JOURNAL OF CLINICAL MICROBIOLOGY, Sept. 1993, p. 2467-2473 Vol. 31, No. 90095-1137/93/092467-07$02.00/0Copyright X 1993, American Society for Microbiology

Species Identification of Oral Viridans Streptococci by RestrictionFragment Polymorphism Analysis of rRNA Genes

JOEL D. RUDNEY* AND CHRISTOPHER J. LARSONDepartment of Oral Science, School ofDentistry, University of Minnesota, 17-252 Moos Tower,

515 Delaware Street, S.E., Minneapolis, Minnesota 55455

Received 1 April 1993/Returned for modification 7 May 1993/Accepted 11 June 1993

Oral streptococci formerly classified as Streptococcus sanguis have been divided into six genetic groups.Methods to identify those species by genotype are needed. This study compared restriction fragmentpolymorphisms of rRNA genes (ribotypes) for seven S. gordonii, three S. sanguis, four S. oralis, three S. mitis,one S. crista, and seven S. parasanguis strains classified in previous DNA hybridization studies, as well as oneclinical isolate. DNA was digested with HindIH, PvuII, HindIH and PvuII combined, EcoRI, BamHI, AatII,AlwNI, and DraII. DNA fragments were hybridized with a digoxigenin-labeled cDNA probe obtained byreverse transcription ofEscherichia coli 16S and 23S rRNA. S. oralis, S. mitis, and S. parasanguis all showedan isolated 2,290-bp band in AatII ribotypes that was absent from S. gordonii, S. sanguis, and S. crista. Thelast three groups showed species-specific bands with AatII and also with PvuH. S. oralis could be distinguishedfrom S. mitis and S. parasanguis in AlwNI and DraH ribotypes. S. mitis and S. parasanguis could not bedistinguished, since they shared multiple bands in PvuH, AlwNI, and EcoRI patterns. The clinical isolate in thepanel was very similar to S. sanguis by all enzymes used. Our findings suggest that ribotyping may be usefulfor genotypic identification of oral viridans streptococci. Initial digests of clinical isolates might be made withAatII, followed by PvuII orAlwNI. Isolates then could be identified by comparing ribotype patterns with thoseof reference strains. This approach could facilitate clinical studies of these newly defined species.

Clinical studies of oral viridans streptococci have beenmade more difficult by recent findings that strains formerlyplaced within Streptococcus sanguis or Streptococcus mitismay represent six distinct genetic groups (6). Those groupshave been assigned the species names Streptococcus gordo-nii, S. sanguis, Streptococcus oralis, S. mitis, Streptococcusparasanguis, and Streptococcus crista (6, 11, 19, 23, 42).Species differences may be of importance in oral ecology,since newly defined groups appear to occupy distinct oralsites. Recent clinical studies have suggested that S. sanguis,S. oralis, and S. mitis biovar 1 strains are early colonizers ofthe tooth surface. Moreover, prevalences of those speciesdiffered in caries-active and caries-inactive individuals. S.gordonii seemed to appear later, as plaque became mature.S. mitis biovar 2 was infrequently seen in plaque but washighly prevalent on the dorsum of the tongue (14, 24, 28, 29).

Sanguis group species are not distinguishable on the basisof colony morphology. Current identification protocolstherefore rely on series of phenotypic tests (2, 23). Onepotential problem with such testing is that not all members ofa given species may be positive for particular traits (2, 23).Another potential problem is that laboratory strains ofstreptococci can undergo phenotypic shifts on passage invitro and in vivo (20, 39). This raises the possibility thatphenotypic classifications based on reference strains do notapply to all clinical isolates.Methods for genotypic identification of newly defined

species could help to resolve such problems. We began toinvestigate those methods in the context of a large popula-tion-based study of saliva antimicrobial protein effects onplaque composition (32). Initial studies focused on restric-tion endonuclease fingerprinting of a panel of strains previ-

* Corresponding author. Electronic mail address: jrudney@staff.tc.umn.edu.

ously assigned to each species on the basis of DNA hybrid-ization (33). We found restriction endonuclease fingerprintsto be strain specific in conventional agarose gels, whichindicated that each species is genetically diverse. Strainfingerprints for rare restriction sites were equally diverse infield inversion gels. However, patterns produced were muchsimpler, which did facilitate comparison of different strains.The study reported here employed ribotyping as an alter-

native approach to genotypic identification of streptococcalspecies. Ribotyping compares patterns produced whenprobes for rRNA genes are hybridized with restrictiondigests of chromosomal DNA (38). rRNA operons containmany highly conserved sequences (21, 25). This propertyallows cross-hybridization of a probe from a convenientspecies across distantly related groups. Commercially avail-able rRNA from Escherichia coli thus has been used toribotype a wide range of gram-negative and gram-positivespecies (1, 3, 4, 38). Ribotype patterns are considerablysimpler than DNA fingerprints from conventional gels, butthey can show diversity within species. That combination ofsimplicity and diversity has made ribotyping a useful tech-nique for tracking strains in epidemiological studies (1, 3, 4,31, 37, 43). However, the high degree of conservation amongrRNA sequences can make ribotypes useful for speciesidentification as well. A number of studies have reportedribotype patterns that contain bands common to all membersof the same species (9, 17, 18, 22, 27, 30, 36). This techniquethus may allow oral streptococci to be identified at the levelof both strain and species. We compared ribotype patternsobtained with seven restriction enzymes in a panel of 25strains assigned to the six newly defined groups in previousDNA hybridization studies. Research questions were asfollows: are ribotype patterns in strains of known affiliationspecies or strain specific, and can bands which are sharedamong members of the same species be identified?

