD NBD 19.1 8.9 NBD 184.6 44.6 NBD NBD NBD 180.5 23.1 1002 938 1003 *p < 0.05; Shaded boxes represent isolates recovered from healthy sites; Clear boxes represent isolates recovered from clinically infected sites; NBD ¼ No binding 939 detected. 1004 940 1005 941 1006 942 with other strains binding to the same ligand). Pathogenic in the initiation of infection. Members of the SAG have 1007 943 strains of S. constellatus showed higher adherence to alveolar previously been shown to adhere to the matrix components 1008 944 bone and gingival proteoglycans compared with the laminin, fibronectin and fibrinogen and weakly to collagens 1009 945 1010 946 commensal stains within the same subgroup. Generally, type I and IV (Allen et al., 2002; Willcox and Knox, 1990). In 1011 947 bacterial strains appeared to demonstrate no preferential this study, we demonstrated the additional ability of the 1012 948 binding to either recombinant decorin or biglycan ( p > 0.05). extracellular matrix proteoglycans decorin and biglycan to 1013 949 Exceptions were seen for S. constellatus 322/95, which bind SAG, initiating the formation of microbial biolayers. 1014 950 showed significant preferential adherence to recombinant Significantly, the results demonstrate that the interaction is 1015 951 decorin ( p < 0.001), while S. intermedius 30C showed mediated almost entirely by the GAG chains conjugated to the 1016 952 significant preferential adherence to recombinant biglycan respective core protein. 1017 953 1018 p < Ò 954 ( 0.001). Removal of the GAG chains from the respective SPR technology using the Biacore 3000 system enabled 1019 955 proteoglycans abolished or significantly reduced bacterial the contribution of specific extracellular matrix components in 1020 956 adherence to the ligand ( p < 0.001for all comparisons). the early events of biofilm formation in real time. The 1021 957 proteoglycan ligands represented either decorin or biglycan or 1022 958 3.3. GAG depolymerisation activity a mixture of the two proteoglycans, and were confirmed to 1023 959 have been purified from other matrix components such as 1024 960 S. constellatus strains 322/95 and 34C and S. intermedius serum proteins and fibronectin. The C1 chip used in this study 1025 961 1026 962 strains 127/95 and 30C were the only strains demonstrated to was devoid of dextran which, as a polysaccharide glucan, 1027 963 possess high chondroitinase and hyaluronidase activity, as could have produced additional potential binding sites. 1028 964 demonstrated by an increase in 232 nm absorbance of 0.5 units Random covalent coupling was utilised to immobilise the 1029 965 or greater compared with no bacterial control (Table 1). proteoglycans at the surface of the sensor chip and to ensure 1030 966 the display of an ensemble of potential multivariant ligand 1031 967 4. Discussion orientations on the proteoglycan. This solid phase ECM 1032 968 adherence assay more closely mimics in vivo conditions, 1033 969 1034 970 An increasing number of microbes have been shown to where matrix components are tethered. Sensograms obtained 1035 971 depend upon extracellular matrix components for adhesion to during the 250 s period indicated a fluctuating SPR signal. 1036 972 host cells and tissues (Menozzi et al., 2002; Rostand and Esko, This could be considered comparable to the in vivo formation 1037 973 1997; Wadstrom and Ljungh, 1999). Moreover, adhesion of of biofilms, where models describe the initial reversible 1038 974 microbes to host components has been proposed as the critical attraction of passively transported bacteria to the protein 1039 975 initial step in the establishment of bacteria as commensals or surface by long-range weak physicochemical forces, such as 1040

