Cell Surface Glycoside Hydrolases of Streptococcus gordonii PromoteGrowth in Saliva

Jinghua Yang,a,b Yuan Zhou,a,c Luxia Zhang,a Nehal Shah,a Cheng Jin,b Robert J. Palmer, Jr.,a John O. Cisara

Microbial Receptors Section, Laboratory of Cell and Developmental Biology, National Institutes of Health, Bethesda, Maryland, USAa; State Key Laboratory of Mycology,Institute of Microbiology, Chinese Academy of Sciences, Beijing, Chinab; State Key Laboratory of Oral Diseases, West China Hospital of Stomatology, Sichuan University,Chengdu, Chinac

ABSTRACT

The growth of the oral commensal Streptococcus gordonii in saliva may depend on a number of glycoside hydrolases (GHs), in-cluding three cell wall-anchored proteins that are homologs of pneumococcal �-galactosidase (BgaA), �-N-acetylglucosamini-dase (StrH), and endo-�-N-acetylglucosaminidase D (EndoD). In the present study, we introduced unmarked in-frame deletionsinto the corresponding genes of S. gordonii DL1, verified the presence (or absence) of the encoded proteins on the resulting mu-tant strains, and compared these strains with wild-type strain DL1 for growth and glycan foraging in saliva. The overnightgrowth of wild-type DL1 was reduced 3- to 10-fold by the deletion of any one or two genes and approximately 20-fold by the dele-tion of all three genes. The only notable change in the salivary proteome associated with this reduction of growth was a down-ward shift in the apparent molecular masses of basic proline-rich glycoproteins (PRG), which was accompanied by the loss oflectin binding sites for galactose-specific Erythrina cristagalli agglutinin (ECA) and mannose-specific Galanthus nivalis aggluti-nin (GNA). The binding of ECA to PRG was also abolished in saliva cultures of mutants that expressed cell surface BgaA alone ortogether with either StrH or EndoD. However, the subsequent loss of GNA binding was seen only in saliva cocultures of differentmutants that together expressed all three cell surface GHs. The findings indicate that the growth of S. gordonii DL1 in saliva de-pends to a significant extent on the sequential actions of first BgaA and then StrH and EndoD on N-linked glycans of PRG.

IMPORTANCE

The ability of oral bacteria to grow on salivary glycoproteins is critical for dental plaque biofilm development. Little is known,however, about how specific salivary components are attacked and utilized by different members of the biofilm community, suchas Streptococcus gordonii. Streptococcus gordonii DL1 has three cell wall-anchored glycoside hydrolases that are predicted to acton host glycans. In the present study, we introduced unmarked in-frame deletions in the corresponding genes, verified the pres-ence (or absence) of encoded proteins on the resulting mutant strains, and compared these strains with wild-type DL1 forgrowth and glycan foraging in saliva. The results indicate that the growth of S. gordonii DL1 depends to a significant extent onsequential action of these cell surface GHs on N-linked glycans of basic proline-rich salivary glycoproteins, which appears to bean essential first step in salivary glycan foraging.

The development of dental plaque in the absence of exogenousnutrients has long been thought to depend on bacterial forag-

ing of host salivary glycans (1, 2). Little is known, however, abouthow different oral species attack and utilize salivary macromole-cules as substrates for growth and whether cross-feeding betweenspecies influences dental plaque biofilm development. Recently,we compared the growth of different oral bacteria in filter-steril-ized saliva that was first heat treated to inactivate endogenousglycoside hydrolases and then dialyzed to remove possible low-molecular-mass growth substrates. Growth through two transfersin this model system was observed elsewhere for strains of Strep-tococcus gordonii, Streptococcus oralis, and Streptococcus mitis butnot for strains of certain other species, including Actinomyces spp.,Abiotrophia defectiva, Rothia spp., Streptococcus salivarius, Strep-tococcus sanguinis, Streptococcus sobrinus, and Streptococcus ves-tibularis, which grew poorly or not at all (3). Importantly, thegrowth of the first three streptococcal species mentioned abovewas accompanied by degradation of basic proline-rich glycopro-teins (PRG), whereas species that failed to grow did not alter theelectrophoretic mobility or lectin reactivity of PRG nor any othersalivary component.

S. gordonii strain DL1 (i.e., Challis) was one of the streptococci

that grew well in saliva and appeared to utilize PRG as a growthsubstrate. Annotation of this strain in the Carbohydrate-ActiveEnzymes database (CAZy; www.cazy.org) revealed 31 putativeglycoside hydrolases (GHs) distributed among 16 GH families.Based on specific GH activities associated with these families, wesurmised that host glycan foraging could involve at least 17 puta-tive GHs of S. gordonii. Of these, three were identified as cellwall-anchored proteins based on the predicted presence of N-terminal signal peptides and C-terminal sorting motifs (4).Moreover, these three proteins were homologues of Streptococcuspneumoniae �-galactosidase (BgaA), �-N-acetylglucosaminidase

Received 29 April 2016 Accepted 14 June 2016

Accepted manuscript posted online 17 June 2016

Citation Yang J, Zhou Y, Zhang L, Shah N, Jin C, Palmer RJ, Jr, Cisar JO. 2016. Cellsurface glycoside hydrolases of Streptococcus gordonii promote growth in saliva.Appl Environ Microbiol 82:5278 –5286. doi:10.1128/AEM.01291-16.

Editor: A. M. Spormann, Stanford University

Address correspondence to Jinghua Yang, jinghuay98@gmail.com.

For a companion article on this topic, see doi:10.1128/AEM.01111-16.

Copyright © 2016, American Society for Microbiology. All Rights Reserved.

crossmark

5278 aem.asm.org September 2016 Volume 82 Number 17Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

(StrH), and endo-�-N-acetylglucosaminidase D (EndoD), whichare known to act on host glycans (5, 6) and, in the case of BgaA andStrH, to promote the growth of S. pneumoniae on the humanserum component �-1 acid glycoprotein (7).

Based on the information summarized above, we hypothesizedthat BgaA, StrH, and EndoD of S. gordonii are critical for salivaryglycan foraging. To test this hypothesis, we deleted the genes forthese proteins in different combinations from S. gordonii DL1 andthen compared the growth of the wild-type and mutant strains insaliva. The results indicate that the growth of S. gordonii DL1depends to a significant extent on sequential action of the threecell surface GHs on N-linked glycans of salivary PRG.

MATERIALS AND METHODSPlasmids and bacterial strains. The plasmids listed in Table 1 were puri-fied by using QIAprep spin miniprep kits (Qiagen) from Escherichia coliDH5� grown in LB broth containing either ampicillin (75 �g/ml) orerythromycin (500 �g/ml) to maintain antibiotic resistance. Wild-type S.gordonii DL1 and glycoside hydrolase deletion mutant (�GH mutant)strains (Table 1) were maintained at �70°C as frozen stock cultures inTodd-Hewitt broth (THB; Oxoid) containing 40% glycerol and weretransferred to THB or another complex medium, as indicated below, atthe beginning of each experiment.