2467

on April 10, 2019 by guest

http://jcm.asm

.org/D

ownloaded from

2468 RUDNEY AND LARSON

TABLE 1. Strains of oral streptococci used in this study

Species and straina Sourceb

S. gordoniiATCC 10558 .............. ATCCATCC 12396 .............. ATCCATCC 33399 .............. ATCCBlackburn .............. P. Fives-Taylor (A. Coykendall)SPED3 .............. P. Fives-Taylor (P. Handley)S7 .............. W. LiljemarkM5 .............. W. Liljemark (B. Rosan)

S. sanguisATCC 10556 .............. ATCCHPC1.............. P. Fives-Taylor (A. Coykendall)804 .............. P. Fives-Taylor (A. Coykendall)

S. oralisATCC 10557 .............. ATCCATCC 35037 .............. ATCCCR834.............. P. Fives-Taylor (P. Handley)9811.............. W. Liljemark (B. Rosan)

S. mitisATCC 903 .............. ATCCNCTC 10712 .............. NCTCNCTC 12261 .............. NCTC

S. parasanguisATCC 15909 .............. ATCCATCC 15911 .............. ATCCATCC 15912 .............. ATCCMGH 145 .............. P. Fives-Taylor (A. Coykendall)MGH 413 .............. P. Fives-Taylor (A. Coykendall)UC 4989 .............. P. Fives-Taylor (A. Coykendall)FW 213 .............. P. Fives-Taylor

S. crista CR3 .............. P. Fives-Taylor (P. Handley)Clinical isolate 13379 ..............M. Herzberg (CDC)

a Species identifications were obtained from previous taxonomic studies(6-8, 11, 16, 19, 23, 35, 41, 42).

b ATCC, American Type Culture Collection; NCrC, National Collection ofType Cultures; CDC, Centers for Disease Control (Atlanta, Ga.). Investiga-tors from whom we obtained strains are listed with the source (in parentheses)from which they obtained those strains (their institutions are noted inAcknowledgements).

MATERIALS AND METHODS

Bacterial strains. Bacterial strains employed in this studyare described in Table 1. Strains were obtained from theAmerican Type Culture Collection (Rockville, Md.), theNational Collection of Type Cultures (London, United King-dom), or culture collections of the University of MinnesotaDepartment of Oral Science faculty or through the generousgift of Paula Fives-Taylor, Department of Microbiology,University of Vermont. Species names given are for strainsclassified in previous taxonomic studies (6-8, 11, 16, 19, 23,35, 41, 42). Clinical isolate 13379 was included for compar-ison with classified strains. That strain had shown greaterphenotypic variation than other strains in our DNA finger-print studies, as it was placed in four different species atdifferent times by the API Rapid Strep system (AnalytabProducts, Plainview, N.Y.). However, it had maintained thesame fingerprint pattern at all times (33).

Culture methods. Stocks were maintained and preparedfor DNA isolation as described by Rudney et al. (33). Thepurity of stocks was checked by a series of tests. Colonymorphology on mitis salivarius agar first was compared withphotomicrographs of variants seen in the past. The APIRapid Strep system then was used to evaluate phenotypiccontinuity. Data on the reliability of API Rapid Strep resultsfor our panel are provided in the work of Rudney et al. (33).API Rapid Strep findings were supplemented by using the

method of Douglas et al. (11) to test for amylase binding.When strains were digested with HindIII or PvulI, finger-print patterns were compared with those seen in our previ-ous study (33).

Preparation of DNA. Chromosomal DNA was prepared inagarose beads as described by Rudney et al. (33). Washedcells were emulsified with low-melting-point agarose (Sigma,St. Louis, Mo.) in paraffin oil. Cells embedded in beads werelysed with a mixture of 2 mg of lysozyme per ml, 1,000 U ofachromopeptidase per ml, and 42.5 U of mutanolysin per ml(all from Sigma), followed by 2% sodium dodecyl sulfate(SDS) and 200 ,ug of proteinase K (GIBCO BRL, Gaithers-burg, Md.) per ml, all at 40°C. We first began to prepareDNA in agarose beads for field inversion gel electrophoresis(33). This method prevents shearing of large DNA fragmentsobtained with enzymes having rare restriction sites (5, 40).In these ribotyping studies, we have found that restrictionendonuclease AatII gives much better results with DNAembedded in beads (see below). The technique does notrequire toxic organic solvents like phenol or chloroform, andsamples may be run in either conventional or field inversiongels. The primary disadvantage is that it is more difficult tocontrol loading of gels. That problem was addressed byrunning replicate samples in different gels. For some gels,DNA was recovered from beads with the Elu-Quik DNApurification kit (Schleicher & Schuell, Keene, N.H.) accord-ing to the manufacturer's instructions. Sample concentra-tions were determined spectrophotometrically (33), andamounts of beads loaded were varied to equalize loadingacross lanes.

