Analysis of relationships among isolates of Citrobacter diversus by using DNA fingerprints generated by repetitive sequence based primers in the polymerase chain reaction

0095-1137/92/112921-09$02.00/0

CopyrightX 1992, AmericanSociety forMicrobiology

Analysis of Relationships among Isolates of Citrobacter diversus by

Using

DNA

Fingerprints

Generated by

Repetitive

Sequence-Based

Primers

in the Polymerase Chain Reaction

CHARLESR. WOODS,

JR.," 2t

JAMES

VERSALOVIC,3

THEARITHKOEUTH,3

ANDJAMES R.

LUPSKI2,3*

C. T. ParkerLaboratory, TexasChildren's Hospital,1and Institute for Molecular

Genetics3

andDepartmentof

Pediatrcs,

2Baylor College of Medicine, Houston, Texas77030

Received 12June 1992/Accepted 27 August 1992

Oligonucleotide probeswhich match consensus sequences of the repetitive extragenic palindromic (REP) elementhybridize to genomic DNA of diverse bacterial species. Primers based on the REP sequence generate complex band patterns with genomic DNA in the polymerase chain reaction (PCR), a technique named REP-PCR. We used REP-PCR with genomic DNA to fingerprint 47 isolates of Citrobacter diversus. Previously, 37 were assigned electrophoretic types (ETs) by multilocus enzyme electrophoresis and 35 were evaluated by using outer membrane protein profiles. Fingerprints were compared by visual inspection and bysimilarity

coefficients(SimCs) based on the number of common bands versus total bands between two given isolates. DNA fingerprints were

highly

similar

visually

for patient pairs and outbreak-related sets. SimCs for these were >0.952. Fingerprints of isolates with different ETs

generally

were distinctive. Among 21 unrelated isolates representing 15 ETs,

only

6 of 210 comparisons had SimCs of 20.952. REP-PCR

rapidly

generated DNA fingerprints which were

highly

similar for

epidemiologically

linked isolates ofC. diversus and distinct for

previously

characterized strains within this species. The

ability

of this method to discriminate between C. diversus isolates with the samebiotype was similar to that of multilocus enzyme electrophoresis and outer membrane protein profiles. REP-PCR may be useful in evaluation ofapparent outbreaks of this or other bacterial species which possess these extragenic, repetitive elements.

Traditionally, once bacterial isolates from an apparent outbreak have been determined to be the same species, further evaluation for similarity or relatedness has been based on phenotypic properties. While often sufficient, these methods are not always adequately sensitive to distinguish unrelated strains with similar phenotypes among the

proper-ties assessed (33). Molecular genetic methods, including

plasmid profiles, restriction endonuclease digestionpatterns ofplasmidand chromosomalDNA, and DNA hybridization, have been useful in evaluation of outbreaks of a variety of bacterial pathogens inrecentyears (1,5, 15, 26, 33, 34).

We describe the application of a recently developed method, repetitive element sequence-based polymerase

chain reaction (rep-PCR) (31), for epidemiologic analysis. This method generates fingerprints from genomic DNA

whichidentify specific strains withinabacterial species(31, 32). The term rep-PCR refers to the general methodology involvingtheapplicationofoligonucleotide primers basedon families of short, extragenic repetitive sequences. These

repetitive DNA sequences are dispersed throughout the

prokaryoticchromosome and appeartobeconservedamong many members ofthefamily

Enterobacteriaceae

aswell as

other bacterial

species

(8, 13,

21,

28, 29,

31).

Consensus

sequence

probes matching

two

repetitive

sequence

families,

the

38-bp

repetitive

extragenic

palindromic

(REP)

element

and the 126-bpenterobacterial repetitive

intergenic

consen-sus(ERIC) sequence,

hybridize

to

genomic

DNAofenteric bacteria, other related

gram-negative bacteria,

and several

distantly related bacterial

species (31).

The

palindromic

*

Corresponding

author.

tPresent address: Department of Pediatrics, Bowman

Gray

SchoolofMedicine, Winston-Salem, NC 27157-1081.

portionsof theserepetitiveelements enable the formation of stablestem-loopstructures,but theirprecise functional role isunclear (21).

