University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

U.S. Department of Veterans Affairs Staff Publications U.S. Department of Veterans Affairs

12-7-1999

Bacterial diversity within the human subgingival crevice

Ian Kroes Stanford University School of Medicine

Paul W. Lepp Stanford University School of Medicine, [email protected]

David A. Relman Stanford University School of Medicine, [email protected]

Follow this and additional works at: https://digitalcommons.unl.edu/veterans

Kroes, Ian; Lepp, Paul W.; and Relman, David A., "Bacterial diversity within the human subgingival crevice" (1999). U.S. Department of Veterans Affairs Staff Publications. 18. https://digitalcommons.unl.edu/veterans/18

This Article is brought to you for free and open access by the U.S. Department of Veterans Affairs at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in U.S. Department of Veterans Affairs Staff Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. Bacterialdiversity within the human subgingivalcrevice

Ian Kroes, Paul W. Lepp, and David A. Relman* Departmentsof Microbiologyand Immunology,and Medicine,Stanford University School of Medicine,Stanford, CA 94305, and VeteransAffairs Palo Alto HealthCare System, Palo Alto,CA 94304 Editedby Stanley Falkow, Stanford University, Stanford, CA, and approvedOctober 15, 1999(received for review August 2, 1999) Molecular, sequence-based environmental surveys of microorgan- associated with disease (9-11). However, a directcomparison isms have revealed a large degree of previously uncharacterized between cultivation-dependentand -independentmethods has diversity. However, nearly all studies of the human endogenous not been described. In this study,we characterizedbacterial bacterial flora have relied on cultivation and biochemical charac- diversitywithin a specimenfrom the humansubgingival crevice terization of the resident organisms. We used molecular methods byusing both methods. Our resultsreveal a significantlybroader to characterize the breadth of bacterial diversitywithin the human diversityof bacterial16S rDNA sequence types(phylotypes) by subgingival crevice by comparing 264 small subunit rDNA se- using the cultivation-independentapproach, although each quences from 21 clone libraries created with products amplified method identifiedpreviously uncharacterized phylotypes and directlyfrom subgingival plaque, with sequences obtained from should be viewed as complementary. that were cultivated from the same specimen, as well as with sequences available in public databases. The majority(52.5%) Materialsand Methods of the directlyamplified 16S rRNAsequences were <99% identical SpecimenCollection. Subgingival plaque was collected fromthe to sequences within public databases. In contrast, only 21.4% of mesial surfaceof tooth 3 and the mesial and distal surfacesof the sequences recovered from cultivated bacteria showed this tooth30 of a 39-year-oldCaucasian male, in accordance witha degree of variability.The 16S rDNA sequences recovered by direct protocol approved by the StanfordAdministrative Panel on amplification were also more deeply divergent; 13.5% of the Human Subjects in Medical Research. The subjectwas a non- amplified sequences were more than 5% nonidentical to any smokerwith mild gingivitis and no recordof antibioticuse during known sequence, a level of dissimilarity that is often found the previous3 months.The plaque materialwas dispersedin 1.3 between members of differentgenera. None of the cultivated ml of reduced Treponeme broth(Anaerobe Systems,San Jose, sequences exhibited this degree of sequence dissimilarity.Finally, CA). The vial was flushedwith an anaerobic gas mixture(80% k2 directamplification of 16S rDNAyielded a more diverse view of the N/10% H2/10% CO2) and was maintainedon ice for1 hrbefore 0 subgingival bacterial flora than did cultivation. Our data suggest processing. -j do that a significantproportion of the resident human bacterial flora a remain poorly characterized, even within this well studied and Cultivationand PhenotypicAnalysis of Bacteriafrom Subgingival 2 familiar microbial environment. Specimen.The plaque suspensionwas diluted 1:2 and 1:20 in reduced Treponeme broth.A 100-,ulaliquot of the 1:2 dilution The endogenousbacterial flora in man is thoughtto playa role was examinedby dark field microscopyimmediately and after in nutrition,carcinogenesis, and resistanceto colonizationby 24-hrincubation for the presence of spirochetes.The diluted pathogens,and to harborpotential opportunistic pathogens (1). An sampleswere used to inoculate a wide range of routineclinical accurate understandingof these roles, and the nature of the microbiologicalmedia (see supplementalmaterial on the PNAS interactionsamong and betweenindividual members and thehost, web site,www.pnas.org). Isolates were chosen forfurther char- requires as a firststep knowledgeof the compositionof the acterizationbased on differencesin colonymorphology and were microbialcommunity. Many studiesof environmentalmicrobial identifiedby using routineclinical microbiologicalassays (see communitieshave demonstratedthe limitationsof cultivation- supplementalmaterial). dependentmethods in determiningcommunity composition. En- vironmentalsurveys based on acquisitionof phylogenetically useful SpecimenDigestion and DNAExtraction. An undiluted100-,ul ali- microbialsequences (2-4) suchas thatof the 16S rRNA gene (16S quot of the suspended plaque materialwas added to an equal rDNA) have revealed a great deal of previouslyunsuspected volume of lysis buffer[400 ,ug/mlproteinase K (Boehringer bacterialand archaeal diversity.In mostinstances, the cultivated Mannheim)/2% Laureth-12 (Mazer Chemicals, Gurnee, IL)/ membersrepresent <1% of the total extantpopulation. Broad 100 mM Tris-HCl, pH 8.5/2 mM EDTA] and was incubated range small subunitrDNA PCR methods have also revealed overnightat 554C.The digestwas thensonicated for 2 minat 120 cultivation-resistantpathogens in disease settings(5). W in a bath sonicator(cell disrupterW185; Branson). One half Despite the limitationsof thisapproach, most surveysof the ofthe digested sample was purifiedand precipitatedwith phenol, human endogenous bacterial flora have relied on cultivation. chloroform,and ethanol by using standardprotocols. The re- Nonetheless,nearly 500 bacterial strainshave been recovered mainderof the digestwas centrifugedat 1,200 x g for2 min,and fromthe subgingival crevice, a particularlywell studiedmicrobial the supernatantwas used directlyin PCR. niche. Many of these strainsare thoughtto be commensals,and a smallernumber, opportunistic pathogens (6, 7). Local disease, including dental caries, gingivitis, and periodontitis, has This paper was submitted directly(Track II) to the PNAS office. promptedmost examinationsof the oral bacterialflora. These Data deposition: The sequences reported inthis paper have been deposited in the GenBank diseases are associated with changes in both local bacterial database (accession nos. AF201965-AF202029). densityand species composition(8). *To whom reprintrequests should be addressed at: Veterans AffairsPalo Alto Health Care Few efforts System 154T, 3801 Miranda Avenue, Palo Alto, CA 94304. E-mail: relman@cmgm. have been undertakento characterizehuman stanford.edu. endogenousmicrobial communities using broad-based molecu- The publication costs of this article were defrayed in part by page charge payment. This lar methods.A smallnumber of 16S rDNA sequences have been article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. characterizedafter direct amplificationfrom oral specimens ?1734 solely to indicate this fact.

