Biofilm-specific surface properties and proteinexpression in oral Streptococcus sanguis

Catherine Black, Iain Allan, Susannah K. Ford, Michael Wilson,Roderick McNab*

Department of Microbiology, Eastman Dental Institute, University College London,256 Gray’s Inn Road, London WC1X 8LD, UKAccepted 12 December 2003

Introduction

Members of the mitis group of streptococci (includ-ing Streptococcus gordonii, Streptococcus mitis,Streptococcus oralis and Streptococcus sanguis(sanguinis))1 are found at most sites in the human

Archives of Oral Biology (2004) 49, 295—304

KEYWORDS

Biofilm;

Streptococcus sanguis;

Protein expression;

Oral streptococcus

Summary Objective: Oral streptococci are primary colonisers of the tooth surface andare abundant in dental plaque biofilms. Bacteria growing in these relatively dense,surface-associated communities are phenotypically quite distinct from their plank-tonic counterparts. The purpose of the present study was to develop a method toinvestigate biofilm-specific surface protein expression by Streptococcus sanguis to helpprovide a better understanding of the critical events in plaque development. Design:Biofilm cells were grown on the surface of glass beads in a biofilm device fed withmucin-containing artificial saliva. Planktonic cells were grown in continuous culture atapproximately the same growth rate. Surface hydrophobicity of biofilm and planktoniccells was determined by hexadecane partitioning, and expression of streptococcalfibronectin adhesin CshA was determined in ELISA using specific antiserum. Antiseraraised to glutaraldehyde-fixed whole biofilm or planktonic grown cells were used toscreen an expression library of S. sanguis genomic DNA, and isolated clones weresequenced. Results: Phenotypic analysis of biofilm and planktonic cells confirmed thatmode of growth affected surface properties of S. sanguis. Thus, hydrophobicity andCshA expression was significantly elevated in biofilm cells. Library screening withbiofilm antiserum yielded 32 recombinant clones representing 21 different S. sanguisproteins involved in adhesion and colonisation, carbohydrate utilisation or bacterialmetabolism. In differential analysis of four selected Escherichia coli clones, biofilmantiserum reacted five times stronger than planktonic antiserum with cell-freeextracts of clones encoding homologues of CshA and Cna collagen adhesin of Staphy-lococcus aureus, suggesting that these surface proteins are up-regulated in biofilmcells. In contrast, both antisera reacted equally strongly with cell-free extracts of theremaining two clones (encoding dihydrofolate synthase and an unknown protein).Conclusions: The method described represents a useful means for determining bacter-ial protein expression in biofilms based on a combination of molecular and immuno-logical techniques. Surface expression of putative fibronectin and collagen adhesinswas up-regulated in biofilm cells.� 2004 Elsevier Ltd. All rights reserved.

*Corresponding author. Present address: GlaxoSmithKline Con-sumer Healthcare, St. George’s Avenue, Weybridge, SurreyKT13 0DE, UK. Tel.: þ44-1932-822257; fax: þ44-1932-822120.E-mail address: rod.x.mcnab@gsk.com (R. McNab).

0003–9969/$ — see front matter � 2004 Elsevier Ltd. All rights reserved.doi:10.1016/j.archoralbio.2003.12.001

oral cavity and are numerically abundant in dentalplaque.2 Colonisation and persistence of strepto-cocci in the oral cavity depends, at least in part, onthe abilities of these bacteria to adhere to oralsurfaces, including the salivary pellicle, other oralbacteria in plaque, and host epithelial cells, and adiverse array of colonisation-associated streptococ-cal cell surface and secreted proteins have beendescribed in some detail (reviewed in3,4). It is theability to adhere to oral surfaces so successfullythat affords these bacteria a critical role in theformation of dental plaque, from which diseasessuch as caries, gingivitis and periodontitis can arise.The mitis group streptococci are also major causa-tive agents of bacterial endocarditis,5,6 a poten-tially life-threatening condition characterised bybacterial colonisation of non-bacterial thromboticvegetations.

In the case of S. sanguis and closely relatedspecies S. gordonii, the wealth of knowledge con-cerning the function and expression of surface pro-teins implicated in adhesion and colonisation hasbeen gathered, on the whole, from analysis ofbatch-culture (planktonic) cells. However in bothoral and cardiovascular sites, these surface colonis-ing bacteria are present in a sessile form, commonlyreferred to as a biofilm. Bacteria in a biofilmencounter very different environmental conditionscompared to those in disseminated planktonic form.Consequently, it is unsurprising that bacteria inbiofilms are phenotypically different from theirplanktonic counterparts, and may demonstratealterations in cellular components such as proteins,fatty acids, and phospholipids associated with thecell envelope, and production of extracellularenzymes and polysaccharides.7 Such responses bythe constituent cells of biofilms may account forthe enhanced resistance to antimicrobial agentsfrequently encountered in biofilms (reviewed byWilson8), although this is has yet to be proven. Whathas now been established, however, is that geneexpression can differ markedly in biofilms comparedto planktonic culture. For example, Pseudomonasaeruginosa has been shown to up-regulate alginatesynthesis when adhering to a surface.9,10 In morerecent work, Schembri et al.11 reported that genesencoding proteins involved in adhesion and auto-aggregation, as well as putative genes encodingtransport proteins, oxidoreductases, and genesassociated with heavy metal resistance, were allup-regulated in biofilms of Escherichia coli.

Elucidating the profile of protein expressionin streptococcal biofilms will provide a betterunderstanding of the critical events in plaquedevelopment, since protein expression in plank-tonic culture is an unreliable representation of

the events in vivo. Such an understanding may helpboost efforts to control oral, and indeed, cardio-vascular diseases.