Outwardly directed primer setsbasedonREPand ERIC consensussequences canbe used to generateclearly resolv-able bandsbyagarosegel electrophoresis after PCR ampli-fication using template genomic DNA from species which

contain these sequences (31). The terms REP-PCR and ERIC-PCR refer to the application of specific repetitive sequences (REP andERIC,respectively)toestablish DNA

fingerprints byPCR. Bands representamplificationof DNA between adjacent repetitive elements within the approxi-mately5-kb limitation of the polymerase extension. Band

patterns provideDNAfingerprintswhich allowdistinctions

between species (31)and betweenstrains withinspecies(3, 31, 32). Differences in band sizes apparently represent

polymorphismsin the distances betweenrepetitivesequence

elementsin different genomes.

Primers of the same base pair length but based on a

randomly assorted repetitive consensus sequence do not generatebandpatternsorgenomic

fingerprints

(32a).

There-fore,

generation

of

rep-PCR

fingerprints

appearsto

require

thatthe

primers

match the

specific repetitive

sequencesand

doesnotrepresentrandom

priming.

Fingerprints

are repro-ducible frommultiplecolonies from thesame

plate

andfrom

daily samples of serial cultures

(32). Samples

of the same

stock strain ofBacillus subtilis used in neonatal

screening

programs at geographically

widespread

institutions also

yield

reproducible,

highly

similar

rep-PCR fingerprints (32).

We

performed

genomic

fingerprinting

of 47 Citrobacter diversus isolates

by

REP-PCR. C. diversuscauses

sporadic

and

epidemic

neonatal

meningitis

which is associated with anunusually

high

frequencyof brain abscess formation

(11,

2921

on April 12, 2020 by guest

http://jcm.asm.org/

2922 WOODS ET AL.

TABLE 1. Characteristics of C. diversus isolates evaluated in this study

Isolate Source and yr Clinicalsource Biotypea ETbyMEE OMPclassification

Houston 1986 Houston 1986 Houston 1986 Houston 1984 Houston1984 Maryland 1986 Kentucky 1966 Kentucky 1976 Kentucky 1976 Florida 1979 Florida 1979 Houston 1985 Houston 1985 Maryland 1986 Houston1986 Houston1986 Maryland 1985 Louisiana1970 Washington,D.C., 1969 Maryland 1983

St. Louis1991 Delaware 1991 Houston1987 Houston1988 Houston 1989 Houston1986 Houston1986 Delaware1966 Maryland 1985 Maryland 1985 Maryland 1983 Maryland 1983 Maryland 1983 Alabama 1984 Alabama1984 Tennessee 1983 Tennessee 1985 Delaware 1975 Kentucky1972 Maryland 1985 Maryland 1986 Dallas1972 Houston 1986 Maryland1985 Houston1985 Tennessee 1985 Maryland 1986 CSF CSF Wound CSF CSF Stool CSF CSF CSF CSF CSF Respiratorytract Respiratorytract Stool Blood Blood Stool CSF CSF Stool CSF CSF Blood CSF Blood Respiratorytract Blood CSF Stool Stool CSF CSF CSF Vagina CSF CSF Urine CSF CSF Stool Stool CSF Superficialabscess Stool Respiratorytract Peritonealfluid Stool

aBiotypesarebasedonfermentationpatternoffoursugars:dulcitol,rhamnose,sorbose, andsucrose.

b_,notdetermined.

16). C. diversus outbreaks have been evaluated with bio-types (6, 9, 10, 20, 25, 34), serotypes (10, 34), and

antibio-grams(6,9, 25, 34). These methods, while helpfultoidentify and track epidemic strains in small outbreaks, correlate poorlywithoneanother when appliedon alarger scale (22).

Previously, 35 of these isolateswere evaluated byusing

outer membrane protein (OMP) profiles (18) and 37 were

evaluatedby multilocusenzymeelectrophoresis(MEE) (19).

The bank includes four pairs of isolates from the same

patients, oneset ofisolates froma mother-infant pair, and

threesets ofisolates from outbreaks of meningitis. Distinct geographic regions of theUnited Statesarerepresentedwith

16 isolates from the Houston, Tex., area during a 5-year

period and 17 from the Maryland-Delaware-Washington,

D.C.,areaduringa25-year period. This studydemonstrates thatgenomic fingerprinting by REP-PCR is a rapid,

repro-ducible, and useful methodto examine isolates ofbacterial species from apparentoutbreaks ofdisease.