Plasmid construction. Unmarked in-frame deletions were intro-duced into S. gordonii DL1 using a nonreplicative allelic exchange vectorsimilar to pCWU3, which was developed for use in Actinomyces oris (11).We found that the kanamycin resistance and mCherry fluorescence mark-ers of pCWU3 were not optimal for allelic exchange in S. gordonii DL1,

and we therefore prepared a similar plasmid, designated pEG (Fig. 1 andTable 1), that contained the ermAM cassette and expressed green fluores-cent protein (GFP) from the constitutive S. gordonii promoter, PrpsJ. TheermAM cassette was amplified by PCR from pKSerm2 using KOD hotstart DNA polymerase (Novagen) and primers ermAM-F/ermAM-R (Ta-ble 2). The resulting PCR product was cloned into the AclI site of dephos-phorylated pUC19c to obtain pUC19c-1. The 198-bp sequence designatedPrpsJ, which is located immediately upstream from rpsJ (i.e., SGO_1986),was amplified from S. gordonii DL1 genomic DNA with primer pair rpsJ-F/rpsJ-R. The 720-bp sequence of gfp (including two TAA stop codons)was codon optimized for expression in S. gordonii and synthesized by BlueHeron Biotechnology, Inc. (Bothell, WA). The optimized sequence wasamplified from a gfp-containing derivative of pBR322 using primer pairgfp-F/gfp-R. The sequences of PrpsJ and gfp were linked by overlap exten-sion PCR performed with primer pair rps-F/gfp-R and using the corre-sponding PCR products as the template. The resulting 0.9-kb PCR prod-uct (PrpsJ-gfp) was cloned into the EcoRI site of ermAM-containingpUC19c-1 to obtain pEG.

Gene exchange cassettes containing deletion constructs of S. gordoniiDL1 bgaA, strH, or endoD (i.e., SGO_1486, SGO_0405, or SGO_0208,respectively) flanked by approximately 1-kb adjacent gene-targeting se-quences were prepared from overlapping PCR products amplified fromgenomic DNA with primer pairs (Table 1) GH2-1/GH2-2 and GH2-3/GH2-4, GH20-1/GH20-2 and GH20-3/GH20-4, and GH85-1/GH85-2and GH85-3/GH85-4. The deletions in the gene exchange cassettes re-moved residues 32 to 2303 from the 2,350-amino-acid-coding sequenceof bgaA, residues 27 to 1129 from the 1,142-amino-acid-coding sequenceof strH, and residues 25 to 1532 from the 1,545-amino-acid-coding se-quence of endoD. Each cassette was cloned into pEG as indicated in Table1 to obtain pEG�bgaA, pEG�strH, or pEG�endoD, respectively.

Isolation of �GH mutants. Transformation of S. gordonii DL1 wasperformed as previously described (12) by incubating early-log-phasecells with 30 to 50 ng of pEG�bgaA, pEG�strH, or pEG�endoD. Reactionmixtures were plated on brain heart infusion (BHI) agar (Oxoid) contain-ing 10 �g/ml erythromycin to select for Emr transformants that also ex-pressed gfp from the integrated plasmid. Integration of each plasmid at thetarget site was confirmed by PCR performed with primers designed fromadjacent plasmid and S. gordonii chromosomal sequences (Table 2, prim-ers UC19-F with GH2-6, GH20-6, or GH85-6 and UC19-R with GH2-5,GH20-5, or GH85-5). Plasmid-bearing clones were then transferred and

TABLE 1 Bacterial strains and plasmids used in this study

Plasmid orstrain Description

Reference orsource

PlasmidspUC19c amp-containing cloning vector (GenBank

accession no. L09137.2)8

pKSerm2 Source of ermAM cassette 9pUC19c-1 pUC19c with ermAM replacing amp in

AclI siteThis study

pBR322gfp Source of codon-optimized gfp This studypEG pUC19c-1 with PrpsJ-gfp in EcoRI site

(GenBank accession no. KT962975.1)This study

pEG�bgaA pEG with bgaA deletion cassette in KpnIsite

This study

pEG�strH pEG with strH deletion cassette in SalIand HindIII sites

This study

pEG�endoD pEG with endoD deletion cassette in KpnIsite

This study

StrainsDL1 Wild-type Streptococcus gordonii (strain

Challis)10

�A �bgaA mutant of strain DL1 This study�H �strH mutant of strain DL1 This study�D �endoD mutant of strain DL1 This study�A/H �bgaA mutant of strain �H (i.e., �bgaA

�strH)This study

�A/D �endoD mutant of strain �A (i.e., �bgaA�endoD)

This study

�H/D �endoD mutant of strain �H (i.e., �strH�endoD)

This study

�A/H/D �strH mutant of strain �A/D (i.e., �bgaA�strH �endoD)

This study

FIG 1 Physical map of pEG, used to construct �GH mutants of S. gordoniiDL1. Key features include a multiple cloning site (MCS) to insert the genedeletion cassette, an antibiotic resistance marker (ermAM cassette) to select forintegration of the resulting plasmid into the streptococcal chromosome, and afluorescence marker (gfp expressed from PrpsJ of S. gordonii) to counterselectfor plasmid excision, which resolves plasmid-bearing cointegrates into either�GH mutants or wild-type clones.

Cell Surface Glycoside Hydrolases of S. gordonii

September 2016 Volume 82 Number 17 aem.asm.org 5279Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

passaged several times in antibiotic-free medium to allow excision of in-tegrated plasmids (accompanied by loss of Emr and GFP fluorescence)prior to plating. Plasmid-free (nonfluorescent) colonies were typicallyseen at frequencies of from 10�2 to 10�3 and were distinguished fromplasmid-bearing (fluorescent) colonies using a Fuji image reader. Thenonfluorescent colonies were screened by PCR with primer pairs thatamplified across deletions (Table 2, GH2-F/GH2-R, GH20-F/GH20-R, orGH85-F/GH85-R) to distinguish wild-type bacteria from �GH mutants.PCR products amplified from �GH mutants were also sequenced to con-firm deletions in these strains.

Mutants with deletions in more than one gene were obtained by re-peating the procedure described above. As summarized in Table 1, �bgaA�strH, �bgaA �endoD, and �strH �endoD double mutants (i.e., strains�A/H, �A/D, and �H/D, respectively) were derived from �strHor �endoD single mutants (i.e., strains �H or �D, respectively), and the�bgaA �strH �endoD triple mutant (i.e., strain �A/H/D) was derivedfrom the �bgaA �endoD double mutant (i.e., strain �A/D).