PNAS I December7, 1999 | vol. 96 | no. 25 | 14547-14552

Proceedings of the National Academy of Sciences of the United States of America,Vol. 96, No. 25 (Dec. 7, 1999), pp. 14547-14552 Table 1. Oligonucleotideprimers and probe Sequences with ?99% identitywere considered as a single Primers/probe* Sequence (5' -- 3') Ref. phylotype.A single representativeclone fromeach phylotype was chosen forfurther phylogenetic analysis. SDBactO338aA18t ACTCCTACGGGAGGCAGC 12 The phylogeneticassociations of all representativesequences SDBactO515aA19 GTGCCSGCMGCCGCGGTAA 13t were determinedfrom 100 bootstrappedreplicates (18) byusing SDBactO515aS19 TTACCGCGGCMGCSGGCAC 13t a maximum-likelihoodalgorithm (19). These associationswere SDBactl371 aS20 AGGCCCGGGAACGTATTCAC 14 confirmedby using a least-squares fit (20) of evolutionary SDBactl 525aS17 AAGGAGGTGATCCAGCC 15$ distances(with Jukes-Cantor correction) and byusing parsimony SDBactl 529aS17 YAKAAAGGAGGTGWTCC Thisstudy algorithms.More detailed phylogeneticanalyses based on 864 SGTrepOO08aA20 AGAGTTTGATCMTGGCTCAG 9 and 894 masked positions of Low GC Gram Positive and SGStrpl264aS22? AGAGATTAGCTTGCCGTCACCG Thisstudy Actinobacteriasequences, respectively, were performed by using the same methods.Coverage of clone librarieswas calculated *Primersand probesare namedaccording to OligonucleotideProbe Database convention(16). The namesincorporate the 5' terminalposition (E. coli 16S accordingto the methodof Good (21), and an estimateof the rRNAnumbering convention) and primerlength in nucleotides. numberof unseen species was obtained by using a parametric tOrientationindicates PCR primer sequence. Reverse complement sequence model (22). used forin situprobing. tModifiedfrom indicated reference. In SituHybridization and Enumeration.A 50-,ulaliquot of subgin- ?Streptococcus-specificprobe. givalplaque suspensionwas fixedand simultaneouslyhybridized with33 ng/plfluorescently labeled S-G-Strp-1264-a-S-22and 1.6 ng/Alfluorescently labeled S-D-Bact-0338-a-S-18as described PCRAmplification. PCR was performedwith broadly conserved (23), but omittingsodium borohydride. Probe S-G-Strp-1264-a- bacterial 16S rDNA primersor spirochete-specific16S rDNA S-22 was designedfrom conserved regions of Streptococcusspp. primers(Table 1). Broad range primerpairs were ("A") S-D- 16S rDNA sequences in the RDP smallsubunit aligned database Bact-0338-a-A-18 + S-D-Bact-1525-a-S-17; ("B") S-D-Bact- v3.0 (24) and GenEMBL database, as well as fromStreptococcus 0515-a-A-19 + S-D-Bact-1371-a-S-20;("C") S-D-Bact-0515-a- spp. 16S rDNA sequences found fromthis study.Probe S-G- A-19 + S-D-Bact-1525-a-S-17;and ("D") S-D-Bact-0515-a-A-l9 Strp-1264-a-S-22was synthesizedwith 5'-Aminolink 2 (Applied + S-D-Bact-1529-a-S-17.A 1-l aliquot of extractedDNA or Biosystems)and was labeled by using the FluoReporterTexas digestsupernatant was added to 99 plIof PCR reactionmix (10 Red-X Oligonucleotide Amine Labeling Kit (Molecular mM Tris HCl, pH 8.5/2mM MgCl2/50mM KCl/200 AM ofeach Probes). The conserved bacterial probe S-D-Bact-0338-a-S-18 dNTP/20 pmol of each primer).The reactionwas heated at 940C was synthesizeddirectly with the FITC-labeled phosphoramidite for 3 min before the addition of 2.5 units of AmpliTaq DNA 6-FAM (401527, Applied Biosystems).Samples were examined polymerase(Perkin-Elmer). 16S rDNA geneswere amplified for on an Olympus (New Hyde Park, NY) IX-70 microscope,and 30 or 40 cycles;each cycleconsisted of 940C (30 s), 550C (30 s), images were collected by using the DeltaVision deconvolution and 720C (30 s), with the exceptionof PCRs using S-D-Bact- system(Applied Precision,Seattle). A series of 32 slices were 1529-a-S-17,for which an annealing temperatureof 500C was collected every0.2 Ktmalong the z axis and was projected by used. The sensitivityof PCRs withprimer pair "B" was deter- usingthe maximumintensity algorithm to forma singleimage. mined by amplifyingserial dilutionsof cloned phylotypeA-38 Streptococcusspp. were enumeratedin 12 representativefields at 16S rDNA in water. x 1,000 magnification.