MATERIALS ANDMETHODS

Bacterialisolates. C. diversusisolates(Table 1) have been collected in the C. T. ParkerLaboratoryofTexasChildren's Hospital since 1984. Isolates with known OMP or MEE

typeshavebeendescribedpreviously (18, 19). Isolate 5595

wasprovided by N. Middelkamp, St. Louis, Mo., and 5596

wasprovided by S. Eppes, Wilmington, Del. Other isolates

notpreviously describedwereobtainedfrom local Houston

b 6 6 6 7 7 16 16 16 16 16 14 12 11 11 11 Fig. 1 4632 4637 4635 4023 4036 4583 4410 4511 4512 4509 4510 4277 4278 4577 4472 4476 4576 4408 4407 4578 5595 Fig. 2 5596 4749 5043 5115 4397 4573 4406 4574 4580 4514 4515 4516 4428 4429 2988 4432 4513 4405 4582 4579 4409 4485 4581 4310 4431 4575 F+ A-F+ D+ D+ B+ B+ B-B+ B- G- D-B+ B+ F-d d d d d e c d d d d d d d a a a a a a a e e d e e e e e e e e e e e e f e c c c e c e c c c C- B-B+ B-A+ A+ C- B-B+ D-D+ G-E+ G- D- G-15 15 15 15 15 15 15 15 15 15 15 15 10 9 9 8 5 4 3 2 2 1

J. CLIN. MICROBIOL.

on April 12, 2020 by guest

http://jcm.asm.org/

hospitalsfrom 1986 to 1989. The isolates are from diverse

clinical sites: cerebrospinal fluid(CSF) (n = 23), stool (n =

10), blood (n = 5),respiratory tract(n = 4), vagina (n = 1), peritoneal fluid (n = 1), urine (n = 1), a wound (n = 1),and

anabscess (n = 1).

Related sets ofisolates are 4023 and 4036 (CSF cultures fromthe sameinfant5 weeks apart); 4277and4278(tracheal aspirate culturesfrom the same patient several hours apart on the same day); 4472 and 4476 (bloodcultures from the sameinfant3daysapart);4632and 4637(CSFcultures from

the same infant 10 days apart) and 4635 (wound culture

registered thesame day as 4637 from adifferentpatient in the samehospital); 4428 and 4429(maternal vaginalcultureand her infant's CSFculture); 4509and 4510(CSF cultures from different infants in the same hospital 5 weeks apart) (10);

4511 and 4512 (CSF cultures from different infants in the same locationin the sameyear); and 4514,4515, and 4516 (CSF cultures from three different infants in the same

hospitalover a4-monthperiod) (20).

All isolates were maintained as -70°C frozen glycerol

stocks. Biotypes of C. diversus isolates not previously reported (17) were assigned according to the scheme of Richard etal. (27).

REP-PCRfingerprints. Genomic DNA samples from iso-lateswereprepared and quantitatedaspreviously described (31). REP-based primersequences REP1R-I (5'-IIIICGICG ICATCIGGC-3') and REP2-I

(5'-ICGICI'rATCIGGCCT

AC-3')were previously described (31). PCR amplifications

wereperformed inanautomatedthermalcycler (first-gener-ation Perkin-Elmer Cetus DNA thermal cycler) with an

initial denaturation

(950C,

7 min) followed by 30 cycles of

denaturation (90°C, 30 s), annealing (40°C, 1 min), and extension

(650C,

8

min)

withasingle final extension

(650C,

16

min).

Five-microliter samples of each PCR mixture were

electrophoresed directly in 1% agarose gels containing 40

mMTris

acetate-i

mM EDTAbuffer and 0.5,gofethidium

bromide perml. Gelswere photographedwith a 60-s expo-sure to Polaroid type55 film.

Analysis of band patterns. Sizes of bands generated by

electrophoresis of the PCR amplificationswere assignedby direct comparison to concurrently run 1-kb DNA ladder

standards (BRL Life

Technologies,

Inc.). Thepresence or

absence of bandswithin agel lanewas determinedby two

observers(C.R.W. and J.V.) withone

fully

blindedtoexact

gel lane locations of

epidemiologically

linked isolates.

DNA fingerprints of isolates first were compared

for

similarity by visualinspection of band patterns. Unlike the

SimCanalyses

below,

entire laneswere

compared

without sizeranges

by

visual

inspection.

Fingerprints

were

consid-ered

highly

similar when all visible bands

represented

in each isolate had the same apparent

migration

distance. Variations in

intensity

and shape did not represent differ-ences. The absence of up to two bands from one isolate when allothervisiblebands in the

fingerprints

had

matching

positions was allowedbefore isolates were considered

dif-ferent

by

visual

inspection.

Bands in the 900-to3,200-bprange of selectedC. diversus

isolateswere

analyzed

more

rigorously by

themethodofvan

Soolingen

et al.