Since the bacterial cell surface is the main forumfor interactions between bacteria and with hostcomponents within the relatively compact andstructurally regulated environment of oral biofilms,in this study we have attempted to identify differ-entially expressed surface components of S. sanguis(with respect to growth mode) using antiserumraised to fixed whole cells of S. sanguis grown inbiofilm or planktonic conditions. Screening withbiofilm antiserum identified a total of 32 clones,representing up to 21 different proteins. Furtheranalysis of selected clones indicated up-regulationin biofilms of proteins with significant homology toCshA and Cna adhesins of S. gordonii and Staphy-lococcus aureus, respectively. In contrast, proteinswith limited homology to SdrD adhesin of S. aureusor to dihydrofolate synthetase of Streptococcuspneumoniae were equally reactive with biofilmand planktonic antisera.

Methods

Bacterial strains and media

S. sanguis NCTC10904 and S. gordonii DL1-Challiswere grown routinely at 37 8C on TSBY agar12 in acandle jar. Liquid cultures were grown withoutshaking in screw-cap tubes or bottles at 37 8C inTSBY medium or in TY-glucose medium.12 For pre-paration of planktonic or biofilm cells, S. sanguiswas grown in mucin-containing medium (MCM; 5 gproteose peptone (Oxoid), 2.5 g hog gastric mucin(Sigma), 2 g yeast extract (Oxoid), 1 g Lab Lemcopowder (Oxoid), 0.2 g NaCl, 0.35 g CaCl2, 0.2 g KCland 0.5 g urea l�1). E. coli strain XL1-Blue wasgrown aerobically at 37 8C in LB13 containing ampi-cillin at 100 mg ml�1 where required.

DNA manipulations

Routine molecular biology techniques were per-formed according to Sambrook et al.13 Chromoso-mal DNA was isolated from S. sanguis as describedpreviously.14 The generation of a partial Sau3AIlibrary of S. sanguis NCTC10904 DNA in lZAPII(Stratagene) and in vivo excision of phagemidpBluescript was performed according to the manu-facturer’s instructions. Phagemid DNA was iso-lated from E. coli using Qiaquick (Qiagen).Restriction and modifying enzymes (from NewEngland Biolabs) were used under the conditionsrecommended by the manufacturer. DNA sequencing

296 C. Black et al.

was performed with cloning vector-derived primers(T3 or T7 primers) in an Applied Biosystems model310A automated DNA analyser. Sequence data wereanalysed with DNASIS (Hitachi Software EngineeringCo.) and the BLAST algorithm15 was utilised for data-base searches.

Preparation of planktonic and biofilmS. sanguis cells

Biofilm cells of S. sanguis were grown in a glasspacked tube (GPT) device.16 The device consisted ofa 100 cm length of silicone rubber tubing (internaldiameter 5 mm) filled with 710—1180 mm-diameterglass beads (Sigma) and was inoculated with anovernight TSBY culture of S. sanguis. Following a5 h static attachment phase, MCM was pumpedthrough the device at a flow rate of 29 ml h�1. Asteady state was achieved after 4 days as deter-mined by plate counts of the GPT flow-througheffluent. Planktonic cells were prepared by growthin continuous culture at 37 8C in MCM. The flow rate(35 ml h�1 for a 250 ml culture) was adjusted to givea mean-specific growth rate similar to that of thebiofilm-grown cells (0.14 h�1). Steady-state cellsfrom each system were harvested and washed twicein phosphate buffered saline (PBS, 200 ml) in pre-paration for further analysis.

Determination of cell-surface properties

Cell-surface hydrophobicity was estimated as thepercentage of total cells partitioning with hexade-cane.14 The immunoreactivity of CshA on intactstreptococcal cells was determined by ELISA aspreviously described17 by using antiserum raisedto a truncated (260 kDa) fragment of S. gordoniiDL1-Challis CshA polypeptide.18 An internal controlwas included on all plates and comprised mutano-lysin-released polypeptides from an overnight cul-ture of S. gordonii DL1-Challis (diluted 1:800 inELISA coating buffer; 50 ml per well).

Antiserum

Antisera to glutaraldehyde-fixed cells were raisedin New Zealand male rabbits as described byJenkinson.19 Briefly, planktonic or biofilm cells(100 mg dry weight) were fixed with 0.25% (w/v)electron microscopy grade glutaraldehyde in PBS atroom temperature for 40 min. Cells were thendiluted to 10 ml with PBS, washed three times withPBS and suspended in PBS at approximately1011 cells ml�1. Portions (0.5 ml) were injectedintramuscularly on four separate occasions at inter-vals of 3 days. Then 20 days after the last of these

injections, the rabbits were boosted and bled aftera further 6 days.

Analysis of bacterial proteins

Cell-wall-associated polypeptides were extractedfrom biofilm or planktonic streptococcal cells fol-lowing mutanolysin treatment in spheroplastingbuffer.20 For analysis of protein expression byrecombinant E. coli clones, cells were grown tomid-exponential phase, IPTG was added to a finalconcentration of 1 mM and cultures were then incu-bated for a further 3 h. Cells from 1.5 ml of culturewere harvested by centrifugation (4000 � g, 10 min,4 8C), washed once with TE buffer, suspended in0.4 ml TE buffer containing 0.2 mg ml�1 lysozymeand incubated at 37 8C for 20 min. The cell suspen-sion was sonicated (10 s on ice at 280 W cm�2) usingan Ikasonic U50 hand-held sonicator fitted with a3 mm-diameter probe (IKA-Werke GMBH, Staufen,Germany) to disrupt cells, and debris was removedby centrifugation (10 000 � g, 10 min, 4 8C) and thesupernatant carefully removed. Protein concentra-tions were determined by using a Bio-Rad proteinassay kit with bovine serum albumin as the stan-dard.