J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12 9

1041 1106 1042 1107 1043 1108 1044 1109 1045 1110 1046 1111 1047 1112 1048 1113 1049 1114 1050 1115 1051 1116 1052 1117 1053 1118 1054 1119 1055 1120 1056 1121 1057 1122 1058 1123 1059 1124 1060 1125 1061 1126 1062 1127 1063 1128 1064 1129 1065 1130 1066 1131 1067 1132 1068 1133 1069 1134 1070 1135 1071 1136 1072 1137 1073 1138 1074 1139 1075 1140 1076 1141 1077 1142 1078 1143 1079 1144 1080 1145 1081 1146 1082 1147 1083 1148 1084 1149 1085 1150 1086 1151 1087 Fig. 4. Graphical comparison of adherence of different SAG isolates to the respective decorin and biglycan preparation. SAG isolates were analysed in triplicate 1152 1088 and injected over the immobilised proteoglycan in a random order. Means and SD were determined. Whilst this method does not provide highly accurate values, it 1153 1089 does allow identification of trends and gross changes in adherence patterns. 1154 1090 1155 1091 1156 1092 the van der Waals forces (Marsh and Martin, 2009). These pathogenic isolates or vice-versa. As a positive control, all 1157 1093 attractions can be broken by net repulsive forces between the strains bound to fibronectin. 1158 1094 bacteria and the protein surface and by shear flow of liquid In this study, all SAG isolates were observed to adhere to 1159 1095 across the surface, which is mimicked within the Biacore 3000 gingival proteoglycans. For S. intermedius and S. constellatus 1160 1096 system. Attachment subsequently becomes less reversible by strains, adherence was observed to be relatively strong and for 1161 1097 stereochemical interactions forming between adhesins on the isolates 670/95, 48C, 305/96, 322/95, 11C, 30C and 84C, 1162 1098 bacteria and proteins within the extracellular matrix. Gener- adherence was statistically significantly greater than adherence 1163 1099 1164 S. intermedius 1100 ally, exhibited moderate to strong adherence to to fibronectin. It is recognised that SAG may interact with 1165 1101 most proteoglycan ligands, whilst the majority of S. anginosus a number of components within the extracellular matrix, such 1166 1102 strains exhibited low adherence of less than 50 RU, particu- as laminin, fibronectin and fibrinogen (Allen et al., 2002; 1167 1103 larly to the C4S-rich proteoglycans of recombinant decorin Willcox and Knox, 1990). However, the strong interactions 1168 1104 and biglycan and alveolar bone. There appeared to be no observed comparable to fibronectin provide support for 1169 1105 preferential adherence of commensal isolates compared to considering proteoglycans, particularly those in gingival 1170

10 J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12

1171 tissues, as important constituents in facilitating SAG biofilm binding proteins (Dbp A, DbpB, Bgp) have been described in 1236 1172 formation. Only 59% bound to alveolar proteoglycans and the spirochete Borrelia burgdorferi sensu lato group (Leong 1237 1173 1238 e S. 1174 70 75% bound to recombinant proteoglycans, with only et al., 1998). The decorin binding site of DbpA has been 1239 1175 intermedius 30C showing significantly greater adherence to mapped to a conserved peptide motif (EAKVRA), with 1240 1176 recombinant biglycan compared to fibronectin ( p < 0.001). binding mediated by the GAG chain, particularly DS (Salo 1241 1177 Characterisation of the GAG chain demonstrated that a high et al., 2011). In those studies, DbpA showed considerably 1242 1178 proportion of ginigival proteoglycans were substituted with stronger binding under the flow conditions provided by SPR 1243 1179 DS, whilst other proteoglycan ligands carried the C4S GAG analysis, compared with stationary incubation adhesion 1244 1180 chain. Removal of the DS or C4S chain with chondroitinase assays. The authors hypothesise that the DbpA adhesion acts 1245 1181 1246 1182 ABC abolished SAG adherence for the majority of isolates as a “catch bond” and sheer force is required to induce 1247 1183 (adherence reduced to very low levels for three S. intermedius stronger interaction with the decorin (Salo et al., 2011). 