(30),

which determines a coefficient of

similarity

(SimC)

between

fingerprints

of two isolates as

follows:

k k

SimCAB

=

(ai

+

bi

-

[ai

-

bi])l

I

(ai

+

bi),

i=1 i=1

where

a,

and

b,,

theintensitiesofband i inpatternsAand B, areeither0(not present) or 1 (present) and k is the number

of distinct bands present among patterns A and B. This

coefficient ranges from 0 to 1.0, where 1.0 represents the

identical presence and position of all bands in the two

fingerprintsbeingcompared.

RESULTS

Evaluation offingerprints of C. diversus isolates from the samepatient or

outbreak

Isolates epidemiologically linked

byrecoveryfrom the same patient or outbreak were used as

the standard of similarity to assess REP-PCR-based DNA

fingerprinting. Isolates of each pair or set have the same

biotype. When available forboth members of a pair (Table

1), electrophoretic types and OMP profiles are identical

within pairs, except for 4277 and 4278, which have OMP

profiles ofB-andB+, respectively.

For eachofthe four pairs of isolates from the same patient

and the isolates from the one mother-infant pair, band

patterns

werehighly similaronvisualinspection(Fig. 1 and 2, bracketed lanes). In several instances one band was

absentfrom the

analysis

rangeof the

pattern

in one member

ofa pair. Inthe mother-infant isolate pair (4428 and 4429; Fig.

2),

one lane had asmear

correlating

withtwobandsin

the other member of the pair. Otherwise, the remaining bandswere present in both members of each of these five

pairs. When differences in band intensity were present in

closely related isolates, these

generally

were evident throughout thefingerprint, suggesting differencesin amounts

of amplified DNA in the samples. Bands were considered

present orabsent; differences in intensitywereignored. Among the three outbreak-related sets,

patterns

again werehighly

similar

onvisualinspection.Among thetriplet of

4514, 4515, and 4516(Fig.2), the last differed from theother twobythe absenceofonebandin analysis range. Thepair of

4511 and 4512(Fig. 1) differedby the absence of one band

from thefingerprintof4511.Thepairof4509 and 4510 (Fig.

1)

showed the most apparentdifferences among the related strains:twobandsgreaterthan3,000 bpwereabsent in4509,

and two other bands in the2,300- to2,700-bprange also were

of low intensity. The remainder of their bands in each

fingerprint were in matching positions. ERIC-PCR-based

DNA fingerprints of 4509 and4510 wereverysimilar (data

not shown).

SimCsfor the related isolates (seven comparisonsamong

pairs and three comparisons among the outbreak

triplet)

rangedfrom0.952to 1.00withone

exception:

theSimCfor

4509 and 4510 was 0.889. SimCs of >0.952 represent no more thanoneband differencebetween two

fingerprints

in the analysis range. We consideredthis

degree

of

similarity

between two isolates to be sufficient and necessary to

identify them as the same

strain,

or

clone,

ofC. diversus. Analysis ofDNA

fingerprints

generated by

ERIC-PCR

sup-ported

thehigh

degree

of

similarity

among

epidemiologically

relatedisolates

(data

not

shown).

Comparison ofrep-PCRfingerprintsof C. diversus isolates with MEE and OMP classifications. Visual

inspection

of

rep-PCR

fingerprints

of the 47C. diversus isolates

suggested

18 different

patterns (Fig.

1 and

2). Among

the 29

epidemi-ologically unrelated isolates for which both ET and OMP

classifications are known

(18,

19),

REP-PCR

provided

a

similar

degree

ofdiscriminationwithin

biotype

groups

ver-sus MEEand OMP

profiles

(Table

2).

Visual

inspection

of

fingerprints of these 29 isolates delineated 16 REP-PCR

on April 12, 2020 by guest

http://jcm.asm.org/

2924 WOODS ET AL.

1 2 3 4 5 6

H]

F

]I

m

n

CIn4

o

CM t

C0

0 LCO l Il

7 8 9 10 11

HF]

Co v- 1 cn o I O L t

I In I Is

12 13 14 15 16

H

_I

H

X - N to

CM CM LO IT T

l*

1 IT

lt

Il 17 18

oL.l 19 20 21 E:

o -o5

u_

- CO LO - n

o r C )-m

vIl

_I~ v LO CD

I) I) a I

I

FIG. 1. REP-PCR-basedDNAfingerprintsof21C. diversus isolates. Bracketed isolatesrepresent pairsfrom the samepatient (4632and

4637;4023 and4036;4277 and4278;4472 and4476)orfrom thesamemeningitisoutbreak(4511and4512;4509and4510). Isolates 4036to

4410 (lanes5to7)areET6;isolates 4511and4512areET7;isolates 4509 to 4577(lanes 10to14)areET16;isolate4472is ET14; isolate

4576is ET12;and isolates 4408to 4578(lanes18 to20)areET11.

types. MEE and OMP profile classifications delineated 14 and 11 types,respectively. Eachofthe threetypingmethods occasionallydiscriminated between isolatesidentifiedasthe

same strainbyboth of the othermethods.