Restriction endonuclease panel. Ribotypes for a given set ofstrains and species tend to vary according to the restrictionenzyme used. Some enzymes may produce patterns that arelargely strain specific, while others may yield patterns withcommon bands. Results are difficult to predict in advance, soit is important to compare findings across a panel of endo-nucleases (18, 22, 36). HindIII and PvuII (GIBCO BRL)were selected for this study because they have been used toribotype other genera (1, 3, 4, 17, 22, 31, 36). They likewiseallowed us to check fingerprints against those seen in ourprevious study (33). Those enzymes also were used in adouble digest. EcoRI and BamHI (GIBCO BRL) were addedto the panel as additional enzymes used by other investiga-tors (1, 4, 9, 18, 22, 30, 36, 37, 43). The panel was completedwith AatII, AlwNI (New England Biolabs, Beverly, Mass.),and DraII (Boehringer Mannheim, Indianapolis, Ind.). Theywere selected by using program SEQ (IntelliGenetics Inc.,Mountain View, Calif.; provided through the MolecularBiology Computing Center, University of Minnesota) to listrestriction sites in six S. sanguis sequences filed in theGenBank data base (10, 12, 13, 15, 42). AatII, AlwNI, andDraII showed only single restriction sites in most sequencesexamined. This suggested that they might yield ribotypeswith relatively simple patterns.

Agarose gel electrophoresis. Washed packed beads wereequilibrated overnight in appropriate restriction buffers.Packed beads then were treated with a selected restrictionendonuclease(s). Digestion was accelerated by heating sam-ples in a microwave oven at full power for three 10-sintervals. There was a 10-s pause after the first and secondheating steps; tubes were rotated 600 around their verticalaxis during each pause. Samples cooled to room temperaturefor approximately 30 min, and digestion then was stopped(26a). Molten agarose was used to seal digests into wells ofhorizontal 0.7% agarose gels (15 by 15 cm, 200 ml). Sizestandards consisted of digoxigenin-labeled DNA molecular

J. CLIN. MICROBIOL.

on April 10, 2019 by guest

http://jcm.asm

.org/D

ownloaded from

RIBOTYPES OF ORAL STREPTOCOCCI 2469

weight marker II (Boehringer Mannheim) loaded into theouter left, middle, and outer right wells (1). Gels were run at30 V in TBE buffer (0.089 M Tris-borate, 20 mM EDTA; pH8.3) for 17 h at 25°C. They then were stained and photo-graphed as described by Rudney et al. (33).

Southern blotting. Gels were depurinated in 0.25 M HCIfor 10 min, washed three times in deionized water, denaturedin 0.5 M NaOH-1.5 M NaCl for 30 min, and then neutralizedwith 0.5 M HC1-1.5 M NaCl for 30 min (1). A model 750vacuum blotter (Bio-Rad, Richmond, Calif.) was used totransfer DNA to Hybond N+ nylon membranes (Amersham,Arlington Heights, Ill.). Transfers were run for 90 min in 10xSSC (3 M NaCl, 0.3 M sodium citrate; pH 7.0) at a pressureof 10 in. of Hg. Membranes then were washed in 6x SSC anddried at 65°C for 1 h. A cDNA probe for rRNA genes wasprepared by reverse transcription of E. coli 16S and 23SrRNA (Boehringer Mannheim; 0.25 ,ul) in the presence ofdigoxigenin-labeled dUTP (Boehringer Mannheim), as de-scribed by Baloga and Harlander (1). All hybridization stepswere carried out in an incubator-shaker (Boekel, Philadel-phia, Pa.), with the membrane in a heat-sealed plastic bag.The membrane was prehybridized overnight at 65°C withhybridization solution consisting of 5 x SSC, 0.1% Sarkosyl,0.02% SDS, and 0.01% blocking agent (provided as part ofthe Genius 3 digoxigenin detection system from BoehringerMannheim). The membrane was then incubated overnight at65°C in hybridization solution with heat-denatured probe.This was followed by two 5-min washes with 4x SSC-0.1%SDS and two washes with 0.1x SSC-0.1% SDS, all at 25°C.Bound digoxigenin-labeled probe was detected by using theGenius 3 kit according to the manufacturer's instructions (1).Comparison of ribotype patterns. Blots were scanned at a

resolution of 1,024 by 1,024 pixels with a Visage 110 imageanalysis system (Bio-Image, Ann Arbor, Mich.). Visageprogram WHOLE BAND was used for determination ofband sizes, as described by Rudney et al. (33). Band sizepatterns were compared among members of the same anddifferent species by using Visage program WB COMPARE.This was done by visual examination of band patterns fromimages and also by computer matching of bands. Strainsbelonging to the same species were run in the same gel forinitial evalqation. Replicates of representative strains wererun in different gels to evaluate the reproducibility of bandpatterns and size estimates. When bands appeared to bespecies specific, results were checked by running differentspecies in the same gel.

RESULTS

Replicate lanes in different gels always could be identifiedby visual examination. The presence or absence of minorbands was influenced by the amount of DNA loaded inreplicate lanes, but this did not interfere with recognition ofthe ribotype patterns. Computer matching of bands by sizewas not always successful when the same strains were run indifferent gels. This problem was particularly apparent withbands larger than 10,000 bp or smaller than 2,000 bp. Speciescould be discriminated by visual identification of consistentband patterns within ribotypes forAatll, PvuII, andAlwNI.Band sizes given below are means plus or minus standarddeviations.AatII typically produced ribotypes containing closely

spaced bands larger than 8,000 bp and smaller bands thatwere widely separated. S. oralis, S. mitis, and S. parasan-guis all showed a band at approximately 2,290 + 17 bp thatwas absent from S. gordonii, S. sanguis, and S. crista strains

S. ;o, rS S. .csas.taS

.s~~~~~~~~~ /

23190 bp

9416 bp - - .. - r i

655D7 bp .th-_.*

4361 bp

aMdkf

2322 bp

2027 bD

FIG. 1. AatII ribotypes for S. gordonu strains, S. sanguisstrains, S. crista CR3, and clinical isolate 13379, compiled from twoblots by using Adobe Photoshop 2.0.1 and Quark Express 3.11software with a Macintosh IIcx computer and a Sharp 450 scanner.Species and strain names are above lanes (species are separated byangled lines). "Size std." refers to DNA molecular weight marker II(Boehringer Mannheimn). Band sizes are given in the left margin.Arrowheads denote bands common to a species. All S. gordoniistrains shared a band at approximately 9,374 bp and a doublet atapproximately 7,086 and 6,587 bp. All S. sanguis strains shared adoublet at approximately 9,776 and 9,324 bp. Clinical isolate 13379shows a pattern similar to that of S. sanguis ATCC 10556.