1248 1184 isolates). This suggests that the GAG chain plays a major role The interaction between bacteria and the extracellular 1249 1185 in adherence of bacterial isolates, and DS chains facilitate matrix is proposed to promote bacterial colonisation, facilitate 1250 1186 stronger binding compared with C4S chains. C4S and DS are bacterial invasion into deeper tissues and facilitate systemic 1251 1187 both composed of a linear polysaccharide assembled as dissemination. The adherence to extracellular matrix compo- 1252 1188 disaccharide repeat units containing N-acetyl-galactosamine nents may also be important in the pathogenesis of infection, 1253 1189 1254 and hexuronic acid that are linked together by b-glycosidic although the precise involvement of extracellular matrix 1190 b b 1255 1191 ( 1,4 or 1,3) linkages. DS is structurally related to CS; components such as decorin and biglycan is unclear. Binding 1256 1192 however, a high proportion of uronic acid of DS is epimerised of Group B streptococcal alpha C protein to host cells has been 1257 1193 to iduronic acid, whilst CS contains mostly glucuronic acid. proposed to involve multiple binding to cell surface heparin 1258 1194 Casu et al. (1988) reported that epimerisation of uronic acid to sulphate proteoglycans, and the competitive inhibition of 1259 1195 iduronic acid provides additional degrees of rotation within the alpha C protein with exogenous GAGs promotes host cell 1260 1196 GAG chain. This additional flexibility is proposed to enable survival and lowers the bacterial burden (Baron et al., 2009). 1261 1197 1262 1198 the negatively charged sulphate and carboxyl groups on the Similar in vitro studies have demonstrated that cell surface 1263 1199 chain an easier search for sites on basic receptors on inter- GAGs facilitate the interaction of Streptococcus pneumo- 1264 1200 acting molecules, thereby improving their adherence potential. coccus (Tonnaer et al., 2006), S. uberis (Almeida et al., 1999a) 1265 1201 Of note, iduronic acid is also present in significant amounts in and Streptococcus pyrogenes (Frick et al., 2003) with host 1266 1202 heparin and heparin sulphate, which have also been reported to cells, which is interrupted by removal of the GAG by chon- 1267 1203 bind readily to streptococci isolates (Almeida et al., 1999a; droitinases, GAG synthase inhibitors and exogenous GAGs. 1268 1204 Frick et al., 2003; Srinoulprasert et al., 2006; Tonnaer et al., This may suggest that other GAGs within the extracellular 1269 1205 1270 1206 2006). Within this study, the interaction of SAG with the matrix act as inhibitors of bacterial/host cell interactions. 