Proteins were separated by sodium dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE) on 10% (w/v) acrylamide gels and werestained with Coomassie brilliant blue or withsilver nitrate21 or were transferred to Hybond-Cnitrocellulose membranes (Amersham) by electro-blotting.22 Nitrocellulose blots were incubated,antiserum diluted 1:500, and antibody bindingwas detected by using peroxidase-conjugated swineimmunoglobulins to rabbit IgG as described else-where.23

To determine antiserum reactivity of recombi-nant proteins expressed in E. coli, cell-free extractswere prepared as described above and diluted to20 mg ml�1 in ELISA coating buffer (15 mM Na2CO3,35 mM NaHCO3, pH 9.6), and 50 ml portions appliedto microtitre plate wells (NUNC MaxiSorp). Addi-tional sites were blocked with BSA and wells wererinsed with TNMC buffer (1 mM Tris—HCl, 150 mMNaCl, 0.1 mM MgCl2, 0.1 mM CaCl2, pH 8.0) andincubated with biofilm or planktonic antiserum(diluted 1:500 in TNMC containing 0.01% (w/v)BSA) for 2 h at room temperature. After washingthe wells three times with TNMC containing 0.05%(v/v) Tween 20, bound antibody was detected withHRP-conjugated anti-mouse IgG (Dako; 1:1000) ando-phenylenediamine as described previously.17

ELISA (A492) values were corrected for backgroundbinding of the cognate pre-immune serum (diluted1:500).

S. sanguis protein expression in biofilm 297

Results

Surface and aggregation properties ofbiofilm-grown cells

To demonstrate that S. sanguis cells growing as abiofilm (attached to beads within the GPT device)were physiologically different from cells grownplanktonically at approximately the same mean-specific growth rate in a continuous culture device,we investigated several surface properties for eachcell population (Fig. 1). The surface hydrophobicityof biofilm and continuous culture planktonic cellsdiffered markedly. Thus, 68.7% of biofilm cellsadhered to the hexadecane phase while no decreasein absorbance was observed when a suspension ofcontinuous culture-derived cells was mixed withhexadecane, indicating that the planktonic cellswere hydrophilic. Cells were continually shed fromthe biofilm and could be harvested from the GPT

flow-through (effluent). The hydrophobicity ofeffluent cells harvested as early as 24 h followinginoculation of the device had decreased to only5% of that of input (batch culture) cells, and thehydrophilic nature of effluent cells was maintainedfor samples harvested throughout the GPT run.Previous work has demonstrated that the surfacehydrophobicity of S. gordonii and S. sanguis isdetermined, at least in part, by the expressionof CshA, a large cell-wall-associated adhesin.22,25

Consequently, we investigated the surface expres-sion of CshA by the various cell populations usingwhole-cell ELISA (see Methods). Biofilm cellsexpressed significantly more CshA than planktoniccells derived from the continuous culture device(approximately 2.6-fold, Fig. 1) and values corre-lated, to a greater or lesser extent, with surfacehydrophobicity for all cell populations (Fig. 1).However, it was clear that CshA is not the soledeterminant of surface hydrophobicity since hydro-philic planktonic cells did express low levels ofCshA.

Finally, when the cell-wall-associated proteinsexpressed by biofilm and planktonic cells werecompared by SDS-PAGE, significant differenceswere observed in the mutanolysin-released proteinprofiles of these cell populations (data not shown).

Library screening with antiserum tobiofilm-derived cells

Having established that the two cell populationsdiffered significantly in cell-surface propertiesand protein expression profiles, we wished to iden-tify genes encoding biofilm-specific or up-regulatedsurface proteins. In order to do this, antiserum wasraised to fixed whole S. sanguis biofilm cells har-vested from the GPT and to planktonic cells har-vested from the continuous culture device (seeMethods). Antiserum to biofilm cells reacted to agreater extent in ELISA with early stationary phasecells of S. sanguis than did the antiserum raised tocontinuous culture-derived cells (data not shown).The initial intention was to absorb the biofilmcell antiserum with whole cells derived from thecontinuous culture device in order to remove anti-bodies to surface proteins expressed by both biofilmand planktonic cells, and thus generate a biofilm-specific antiserum preparation. However, we wereunable to obtain sufficient cells for thoroughabsorbtion and removal of antibodies to sharedsurface proteins. Consequently, the S. sanguisgenomic library was screened directly with thebiofilm antiserum to obtain a bank of clones encod-ing biofilm-expressed proteins. Approximately20 000 plaques were screened and 32 reactive

0

0.1

0.2

0.3

0

20

40

60

80

100

Batch

5 deffl.

24 heffl.

Batch 24 heffl.

5 deffl.

Cont.

Bio.

Bio.

Cont.

(A)

(B)

% h

ydro

ph

ob

icit

yE

LIS

A r

eact

ivit

y (A

492)

Figure 1 Surface properties of biofilm and planktonicgrown S. sanguis cells. (A) Cell-surface hydrophobicity.Approximately 2 � 109 cells of S. sanguis (2.5 ml) weremixed with 0.25 ml of hexadecane and the percentage ofcells associated with the hexadecane phase was calcu-lated from optical density measurements taken beforeand after mixing. (B) Cell-surface expression of CshA-likeproteins. A total of 2 � 107 S. sanguis cells per well wereimmobilised. CshA antiserum was diluted 1:2000 andELISA values (A492) were corrected for values obtainedwith pre-immune serum (diluted 1:2000): effl., effluentcells harvested after 24 h or 5 days (5 d); Bio., biofilmcells; Cont., continuous culture cells.