Cloningof Amplified16S rRNAGenes and Sequencing.Amplicons Results were ligatedinto pCR 2.1 (Invitrogen).One halfof the product DirectAmplification of 16S rDNAfrom Subgingival Plaque. Initial from each ligation reaction was restrictedwith Bsu36I (New amplificationand cloningwith four broad rangebacterial primer England BioLabs) (see Results). ToplOF' (In- pairs (see supplementalmaterial) revealed that46 of 50 ampli- vitrogen)was transformedwith either restricted or unrestricted fied 16S rRNA genes were of streptococcalorigin. This high plasmid DNA. 16S rDNA was amplifiedand purifiedas previ- percentageof streptococcal16S rDNA appeared to conflictwith ously described (17) from randomlychosen E. coli ToplOF' reportedmicroscopic counts in which streptococcicomprised clones and all cultivatedisolates with unique phenotypes.16S -30% of plaque-associated bacterial communities(25). The rDNA was sequenced by using the Prism AmpliTaq FS relativeabundance of Streptococcusspecies to total bacteria in DyeDeoxy terminatorcycle sequencing kit and a 373A se- the original sample was evaluated by using a fluorescently quencer (Applied Biosystems).GenBank database accession labeled Streptococcus-specificprobe and a nonspecificbacterial numbersare AF201965-AF202029. probe (S-D-Bact-0338-a-S-18).Streptococcus spp. comprised79 of 268 (29.5%) bacteriain thisspecimen as determinedby in situ PhylogeneticAnalysis. After determiningpreliminary plhyloge- hybridization(Fig. 1). neticassociations, the integrity of amplifiedsequences displaying The over-representationof Streptococcusspp. was reduced in <99% identityto database sequenceswas confirmedby using the subsequentclone librariesby restriction of plasmidswith Bsu36J CHECKCHIMERA programof the Ribosomal Database Project. beforetransformation. This restrictionendonuclease recognizes Chimeras were defined as sequences in which subfragments a site presentin the 16S rDNA of nearlyall Streptococcusspp. possessed a highersimilarity to database sequences otherthan Transformantcolony counts were reduced by >7-fold after thedatabase sequence mostsimilar to theentire query sequence. Bsu36Jrestriction of equivalentamounts of DNA. Having solved Initial alignmentof amplifiedsequences was performedby the problemof over-representationof theStreptococcus spp., we using the automated 16S rRNA sequence aligner of the ARB created 21 clone librariesfrom 15 separate PCR amplifications softwarepackage (Technical Univ. of Munich, Munich, Ger- of rDNA fromdigested subgingival plaque (see supplemental many) againsta database of 7,916 complete and partial rRNA material);245 randomlychosen clones from14 Bsu361-restricted sequences (March 1997 ARB small subunit rRNA database librariesand 25 clones from 7 unrestrictedlibraries were se- release). Ambiguouslyand incorrectlyaligned positionswere quenced. Six of the 270 amplifiedsequences were chimerasand aligned manuallyon the basis of conserved primarysequence were removedfrom subsequent analyses. and secondary structure.Similarity matrices were generated The percent sequence identityamong the 264 remaining from 516 to 961 masked (unambiguouslyaligned) positions. clones was determinedby comparisonof 489-690 homologous

14548 | www.pnas.org Kroes et a!. by 14 of 124 clones (11.3%) in librariescreated with primer pair "A" and 5 of52 clones (9.6%) in librariescreated by using primer pair "D." None of the Capnocytophagasequences within the ARB 16S rDNA database contained mismatcheswith the S-D-Bact- 0515-a-A-19primer. Similarly, Actinomyces spp. were observed frequentlyby using primer pair "C" (20 of 86 clones) butnot with primerpair "D" (1 of 52 clones). There were no mismatches between the Actinomycessequences and the S-D-Bact-1529-a- S-17 primer.