Enzyme assays. Bacteria cultured in THB, heart infusion broth (HIB;Becton, Dickinson), or whole saliva were washed in saliva-salts buffer (3)and adjusted to final cell densities of approximately 1 � 109/ml. Enzymeassays were performed by mixing 180 �l of bacterial suspension with 20 �lof either 10 mM 2-nitrophenyl-�-D-galactopyranoside (2-Np-�-Gal) or4-nitrophenyl-N-acetyl-�-D-glucosamine (4-Np-�-GlcNAc) (both fromSigma) for 0, 30, 60, 90, or 120 min. Following the addition of NaOH tostop the reaction and centrifugation to remove bacteria, the A405 of eachreaction mixture was measured with a Wallac Victor3 plate reader(PerkinElmer). Similar assays were performed by mixing 30 �l of bacterialsuspension with 100 �l of 100 �M 4-methylumbelliferyl-N-acetyl-�-D-

glucosamine (4-MU-�-GlcNAc) (Sigma) in a total volume of 200 �l for 0,30, 60, 90, or 120 min prior to measurement of fluorescence at 460 nm(from excitation at 355 nm) with a Wallac Victor3 plate reader.

Immunological methods. N-terminally polyhistidine (6�His)-taggedrecombinant proteins containing the putative catalytic region of each cellsurface GH were prepared for use as immunogens. The nucleotide se-quences for amino acid residues 80 to 609 of BgaA (13, 14), 136 to 507 ofStrH (15, 16), and 113 to 762 of EndoD (17), including in-frame 5=-BamHI and 3=-HindIII cloning sites and the requisite start (ATG) andstop codons (TAA or TGA), were optimized for expression in E. coli,synthesized, and cloned into pTrcHisB (Invitrogen) by Blue Heron Bio-technology, Inc. Plasmid-bearing E. coli EC100 bacteria were grown at37°C with aeration in Luria-Bertani broth supplemented with ampicillin(150 �g/ml). At an A600 of 0.4, the cultures were rapidly chilled in icewater, and isopropyl thiogalactopyranoside (IPTG) was added to a finalconcentration of 1 mM. Cultures were then incubated for an additional 6h at 24°C (to preclude the formation of inclusion bodies). Cells wereharvested by centrifugation (13,000 � g for 15 min at 5°C), washed byresuspension and centrifugation in 25 mM Tris-HCl (pH 7.5) buffer,homogenized in 2.5 volumes of buffer, and disrupted by sonication on ice.The extracts were clarified by high-speed centrifugation (180,000 � g for1 h at 5°C) and applied to small columns of Ni2�-nitrilotriacetic acid(NTA)-agarose. The columns were rinsed with wash buffer (50 mMTris-HCl, 0.5 M NaCl, and 20 mM imidazole, pH 8.0) to remove unboundmaterial and eluted with 0.5 M imidazole (in wash buffer) to obtain6�HisBgaA (60.4 kDa), 6�HisStrH (41.8 kDa), or 6�HisEndoD (73.8 kDa).

Rabbit antisera against keyhole limpet hemocyanin (KLH; Pierce)-conjugated 6�HisBgaA, KLH-conjugated 6�HisStrH, or 6�HisEndoD alone

TABLE 2 Primers used in this study

Primer Sequencea

ermAM-F GCGCAACGTTAAGAGTGTGTTGATAGTGCAG (AclI)ermAM-R TCCTAACGTTCATACCTAATAATTTATCTACATTCCC (AclI)rpsJ-F GGCGGAATTCTAAAAATAGTTAAAAAAATCTCTCAA (EcoRI)rpsJ-R AATAATTCTTCGCCTTTACGCATCTTTTTCTCCTTTTCGTCTATTTAAGgfp-F AAGGAGAAAAAGATGCGTAAAGGCGAAGAATTATTgfp-R CTTGAATTCTTATTATTTATAGAGTTCGTCCATTCC (EcoRI)GH2-1 GGCCGGTACCCAACCAATACGGTACTGAAATTGGTAGTGCA (KpnI)GH2-2 GCTACAAGACTCGCGCCTAAGCCCAAGACTGCACAAGTTCCAATAAGCACTGAGCGH2-3 GCTCAGTGCTTATTGGAACTTGTGCAGTCTTGGGCTTAGGCGCGAGTCTTGTAGCGH2-4 GATGGGTACCGCCACTGCAACCCGCTGCTTCTGACCACCG (KpnI)GH2-5 GGTTACCTCAAAGTGACCGACATCACGH2-6 CCGCTTGGCACGGCTGGCAGCUC19-F GACACATGCAGCTCCCGGAGACUC19-R GCGGGCAGTGAGCGCAACGCGH2-F GCGACAGAACTGATCGGATAGAGATGGH2-R GCTCTTATCAAGGGAAACTTTCTTTCCCGH20-1 CCAGGTCGACGTGGACGCTGTCATGGCTCAGGTGTCATC (SalI)GH20-2 GCTGTTCCCAGAAGAGTTATAGGCCAGATTCCAATTAGAACAGAAGCTGCTCCGH20-3 GGAGCAGCTTCTGTTCTAATTGGAATCTGGCCTATAACTCTTCTGGGAACAGCGH20-4 GCTCAAGCTTCGACCTGCACCATGAATGACCACAGGCTG (HindIII)GH20-5 GGTGCTAAGCCTGGGACAAGTCCAGTGACGH20-6 CTCCATCCCGCTCTTCTGGTAAAGH20-F CCGTGGTGGAGGTATGGGCGACCGH20-R CTTGCGTCCACCCAAACCGGH85-1 GGCCGGTACCGAAACAGTTATCTCAGGTGCTTCATGTAC (KpnI)GH85-2 CCTGCTAGGAGAAGGCCACCGAGTCCTACTAAAACAGAACAAACTCCCAAGCGH85-3 GCTTGGGAGTTTGTTCTGTTTTAGTAGGACTCGGTGGCCTTCTCCTAGCAGGGH85-4 CAACGGTACCCACGAATGGACGTTCTGGAGCTTC (KpnI)GH85-5 CGGTCGTCTTGCATTCCGTCGTATCGH85-6 GGACCGATGTCCAAACCAAGGAAGCGH85-F GGTGGAAGGGATTGACAATGAAAGGH85-R TATCCCTTTCATCAGCAAAGTAGTCa Complementary sequences in overlapping primer pairs are shown in boldface, and cleavage sites for restriction endonucleases (parentheses) are underlined.

Yang et al.

5280 aem.asm.org September 2016 Volume 82 Number 17Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

were prepared at Covance Research Products (Denver, PA) by subcuta-neous injections of the antigen in Freund’s adjuvant following standardprotocols. Dot immunoblotting was performed with bacteria from over-night HIB cultures that were spotted (6 � 107 bacteria per 3-mm spot)on nitrocellulose membranes with a Bio-Dot microfiltration apparatus(Bio-Rad). Membranes were blocked in Tris-buffered saline containing0.1% Tween 20 and 2% skim milk and then incubated with 1/1,000anti-GH antiserum that was first preabsorbed at a dilution of 1/10 withtriple mutant strain �A/H/D to remove background anti-S. gordonii an-tibody. Bound anti-GH antibody was detected using horseradish peroxi-dase-conjugated goat anti-rabbit IgG (Bio-Rad) and a metal-enhanced3,3=-diaminobenzidine (DAB) substrate kit (Life Technologies).