298 C. Black et al.

clones were purified by re-screening and phagemidsrescued for DNA sequence analysis.

DNA sequence analysis of immunopositiveclones

Approximately 400 bp of sequence was obtained foreach clone using T3 vector primer and this was usedto interrogate the NCBI protein database usingBLASTX search tool.15 Sequences of 31 of the 32clones showed similarity to bacterial proteinsinvolved in adhesion and colonisation, cell-wallbiosynthesis, carbohydrate utilisation and in bac-terial metabolism. Representative clones are listedin Table 1. The remaining clone (R6) demonstrated73% identity with a streptococcal protein ofunknown function (Table 1).

Adhesion and colonisationA total of seven clones were isolated that encodedfragments of CshA. Of these, six were revealed by

sequence analysis to be clones encoding a portion ofthe amino acid repeat region of CshA with bestmatch (75% identity) within repeat block 3 of S.gordonii CshA (starting at amino acid residue 1050).The 50-cloning site was the same for all six clones,however the 30-cloning site differed, resulting in arange of insert sizes from 0.9 to 6.2 kb. Clone C2(insert size 5.3 kb) and clone C11 (insert size 1.6 kb)were selected for further analysis. The seventhclone, R12, encoded a polypeptide fragment with54% identity with the non-repetitive region of CshA,starting at amino acid residue 452 of the S. gordoniiprotein.

Several other clones were identified thatencoded proteins involved in adhesion and coloni-sation of bacteria. Identical clones R5 and R11were isolated that demonstrated homology toCna collagen adhesin of S. aureus26 and siblingclones R1, C1 and C3 encoded a protein fragmentwith limited similarity (27/79 amino acid residueidentity) to a fibrinogen binding protein homologue

Table 1 Identity of clones reacting with antiserum raised to biofilm-grown S. sanguis cells.

Clone Sequence ID Function Amino acidsequence identity

Adhesion and colonisationR12 CshA S. gordonii, CAA46281 (non-repetitive region) Adhesion, fibrils 46/84 (54%)C2 CshA S. gordonii, CAA46281 (repeat region) Adhesion, fibrils 80/110 (73%)R5 Cna S. aureus, A42404 Collagen adhesin 42/110 (38%)C3 Putative SdrD S. aureus, CAA06651 Fibrinogen binding protein

homologue27/79 (34%)

C8 Iga S. sanguis, CAA73856 IgA1 protease 115/116 (99%)

Carbohydrate utilisationR4 GtfR S. oralis, BAA95201 Glucosyltransferase 87/90 (96%)C7 GtfG S. gordonii, AAC43483 Glucosyltransferase 76/80 (95%)R15 NplT S. pneumoniae, AAK99752 Neo-pullulanase 64/70C9 PulT Bacillus stearothermophilus, 1808262A Thermostable pullulanase 66/117 (56%)C23 Putative S. pneumoniae, AAK74446 Putative alkaline pullulanase 81/112 (72%)C6 DeoR S. pneumoniae, AAK74425 Transcriptional regulator 84/117 (72%)

Metabolism and growthR18 PenX S. mitis, CAC08461 Penicillin binding protein 2� 44/76 (58%)R3 GlpK S. pneumoniae AAL00793 Glycerol kinase 94/94 (100%)C17 Orf7 S. mutans, BAA32095 Surface presentation of

polysaccharide58/89 (65%)

C20 Putative S. pyogenes, AAK33397 Putative arylalkylaminen-acetyltransferase

53/89 (60%)

R9 NtpI S. pyogenes, AAK33253 V-type Naþ ATPase subunit I 58/75 (77%)R17 HemK S. pneumoniae, AAK99729 Possible adenine methyltransferase 51/66 (77%)C4 SulB S. pneumoniae, AAK74468 Dihydrofolate synthetase 61/109 (63%)C15 Sp1159 S. pneumoniae, AAK75646 Phosphoglucomutase/

phosphomannomutase108/124 (87%)

C18 DhaM L. lactis, AAK04347 Dihydroxyacetone kinase 60/119 (50%)C19 GidA S. pneumoniae, AAK98928 Glucose inhibited division protein 116/121 (96%)

UnknownR6 Sp0731 S. pneumoniae, AAK74872 Conserved domain protein 43/59 (73%)

S. sanguis protein expression in biofilm 299

SdrD of S. aureus.27 Clones R5 (Cna) and C3(SdrD) were selected for further study. In order toconfirm the identity of clones R5 and C3, sequencedata was used to interrogate the incomplete S.sanguisgenomeatVirginiaCommonwealthUniversity(www.sanguis.mic.vcu.edu). In both instances, thesearch yielded sequence information that extendedthe query sequence at both 30- and 50-ends (data notshown). A BLAST search of NCBI databases with theextended R5 sequence confirmed that the R5 cloneencoded part of a larger Cna-like protein (data notshown). The protein encoded by the extended S.sanguis sequence also demonstrated homology toCNE, a collagen adhesin of Streptococcus equi.28 Incontrast, the extended sequence of clone C3obtained from the S. sanguis incomplete genomefailed to confirm the SdrD homology (data notshown). Indeed, there was no significant matchwithin the NCBI databases. Finally, clone C8encoded a polypeptide fragment of IgA proteasewith almost 100% identity to a previously reportedIgA protease of S. sanguis (Table 2).