(Fig. 1 and 2). All S. gordonii strains shared a band atapproximately 9,374 + 113 bp and a doublet at approxi-mately 7,086 ± 40 and 6,587 ± 49 bp (Fig. 1). S. sanguisstrains shared a doublet at approximately 9,776 ± 168 and9,324 + 46 bp (Fig. 1). The lower band in that doubletappeared to correspond to the 9,374-bp band noted for S.gordonii (Fig. 1). S. crista CR3 displayed a band at approx-imately 2,440 bp that was not present in other species. S.oralis, S. mitis, and S. parasanguis strains could not readilybe distinguished from one another. All showed the commonband noted above, and common bands were seen amongsubsets of strains. However, no common bands were spe-cific for all members of a given species (Fig. 2). AatIIribotypes thus were primarily useful for identification of S.gordonii, S. sanguis, and S. crista. Unique bands werepresent in mostAatIl patterns, so that enzyme could be usedto identify strains within species.PvuII ribotypes also were useful for discriminating among

S. gordonii, S. sanguis, and S. crista strains (Fig. 3), butpatterns were less strain specific than those seen withAatII.Identical patterns were seen for S. gordonii S7, M5, ATCC12396, and SPED3; strain ATCC 33399 was only slightlydifferent from that group. S. gordonii Blackburn and ATCC10558 appeared to form a distinct set which shared bandswith the five strains above at approximately 4,500 ± 50,3,475 ± 60, and 3,020 ± 6 bp. S. sanguis PvuII patterns werestrain specific, but that species could be distinguished by thepresence of a common band at approximately 2,100 ± 50 bp.S. crista CR3 showed a PvuII pattern which was differentfrom all S. gordonii or S. sanguis strains. PvuII patterns forS. oralis, S. mitis, and S. parasanguis were strain specific;no bands seen were shared by all members of a species (not

VOL. 31, 1993

on April 10, 2019 by guest

http://jcm.asm

.org/D

ownloaded from

2470 RUDNEY AND LARSON

S. tra!s S p.arasanqL;As S rritis

V.;4 A...s

S. gordonhi

cO'S 4 sS mAM g teS gC.)81 Q

"Io' C

S. sarnguis S crista

h /

AI,.-C)a o~ 52 4Q;y440 - C

*F' 6557 bp

M 44361 bp

--.

0,

:.

w mmai--:_.! ------ -._.* ... .: : .

.: .:

.. ....

FIG. 2. AatII ribotypes for representative S. oralis, S. mitis, andS. parasanguis strains, compiled from two blots by using AdobePhotoshop 2.0.1 and Quark Express 3.11 software with a MacintoshIIcx computer and a Sharp 450 scanner. All showed a band atapproximately 2,290 bp absent from S. gordonii and S. sanguisstrains.

shown). However, S. mitis strains were observed to sharePvuII bands with S. parasanguis strains. Similarities were

particularly apparent for S. mitis ATCC 903 and S. parasan-guis ATCC 15911 and ATCC 15912 (not shown; see below).

S. oralis could be distinguished from S. mitis and S.parasanguis by AlwNI ribotypes (Fig. 4 and 5). Each S.oralis strain displayed a unique pattern. However, all hadthree bands in common at approximately 6,025 + 61, 5,314 +54, and 4,583 ± 86 bp and lacked a band at approximately7,616 ± 95 bp present in all S. mitis and S. parasanguisstrains. Those two species showed other common bands. S.mitis NCTC 10712 and NCTC 12261 could be grouped withS. parasanguis ATCC 15909, FW 213, UC 4989, and ATCC15911 (Fig. 4), while S. mitis ATCC 903 appeared similar toS. parasanguis ATCC 15911, MGH 145, and MGH 413 (Fig.5). AlwNI ribotypes were not useful for identification of S.gordonii, S. sanguis, and S. crista, since no bands wereshared by all members of each species (not shown).

Ribotypes with other restriction enzymes supplementedinformation provided by AatII, PvuII, and AlwNI. Resultsaccordingly are reported without illustrations. The set of fiveS. gordonii strains grouped together by PvuII was separatedby BamHI, DraII, and HindIII-PvuII double digests intothree subsets including ATCC 33399, M5 and S7, and SPED3and ATCC 12396, respectively (the same subsets could beseen among AatII ribotypes of S. gordonii strains presentedin Fig. 1). S. gordonii strains showed many bands in BamHIribotypes, while other species showed only a few largebands. S. oralis could be distinguished from S. mitis and S.parasanguis strains by the presence of a band at approxi-mately 3,791 ± 15 bp in DraII ribotypes. Only three S.gordonii strains were digested by EcoRI (S7, Blackburn, andATCC 10558). Similarities among EcoRI patterns for S. mitis

FIG. 3. PvuII ribotypes for S. crista, S. gordonii, and S. sanguisstrains and for clinical isolate 13379, compiled from two blots byusing Adobe Photoshop 2.0.1 and Quark Express 3.11 software witha Macintosh IIcx computer and a Sharp 450 scanner. All S. gordoniistrains shared bands at approximately 4,500, 3,475, and 3,020 bp. S.sanguis strains had a common band at approximately 2,100 bp.Ribotypes for clinical isolate 13379 and S. sanguis ATCC 10556were very similar.