Comparisonof Cultivatedand AmplifiedPhylotypes. Cultivation methodsyielded 56 isolates fromsubgingival plaque thatwere distinctbased on antigenicor biochemicaltraits; 37.5% of these isolatescould notbe definitivelyidentified at thegenus level, and 55% could not be identifiedat the species level, using conven- tional phenotypicmethods. Sequence analysisof 489-690 ho- mologous positions from the 16S rDNA of these 56 isolates indicated28 unique phylotypesbased on our 1% identitydefi- nition(Figs. 2-4). The phylotypesrecovered by amplificationand cloningexhib- ited greaterdivergence and diversitythan phylotypes recovered bycultivation. The use of a strictdefinition of species boundaries (?97% sequence identity)(36) did notalter this finding. The 16S rDNA from 22 of the 28 cultivated isolates showed ?99% identityto sequenceswithin public databases. Conversely,52.5% of the 59 amplifiedphylotypes had not been previouslycharac- terized (<99% identity)by 16S rDNA analysis(Table 2). Fifty percentof the cultivatedphylotypes were also presentwithin the Fig.1. Streptococcusspp. (yellow)in a thinsmear of the gingivalspecimen 59 phylotypesrecovered by direct,broad-range amplification revealedby in situ hybridizationwith a Streptococcus-specificTexas Red- whereas only 24% of the directlyamplified phylotypes were labeledoligonucleotide probe (red) and a broadrange bacterial FITC-labeled recoveredby cultivation. Although cultivation did not reveal the oligonucleotideprobe (green). (x500.) same degree of divergenceas amplification,selected groups of 0

bacteriawere more readilyrecovered by cultivation.Cultivation 0 0 to 16S rDNA positionsbetween members of monophyleticgroups. recovered four Prevotellaphylotypes and two phylotypesof 0 Sequences differingby <1% were consideredto be representa- Staphylococcus,Rothia dentocariosa and Micrococcus luteus, tiveof a singlephylotype. This phylotypedefinition was chosen whereasonly a singlePrevotella phylotype and none of the other were detected direct afterevaluating the heterogeneityamong multiple copies of 16S organisms by amplification. Approximately48% of the 77 phylotypes(59 amplifiedand 28 rRNA genes in 6 fullysequenced bacterialgenomes containing cultivated)were previouslyuncharacterized sequences (<99% multiplerrn operon copies: SynechocystisPCC 6803, 0% among identicalto anyof the nearly 5,000 bacterial 16S rRNA sequences 2 operons; Haemophilusinfluenzae, 0% among 6 operons; Hel- >1,000 nucleotides in length within the ARB database). Se- icobacterpylori,<0.1% sequence dissimilarityamong 3 operons; quences obtained by direct amplificationand cloning repre- Treponemapallidum, 0.06% among 2 operons; Bacillus subtilis, sented -83% of these novel phylotypeswhereas -10.5% were ?0.6% among 10 operons; and E. -<1.1% among7 operons. coli, discoveredonly by cultivation; the remaining 6.5% wererevealed the diversityfound This definitionmay in factunder-represent by both approaches (Figs. 2-4). Many of the novel phylotypes among closely related taxa. Based on our definition,the 264 withthe lowest similarity to publicdatabase sequences fellwithin clones represented59 unique phylotypes(Figs. 2-4). These 59 the Low GC Gram PositiveClostridia and Actinobacteriadivi- phylotypesconstitute -88% of all sequences withinthe cloned sions (Figs. 3 and 4). population[(number of phylotypesobserved once/total number The Clostridia group of Low GC Gram Positives displays of observations)x 10] (21). The remaining12% of the cloned extensivephenotypic and phylogeneticdiversity (26). In our populationcould potentiallybe composed of hundredsof addi- study,the majorityof phylotypeswithin this group were recov- tional unique phylotypes.The numberof unique unseen species ered only by amplification(Fig. 3). Among the 15 phylotypes maybe estimatedby using a parametricmodel thatexamines the identified,9 were previouslyuncharacterized by 16S rDNA numberof phylotypesobserved only once versusthose observed analysis,and 5 of these displayed<93% sequence identitywith multipletimes (At = nit - n2t2+ n3t3..... ; where At is the organismswithin the ARB database and to each other.Phylotype expectednumber of newphylotypes, t is thenumber of additional A-35 exhibitedthe greatest divergence among the clone libraries: sample sets and n, is the numberof phylotypesappearing x times 89.0% sequence identitywith its closest relative, a ruminal in the originalsample set). Based on methodsfor estimating the bacteriumsequence (GenBank accession no. AF001743). Sim- numberof unseen species (22), these 59 phylotypesconstitute ilarly,A-36 (not shown) displayed89.4% sequence identitywith <70% ofthe total predicted number of unique phylotypeswithin a differentruminal bacterium sequence (GenBank accession no. the clone libraries. AF001716). A-38 and A-39 were 98% identicalto each otherand The creation of 15 individual libraries by using different 93% identicalto theirclosest relative,Schwartzia succinivorans. methodsallowed us to examine the effectof DNA purification S. succinivoransis an obligatesuccinate fermenter that has been and primer pairs on librarycomposition. Phenolchloroform isolated from the rumen of a cow. Veillonellaspp. are early extractionof DNA had no significanteffect. Primer selection was plaque colonizers(27) and are instrumentalin the formationof associated with several notable differences:Capnocytophaga the matrixto which late colonizers such as Selenomonasand were representedby only 1 of 86 clones (1.2%) in libraries Eubacteriumadhere. The A-44 sequence is identical to one createdby using primer pair "C" whereasthey were represented recentlycloned froma dentoalveolarabcess (11). Finally,A-33,