Carbohydrate utilisationGlucosyltransferase enzymes of oral streptococcicatalyze the polymerisation of glucose moietiesfrom sucrose to form high molecular weight glucansthat have been implicated in enhancing attachmentand accumulation of streptococci. Two indepen-dent clones, C21 and C7, were obtained thatencoded significant homology to glucosyltrans-ferases of S. gordonii and S. oralis (Table 1). Twosiblings of C21 were also obtained but were notstudied further. Clones C7 and C21 matched withdifferent regions of the glucosyltransferase fromS. gordonii (starting at amino acid residue 702and 878 of S. gordonii GtfG, respectively). How-ever, sequence analysis of the 30-end of clone C7

indicated that these clones overlapped andexpressed fragments of the same S. sanguis gluco-syltransferase. Neither C3 nor C21 GTF open readingframes were in-frame with respect to the a-peptideof b-galactosidase.

Three unique clones were isolated that expressedfragments of pullulanase amylolytic enzymes. Eachclone demonstrated homology with a different pull-ulanase of S. pneumoniae identified by genomicsequencing. Thus, the best matches for clonesC9, C23, and R15 were thermostable pullulanase,alkaline amylopullulanase and neo-pullulanase,respectively (Table 1). Thermostable and alkalinepullulanases of S. pneumoniae demonstrate signifi-cant sequence identity and it remains possible thatclones C9 and C23 encode fragments of the same S.sanguis protein.

Finally, clone C6 demonstrated significanthomology with deoR of Lactococcus lactis encodinga transcriptional repressor of carbohydrate cata-bolic systems.

Metabolism and growthA total of 9 clones were identified that encodedproteins with homology to bacterial componentsinvolved in cell-wall biosynthesis or in metabolismand growth (Table 1). A significant proportion ofthese proteins would be expected to be intracellu-lar. Clone C4 encoding a homologue of dihydrofo-late synthetase was selected for further study.

Relative reactivity of E. coli clones withbiofilm and planktonic sera

Whole-cell lysates from selected E. coli clonesencoding adhesion-related proteins were analysedin ELISA to determine their relative reactivity withbiofilm and planktonic sera (Table 2). As expected

Table 2 Differential reactivity of biofilm and planktonic sera with selected clones.

Clone Best match Antiserum reactivitya Ratiob

Biofilm Planktonic

R12 N-CshA 0.261 � 0.039 0.054 � 0.011 4.8C2 C-CshA (5.3 kb) 0.379 � 0.051 0.281 � 0.009 1.3C11 C-CshA (1.6 kb) 0.349 � 0.028 0.355 � 0.041 1.0R5 Cna 0.278 � 0.043 0.055 � 0.009 5.0C3 SdrD 0.285 � 0.022 0.278 � 0.022 1.0C4 SulB 0.227 � 0.030 0.185 � 0.029 1.2

Values are the means � S:D: of quadruplicate determinations from a representative experiment repeated threetimes.a One microgram protein per well was immobilised. Biofilm and planktonic antisera were diluted 1:500 and ELISAvalues (A492) were corrected for values obtained with the relevant pre-immune serum (diluted 1:500).b The biofilm:planktonic expression ratio was calculated by dividing the biofilm ELISA value with that obtained usingplanktonic antiserum.

300 C. Black et al.

from ELISA analysis of whole cells of S. sanguis(above), biofilm antiserum reacted better with pro-tein extracts from clone R12 encoding a portionof the non-repetitive region of CshA (biofilm:plank-tonic ratio of 4.8). Less expected was the obser-vation that reactivity of sera to protein extracts ofclone C2, encoding CshA repeat region fragment,was closer to parity (ratio 1.3). A similar result wasobtained for the related but smaller clone C11(Table 2). The biofilm:planktonic ratio for cloneR5 encoding the Cna collagen adhesin homologuewas 5.0 suggesting that this protein was up-regu-lated in biofilm-grown cells. In contrast, the reac-tivity of both biofilm and planktonic sera withextracts of E. coli clones C3 and C4, encoding theputative SdrD homologue and dihydrofolate synthe-tase, respectively, was similar (Table 2).

Discussion

Increased awareness of the fundamental differencebetween planktonic and biofilm cells has spurredthe development of new approaches to understand-ing the biofilm phenotype. On the whole, thesemethodologies focus on mono-species biofilms,which, although falling short of in vivo reality (some500 different species inhabit the human oral cav-ity), do nevertheless offer a quantum step overplanktonic systems in terms of understanding nat-ural biofilms. A number of studies have employedmutagenic approaches to identify genes whoseexpression is required for initiation of the biofilmlifestyle of oral bacteria.29—33 A wide range of geneswas identified whose products generally fell intoone of the following categories: peptidoglycansynthesis, cell-cell communication, environmentalsensing and signalling, or adhesion.29—33 In contrast,proteomics and real-time RT-PCR have allowedmeasurement of relative expression of genes inestablished biofilms,34,35 and although there is someoverlap with the genes identified by mutagenesis,on the whole a different subset of genes was iden-tified. In one study, diminished expression of gly-colytic enzymes was noted for S. mutans grown in abiofilm, whereas biosynthetic functions wereenhanced.34 In a second study, the majority oftranscripts analysed were seen to be down regu-lated in biofilms of S. gordonii, including thoseencoding transport and adhesion functions.35 Inter-estingly, this study demonstrated a 5.5-fold lowerabundance of abpA transcripts (encoding a-amy-lase-binding protein) in biofilm cells of S. gordoniiwhereas, in the study by Loo et al.29 mutation inabpA abrogated biofilm formation of S. gordonii onabiotic surfaces, a result seen also for S. mutans.36

Clearly, the methodological approach to identifybiofilm-related gene influences the resultsobtained. In the current study, we focussed onthe cell-surface protein expression of establishedS. sanguis biofilms.