and S. parasanguis were comparable to those observed forAlwNI, and strains could be segregated into the same broadsubsets. HindIII ribotypes were least useful for speciesdiscrimination, because patterns were complex and difficultto interpret.Our panel of strains included one clinical isolate, strain

13379. The AatII ribotype pattern for 13379 was very similarto that for S. sanguis HPC1 and ATCC 10556, although theupper band of the 9,750-bp-9,250-bp doublet seen in thosestrains was missing, as was a prominent band at 6,886 bp(Fig. 1). Strains ATCC 10556 and 13379 shared three majorbands in PvuII ribotypes (Fig. 3), while BamHI ribotypes forthose strains were identical (Fig. 5). Comparable resultswere observed for EcoRI patterns (Fig. 5), but in that case13379 was very similar to S. sanguis 804 and less similar toATCC 10556 (strain 13379 was not examined with enzymeAlwNI or DraII). Those findings suggested that 13379 mighttentatively be classified as S. sanguis.

DISCUSSION

The objective of this study was to investigate the value ofribotyping for identification of oral viridans streptococci.Agreement with established taxonomic schemes can providea measure of the validity of ribotype results. Previoustaxonomic studies have suggested that four currently recog-nized sanguis group species can be divided into two distantlyrelated lineages. S. gordonii and S. sanguis represent one

lineage; S. oralis and S. mitis represent the other. Thisdistinction is based on genetic distances obtained through

23190 bp

9416 bp23190 bp

9416 bp 2-s-

6557 bp

4361 bp

2322 bp _

2027 bp

2322 bp

2027 bp

J. CLIN. MICROBIOL.

on April 10, 2019 by guest

http://jcm.asm

.org/D

ownloaded from

RIBOTYPES OF ORAL STREPTOCOCCI 2471

S. oralls

RI' C'

CI1 C

c

S. rnit'fs

C~ qI?

S. varasan;uis

ol"I (b-

I

-!;. C)CZ,, 15", / 41C -

23190 bp S?!.

9416 bp

6557 bp..

- ...M O .

431Pb- -4361 bp *-

-m Om

-NW ~ ~ U

2322 bp _2027 bp _

FIG. 4. AlwNI ribotypes for S. oralis and S. mitis strains, with representative S. parasanguis strains, compiled from two blots by usingAdobe Photoshop 2.0.1 and Quark Express 3.11 software with a Macintosh IIcx computer and a Sharp 450 scanner. All S. oralis strains hadthree bands in common at approximately 6,025, 5,314, and 4,583 bp and lacked a band at approximately 7,616 bp present in all S. mitis andS. parasanguis strains.

A At N Ba"T' H; Ecc' RI

DNA hybridization and multilocus enzyme electrophoresis(6, 16, 23), and it also is supported by phenotypic differencesin cell wall structure (35). AatII ribotype results are consis-tent with a two-lineage concept, since a prominent 2,250-bpband was present in S. oralis and S. mitis and absent in S.gordonii and S. sanguis. Ribotype results also agree withestablished divisions of species within each lineage. Com-mon bands in AatII and PvuII ribotypes could be used toseparate S. gordonii strains from S. sanguis strains; commonbands in AlwNI and DraII ribotypes could be used toseparate S. oralis strains from S. mitis strains.The utility of ribotyping for identification of S. crista and

S. parasanguis strains is more difficult to establish. Ri-botypes for the S. crista strain we used were distinct fromthose seen for other groups. However, additional S. cristastrains must be examined to determine whether commonbands exist for the enzymes used here. S. parasanguispresents a different problem. AatII ribotypes for the sevenstrains used all showed the 2,290-bp band seen for S. oralisand S. mitis. Moreover, S. parasanguis strains showedPvuII, AlwNI, and EcoRI ribotypes that were very similar tothose of S. mitis strains. Those findings suggest that it maybe difficult to distinguish S. parasanguis from S. mitis byribotyping. One explanation for that difficulty might be thatthose species are very closely related.

Ribotypes describe sequence diversity in only small por-tions of the rRNA operon, and apparent diversity will varyaccording to restriction enzymes used. DNA hybridizationand rRNA sequencing provide more-complete informationon similarity between genotypes, although it is important tonote that results of those methods do not always agree (26).Neither approach has been used to compare current S.

N B

up

1,4.23 1 9C C

r- -- ...1- a"

_ -

low

2322 cr

2C2. Do

FIG. 5. Single blot labeled by using Adobe Photoshop 2.0.1 andQuark Express 3.11 software with a Macintosh IIcx computer and aSharp 450 scanner. AlwNI lanes show AlwNI ribotypes for S. mitisATCC 903 and S. parasanguis strains that appear similar to it.BamHI and EcoRI lanes showBamHI and EcoRI ribotypes (respec-tively) for S. sanguis strains and clinical isolate 13379. The BamHIpattern for 13379 is very similar to that for ATCC 10556; the EcoRIpattern for 13379 is more similar to that for 804 than to that forATCC 10556.