Kroes et al. PNAS I December 7, 1999 | vol. 96 | no. 25 | 14549 AC-31 Streptococcus mitis AC-24 Streptococcus oralis A-28 Streptococcus gordonii A-30 Streptococcus sanguis AC-25 Streptococcus parasanguis A-26 A-27 C-23 Streptococcus salivarius C29 Streptococcus mutans A-21 Abiotrophiadefectivus 1 A-22 Abiotrophiaadjacens Low GC Gram Gemella morbillorum Positives Al 7 Gemella haemolysans A-18 0-19 Staphylococcus epidermidis C-20 Staphylococcus haemolyticus

Clostridia-group

AC-59 Fusobacterium nucleatum Leptotrichiabucalis Fusobacteria AC-60 |15 Actinobacteria ] Actinobacteria AC-48 parainfluenzae { A-49I A4Gamma A-50 A-51 Thiobacillus cuprinus A-52 Lautropia mirabilis C-53 Neisseria flavescens Beta Proteobacteria rCAC-54 Neisseria macaca AC-55 Kingella oralis A-57 A-56 A-58 hominis A-45 Campylobacter gracilis A-46 Campylobacter rectus -47 Campylobacter concisus CmpylobacterjejuniI Epsilon Proteobacteria I _ ~~~~~~~Rickettsiarickettsdii J Alpha Proteobacteria C-64 Prevotella nigrescens Prevotella oris I~C-65 C-66 Prevotella denticola Prevotella loescheii -=A-68 Porphyromonas catoniae Prevotella & Bacteroides C-67 C-61 Capnocytophaga ochracea AC-62 Capnocytophaga granulosa --F Bergeyellazoohelcum l ~~~~~~~~~A-63- Borrelia burgdorferi SSpirochetes I I Treponema pallidum I~~~~~~~~~~~~~~~ | Chlamydia psittaci C Chlamydia trachomatis

0.0 0.1 0.2 0.3 evolutionarydistance

Fig.2. Phylogeneticrelationships of bacteria detected in a subgingivalspecimen, inferred from 16S rDNA sequence analysis. This unrooted tree was constructed byusing 489 homologoussequence positions (534-1,050; E. coli numbering)and a maximum-likelihoodalgorithm. Sequences were given a coloredalphanumeric designationaccording to whetherthey were amplified directly from the specimen(A; red),obtained from organisms cultivated from the specimen(C; green), orfound by both methods (AC; blue). An organism name is providedadjacent to an alphanumericdesignation when a sequencegenerated from this study was -99.0% identicalto a sequence in a publicdatabase. Sequences selected from a publicdatabase for purposes of reference(not detected in the specimen)are indicatedwiththeorganism name in black.The numberwithin each shadedfan indicatesthetotal numberof unique (amplified and cultivated)phylotypeswithin the group.

A-34, A-72, and C-73 are related to Eubacteriumsaburreum, a lum schaalii and Actinomycessuis. A. schaalii has been isolated commonoral isolate.Many of the Eubacterium species have been from human blood and urine (28) whereas A. suis has been isolatedfrom the rumen and/or are capable ofdegrading various associated withporcine cystitisand pyelonephritis. plant polymers(saccharolytic, xylanotic, or cellulolytic). Several interestingphylotypes were not assigned to the Clos- Phylogeneticanalysis revealed 18 phylotypeswithin the Acti- tridiaor Actinobacteriagroups and are listed in Table 3. For nobacteriagroup, 12 ofwhich were previously unidentified. Nine example,the deeply divergentphylotype A-63 was most similar of these twelve novel phylotypesdisplayed <93% sequence to Weeksellazoohelcum, a normalinhabitant of the canine upper identitywith organismswithin the ARB database (Fig. 4). Of respiratorytract that has been culturedfrom human dog bite these,A-11 had the lowestsimilarity value to its closestrelative wounds (29). However,drawing conclusions about an organism (-90%). This phylotypeclustered consistently with Actinobacu- whose existenceis inferredfrom only sequence is problematic