Clear phenotypic differences between biofilmand planktonic cells of S. sanguis were observed,where biofilm cells grown in the GPT were found tobe markedly more hydrophobic than planktoniccells grown at approximately the same mean-spe-cific growth rate. In contrast, cells shed from thebiofilm in the GPT device were only slightly morehydrophobic than the planktonic cells, an observa-tion noted previously for biofilms of P. aeruginosa37

and E. coli.38 Since surface hydrophobicity has beenlinked with adhesive potential of oral strepto-cocci,25 it may be hypothesised that mature biofilmssheds cells of reduced hydrophobicity in order forthese cells to dissipate to other potential niches as asurvival or reproductive strategy. The observationthat CshA expression was greater in biofilm-grown S.sanguis cells correlated with increased hydropho-bicity of these cells and supported the hypothesisthat the expression of specific proteins could differmarkedly between biofilm and planktonic cells.Antiserum raised to fixed whole cells of S. sanguisderived from biofilm or planktonic grown cellsshould reflect these differences. Consequently weused differential antiserum screening to identifycell-surface-associated proteins of S. sanguis thatwere differentially expressed in biofilm cells. Sim-plistically, differential reactivity of E. coli cloneswith biofilm and planktonic cell antisera was takenas a measure of the relative levels of expression ofthat protein on the surface of S. sanguis cells used toimmunise the rabbits. Differences in immuneresponses by the rabbits may impact on the data,and further planned studies will attempt to confirmdifferential expression through RNA isolation andRT-PCR using specific primer pairs.

As expected, the majority of clones isolated fromfirst screening of the S. sanguis library with biofilmcell antiserum encoded cell surface or secretedproteins (20/32 clones), although among the iso-lated clones were proteins that have not been pre-viously described in oral streptococci (Cna, putativeSdrD homologues; see below). The remaining clonesencoded proteins that would be expected, on thebasis of function or lack of an N-terminal leaderpeptide, to be intracellular in nature and theirisolation may reflect a degree of cell lysis duringpreparation of the inoculum. Recently, however, agrowing number of streptococcal proteins thatdo not contain a recognised primary sequenceassociated with Sec-dependent export, havebeen demonstrated as being presented at the cell

S. sanguis protein expression in biofilm 301

surface, suggesting that the streptococci have an asyet uncharacterised mechanism of protein export.These include Group A streptococcal surface dehy-drogenase,39 enolase of S. pyogenes40 and S. pneu-moniae,41 and pneumococcal PavA fibronectinadhesin,24 and it remains possible that the detec-tion of some of the clones described in this studymay in fact reflect genuine surface presentation onintact cells.

The 2.6-fold difference in CshA expression onwhole cells of biofilm versus planktonic grownS. sanguis, as measured using polyclonal antiserumto a 260 kDa fragment of CshA, was comparable withthe 4.8-fold difference determined by analysis oflysates of E. coli clone R12 (N-CshA) using biofilmand planktonic antiserum, thus validating theexperimental approach. In contrast, the relativereactivity of C-CshA clones C2 and C11 was closerto parity, and this was independent of the cloneinsert size. The presence of repeated epitopesexpressed by the C-CshA clones might have a bear-ing although this remains to be tested experimen-tally. Certainly, whole-cell ELISA has demonstratedthat both non-repetitive and repeat regions of CshAare available on the surface of whole cells forrecognition and binding by antibodies raised tothe respective peptide regions.42 However, theimmunogenicity of different CshA epitopes maybe subject to modulation, when presented in thecontext of the complex bacterial cell surface. Anadditional complicating factor is the presence ofmore than one CshA-like protein on S. sanguisNCTC10904 (McNab, unpublished data). The expres-sion of the multifunctional adhesin CshA of S. gor-donii is maximal in late exponential phase of growthand regulation of cshA promoter activity mayinvolve production and detection of oligopeptidesignal(s).43 Clearly, the constrained environmentof a biofilm may allow the rapid accumulationand maintenance of signalling molecules at athreshold level, and this may account for the ele-vated expression of CshA in S. sanguis biofilms.Indeed, competence in S. mutans, a density depen-dent phenomenon based on the secretion anddetection of peptide pheromones, is markedlyenhanced in S. mutans biofilms.44 Differential ana-lysis of clone R12 indicated that the Cna homologueof S. sanguis was also up-regulated approximatelyfive-fold in biofilm cells. Cna of S. aureus is a multi-domain protein with a molecular architecture thatis similar to a number of Gram-positive bacterialsurface proteins. The protein fragment expressedby clone R12 demonstrated 38% identity (54%homology) to a B repeat sequence of Cna. The Brepeat region is proposed to act as a ‘stalk’ to holdthe collagen binding A domain distal from the bac-

terial surface.45 Although analysis of the S. sanguisincomplete genome database confirmed thesequence homology, whether or not the native S.sanguis gene encodes a collagen binding proteinawaits further sequence and experimental analysis.Collagen binding by oral streptococci has beenattributed, at least in part, to the antigen I/II familyof adhesins,46 although no antigen I/II homologueswere identified in this study.

In contrast to CshA and the Cna homologues,differential antiserum analysis of clones C3 andC4 suggested that these proteins were expressedequally by biofilm and planktonic S. sanguis cells.The identity of the protein partly encoded by cloneC3 needs to be confirmed by further sequenceanalysis. Nevertheless, the ELISA data presentedin Table 2 strongly suggest that this protein is sur-face exposed on S. sanguis cells and is immunogenicin rabbits.