1271 1207 proteoglycan ligand was assessed at a physiological pH of 7.4. However, in providing a novel mechanistic viewpoint, studies 1272 1208 Increases or decreases in pH would lead to respectively have also suggested that GAGs may act as molecular bridges 1273 1209 increased ionisation or increased protonation of dissociable in directing binding to other matrix components such as 1274 1210 groups, on both the GAGs and the bacterial cell surface vitronectin (Duensing et al., 1999). Most notably, pre- 1275 1211 adhesins, the level of which depended upon the dissociation incubation of S. uberis with GAGs, particularly the iduronic 1276 1212 constant of the functional groups involved in facilitating the acid-rich GAGs, enhanced beta-casein-mediated adherence to 1277 1213 1278 1214 interaction. During inflammation, the pH of the gingival and internalisation in mammary epithelial cells (Almeida 1279 1215 pocket can vary from pH 6.5 to 8.5 (Leblebicioglu et al., et al., 2003). In relation to the study reported herein, it is 1280 1216 1996), which could potentially influence GAG e bacterial unclear at present whether binding of bacteria to the GAG 1281 1217 adherence, although in the absence of any observable extremes chains of decorin or biglycan as an extracellular matrix protein 1282 1218 in pH changes in inflamed tissues, these effects are likely to be would inhibit subsequent cell binding or facilitate it. Of note, 1283 1219 small. selected isolates of S. constellatus and S. intermedius strains 1284 1220 The nature of the bacterial adhesins on the SAG that (322/95, 34C, 127/95, 30C) possessed high chondroitinase 1285 1221 1286 S. 1222 facilitate binding to proteoglycans is unknown. Numerous activity, which was absent from the more poorly-adhering 1287 1223 matrix binding adhesins have been described in streptococcal anginosus isolates which might be important in release of the 1288 1224 species and, in general, fall into two groups (Lofling et al, GAG chain from the parent molecule and subsequent inter- 1289 1225Q2 2010); the MSCRAMMs (microbial surface cell recognition actions, be it as an enhancer or an inhibitor. 1290 1226 adhesion matrix molecules) and a class of adhesins which have In conclusion, this study has demonstrated the ability of 1291 1227 been reported to preferentially recognize sialylated glyco- SAGs, particularly S. constellatus and S. intermedius,to 1292 1228 conjugates. Within the MSCRAMMs, M protein and protein F adhere to the GAG moiety of the extracellular matrix 1293 1229 1294 S. pyogenes 1230 on the surface of have been shown to adhere to proteoglycans decorin and biglycan. Adherence was greater 1295 1231 DS, heparin sulphate and heparin (Frick et al., 2003), the alpha for the DS proteoglycans of the gingival proteoglycan. This 1296 1232 C protein adheres to heparin sulphate (Wang et al., 2011) and may be of significance in formation of dental abscesses, since 1297 1233 the FOG (fibrinogen-binding protein) binds DS and heparin viridians-group streptococci are frequently encountered deep 1298 1234 sulphate in group G streptococci (Egesten et al., 2011). inside the gingival connective tissue and associated peri- 1299 1235 Outside of the streptococcus group of bacteria, decorin odontal ligament, which contains similar DS proteoglycans. 1300

J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12 11

1301 The ability of isolates of S. constellatus and S. intermedius to Han, J.K., Kerschner, J.E., 2001. Streptococcus milleri: an organism for head 1366 1302 adhere to C4S proteoglycans of alveolar bone may enhance and neck infections and abscess. Arch. Otolaryngol. Head Neck Surg. 127, 1367 1303 e 1368 their survival and potential virulence in the periapical region 650 654. 1304 Hoepelman, A.I., Tuomanen, E.I., 1992. Consequences of microbial attach- 1369 1305 of a non-vital tooth, leading to infection and abscess forma- ment: directing host cell functions with adhesins. Infect. Immun. 60, 1370 1306 tion. As with other bacterial ECM models, interactions formed 1729e1733. 1371 1307 with the extracellular matrix may be significant in influencing Iozzo, R.V., 1999. The biology of the small leucine-rich proteoglycans. 