VOL. 31, 1993

on April 10, 2019 by guest

http://jcm.asm

.org/D

ownloaded from

2472 RUDNEY AND LARSON

parasanguis and S. mitis strains. S. parasanguis MGH 413was used as a reference strain in a DNA hybridization studyof S. crista (19). However, that strain was not tested againstany of the S. mitis strains included in that panel. Lowhomology has been reported for S. parasanguis SS 898(ATCC 15912) and NCTC 3165 (ATCC 33399), the originaltype strain of S. mitis (42). However, NCTC 3165 is nowconsidered to be a member of S. gordonii (6, 16, 23). Thatreclassification is supported by this study, since strainATCC 33399 showed a PvuII ribotype very similar to thoseof other S. gordonii strains. Other S. mitis strains were notused in that DNA hybridization study (42). A partial 16SrRNA sequence from S. parasanguis was found to be mostsimilar to a corresponding sequence from S. sanguis. How-ever, no S. mitis strain was included in sequencing studies(42). Ribotyping cannot be used to establish whether S. mitisand S. parasanguis are the same or different species, but ourresults suggest that further DNA hybridization or rRNAsequencing studies of both groups may be appropriate.Although species identification was of primary interest in

this study, ribotyping also may be useful for defining sets ofstrains within species. Visual comparison of common bandsin PvuII patterns suggested two major sets of S. gordoniistrains, with the larger set being divisible into three consis-tent subsets by AatII, BamHI, and DraII. Those groupscorrespond in part to biovars 1, 2, and 3 of Kilian et al. (23).However, the ribotype for strain M5 (biovar 2) was moresimilar to that for S7 (biovar 3) than to that for ATCC 10558and Blackburn (biovar 2). Strain ATCC 12396 (biovar 1) alsoappeared more similar to S7 and M5 than did strain ATCC33399 (biovar 3). A similar division of S. mitis and S.parasanguis strains into two sets could be made with AlwNIand EcoRI. The distribution of S. mitis strains in each setcorresponds to biovar 1 (NCTC 10712 and NCITC 12261) andbiovar 2 (ATCC 903) of Kilian et al. (23). Their separation ofS. sanguis ATCC 10556 (biovar 1) and 804 (biovar 2) also isconsistent with AatII, PvuII, and EcoRI ribotypes. How-ever, it is important to note that similarity of clinical isolate13379 to strain ATCC 10556 or 804 depended on whichenzyme was used (although it was most similar to S. sanguisstrains with all enzymes). This suggests that subgroupsestablished on the basis of ribotyping should be consideredtentative until confirmed with multiple restriction enzymesand other methods.Some authors have used the presence or absence of

ribotype bands as a basis for quantitative measures of strainsimilarity (1, 22, 31, 43). We did not attempt that in thisstudy, because estimated band sizes for replicates of thesame strain run in different gels did not always agree.Inconsistency in size estimates may be related to variation inrunning conditions, but disagreements were most pro-nounced for large and small bands. This suggests that effectsof minor variations may be magnified by the logarithmicscale used for size estimation in conventional agarose gels(34). Improvements in band-matching software may facili-tate quantitative analysis of ribotype patterns. However, ourresults suggest that useful information can be gained byvisual comparison of a panel of reference strains.A well-defined reference panel may provide a basis for

genotypic identification of clinical isolates of oral viridansstreptococci. Our results suggest one possible strategy.AatII ribotypes for unknown isolates first could be comparedwith the patterns for classified strains shown in Fig. 1 and 2.The presence or absence of the 2,290-bp band then could beused to distinguish members of the S. oralis-S. mitis and S.gordonii-S. sanguis lineages, while S. crista strains may be

identifiable as distinct from either group. Unique bands inAatII ribotypes also may allow strain identification at thisstage. AatII ribotypes shown in Fig. 1 might be sufficient todiscriminate among S. gordonii and S. sanguis isolates, butPvuII ribotypes would allow preliminary identifications to bechecked by comparison with the patterns in Fig. 3. AlwNIribotypes of isolates showing the 2,290-bp AatII band thencould be compared with Fig. 4 and 5 to separate S. oralisstrains from members of the S. mitis-S. parasanguis group.Phenotypic tests described by Beighton et al. (2) then mightbe used to separate those two species.The above strategy must be regarded as tentative, since

our panel may not provide a representative sample ofribotypes within streptococcal species. However, all strainsused here were previously classified by DNA hybridization,and that may provide added confidence in the suitability ofthe panel. It is encouraging that this set of reference patternscould be used to assign clinical isolate 13379 to S. sanguis.Our current goals are to supplement the panel with addi-tional strains of known affiliation while attempting to identifya large set of oral isolates previously assigned to the sanguisgroup on the basis of colony morphology (32). Successfulclassification of a large proportion of those isolates wouldsuggest that ribotyping can provide a useful approach togenotypic identification of oral streptococci. Ribotyping thencould be applied as a supplement to phenotypic identificationprotocols in clinical studies.

ACKNOWLEDGMENTSThis work was supported by Public Health Service grant DE

08505 from the National Institute for Dental Research.Strains were provided by Mark C. Herzberg and William F.

Liljemark of the School of Dentistry at the University of Minnesotaand by Paula Fives-Taylor and Pamela Wesbecher of the Depart-ment of Microbiology at the University of Vermont. We also thankCharles Schachtele and Jolene Johnson of the School of Dentistry atthe University of Minnesota and Susan K. Harlander and AndreaBaloga of the Department of Food Science and Nutrition at theUniversity of Minnesota for helpful discussions.

REFERENCES1. Baloga, A. O., and S. K. Harlander. 1991. Comparison of

methods for discrimination between strains of Listeria monocy-togenes from epidemiological surveys. Appl. Environ. Micro-biol. 57:2324-2331.

2. Beighton, D., J. M. Hardie, and R. A. Whiley. 1991. A schemefor the identification of viridans streptococci. J. Med. Micro-biol. 35:367-372.