In summary, the results presented in this studyprovide strong evidence for the utility of a com-bined immunological and molecular approach tobetter understand the biofilm phenotype. Theapproach has focused on surface protein differ-ences and, by screening first with biofilm antiserum,has biased somewhat towards identification ofproteins up-regulated in biofilm mode of growth.Nevertheless, the approach has two advantagesover other methodologies such as differential ana-lysis of gene promoter activity. Firstly, differencesin protein levels are determined, and not differ-ences in mRNA levels, which may only reflect thepotential for protein expression. Secondly, thesystem is biased towards identification of surface-associated proteins, which may have direct rele-vance in disease intervention through passive oractive immunisation, for example. The role in bio-film formation of differentially expressed proteinsidentified in this study will be investigated in moredetail in future studies.

Acknowledgements

We thank J.V. Embleton and S.P. Nair, UniversityCollege London, for advice and technical assistance,and B. Cover, University of Canterbury (UK), forgenerating rabbit antiserum to S. sanguis cells.

References

1. Kawamura Y, Hou X-G, Sultana F, Miura H, Ezaki T.Determination of 16S RNA sequences of Streptococcus mitisand Streptococcus gordonii and phylogenetic relationshipsamongst members of the genus Streptococcus. Int J SystBacteriol 1995;45:406—8.

302 C. Black et al.

2. Frandsen EVG, Pedrazzoli V, Kilian M. Ecology of viridansstreptococci in the oral cavity and pharynx. Oral MicrobiolImmunol 1991;6:129—33.

3. Whittaker CJ, Klier CM, Kolenbrander PE. Mechanismsof adhesion by oral bacteria. Ann Rev Microbiol 1996;50:513—52.

4. Jenkinson HF, Lamont RJ. Streptococcal adhesion andcolonization. Crit Rev Oral Biol Med 1997;8:175—200.

5. Douglas WW, Heath J, Hampton KK, Preston FE. Identity ofviridans streptococci isolated from cases of infectiveendocarditis. J Med Microbiol 1993;39:179—82.

6. Johnson AP, Warner M, Broughton K, James D, Efsratiou A,George RC, et al. Antibiotic susceptibility of streptococciand related genera causing endocarditis: analysis of UKreference laboratory referrals, January 1996 to March 2000.Brit Med J 2001;322:395—6.

7. Gilbert P, Das J, Foley I. Biofilm susceptibility to anti-microbials. Adv Dent Res 1997;11:160—7.

8. Wilson M. Susceptibility of oral bacterial biofilms toantimicrobial agents. J Med Microbiol 1996;44:79—87.

9. Davies DG, Chakrabarty AM, Geesey GG. Exoploysaccharideproduction in biofilms: substratum activation of alginategene expression by Pseudomonas aeruginosa. Appl EnvironMicrobiol 1993;59:1181—6.

10. Hoyle BD, Williams LJ, Costerton WJ. Production of mucoidexopolysaccharide during development of Pseudomonasaeruginosa biofilms. Infect Immun 1993;61:777—80.

11. Schembri MA, Kjaergaard K, Klemm P. Global gene expressionin Escherichia coli biofilms. Mol Microbiol 2003;48:253—67.

12. Jenkinson HF, Terry SD, McNab R, Tannock GW. Inactivationof the gene encoding surface protein SspA in Streptococcusgordonii DL1 affects cell-interactions with human salivaryagglutinin and oral actinomyces. Infect Immun 1993;61:3199—208.

13. Sambrook J, Fritsch EF, Maniatis T. Molecular cloning: alaboratory manual. 2nd ed. Cold Spring Harbour, NY: ColdSpring Harbour Laboratory Press; 1989.

14. Jenkinson HF. Novobiocin-resistant mutants of Streptococ-cus sanguis with reduced cell hydrophobicity and defectivein coaggregation. J Gen Microbiol 1987;133:1909—18.

15. Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z,Miller W, et al. Gapped BLAST and PSI-BLAST: a newgeneration of protein database search programs. NucleicAcids Res 1997;25:3389—402.

16. Embleton JV, Newman HN, Wilson M. Influence of growthmode and sucrose on susceptibility of Streptococcus sanguisto amine fluorides and amine fluoride-inorganic fluoridecombinations. Appl Environ Microbiol 1998;64:3503—6.

17. Holmes AR, Gopal PK, Jenkinson HF. Adherence of Candidaalbicans to a cell surface polysaccharide receptor onStreptococcus gordonii. Infect Immun 1995;63:1827—2183.

18. McNab R, Jenkinson HF, Loach DM, Tannock GW. Cell-surface-associated polypeptides CshA and CshB of highmolecular mass are colonization determinants in the oralbacterium Streptococcus gordonii. Mol Microbiol 1994;14:743—54.

19. Jenkinson HF. Cell-surface proteins of Streptococcus san-guis associated with cell hydrophobicity and coaggregationproperties. J Gen Microbiol 1994;132:1575—89.

20. Demuth DR, Duan Y, Brooks W, Holmes AR, McNab R,Jenkinson HF. Tandem genes encode cell-surface polypep-tides SspA and SspB which mediate adhesion of the oralbacterium Streptococcus gordonii to human and bacterialreceptors. Mol Microbiol 1996;20:403—13.

21. Morrissey JH. Silver stain for proteins in polyacrylamidegels: a modified procedure with enhanced uniform sensitiv-ity. Anal Biochem 1981;117:307—10.