14550 1 www.pnas.org Kroes et at. Table 2. Distributionof similarityscores of sequences to the 86 Nocardiaasteroides most closely related public database entry Mycobacteriumtuberculosis AC-69 Corynebacterlummatruchotil A-10 Isolates Amplicons 76 A-9 Actinomycesgeorgiae Percentidentity* No. Percentof total No. Percentof total AC-8 Actinomycesodontolyticus 86 69LAC-1 Actinomycessp. AC-3 ?99.0% 22 78.6 28 47.5 100 52 Actinomycesgerencseriae 97.0-98.9% 3 10.7 13 22.0 ^ C-5 95.0-96.9% 3 10.7 10 17.0 Actinomycesdenticolens 90.0-94.9% 0 0.0 6 10.1 <89.9% 0 0.0 2 3.4 _. [ A-14A-15 Valuesare based on an analysisof 489 or morehomologous positions (see 74 Actinomycesbovis text). 771 A-11 *Degreeof identitywith most closely related database sequence. 10 Actinobaculumschaalii Actinomycessuis 83 Arthrobacterglobiformis ~10~0 0C-71 Rothiadentocariosa when the nearest-knownrelative is so distant.Amplification C-70~ Micrococcusmucilaginosus reactionsusing Treponema-specific 16S rDNA primers(9) failed to reveal treponemal sequences. The sensitivityof detection using conserved bacterialprimers and cloned bacterial (phylo- 0.0 0.1 typeA-38) 16S rDNA was 4-40 gene copies. evolutionarydistance

Discussion Fig.4. Phylogeneticrelationships of bacteriathat belong within the Acti- nobacteriadivision, inferred from 16S rDNAsequence analysis. This tree was Sequence-based environmentalmicrobial surveys have taughtus constructedby using 894 homologouspositions (534-1,438; E. coli number- that cultivationmethods woefully under-represent the true ex- ing) and a maximum-likelihoodalgorithm; it was rooted by usingthe se- tent of bacterial diversity.The gross outlines of this diversity quence of Streptomycescoelicolor. Confidence values >50% fromboot- have begunto emergein recentyears from comparative analyses strappeddata are displayed.For sequence designationscheme, see Fig. 2 of 16S rRNAs. From thesesurveys, 36-40 higherorder bacterial legend. divisionscan be discerned(30). Our studysought to examinethe extentof bacterialdiversity within a well studiedhuman endog- tified.Some of the phylotypesrecovered by rDNA amplification enous niche, the subgingivalcrevice, and to compare the rep- directlyfrom the plaque specimenconformed well with previous O resentationsof diversityprovided by two approaches,that based 0 descriptionsof the bacterial flora of the subgingivalcrevice, .j on cultivationand thatbased on amplificationof rDNA directly includingmembers associated with local and systemicdisease 0 from the same specimen.A priori,one mightpostulate that, such as bacteremia,and endocarditis.However, we found that 0 withina relativelyfamiliar ecological niche that has been sub- u the diversityrevealed by directamplification of rDNA directly 2 jected to numerouscultivation procedures-prompted in partby froma subgingivalplaque specimensubstantively exceeded that the search forcausative agents of local and systemicdisease- revealed by cultivationfrom the same specimen.Furthermore, relativelyfew previouslyunrecognized bacteria would be iden- manyof the more divergentphylotypes have not been detected previouslywithin humans.

74 A-72 Overall,we detectedmembers of fiveof thebacterial divisions 100 AC-32 Eubacteriumsaburreum proposed by Hugenholtzet al. (30) withinsubgingival plaque. C-73 These five divisions (Low GC Gram Positives,Fusobacteria, 75 A-33 A-34 Cluster Actinobacteria,Proteobacteria, and Cytophagales)all appear to Eubacteriumcellulosolvens XlVa be cosmopolitan in nature and are well representedamong 76 Butyrivibriofibrisolvens Unidentifiedrumen bacterium AF001716 Ruminococcus gnavus iou Peptostreptococcusmicros 1 Cluster Table 3. Phylotypes with <99.0% identityto a public database _0 L A-44 Clone fromdentoalveolar abcess XiII sequence, excluding the Actinobacteria and the Low GC Gram Ruminococcus albus Positive Clostridia groups Cluster niedentified rumenbacterium AF001743 I A-35 Percent iou A-42 Veillonelladispar Phylotypes* Related organism(s) Identity Nucleotides 61 A-43 Veillonellaatypica 99 A-40 Dialisterpneumosintes A-37 Selenomonas sputigena Cluster A50 Haemophilussegnis 98.9 541

53j { Zymophiluspaucivorans IX A57 Kingella oralis 98.5 545 Schwartzia succinivorans A27 Streptococcusparasanguis 98.3 529 541 A-38 A49 Haemophilusparainfluenzae 97.6 A26 Streptococcusparasanguis 97.5 529 A68 Prevotellaloescheii 97.2 702 0.0 0.1 0.2 0.3 A56 K. oralis 97.1 545 evolutionarydistance AC60 Leptotrichiabucallis 97.1 1,370 Fig.3. Phylogeneticrelationships of bacteria that belong within the LowGC C65 Prevotellaoris 96.3 814 GramPositive division, inferred from 16S rDNAsequence analysis.This tree C67 Porphyromonascatoniae 96.1 818 was constructedby using 860 homologoussequence positions (534-1,400; E. A63 W. zoohelcum 95.5 663 coilinumbering) and a maximum-likelihoodalgorithm; it was rootedby using A18 Gemellamorbillorum 95.3 953 the sequence of Peptococcusniger. Confidence values >50% fromboot- A51 T. cuprinus 93.9 933 strappeddata are displayed.For sequence designationscheme, see Fig. 2 legend.Cluster designations are based on the proposedphylogenetic scheme *Phylotypeswere obtained by either direct amplification (A), cultivation (C), of Collinset a/. (26). or both(AC).