Growth of bacteria in saliva. The protocol used to measure plank-tonic growth of bacteria in saliva is described elsewhere (3). Briefly, wholesaliva from a single donor was periodically collected on ice, reduced withdithiothreitol to facilitate sterile filtration, heat treated (30 min at 70°C) toinactivate endogenous GH activities, and dialyzed against saliva-salt buf-fer to increase pH stability and remove possible low-molecular-massgrowth substrates. Growth studies in saliva were initiated with bacteriafrom overnight THB cultures of S. gordonii wild-type and mutant strainsthat were washed and suspended in saliva to an optical density at 640 nm(OD640) of 1 (1 � 109/ml). Primary saliva cultures were inoculated by1/1,000 dilution of these suspensions into fresh saliva and incubated over-night at 37°C. Each primary saliva culture was then diluted 1/1,000 intothree aliquots (2 ml each) of fresh saliva, and the resulting secondary salivacultures were incubated for 24 h at 37°C, which was sufficient for maxi-mum growth. Genomic DNA templates were prepared from bacteria thatwere pelleted from 1-ml aliquots of each primary or secondary salivaculture by high-speed centrifugation. Quantitative PCR (qPCR) was per-formed in triplicate on each template.

S. gordonii DL1 and mutant strain �A/H/D were also cultured, aspreviously described (18), in parallel tracks of flow cells perfused withwhole saliva at a flow rate of 7 �l/min. Flow cells were inoculated withdilute cell suspensions (OD540 of 0.05) of the two strains and incubated at37°C for 0, 4, or 18 h. At each time point, the biofilm in a replicate flow cellwas stained with a LIVE/DEAD BacLight bacterial viability kit (ThermoFisher Scientific) and examined with a Meta 510 confocal microscope(Zeiss) to compare biofilm formation. Following microscopy, the stainingsolution was drained from the flow cells and replaced with 250 �l lysisbuffer for extraction and subsequent purification of genomic DNA tem-plates for qPCR.

Biochemical analysis of saliva. Saliva from uninoculated control andsecondary growth cultures was filtered (0.22-�m-pore-size cellulose ni-trate membrane) and assayed in triplicate for protein and carbohydrate,using a bicinchoninic acid (BCA) protein assay kit (ThermoFisher) forprotein, with bovine serum albumin as the standard, and the phenol-sulfuric acid assay for carbohydrate (19), with glucose as the standard.SDS-PAGE of reduced saliva was performed as described elsewhere (3),using NuPAGE 4 to 12% Bis-Tris minigels in morpholinepropanesulfonicacid (MOPS) buffer, followed by silver staining (SilverQuest silver stain-ing kit; Invitrogen). An 70-kDa band was excised from a silver-stainedgel of control saliva and destained by following the recommendations inthe SilverQuest silver staining kit. The excised band was rinsed with 50mM bicarbonate, reduced with dithiothreitol, and alkylated with iodoac-etamide prior to overnight digestion with trypsin (Promega). Tryptic pep-tides were separated and identified by nanoscale liquid chromatographycoupled to tandem mass spectrometry (NanoLC-MS/MS) performed byApplied Biomics, Inc. (Hayword, CA). The resulting peptide mass andassociated fragmentation spectra were submitted to a GPS Explorer work-station equipped with the MASCOT search engine (Matrix Science, Lon-don, United Kingdom) to search the Swiss-Prot database. SDS-PAGE-separated salivary proteins were transferred to nitrocellulose membranesfor lectin blotting, which was performed as described elsewhere (3), usingfluorescein-labeled Sambucus nigra agglutinin (SNA; Vector Laborato-ries) at 3 �g/ml, Erythrina cristagalli agglutinin (ECA; Vector Laborato-

ries) at 2 �g/ml, succinylated wheat germ agglutinin (SWGA; VectorLaboratories) at 5 �g/ml, Galanthus nivalis agglutinin (GNA; EY Labo-ratories) at 5 �g/ml, or Aleuria aurantia lectin (AAL; Vector Laboratories)at 2 �g/ml. The methods for two-dimensional gel electrophoresis (iso-electric focusing on pI 3 to 10 strips, followed by SDS-PAGE on 4 to 12%Bis-Tris gels) have also been described (3).

RESULTSCell surface phenotypes of wild-type and �GH mutant strains.In previous studies of S. pneumoniae (6), deletion of bgaA or strHabolished the hydrolysis of 2-Np-�-Gal or 4-Np-�-GlcNAc, re-spectively, in assays performed with lysates of this streptococcus.In the present study, deletion of each gene from S. gordonii re-sulted in a threefold reduction in hydrolysis of the correspondingsubstrate in assays performed with washed HIB-grown cells. Not-withstanding contributions of other cellular enzyme activities, hy-drolysis of 2-Np-�-Gal was reduced for all mutants that lackedbgaA (strains �A, �A/H, �A/D, and �A/H/D). Likewise, hydro-lysis of 4-Np-�-GlcNAc was reduced for all mutants that lackedstrH (strains �H, �A/H, �H/D, and �A/H/D). However, the ac-tivities of certain bgaA- and strH-expressing mutants were notidentical to those of the wild type. Most notably, the BgaA activity(2-Np-�-Gal hydrolysis) of mutants that expressed bgaA butlacked endoD (strains �D and �H/D) was about half that of thewild type, and the StrH activity (4-NP-�-GlcNAc hydrolysis) ofmutants that expressed strH but lacked bgaA (strains �A and�A/D) was about 4 times that of the wild type. While the basis ofthese observations remains to be explained, it is important to notethat comparable differences in cell surface GH expression werealso seen between wild-type S. gordonii DL1 cells grown in differ-ent culture media. Cells grown in HIB (no added glucose) typi-cally had about 3 times more StrH activity than cells grown in THB(0.2% glucose). More importantly, wild-type cells grown in salivahad up to 10 times more StrH activity than cells from HIB (datanot shown). Comparable studies of cell surface EndoD activitywere not possible because colorimetric or fluorogenic substratesare not commercially available.

The presence of BgaA, StrH, and EndoD on strain DL1, as wellas the expected presence or absence of these proteins on different�GH mutant strains, was firmly established by the reactions ofGH-specific rabbit antisera with bacteria grown in HIB (Fig. 2).Importantly, the deletion of each gene abolished the binding of thecorresponding GH-specific antiserum. Reduced binding of anti-BgaA antiserum to mutant strains �D and �H/D was also noted,in accordance with the reduced BgaA activities of these strainsmentioned above.

Growth of wild-type and �GH mutant strains in saliva. Priorto growth studies in saliva, we compared the growth of S. gordoniiDL1 and the triple mutant strain �A/H/D in THB. The generation

FIG 2 Cell surface GH phenotypes of S. gordonii DL1 wild-type and �GHmutant strains determined by dot immunoblotting of bacteria with rabbitanti-BgaA, anti-StrH, or anti-EndoD antiserum.