3. Bingen, E. H., E. Denamur, N. Y. Lambert-Zechovsky, and J.Elion. 1991. Evidence for the genetic unrelatedness of nosoco-mial vancomycin-resistant Enterococcus faecium strains in apediatric hospital. J. Clin. Microbiol. 29:1888-1892.

4. Blumberg, H. M., D. Rimland, J. A. Kiehlbauch, P. M. Terry,and I. K. Wachsmuth. 1992. Epidemiologic typing of Staphylo-coccus aureus by DNA restriction fragment length polymor-phisms of rRNA genes: elucidation of the clonal nature of agroup of bacteriophage-nontypeable, ciprofloxacin-resistant,methicillin-susceptible S. aureus isolates. J. Clin. Microbiol.30:362-369.

5. Cook, P. R. 1984. A general method for preparing intact nuclearDNA. EMBO J. 3:1837-1842.

6. Coykendall, A. L. 1989. Classification and identification of theviridans streptococci. Clin. Microbiol. Rev. 2:315-328.

7. Coykendall, A. L., and A. J. Munzenmaier. 1978. Deoxyribonu-cleic acid base sequence studies on glucan-producing and glu-can-negative strains of Streptococcus mitior. Int. J. Syst. Bac-teriol. 28:511-515.

8. Coykendall, A. L., and P. A. Specht. 1975. DNA base sequence

J. CLIN. MICROBIOL.

on April 10, 2019 by guest

http://jcm.asm

.org/D

ownloaded from

RIBOTYPES OF ORAL STREPTOCOCCI 2473

homologies among strains of Streptococcus sanguis. J. Gen.Microbiol. 91:92-98.

9. De Buyser, M.-L., A. Morvan, F. Grimont, and N. El Solh. 1989.Characterization of Staphylococcus species by ribosomal RNAgene restriction patterns. J. Gen. Microbiol. 135:989-999.

10. Demuth, D. R., E. E. Golub, and D. Malamud. 1990. Strepto-coccal-host interactions: structural and functional analysis of aStreptococcus sanguis receptor for a human salivary glycopro-tein. J. Biol. Chem. 265:7120-7126.

11. Douglas, C. W. I., A. A. Pease, and R. A. Whiley. 1990.Amylase-binding as a discriminator among oral streptococci.FEMS Microbiol. Lett. 66:193-198.

12. Dowson, C. G., A. Hutchison, N. Woodford, A. P. Johnson,R. C. George, and B. G. Spratt. 1990. Penicillin-resistant viri-dans streptococci have obtained altered penicillin-binding pro-tein genes from penicillin-resistant strains of Streptococcuspneumoniae. Proc. Natl. Acad. Sci. USA 87:5858-5862.

13. Fenno, J. C., D. J. LeBlanc, and P. M. Fives-Taylor. 1989.Nucleotide sequence analysis of a type 1 fimbrial gene ofStreptococcus sanguis FW213. Infect. Immun. 57:3527-3533.

14. Frandsen, E. V. G., V. Pedrazzoli, and M. Kilian. 1991. Ecologyof viridans streptococci in the oral cavity and pharynx. OralMicrobiol. Immunol. 6:129-133.

15. Ganeshkumar, N., P. M. Hannam, P. E. Kolenbrander, andB. C. McBride. 1991. Nucleotide sequence of a gene coding fora saliva-binding protein (SsaB) from Streptococcus sanguis 12and possible role of the protein in coaggregation with actinomy-ces. Infect. Immun. 59:1093-1099.

16. Gilmour, M. N., T. Whittam, M. Kilian, and R. K. Selander.1987. Genetic relationships among the oral streptococci. J.Bacteriol. 169:5247-5257.

17. Grimont, F., M. Lefevre, E. Ageron, and P. A. D. Grimont.1989. rRNA gene restriction patterns of Legionella species: amolecular identification system. Res. Microbiol. 140:615-626.

18. Hall, L. M., B. Duke, M. Guiney, and R. Williams. 1992. Typingof Enterococcus species by DNA restriction fragment analysis.J. Clin. Microbiol. 30:915-919.

19. Handley, P., A. Coykendall, D. Beighton, J. M. Hardie, andR. A. Whiley. 1991. Streptococcus crista sp. nov., a viridansstreptococcus with tufted fibrils, isolated from the human oralcavity and throat. Int. J. Syst. Bacteriol. 41:543-547.

20. Hillman, J. D., S. W. Andrews, S. Painter, and P. Stashenko.1989. Adaptive changes in a strain of Streptococcus mutansduring colonization of the human oral cavity. Microb. Ecol.Health Dis. 2:231-239.

21. Hopfl, P., W. Ludwig, K. H. Schleifer, and N. Larsen. 1989. The23S ribosomal RNA higher-order-structure of Pseudomonascepacia and other prokaryotes. Eur. J. Biochem. 185:355-364.

22. Kiehlbauch, J. A., B. D. Plikayatis, B. Swaminathan, D. N.Cameron, and I. K. Wachsmuth. 1991. Restriction fragmentlength polymorphisms in the ribosomal genes for species iden-tification and subtyping of aerotolerant Campylobacter species.J. Clin. Microbiol. 29:1670-1676.

23. Kilian, M., L. Mikkelsen, and J. Henrichsen. 1989. Taxonomicstudy of viridans streptococci: description of Streptococcusgordonii sp. nov. and emended descriptions of Streptococcussanguis (White and Niven 1946), Streptococcus oralis (Bridgeand Sneath 1982), and Streptococcus mitis (Andrewes andHorder 1906). Int. J. Syst. Bacteriol. 39:471-484.