Kroes et al. PNAS I December 7, 1999 | vol. 96 | no. 25 | 14551 cultivatedmicroorganisms. Initially, most of the amplified16S undetectedphylotypes were presentwithin the clone libraries rDNA was of streptococcalorigin, a findingthat was contra- and comprised -12% of the clones. In ecological terms,the dicted by previouslypublished data (25) and by our in situ clone libraries,which are assumed to representthe subgingival hybridizationdata. This discrepancypointed to a bias in PCR microbialflora from this specimen, are uneven (a small number amplification.Previously reported PCR biases have been attrib- of species dominatethe population) but contain a large degree uted to variationin bacterialgenome size, 16S rDNA allele copy of richness(a large numberof unique phylotypes).This conclu- number,G+C content,rDNA secondary structure,and tem- sion is consistentwith the cultivation-basedfindings of Moore plate concentration(31-33). Our subsequentstrategy to reduce (8) in which 10-16 species accounted for 60-71% of the culti- the numberof streptococcal16S rDNA clones by usingBsu36I vatable oral bacteria in adult gingivitis.The transferof mathe- was successful;however, it probablyreduced the representation maticalconstructs, such as the Good coverage estimatoror the of otherbacterial sequences within the libraries. In fact,there are Shannon-Weaverindex, from macroecology to humanmicrobial Bsu36I endonuclease restrictionsites within16S rDNAs from ecology should facilitatea more meaningfuland quantifiable cyanobacteriaand spirochetes.Another obvious limitationin comparisonof therichness and evennessbetween different hosts our studywas the use of just one specimenfrom one individual. and sites over time. High-throughputsequencing methods and thepossibility of high Overall, direct amplificationof 16S rDNA yielded a more densitymicroarrays for microbial surveying should make analysis diverseview of the bacterial flora associated with subgingival of multiplespecimens from numerous individuals increasingly plaque than that found by cultivation.Our findingsraise the practicable. possibilitythat previously unidentified higher order taxonomic Typically,large numbersof Treponemahave been associated divisionsstill existwithin the human ecosystem.Direct ampli- withperiodontitis (9, 34, 35). Choi et al. foundthat only 7% of ficationand cultivationmay be viewed as complementarytech- 16S rDNA clones obtained by amplificationusing broad-range niques forthe studyof microbialdiversity and identificationof bacterial primerswere of treponemal origin whereas phase- potentialopportunistic pathogens. Obviously, the physiological contrastmicroscopy indicated that spirochetes comprised -20% and clinical relevance of the organismsdetected in this study of the bacterialpopulation (9). Attemptsto amplifyTreponema remainunclear. At one extreme,some of the phylotypesmay be 16S rDNA fromour specimenwere unsuccessful.This failure transientresidents acquired from the environment.For example, does not appear to be explainedby the choice of primersbecause A-51 is related to metal-oxidizingy-proteobacteria, such as the the predicted annealing sites are well conserved among the sulfur-oxidizerThiobacillus cuprinus, a bacteriumtypically iso- Treponemasequences withinthe ARB 16S rRNA database. It is lated fromlake sediments.A-51 maybe a transientoral resident probable that treponemes,if present in our specimen,were obtained fromdrinking water. Other organismsmay be critical presentin numbersbelow the detectionlimits of our PCR. The formaintenance of ecosystemstability and oral healthor maybe absence of spirochetesdetected by darkfieldmicroscopy sup- occasional pathogens.Large-scale ecosystemanalyses at multi- portsthis presumption. ple intraoralsites in multiplehosts and at multipletimes should A crude estimateof the populationcoverage by the 59 unique help to clarifysome of these issues. phylotypeswe identifiedby direct amplification may be obtained byusing the estimatorproposed by Good (21). By thisestimate, We thankPat Mickelsenand ArleneMorton from the StanfordUni- versityHospital Clinical Microbiology Laboratory for bacterial cultiva- the 59 phylotypesaccounted for 88% of all cloned sequences. tion;Victor T. Lee andPenny Asami Wong for specimen collection; the Usinga parametricmodel for estimating the number of "unseen" StanfordUniversity Digestive Diseases Center;and Doug Smithand species,the 59 unique phylotypesrepresented <70% of the total DeborahDodge fromPerkin-Elmer for DNA sequencingtechnology numberof unique phylotypeswithin the clone population (22). support.This work was supported by the Donald E. andDelia B. Baxter By inference,an additional 25 [(59/0.7) - 59] unique but Foundationand theLucille P. MarkeyCharitable Trust.