Cell Surface Glycoside Hydrolases of S. gordonii

September 2016 Volume 82 Number 17 aem.asm.org 5281Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

times, calculated from measurements of culture turbidity duringthe mid-log phase of growth, were nearly identical (54 min foreach), as were the overnight cell densities (109/ml for each). Instriking contrast, the endpoint cell density of the triple mutant

strain �A/H/D in 24-h saliva cultures was approximately 20-foldlower than that of S. gordonii DL1 (Fig. 3A). The endpoint celldensities of mutants that lacked one or two cell surface GHs were3- to 10-fold lower than that of the wild type. Thus, the actions ofcell surface BgaA, StrH, and EndoD were required for maximumgrowth of S. gordonii DL1 in saliva.

The growth and biofilm formation of wild-type strain DL1 andmutant strain �A/H/D were also compared in parallel tracks offlow cells perfused with saliva (Fig. 4). The biofilm formation byeach strain was similar after 4 h of incubation at 37°C but wasconsistently greater for wild-type strain DL1 than for the triplemutant strain �A/H/D after 18 h of incubation. Biovolume mea-surements after 18 h (Fig. 4) revealed 3.9 times more wild-typethan mutant cells, in good agreement with the results from qPCR,which indicated a 3.5-fold difference in genomic DNA extractedfrom the parallel tracks. The lack of propidium iodide staining(red) in the mutant biofilm (Fig. 4B) is also noteworthy, as itindicates that the deletions in strain �A/H/D do not grossly affectmembrane potential or permeability.

Biochemical characterization of saliva from growth cultures.The growth of wild-type S. gordonii DL1 in secondary saliva cul-tures (Fig. 3A) was associated with a 23% loss of salivary protein asmeasured by the BCA assay (Fig. 3B) and a 46% loss of salivarycarbohydrate detected by the phenol-sulfuric acid assay (Fig. 3C).The losses of protein and carbohydrate were significantly less insaliva cultures of all �GH mutant strains, and for certain mutants,such as the triple mutant strain �A/H/D, the concentrations ofsalivary protein and carbohydrate were the same as in controlsaliva. The SDS-PAGE profiles of these samples (Fig. 3D) weregenerally similar, except for the position of a salivary componentthat migrated as a diffuse 70-kDa band in the electrophoretic pat-tern of control saliva. This component appeared at the same po-sition in the salivary profiles of mutants that expressed one or nocell surface GHs (strains �H/D, �A/D, �A/H, and �A/H/D) butwas shifted downward in the profiles of mutants that expressedtwo cell surface GHs (strains �A, �H, and �D). The same bandwas noticeably absent from the salivary profile of wild-type strainDL1, which instead contained additional bands that migrated be-

FIG 3 (A) Growth of S. gordonii DL1 wild-type and �GH mutant strainsmeasured as genomes/ml by qPCR of saliva cultures incubated for 1 day at37°C. (B to D) Filtered saliva samples from uninoculated control cultures (Sc)or cultures of S. gordonii DL1 wild-type or �GH mutant strains were assayedfor BCA-reactive salivary protein (B) and phenol-sulfuric acid-reactive sali-vary carbohydrate (C) and by SDS-PAGE (5 �g salivary protein/lane) followedby silver staining (D); the 70-kDa band was excised (white box) from theuninoculated control saliva lane for NanoLC-MS/MS. GH genotypes of allstrains are indicated below panel D.

FIG 4 Contribution of S. gordonii DL1 cell surface GHs to growth and biofilmformation in flowing saliva. Live (green)/dead (red) staining of bacteria re-vealed significantly more DL1 wild-type (A) than �A/H/D mutant (B) cells in18-h biofilms. Scale bars 30 �m.

Yang et al.

5282 aem.asm.org September 2016 Volume 82 Number 17Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

tween 40 and 60 kDa. NanoLC-MS/MS of the trypsin-digested70-kDa band, excised from one-dimensional gels of control salivaas indicated in Fig. 3D, identified the specific decapeptide RPQGGNQPQR derived from the nonglycosylated N-terminal regionsof two salivary proline-rich proteins, PRB3 and PRB4 (20), alongwith peptides from other salivary components (e.g., �-amylase,IgA heavy chain, and albumin) that migrate below PRB3 and -4 inSDS-PAGE.

Samples of control saliva and saliva from cultures of wild-typestrain DL1 and mutant strain �A/H/D were also compared bytwo-dimensional (2-D) gel electrophoresis. The most prominentdifference between these samples involved migration of basicPRGs, which appeared in the 40- to 60-kDa region of the DL1pattern, rather than in the 70- to 80-kDa region, as in saliva fromthe uninoculated control or the culture of triple mutant strain�A/H/D (Fig. 5, boxed region). A comparable shift in the appar-ent molecular mass of PRG isoforms that migrated between pI 8and 10 was also seen between different samples. Thus, basic PRGs

appeared to be major substrates for GH-dependent growth of S.gordonii DL1 in saliva.

To assess changes in PRG glycosylation, we transferred SDS-PAGE-separated saliva samples (Fig. 3D) to nitrocellulose andprobed the resulting transfers with fluorescein isothiocyanate(FITC)-labeled lectins that bind different features of N-linked gly-cans. In these experiments, PRG from control saliva appeared as aprominent 70-kDa band on blots incubated with Sia-reactiveSNA, Gal-reactive ECA, GlcNAc-reactive SWGA, Man-reactiveGNA, and Fuc-reactive AAL (Fig. 6A). Other, less-prominent lec-tin-reactive salivary components were also detected, includingone at 80 kDa and another between 50 and 60 kDa. Unlike PRGfrom control saliva, PRG from saliva cultures of S. gordonii DL1appeared in the 40- to 60-kDa region of blots and was labeled bySia-reactive SNA, GlcNAc-reactive SWGA, and Fuc-reactive AALbut not by either Gal-reactive ECA or Man-reactive GNA. In con-trast, PRG from saliva cultures of mutants that lacked bgaA (i.e.,strains �A, �A/H, �A/D, and �A/H/D) migrated as a diffuse70-kDa band and reacted strongly with all five lectins, whereasPRG from cultures of mutants that expressed bgaA (i.e., strains�H, �D, and �H/D) migrated slightly faster than PRG from con-trol saliva and reacted with all lectins except Gal-reactive ECA.

The loss of GNA binding to PRG was only seen in saliva cul-tures of S. gordonii DL1, which suggested that the actions of BgaA,StrH, and EndoD were all required for this effect. To test thishypothesis, we set up mono- and cocultures of different mutantstrains that, in combination, expressed all three enzymes. In theseexperiments (Fig. 6B), PRG from monocultures of mutant strains�H, �A/D, �D, and �A/H migrated and were detected (as in Fig.6A) by binding Man-reactive GNA and other lectins, with theexception of ECA, which failed to bind PRG from saliva cultures ofmutants that expressed cell surface BgaA (strains �H and �D). Incontrast, PRG from cocultures of strains �H plus �A/D or ofstrains �D plus �A/H migrated in the 40- to 60-kDa region, likePRG from saliva cultures of wild-type DL1, and bound SNA,SWGA, and AAL but not ECA or GNA. Thus, the loss of GNAreactivity resulted from the action first of BgaA and then those ofStrH and EndoD.