1372 1308 the metabolic activity of the host tissue and the phenotype/ Functional network of interactive proteins. J. Biol. Chem. 274, 1373 e 1309 genotype of the bacterial cell, thus determining whether or not 18843 18846. 1374 Jacobs, J.A., Schot, C.S., Schouls, L.M., 2000. Haemolytic activity of the 1310 these “commensal” organisms initiate an infective process. 1375 1311 Streptococcus milleri group’ and relationship between haemolysis 1376 1312 restricted to human red blood cells and pathogenicity in S. intermedius. 1377 J. Med. Microbiol. 49, 55e62. 1313 1378 Acknowledgements Jacobs, J.A., Tjhie, J.H., Smeets, M.G., Schot, C.S., Schouls, L.M., 2003. 1314 1379 Genotyping by amplified fragment length polymorphism analysis reveals 1315 1380 persistence and recurrence of infection with Streptococcus anginosus 1316 This study was supported by the Wales Office for Research 1381 group organisms. J. Clin. Microbiol. 41, 2862e2866. 1317 and Development for Health and Social Care. Antibodies 1382 Jockusch, B.M., Bubeck, P., Giehl, K., Kroemker, M., Moschner, J., 1318 1383 used in this study were a generous gift of L. Fisher, NICDR, Rothkegel, M., et al., 1995. The molecular architecture of focal adhesions. 1319 NIH, USA. Annu. Rev. Cell. Dev. Biol. 11, 379e416. 1384 1320 1385 Kim, C.W., Goldberger, O.A., Gallo, R.L., Bernfield, M., 1994. Members of 1321 the syndecan family of heparan sulfate proteoglycans are expressed in 1386 1322 References distinct cell-, tissue-, and development-specific patterns. Mol. Biol. Cell. 5, 1387 1323 797e805. 1388 1324 Aderhold, L., Knothe, H., Frenkel, G., 1981. The bacteriology of dentogenous Kitada, K., Inoue, M., Kitano, M., 1997. Experimental endocarditis induction 1389 1325 pyogenic infections. Oral Surg. Oral Med. Oral Pathol. 52, 583e587. and platelet aggregation by Streptococcus anginosus, Streptococcus con- 1390 1326 Allen, B.L., Katz, B., Hook, M., 2002. Streptococcus anginosus adheres to stellatus and Streptococcus intermedius. FEMS Immunol. Med. Microbiol. 1391 1327 vascular endothelium basement membrane and purified extracellular 19, 25e32. 1392 1328 matrix proteins. Microb. Pathog. 32, 191e204. Leblebicioglu, B., Lim, J.S., Cario, A.C., Beck, F.M., Walters, J.D., 1996. pH 1393 1329 Almeida, R.A., Fang, W.H., Oliver, S.P., 1999a. Adherence and internalization changes observed in the inflamed ginigival crevice modulate human 1394 1330 of Streptococcus uberis to bovine mammary epithelial cells are mediated polymorphonuclear leukocyte activation in vitro. J. Periodontol. 67, 1395 1331 by host cell proteoglycans. FEMS Microbiol. Lett. 177, 313e317. 472e477. 1396 1332 Almeida, R.A., Luther, D.A., Oliver, S.P., 1999b. Incubation of Streptococcus Ledezma-Rasillo, G., Flores-Reyes, H., Gonzalez-Amaro, A.M., Garrocho- 1397 1333 uberis with extracellular matrix proteins enhances adherence to and Rangel, A., Ruiz-Rodriguez Mdel, S., Pozos-Guillen, A.J., 2010. Identi- 1398 1334 internalization into bovine mammary epithelial cells. FEMS Microbiol. fication of cultivable microorganisms from primary teeth with necrotic 1399 e 1335 Lett. 178, 81e85. pulps. J. Clin. Pediatr. Dent. 34, 329 333. 1400 1336 Almeida, R.A., Luther, D.A., Nair, R., Oliver, S.P., 2003. Binding of host Leong, J.M., Robbins, D., Rosenfeld, L., Lahiri, B., Parveen, N., 1998. 1401 Structural requirements for glycosaminoglycan recognition by the 1337 glycosaminoglycans and milk proteins: possible role in the pathogenesis of 1402 e Lyme disease spirochete, Borrelia burgdorferi. Infect. Immun. 66, 1338 Streptococcus uberis mastitis. Vet. Microbiol. 94, 131 141. 1403 6045e6048. 