1. Mackowiak,P. A. (1982) N. Engl. J. Med. 307, 83-93. 18. Felsenstein,J. (1985) Evoluttion(Lawrence, Kans.) 39, 783-791. 2. Barns,S. M., Fundyga,R. E., Jeffries,M. W. & Pace, N. R. (1994) Proc. Natl. 19. Olsen, G. J.,Matsuda, H., Hagstrom,R. & Overbeek,R. (1994) Comput.Appl. Acad. Sci. USA 91, 1609-1613. Biosci. 10, 41-48. 3. Giovannoni,S. J.,Britschgi, T. B., Moyer,C. L. & Field, K. G. (1990) Nature 20. DeSoete, G. (1983) Psychometrika48, 621-626. (London) 345, 60-63. 21. Good, I. J. (1953) Biometrika40, 237-264. 4. Stackebrandt,E., Liesack, W. & Goebel, B. M. (1993) FASEB J. 7, 232-236. 22. Efron,B. & Thisted,R. (1976) Biometrika63, 435-447. 5. Relman, D. A., Loutit, J. S., Schmidt,T. M., Falkow, S. & Tompkins,L. S. 23. DeLong, E. F., Wickham,G. S. & Pace, N. R. (1989) Science 243, 1360-1363. (1990) N. Engl. J. Med. 323, 1573-1580. 24. Maidak, B. L., Olsen, G. J.,Larsen, N., Overbeek,R., McCaughey,M. J. & 6. Thoden van Velzen, S. K., Abraham-Inpijn,L. & Moorer,W. R. (1984) J.Clin. Woese, C. R. (1997) NucleicAcids Res. 25, 109-11. Periodontol.11, 209-220. 25. Socransky,S. S., Gibbons, R. J., Dale, A. C., Bortnick,L., Rosenthal, E. & 7. Meyer,D. H. & Fives-Taylor,P. M. (1998) Curr.Opin. Microbiol.1, 88-95. Macdonald, J. B. (1963) Arch. Oral Biol. 8, 275-280. 8. Moore, W. E., Burmeister,J. A., Brooks, C. N., Ranney,R. R., Hinkelmann, 26. Collins,M. D., Lawson,P. A., Willems,A., Cordoba,J. J., Fernandez, G. J.,Garcia, K. H., Schieken,R. M. & Moore, L. V. (1993) Infect.Immun. 61, 2891-2898. P., Cai, J.,Hippe, H. & Farrow,J. A. (1994) Int.J. Syst.Bacteriol. 44, 812-826. B. B. Dewhirst,F. E. & Gobel, U. B. (1994) Infect.Immun. 9. Choi, K., Paster, J., 27. Kolenbrander,P. E. & London, J. (1993) J. Bacteriol.175, 3247-3252. 62, 1889-1895. 28. Lawson, P. A., Falsen, E., Akervall,E., Vandamme,P. & Collins,M. D. (1997) 10. Ashimoto,A., Chen, C., Bakker,I. & Slots,J. (1996) Oral Microbiol.Immunol. Int. J. Syst.Bacteriol. 47, 899-903. 11, 266-273. 29. Talan, D. A., Citron,D. M., Abrahamian,F. M., Moran, G. J. & Goldstein, 11. Dymock, D., Weightman,A. J., Scully,C. & Wade, W. G. (1996) J. Clin. E. J. C. (1999) N. Engl. J. Med. 340, 85-92. Microbiol.34, 537-542. P., Goebel, B. M. & Pace, N. R. (1998) J.Bacter-iol. 180, 4765-4774. 12. Amann,R. I., Krumholz,L. & Stahl, D. A. (1990) J. Bacteriol.172, 762-770. 30. Hugenholtz, 13. Lane, D. J., Pace, B., Olsen, G. J., Stahl, D. A., Sogin, M. L. & Pace, N. R. 31. Farrelly,V., Rainey,F. A. & Stackebrandt,E. (1995) Appl. Environ.Microbiol. (1985) Proc. Natl. Acad. Sci. USA 82, 6955-6959. 61, 2798-2801. 14. Fredricks,D. N. & Relman, D. A. (1998) J. Clin. Microbiol.36, 2810-2816. 32. Suzuki,M. T. & Giovannoni,S. J.(1996)Appl. Environ.Microbiol. 62, 625-630. 15. Edwards,U., Rogall, T., Blocker,H., Emde, M. & Bottger,E. C. (1989) Nucleic 33. Chandler,D. P., Fredrickson,J. K. & Brockman,F. J. (1997) Mol. Ecol. 6, Acids Res. 17, 7843-7853. 475-482. 16. Alm, E. W., Oerther,D. B., Larsen, N., Stahl,D. A. & Raskin,L. (1996) Appl. 34. Riviere, G. R., Smith,K. S., Carranza, N. J.,Tzagaroulaki, E., Kay, S. L. & Environ.Microbiol. 62, 3557-3559. Dock, M. (1995) J. Periodontol.66, 829-837. 17. Relman, D. A. (1993) in DiagnosticMolecular Micr-obiology:Principles and 35. Kigure,T., Saito, A., Seida, K., Yamada, S., Ishihara,K. & Okuda, K. (1995) Applications,eds. Persing,D. H., Smith,T. F., Tenover,F. C. & White,T. J. J. Periodontal.Res. 30, 332-341. (Am. Soc. Microbiol.,Washington, DC), pp. 489-495. 36. Stackenbrandt,E. & Groebel, B. M. (1994) Int.J. Syst.Bacteriol. 44, 846-849.

14552 | www.pnas.org Kroes et al.