DISCUSSION

The previous observation that the number of genes for cell wall-anchored GHs was generally greater in oral streptococci that grewwell in saliva, such as S. gordonii and S. oralis, than in species thatfailed to grow suggested an important role for these enzymes insalivary glycan foraging (3). Four putative cell wall-anchored GHswere identified from S. gordonii DL1, of which three were pre-dicted to act on host glycans based on their homology with pneu-mococcal BgaA (13, 14), StrH (15, 16), and EndoD (17). In thepresent study, we introduced unmarked, in-frame deletions inthe corresponding genes and verified cell surface expression of theencoded proteins on the resulting mutant strains by their reac-tions with anti-BgaA, anti-StrH, or anti-EndoD antisera (Fig. 2).Deletion of these genes did not affect the growth of S. gordonii DL1in glucose-containing THB medium but resulted in an 20-foldreduction of cell densities in 24-h planktonic saliva cultures, aswell as an 4-fold reduction of biofilm formation in flowing sa-liva (Fig. 3 and 4). The only obvious change in the salivary pro-teome associated with the growth of S. gordonii DL1 was a reduc-tion in the apparent molecular mass of basic PRG, which wasaccompanied by the loss of lectin-binding sites for Gal-specific

FIG 5 Silver-stained 2-D gels (50 �g protein/gel) of saliva from uninoculatedcontrol culture (A), wild-type S. gordonii DL1 culture (B), and mutant strain�A/H/D culture (C). GH-dependent growth of strain DL1 was associated witha downward shift in migration of basic PRGs (boxed region).

Cell Surface Glycoside Hydrolases of S. gordonii

September 2016 Volume 82 Number 17 aem.asm.org 5283Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

ECA and Man-specific GNA. Importantly, these effects were notseen in saliva from cultures of mutant strain �A/H/D (Fig. 5 and6), which lacked all three cell surface GHs. The removal of ECAbinding sites from PRG in saliva cultures of different mutantstrains correlated with the cell surface expression of BgaA. How-ever, the subsequent loss of GNA binding sites was seen only insaliva cocultures of �GH mutant strains that in combination ex-pressed all three cell surface GHs. Considered together, these find-ings indicate that the growth of S. gordonii DL1 in saliva dependsto a considerable extent on the actions first of BgaA and then ofStrH and EndoD on N-linked glycans of basic PRG.

Insight into the structural basis of salivary glycan foraging by S.gordonii was gained from NanoLC-MS/MS of the PRG band fromcontrol saliva (Fig. 3D), which associated this component withPRB3 and PRB4. The 30-kDa protein backbones of these genet-ically distinct proline-rich basic salivary glycoproteins (20, 21)contain proline-rich tandem-repeat regions heavily glycosylatedwith N-linked and, to a lesser extent, O-linked oligosaccharides(22, 23). Twenty-seven structurally related N-linked oligosaccha-rides were previously identified from PRB3, the most abundant ofwhich was a biantennary asialo-oligosaccharide with substitutionsof three fucose residues (24). The proposed sequential cleavage of

this structure by BgaA, StrH, and EndoD of S. gordonii (Fig. 7) isconsistent with previously determined substrate specificities ofthese enzymes from S. pneumoniae, as summarized by King (5),and was demonstrated in the present study by lectin blotting ofPRG from saliva cultures of S. gordonii wild-type and �GH mu-tant strains (Fig. 6). The binding of Sia-reactive SNA to PRG fromall saliva cultures, although not expected based on the asialo struc-ture shown in Fig. 7, is readily explained by the presence of termi-nal �2-6-linked Sia on a similar oligosaccharide previously iden-tified from PRB3 (24). The fact that the growth of S. gordonii DL1does not affect SNA binding indicates that the presence of termi-

FIG 6 Lectin blots of saliva (10 �g protein/lane) from uninoculated control cultures (Sc), monocultures of S. gordonii DL1 wild-type and �GH mutant strains(A), or mono- and cocultures of �GH mutant strains (B). GH genotypes of all strains are indicated below the panels, and lectin probes used are listed betweenthe panels.

FIG 7 Proposed sequential degradation of the major N-linked glycan of PRB3(24) by BgaA, StrH, and EndoD of S. gordonii DL1.

Yang et al.

5284 aem.asm.org September 2016 Volume 82 Number 17Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

nal sialic acid protects certain N-linked glycan chains of PRG fromglycan foraging by this species, which unlike S. pneumoniae (6)and certain other viridans group streptococci (25), lacks sialidaseactivity.

ECA reacts well with both Gal�1-4GlcNAc and Fuc�1-2Gal�1-4GlcNAc termini (26, 27). Thus, the ability of all BgaA-expressing wild-type and �GH mutant strains to abolish bindingof ECA to PRG (Fig. 6) suggests cleavage of Gal from terminalGal�1-4GlcNAc to expose terminal GlcNAc�1-2Man� and cleav-age of Fuc�1-2Gal from terminal Fuc�1-2Gal�1-4GlcNAc to ex-pose terminal Fuc�1-3GlcNAc�1-2Man� (Fig. 7). TerminalGlcNAc�1-2Man� is a preferred substrate for StrH of S. pneu-moniae (15), but it remains to be determined whether this enzymeof S. gordonii also acts on the same linkage in Fuc�1-3GlcNAc�1-2Man� to expose terminal Man�1-3Man (Fig. 7). Our ability todetect the action of StrH by lectin blotting of PRG was limited bythe specificities of currently available GlcNAc- and Fuc-bindinglectin probes. Thus, GlcNAc-specific SWGA (28) has been shownto bind internal GlcNAc residues (29) and Fuc-specific AAL (30)reacts well with both �1-3- and �1-6-linked Fuc branches. Al-though not revealed by lectin blotting, the action of StrH wasclearly required for the subsequent removal of GNA-bindingstructures from PRG in cocultures of mutants that expressed thisenzyme in combination with BgaA and EndoD (Fig. 6B).

The ability of GNA to detect sequential action of BgaA, StrH,and EndoD on PRG undoubtedly depends on the specificity of thislectin for terminal Man residues in branched oligosaccharides(31) like those that decorate PRB3 (Fig. 7). This finding does not,however, imply that all GNA-reactive glycans of PRG are actedupon by EndoD. Indeed, binding of GNA to PRG from controlsaliva was not abolished by the growth of any mutant that ex-pressed EndoD (i.e., strains �A, �H, and �A/H) (Fig. 6). A pos-sible explanation for this puzzling result comes from the previousidentification of complex Man4- and Man5-containing oligosac-charides from PRB3 (24). Such structures may bind GNA but notbe substrates of EndoD, which has a strict specificity for the oligo-mannosyl cores of glycoproteins (32). The removal of these high-mannose-type glycans from PRG during the growth of S. gordoniiDL1 can be explained by the actions of BgaA and StrH on adjacentbranches of the same oligosaccharide to expose �1-3-linkedterminal Man, which is essential for the glycan recognition andsubsequent cleavage by EndoD of S. pneumoniae (33). A previ-ously described mannosidase activity of S. gordonii and otherviridans group streptococci (34) that acts on Man5 glycoformsmay also contribute to the removal of high-mannose-type gly-cans from PRG.