22. McNab R, Jenkinson HF. Gene disruption identifies a 290 kDacell-surface polypeptide conferring hydrophobicity andcoaggregation properties in Streptococcus gordonii. MolMicrobiol 1992;6:2939—49.

23. Jenkinson HF, Easingwood RA. Insertional inactivation of thegene encoding a 76-kilodalton cell surface polypeptide inStreptococcus gordonii Challis has a pleiotropic effect oncell surface composition and properties. Infect Immun1990;58:3689—97.

24. Holmes AR, McNab R, Millsap KW, Ohde M, HammerschmidtS, Mawdsley JL. The pavA gene of Streptococcus pneumo-niae encodes a fibronectin-binding protein that is essentialfor virulence. Mol Microbiol 2001;41:1395—408.

25. McNab R, Holmes AR, Jenkinson HF. Cell-surface polypep-tides as determinants of hydrophobicity in Streptococcusgordonii and Streptococcus sanguis. Colloids Surf 1995;5:135—42.

26. Patti JM, Jonsson H, Guss B, Switalski LM, Wiberg K,Lindberg M, et al. Molecular characterization and expressionof a gene encoding a Staphylococcus aureus collagenadhesin. J Biol Chem 1992;267:4766—72.

27. Josefsson E, McCrea KW, Ni Eidhin D, O’Connell D, Cox J,Hook M, et al. Three new members of the serine-aspartatereeat protein multigene family of Staphylococcus aureus.Microbiol 1998;144:3387—95.

28. Lannergard J, Frykberg L, Guss B. CNE a collagen-bindingprotein of Streptococcus equi. FEMS Microbiol Lett 2003;222:69—74.

29. Loo CY, Corliss DA, Ganeshkumar N. Streptococcus gordoniibiofilm formation: identification of genes that code forbiofilm phenotypes. J Bacteriol 2000;182:1374—82.

30. Froeliger EH, Fives-Taylor P. Streptococcus parasanguisfimbria-associated adhesin Fap1 is required for biofilmformation. Infect Immun 2001;69:2512—9.

31. Bhagwat SP, Nary J, Burne RA. Effects of mutating putativetwo-component systems on biofilm formation by Streptococ-cus mutans UA159. FEMS Microbiol Lett 2001;205:225—30.

32. Wen ZT, Burne RA. Functional genomics approach toidentifying genes required for biofilm development byStreptococcus mutans. Appl Environ Microbiol 2002;68:1196—203.

33. Yoshida A, Kuramitsu HK. Multiple Streptococcus mutansgenes are involved in biofilm formation. Appl EnvironMicrobiol 2002;68:6283—91.

34. Svensater G, Welin J, Wilkins JC, Beighton D, Hamilton IR.Protein expression by planktonic and biofilm cells ofStreptococcus mutans. FEMS Microbiol Lett 2001;205:139—46.

35. Gilmore KS, Srinivas P, Akins DR, Hatter KL, Gilmore MS.Growth, development, and gene expression in a persistentStreptococcus gordonii biofilm. Infect Immun2003;71:4759—66.

36. Hazlett KRO, Mazurkiewicz JE, Banas JA. Inactivation of thegbpA gene of Streptococcus mutans alters structural andfunctional aspects of plaque biofilm which are compensatedby recombination of the gtfB and gtfC genes. Infect Immun1999;67:3909—14.

37. Allison DG, Brown MR, Evans DE, Gilbert P. Surfacehydrophobicity and dispersal of Pseudomonas aeruginosafrom biofilms. FEMS Microbiol Lett 1990;59:101—4.

38. Allison DG, Evans DJ, Brown MR, Gilbert P. Possibleinvolvement of the division cycle in dispersal of Escherichiacoli from biofilms. J Bacteriol 1990;172:1667—9.

39. Pancholi V, Fischetti VA. A major surface protein of group Astreptococci is a glyceraldehydes-3-phosphate-dehydrogen-ase with multiple binding activity. J Exp Med 1992;176:415—26.

S. sanguis protein expression in biofilm 303

40. Pancholi V, Fischetti VA. a-Enolase, a novel strong plasmi-n(ogen) binding protein on the surface of pathogenicstreptococci. J Biol Chem 1998;273:14505—15.

41. Bergmann S, Rohde M, Chhatwal GS, Hammerschmidt S. a-Enolase of Streptococcus pneumoniae is a plasmin(ogen)-binding protein displayed on the bacterial cell surface. MolMicrobiol 2001;40:1273—87.

42. McNab R, Holmes AR, Clarke JM, Tannock GW, JenkinsonHF. Cell surface polypeptide CshA mediates binding ofStreptococcus gordonii to other oral bacteria and toimmobilized fibronectin. Infect Immun 1996;64:4204—10.

43. McNab R, Jenkinson HF. Altered adherence properties of aStreptococcus gordonii hppA (oligopeptide permease) mu-

tant result from transcriptional effects on cshA adhesingene expression. Microbiology 1998;144:127—36.

44. Li YH, Lau PC, Lee JH, Ellen RP, Cvitkovitch DG. Naturalgenetic transformation of Streptococcus mutans in growingbiofilms. J Bacteriol 2001;183:897—908.

45. Deivanayagam CCS, Rich RL, Carson M, Owens RT, Danthu-luri S, Bice T, et al. Novel fold and assembly of therepetitive B region of the Staphylococcus aureus collagen-binding surface protein. Structure 2000;8:67—78.

46. Love RM, McMillan MD, Jenkinson HF. Invasion of dentinaltubules by oral streptococci is associated with collagenrecognition mediated by the antigen I/II family of polypep-tides. Infect Immun 1997;65:5157—64.

304 C. Black et al.