24. Kilian, M., J. Reinholdt, B. Nyvad, E. V. G. Frandsen, and L.Mikkelsen. 1989. IgAl proteases of oral streptococci: ecologicalaspects. Immunol. Invest. 18:161-170.

25. Lane, D. J., B. Pace, G. J. Olsen, D. A. Stahl, M. L. Sogin, andN. R. Pace. 1985. Rapid determination of 16S ribosomal RNAsequences for phylogenetic analyses. Proc. Natl. Acad. Sci.USA 82:6955-6959.

26. Martinez-Murcia, A. J., S. Benlloch, and M. D. Collins. 1992.Phylogenetic interrelationships of members of the genera Aero-monas and Plesiomonas as determined by 16S ribosomal DNAsequencing: lack of congruence with results of DNA-DNA

hybridizations. Int. J. Syst. Bacteriol. 42:412-421.26a.McHale, M. W. (Department of Biochemistry, College of Biolog-

ical Sciences, University of Minnesota). Written communication.27. Moureau, P., I. Derclaye, D. Gregoire, M. Janssen, and G. M.

Cornelis. 1989. Campylobacter species identification based onpolymorphism of DNA encoding rRNA. J. Clin. Microbiol.27:1514-1517.

28. Nyvad, B., and M. Kilian. 1987. Microbiology of the earlycolonization of human enamel and root surfaces in vivo. Scand.J. Dent. Res. 95:369-380.

29. Nyvad, B., and M. Kilian. 1990. Comparison of the initialstreptococcal microflora on dental enamel in caries-active and incaries-inactive individuals. Caries Res. 24:267-272.

30. Perolat, P., F. Grimont, B. Regnault, P. A. D. Grimont, E.Fournie, H. Thevenet, and G. Baranton. 1990. rRNA restrictionpatterns of Leptospira: a molecular typing system. Res. Micro-biol. 141:159-171.

31. Pitcher, D., A. Johnson, F. Allerberger, N. Woodford, and R.George. 1990. An investigation of nosocomial infection withCorynebacteriumjeikeium in surgical patients using a ribosomalRNA gene probe. Eur. J. Clin. Microbiol. Infect. Dis. 9:643-648.

32. Rudney, J. D., M. A. Krig, E. K. Neuvar, A. H. Soberay, and L.Iverson. 1991. Antimicrobial proteins in human unstimulatedwhole saliva in relation to each other, and to measures of healthstatus, dental plaque accumulation and composition. Arch. OralBiol. 36:497-506.

33. Rudney, J. D., E. K. Neuvar, and A. H. Soberay. 1992. Restric-tion endonuclease-fragment polymorphisms of oral viridansstreptococci, compared by conventional and field-inversion gelelectrophoresis. J. Dent. Res. 71:1182-1188.

34. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory Press, Plainview, N.Y.

35. Schmidhuber, S., R. Kilpper-Balz, and K. H. Schleifer. 1987. Ataxonomic study of Streptococcus mitis, S. oralis, and S.sanguis. Syst. Appl. Microbiol. 10:74-77.

36. Smith, C. J., and D. R. Calihan. 1992. Analysis of rRNArestriction fragment length polymorphisms from Bacteroidesspp. and Bacteroidesfragilis isolates associated with diarrhea inhumans and animals. J. Clin. Microbiol. 30:806-812.

37. Snipes, K. P., D. C. Hirsh, R. W. Kasten, L. M. Hansen, D. W.Hird, T. E. Carpenter, and R. H. McCapes. 1989. Use of anrRNA probe and restriction endonuclease analysis to fingerprintPasteurella multocida isolated from turkeys and wildlife. J.Clin. Microbiol. 27:1847-1853.

38. Stull, T. L., J. J. LiPuma, and T. D. Edlind. 1988. A broad-spectrum probe for molecular epidemiology of bacteria: riboso-mal RNA. J. Infect. Dis. 157:280-286.

39. Tardiff, G., M. C. Sulavik, G. W. Jones, and D. B. Clewell. 1989.Spontaneous switching of the sucrose-promoted colony pheno-type in Streptococcus sanguis. Infect. Immun. 57:3945-3948.

40. Tudor, J. J., L. Marri, P. J. Piggott, and L. Daneo-Moore. 1990.Size of the Streptococcus mutans GS-5 genome as determinedby pulsed-field gel electrophoresis. Infect. Immun. 58:838-840.

41. Welborn, P. P., W. K. Hadley, E. Newbrun, and D. M. Yajko.1983. Characterization of strains of viridans streptococci bydeoxyribonucleic acid hybridization and physiological tests. Int.J. Syst. Bacteriol. 33:293-299.

42. Whiley, R. A., H. Y. Fraser, C. W. I. Douglas, J. M. Hardie,A. M. Williams, and M. D. Collins. 1990. Streptococcusparasanguis sp. nov., an atypical viridans Streptococcus fromhuman clinical specimens. FEMS Microbiol. Lett. 68:115-122.

43. Woods, T. C., L. 0. Helsel, B. Swaminathan, W. F. Bibb, R. W.Pinner, B. G. Gellin, S. F. Collin, S. H. Waterman, M. W.Reeves, D. J. Brenner, and C. V. Broome. 1992. Characteriza-tion of Neisseria meningitidis serogroup C by multilocus en-zyme electrophoresis and ribosomal DNA restriction profiles(ribotyping). J. Clin. Microbiol. 30:132-137.

VOL. 31, 1993

on April 10, 2019 by guest

http://jcm.asm

.org/D

ownloaded from