1339 Baron, M.J., Wong, S.L., Nybakken, K., Carey, V.J., Madoff, L.C., 2009. Host 1404 glycosaminoglycan confers susceptibility to bacterial infection in Lewis, M.A.O., MacFarlane, T.W., McGowan, D.A., 1986. Quantitative 1340 1405 Drosophila melanogaster. Infect. Immun. 77, 860e866. bacteriology of acute dento-alveolar abscess. J. Med. Microbiol. 21, 1341 1406 Bartie, K.L., Wilson, M.J., Williams, D.W., Lewis, M.A.O., 2000. Macro- 101e104. 1342 1407 restriction fingerprinting of "Streptococcus milleri" group bacteria by Marsh, P., Martin, M., 2009. In: Marsh, P., Martin, M. (Eds.), Dental Plaque in 1343 1408 pulsed-field gel electrophoresis. J. Clin. Microbiol. 38, 2141e2149. Oral Microbiology. Churchill Livingstone, Edinburgh, pp. 74e102. 1344 Casu, B., Petitou, M., Provasoli, M., Sinay, P., 1988. Conformational flexi- Menozzi, F.D., Pethe, K., Bifani, P., Soncin, F., Brennan, M.J., Locht, C., 1409 1345 bility - a new concept for explaining binding and biological properties of 2002. Enhanced bacterial virulence through exploitation of host glycos- 1410 1346 iduronic acid-containing glycosaminoglycans. Trends Biochem. Sci. 13, aminoglycans. Mol. Microbiol. 43, 1379e1386. 1411 1347 221e225. Nagashima, H., Takao, A., Maeda, N., 1999. Abscess forming ability of 1412 1348 Duensing, T.D., Wing, J.S., van Putten, J.P., 1999. Sulfated polysaccharide- streptococcus milleri group: synergistic effect with Fusobacterium nucle- 1413 1349 directed recruitment of mammalian host proteins: a novel strategy in atum. Microbiol. Immunol. 43, 207e216. 1414 1350 microbial pathogenesis. Infect. Immun. 67, 4463e4468. Okayama, H., Nagata, E., Ito, H.O., Oho, T., Inoue, M., 2005. Experimental 1415 1351 Egesten, A., Frick, I.M., Morgelin, M., Olin, A.I., Bjorck, L., 2011. Binding of abscess formation caused by human dental plaque. Microbiol. Immunol. 1416 1352 albumin promotes bacterial survival at the epithelial surface. J. Biol. 49, 399e405. 1417 1353 Chem. 286, 2469e2476. Rostand, K.S., Esko, J.D., 1997. Microbial adherence to and invasion through 1418 e 1354 Eifukukoreeda, H., Yakushiji, T., Kitada, K., Inoue, M., 1991. Adherence of proteoglycans. Infect. Immun. 65, 1 8. 1419 1355 oral streptococcus-milleri cells to surfaces in broth cultures. Infect. Salo, J., Loimaranta, V., Lahdenne, P., Viljanen, M.K., Hytonen, J., 2011. 1420 e 1356 Immun. 59, 4103 4109. Decorin binding by DbpA and B of Borrelia garinii, Borrelia afzelii, and 1421 Borrelia burgdorferi sensu Stricto. J. Infect. Dis. 204, 65e73. 1357 Fisher, L.W., Stubbs, J.T., Young, M.F., 1995. Antisera and cDNA probes to 1422 Schuman, N.J., Turner, J.E., 1999. The clinical significance of beta hemolytic 1358 human and certain animal model bone matrix non-collagenous proteins. 1423 e streptococci of the milleri group in oral abscesses. J. Clin. Pediatr. Dent. 1359 Acta Orthop. Scand. Suppl. 266, 61 65. 1424 Frick, I.M., Schmidtchen, A., Sjobring, U., 2003. Interactions between M 23, 137e142. 1360 1425 proteins of Streptococcus pyogenes and glycosaminoglycans promote Shain, H., Homer, K.A., Beighton, D., 1996. Degradation and utilisation of 1361 1426 bacterial adhesion to host cells. Eur. J. Biochem. 270, 2303e2311. chondroitin sulphate by Streptococcus intermedius. J. Med. Microbiol. 44, 1362 1427 Gossling, J., 1988. Occurrence and pathogenicity of the Streptococcus milleri 372e380. 1363 1428 group. Rev. Infect. Dis. 10, 257e285. Shinzato, T., Saito, A., 1994. A mechanism of pathogenicity of "Streptococcus 1364 Green, A.W., Flower, E.A., New, N.E., 2001. Mortality associated with milleri group" in pulmonary infection: synergy with an anaerobe. J. Med. 1429 1365 odontogenic infection. Br. Dent. J. 190, 529e530. Microbiol. 40, 118e123. 1430