The sequential actions of BgaA, StrH, and EndoD on N-linkedglycans of PRG are predicted to yield free Gal and GlcNAc, both ofwhich support the growth of S. gordonii DL1 (3). As discussedabove, the products of BgaA and EndoD are also likely to includefucosylated oligosaccharides. It is noteworthy that S. gordoniiDL1, which does not grow on L-fucose, has genes for two putative�-L-fucosidases (i.e., SGO_0150 and SGO_1771) that are pre-dicted to be cytoplasmic proteins. Recombinant proteins encodedby these genes were recently identified, respectively, as �1-2- and�1-3- and -4-specific �-L-fucosidases, based on their abilities tocleave fucosylated oligosaccharides (E. Guthrie, unpublisheddata). Thus, SGO_0150, which occurs as an isolated gene in S.gordonii DL1, may act on Fuc�1-2Gal that is cleaved from PRG bythe action of BgaA. Moreover, SGO_1771, which occurs in the

previously described gom regulon (35) along with genes for astructurally characterized N-acetylglucosaminidase (36) andthree putative mannosidases, may function in the metabolism oflarger, Man-containing branched oligosaccharides that resultfrom the action of EndoD. Indeed, inactivation of certain genes inthe gom locus was found to reduce the growth of S. gordonii onboth �1-6-D-mannobiose and GlcNAc�1-2Man (35). Furtherstudies are needed to determine whether oligosaccharides releasedfrom basic PRG by cell surface GHs during the growth of S. gor-donii DL1 in saliva are natural substrates of the intracellular gly-coside hydrolases encoded by genes in the gom locus and at otherloci.

ACKNOWLEDGMENTS

This work was supported by the Intramural Research Program of the NIH,NIDCR.

We gratefully acknowledge the Glycobiology Group of New EnglandBioLabs for performing fucosidase assays and John Thompson and An-dreas Pikis for extensive help in the preparation of recombinant glycosidehydrolases, and Y.Z. thanks Xuedong Zhou of the State Key Laboratory ofOral Diseases, Sichuan University, Chengdu, China, for support.

FUNDING INFORMATIONThis work, including the efforts of John O. Cisar, was funded by HHS |National Institutes of Health (NIH).

This work was supported the Intramural Research Program of the NIH,NIDCR.

REFERENCES1. De Jong MH, Van der Hoeven JS. 1987. The growth of oral bacte-

ria on saliva. J Dent Res 66:498 –505. http://dx.doi.org/10.1177/00220345870660021901.

2. Wickström C, Hamilton IR, Svensäter G. 2009. Differential metabolicactivity by dental plaque bacteria in association with two preparations ofMUC5B mucins in solution and in biofilms. Microbiology 155:53– 60.http://dx.doi.org/10.1099/mic.0.022111-0.

3. Zhou Y, Yang J, Zhang L, Zhou X, Cisar JO, Palmer RJ, Jr. 2016.Differential utilization of basic proline-rich glycoproteins during growthof oral bacteria in saliva. Appl Environ Microbiol 82:5249 –5258. http://dx.doi.org/10.1128/AEM.01111-16.

4. Ton-That H, Marraffini LA, Schneewind O. 2004. Protein sorting to thecell wall envelope of Gram-positive bacteria. Biochim Biophys Acta 1694:269 –278. http://dx.doi.org/10.1016/j.bbamcr.2004.04.014.

5. King SJ. 2010. Pneumococcal modification of host sugars: a major con-tributor to colonization of the human airway? Mol Oral Microbiol 25:15–24. http://dx.doi.org/10.1111/j.2041-1014.2009.00564.x.

6. King SJ, Hippe KR, Weiser JN. 2006. Deglycosylation of human glyco-conjugates by the sequential activities of exoglycosidases expressed byStreptococcus pneumoniae. Mol Microbiol 59:961–974. http://dx.doi.org/10.1111/j.1365-2958.2005.04984.x.

7. Burnaugh AM, Frantz LJ, King SJ. 2008. Growth of Streptococcus pneu-moniae on human glycoconjugates is dependent upon the sequential ac-tivity of bacterial exoglycosidases. J Bacteriol 190:221–230. http://dx.doi.org/10.1128/JB.01251-07.

8. Yanisch-Perron C, Vieira J, Messing J. 1985. Improved M13 phagecloning vectors and host strains: nucleotide sequences of the M13mp18and pUC19 vectors. Gene 33:103–119. http://dx.doi.org/10.1016/0378-1119(85)90120-9.

9. Lunsford RD, London J. 1996. Natural genetic transformation in Strep-tococcus gordonii: comX imparts spontaneous competence on strainwicky. J Bacteriol 178:5831–5835.

10. Hsu SD, Cisar JO, Sandberg AL, Kilian M. 1994. Adhesive properties ofviridans group streptococcal species. Microb Ecol Health Dis 7:125–137.http://dx.doi.org/10.3109/08910609409141342.

11. Wu C, Ton-That H. 2010. Allelic exchange in Actinomyces oris withmCherry fluorescence counterselection. Appl Environ Microbiol 76:5987–5989. http://dx.doi.org/10.1128/AEM.00811-10.

Cell Surface Glycoside Hydrolases of S. gordonii

September 2016 Volume 82 Number 17 aem.asm.org 5285Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from

12. Lunsford RD. 1995. A Tn4001 delivery system for Streptococcus gordonii(Challis). Plasmid 33:153–157. http://dx.doi.org/10.1006/plas.1995.1016.

13. Singh AK, Pluvinage B, Higgins MA, Dalia AB, Woodiga SA, Flynn M,Lloyd AR, Weiser JN, Stubbs KA, Boraston AB, King SJ. 2014. Unrav-elling the multiple functions of the architecturally intricate Streptococcuspneumoniae beta-galactosidase, BgaA. PLoS Pathog 10:e1004364. http://dx.doi.org/10.1371/journal.ppat.1004364.

14. Zahner D, Hakenbeck R. 2000. The Streptococcus pneumoniae beta-galactosidase is a surface protein. J Bacteriol 182:5919 –5921. http://dx.doi.org/10.1128/JB.182.20.5919-5921.2000.

15. Clarke VA, Platt N, Butters TD. 1995. Cloning and expression of thebeta-N-acetylglucosaminidase gene from Streptococcus pneumoniae. Gen-eration of truncated enzymes with modified aglycon specificity. J BiolChem 270:8805– 8814. http://dx.doi.org/10.1074/jbc.270.15.8805.