12 J. Landrygan-Bakri et al. / Research in Microbiology xx (2012) 1e12

1431 Siqueira, J.F., Rocas, I.N., Moraes, S.R., Santos, K.R., 2002. Direct amplifi- dentine-pulp complex in vitro. Biochem. Biophys. Acta - Mol. Basis Dis. 1457 1432 cation of rRNA gene sequences for identification of selected oral patho- 1689, 142e151. 1458 1433 gens in root canal infections. Int. Endod. J. 35, 345e351. Wadstrom, T., Ljungh, A., 1999. Glycosaminoglycan-binding microbial 1459 1434 Srinoulprasert, Y., Kongtawelert, P., Chaiyaroj, S.C., 2006. Chondroitin sulfate proteins in tissue adhesion and invasion: key events in microbial patho- 1460 1435 B and heparin mediate adhesion of Penicillium marneffei conidia to host genicity. J. Med. Microbiol. 48, 223e233. 1461 1436 extracellular matrices. Microb. Pathog. 40, 126e132. Wang, F., Su, C., Hollfelder, K., Waddington, D., Pan, Y.C., 1993. Electro- 1462 1437 Sugars, R.V., Waddington, R.J., Embery, G., 2002. The interaction of blotting proteolytic products from native gel for direct N-terminal 1463 1438 recombinant decorin with a2HS-glycoproteindimplications for structural sequence analysis: an approach for studying protein-protein interaction. 1464 1439 and functional investigations. Protein Expr. Purif. 25, 180e188. Electrophoresis 14, 847e851. 1465 1440 Tipler, L.S., Embery, G., 1985. Glycosaminoglycan-depolymerizing enzymes Wang, Z., Flax, L.A., Kemp, M.M., Linhardt, R.J., Baron, M.J., 2011. Host and 1466 1441 produced by anaerobic bacteria isolated from the human mouth. Arch. Oral pathogen glycosaminoglycan-binding proteins modulate antimicrobial 1467 1442 Biol. 30, 391e396. peptide responses in Drosophila melanogaster. Infect. Immun. 79, 606e616. 1468 1443 Tonnaer, E.L., Hafmans, T.G., Van Kuppevelt, T.H., Sanders, E.A., Whiley, R.A., Fraser, H., Hardie, J.M., Beighton, D., 1990. Phenotypic 1469 1444 Verweij, P.E., Curfs, J.H., 2006. Involvement of glycosaminoglycans in the differentiation of Streptococcus intermedius, Streptococcus constellatus, 1470 1445 attachment of pneumococci to nasopharyngeal epithelial cells. Microbes and Streptococcus anginosus strains within the ‘Streptococcus milleri 1471 1446 Infect. 8, 316e322. group’. J. Clin. Microbiol. 28, 1497e1501. 1472 1447 Waddington, R.J., Embery, G., 1991. Structural characterization of human Whiley, R.A., Beighton, D., Winstanley, T.G., Fraser, H.Y., Hardie, J.M., 1992. 1473 1448 alveolar bone proteoglycans. Arch. Oral Biol. 36, 859e866. Streptococcus intermedius, Streptococcus constellatus, and Streptococcus 1474 1449 Waddington, R.J., Embery, G., Hall, R.C., 1993. The influence of fluoride on anginosus (the Streptococcus milleri group): association with different 1475 1450 proteoglycan structure using a rat odontoblast invitro system. Calcif. body sites and clinical infections. J. Clin. Microbiol. 30, 243e244. 1476 1451 Tissue Int. 52, 392e398. Willcox, M.D., 1995. Potential pathogenic properties of members of the 1477 1452 Waddington, R.J., Hall, R.C., Embery, G., Lloyd, D.M., 2003. Changing "Streptococcus milleri" group in relation to the production of endocarditis 1478 1453 profiles of proteoglycans in the transition of predentine to dentine. Matrix and abscesses. J. Med. Microbiol. 43, 405e410. 1479 1454 Biol. 22, 153e161. Willcox, M.D.P., Knox, K.W., 1990. Surface-associated properties of 1480 1455 Waddington, R.J., Moseley, R., Smith, A.J., Sloan, A.J., Embery, G., 2004. Streptococcus-milleri group strains and their potential relation to patho- 1481 1456 Fluoride-induced changes to proteoglycan structure synthesised within the genesis. J. Med. Microbiol. 31, 259e270. 1482