16. Jiang YL, Yu WL, Zhang JW, Frolet C, Di Guilmi AM, Zhou CZ, VernetT, Chen Y. 2011. Structural basis for the substrate specificity of a novelbeta-N-acetylhexosaminidase StrH protein from Streptococcus pneu-moniae R6. J Biol Chem 286:43004 – 43012. http://dx.doi.org/10.1074/jbc.M111.256578.

17. Abbott DW, Macauley MS, Vocadlo DJ, Boraston AB. 2009. Strep-tococcus pneumoniae endohexosaminidase D, structural and mechanis-tic insight into substrate-assisted catalysis in family 85 glycoside hy-drolases. J Biol Chem 284:11676 –11689. http://dx.doi.org/10.1074/jbc.M809663200.

18. Palmer RJ, Jr, Kazmerzak K, Hansen MC, Kolenbrander PE. 2001.Mutualism versus independence: strategies of mixed-species oral biofilmsin vitro using saliva as the sole nutrient source. Infect Immun 69:5794 –5804. http://dx.doi.org/10.1128/IAI.69.9.5794-5804.2001.

19. Dubois M, Gilles KA, Hamilton JK, Rebers PA, Smith F. 1956. Color-imetric method for determination of sugars and related substances. AnalChem 28:350 –356. http://dx.doi.org/10.1021/ac60111a017.

20. Lyons KM, Azen EA, Goodman PA, Smithies O. 1988. Many proteinproducts from a few loci: assignment of human salivary proline-rich pro-teins to specific loci. Genetics 120:255–265.

21. Kim HS, Lyons KM, Saitoh E, Azen EA, Smithies O, Maeda N. 1993.The structure and evolution of the human salivary proline-rich pro-tein gene family. Mamm Genome 4:3–14. http://dx.doi.org/10.1007/BF00364656.

22. Carpenter GH, Proctor GB. 1999. O-linked glycosylation occurs on basicparotid salivary proline-rich proteins. Oral Microbiol Immunol 14:309 –315. http://dx.doi.org/10.1034/j.1399-302X.1999.140507.x.

23. Oho T, Rahemtulla F, Mansson-Rahemtulla B, Hjerpe A. 1992. Purifi-cation and characterization of a glycosylated proline-rich protein fromhuman parotid saliva. Int J Biochem 24:1159 –1168. http://dx.doi.org/10.1016/0020-711X(92)90387-G.

24. Gillece-Castro BL, Prakobphol A, Burlingame AL, Leffler H, Fisher SJ.1991. Structure and bacterial receptor activity of a human salivary proline-rich glycoprotein. J Biol Chem 266:17358 –17368.

25. Kilian M, Mikkelsen L, Henrichsen J. 1989. Taxonomic study of viridansstreptococci: description of Streptococcus gordonii sp. nov. and emended

descriptions of Streptococcus sanguis (White and Niven, 1946), Streptococ-cus oralis (Bridge and Sneath, 1982), Streptococcus mitis (Andrewes andHorder, 1906). Int J Syst Bacteriol 39:471– 484. http://dx.doi.org/10.1099/00207713-39-4-471.

26. Blixt O, Head S, Mondala T, Scanlan C, Huflejt ME, Alvarez R, BryanMC, Fazio F, Calarese D, Stevens J, Razi N, Stevens DJ, Skehel JJ, vanDie I, Burton DR, Wilson IA, Cummings R, Bovin N, Wong CH,Paulson JC. 2004. Printed covalent glycan array for ligand profiling ofdiverse glycan binding proteins. Proc Natl Acad Sci U S A 101:17033–17038. http://dx.doi.org/10.1073/pnas.0407902101.

27. Teneberg S, Angström J, Jovall PA, Karlsson KA. 1994. Characterizationof binding of Gal beta 4GlcNAc-specific lectins from Erythrina cristagalliand Erythrina corallodendron to glycosphinogolipids. Detection, isolation,and characterization of a novel glycosphinglipid of bovine buttermilk. JBiol Chem 269:8554 – 8563.

28. Monsigny M, Roche AC, Sene C, Maget-Dana R, Delmotte F. 1980.Sugar-lectin interactions: how does wheat-germ agglutinin bind sialogly-coconjugates? Eur J Biochem 104:147–153. http://dx.doi.org/10.1111/j.1432-1033.1980.tb04410.x.

29. Goldstein IJ, Hammarström S, Sundblad G. 1975. Precipitation andcarbohydrate-binding specificity studies on wheat germ agglutinin.Biochim Biophys Acta 405:53– 61. http://dx.doi.org/10.1016/0005-2795(75)90313-X.

30. Iskratsch T, Braun A, Paschinger K, Wilson IB. 2009. Specificity analysisof lectins and antibodies using remodeled glycoproteins. Anal Biochem386:133–146. http://dx.doi.org/10.1016/j.ab.2008.12.005.

31. Shibuya N, Goldstein IJ, Van Damme EJ, Peumans WJ. 1988. Bindingproperties of a mannose-specific lectin from the snowdrop (Galanthusnivalis) bulb. J Biol Chem 263:728 –734.

32. Koide N, Muramatsu T. 1974. Endo-beta-N-acetylglucosaminidase act-ing on carbohydrate moieties of glycoproteins. Purification and propertiesof the enzyme from Diplococcus pneumoniae. J Biol Chem 249:4897– 4904.

33. Tai T, Yamashita K, Ogata-Arakawa M, Koide N, Muramatsu T,Iwashita S, Inoue Y, Kobata A. 1975. Structural studies of two ovalbuminglycopeptides in relation to the endo-beta-N-acetylglucosaminidase spec-ificity. J Biol Chem 250:8569 – 8575.

34. Homer KA, Roberts G, Byers HL, Tarelli E, Whiley RA, Philpott-Howard J, Beighton D. 2001. Mannosidase production by viridans groupstreptococci. J Clin Microbiol 39:995–1001. http://dx.doi.org/10.1128/JCM.39.3.995-1001.2001.

35. Kilic AO, Tao L, Zhang Y, Lei Y, Khammanivong A, Herzberg MC.2004. Involvement of Streptococcus gordonii beta-glucoside metabolismsystems in adhesion, biofilm formation, and in vivo gene expression. JBacteriol 186:4246 – 4253. http://dx.doi.org/10.1128/JB.186.13.4246-4253.2004.

36. Langley DB, Harty DW, Jacques NA, Hunter N, Guss JM, Collyer CA.2008. Structure of N-acetyl-beta-D-glucosaminidase (GcnA) from the en-docarditis pathogen Streptococcus gordonii and its complex with the mech-anism-based inhibitor NAG-thiazoline. J Mol Biol 377:104 –116. http://dx.doi.org/10.1016/j.jmb.2007.09.028.

Yang et al.

5286 aem.asm.org September 2016 Volume 82 Number 17Applied and Environmental Microbiology

on January 5, 2021 by guesthttp://aem

.asm.org/

Dow

nloaded from