REP-PCR types determined by visual inspection

corre-lated directly with ETs with the following exceptions: (i) fingerprintsof ETs 1 and 2werethesame, (ii)fingerprintsof each of the three ET 6 isolates were different, (iii)

finger-prints of ETs 8 and 9 were the same, and (iv) therewere

threefingerprintsamongthe 12 ET 15 isolates. Whenmore

thanoneETwasrepresentedbyasingleREP-PCR type, the ETs alwayshad the same biotypebut notalways thesame OMPprofile. ETs 1 and 2 and ETs 8 and 9 differ byallelic polymorphismofonly1of the20enzymesused in the MEE analysis (19). REP-PCR corresponded better with MEE than did either method with the OMP profile classification: mul-tiple ETs and REP-PCR types werepresent in each ofthe four OMPprofilegroupswith morethan twoisolates.

RelationshipsamongC.diversus isolateswithdifferent ETs.

C. diversus isolates with different ETs by MEE (19) were

usedasthestandard ofdifferencetoassessREP-PCR-based

fingerprinting within this one species. The results of the

visualinspection described above suggested thatREP-PCR

wasabletodistinguishbetweenmostMEE-defined strainsof C. diversus. SimCs were determined for the 210 possible

comparisons among a 21-isolate subset, which represented the15 ETs availabletous(wedidnothaveanET13 isolate

in our strain bank). All three ET 6 isolates were included since these had different biotypes and a SimC of <0.950

amongthemselves. Two unrelated isolates from each of ETs

11, 15, and 16 and both ET 9 isolateswere included. Only

oneoftwoET 2isolateswasincluded since these hadhighly

similar fingerprintsonvisual inspection and a SimC of1.0 relativeto each other.

Only 6 of 210 relationships had SimCs of >0.952 (Table3).

One was between isolates with the same ET. Four others werebetweenisolates with thesamebiotype. ETs oftwoof thesepairs (4575 and 4431; 4579 and 4582) again are

poly-morphicatonly 1 of 20enzymesusedin the MEE classifi-cation(19). This confirmed the visual impression that REP-PCRwasabletodistinguishmostMEE-defined strains of C.

diversus.

RelationshipsofC. diversus isolates with thesameET.As described above, isolates with the same ET generallyhad

similarDNA fingerprintsonvisualinspection (Fig. 1 and 2;

see Table 1 for ETs of isolates). Among the ET 15 group

(Fig. 2), all isolates had similarREP-PCR-based fingerprints exceptforisolate 4406 (from 1966; the oldest member ofthe group). The fingerprints of four isolates, 4428, 4429, 2988, and 4432 (from Alabama and Tennessee during a 3-year

period), alsoappearedslightly different relativetotheother ET 15 isolates. These had SimCs of >0.956 among

them-selves and hadSimCs of<0.917 with all other ET 15isolates except4574 (Table 4). SimC analysis suggestedtwo strains

a1)

-o ~0

nu

-6108

bp

5090 bp

-

4072

bp

3054 bp

-2036

bp

-1636

bp

W-1018

bp

J.CLIN. MICROBIOL.

on April 12, 2020 by guest

http://jcm.asm.org/

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

CD Cn to Il cn CD v 0 v o tOCDco cOco N mo

a) v v_ o 0) I0N TO 1- (N a:)-n 0

N- tCln " LO tO to aw 0D LO v

i q et v v1 q Iv q q Iv Ir q

I I I I I I I

I I

I I

I

I

19 20 21 22 23 24 25 26

co rN

I

0) to) v- 0 v- LO

c0 Co XC

N-t I I Io

i

- 6108bp

-- 5090 bp

-- 4072 bp

3054 bp

2036bp

1636bp

1018bp

-506,517 bp

FIG. 2. REP-PCR-based DNA fingerprintsof26 C.diversus isolates. Bracketed isolatesrepresentasetfromameningitis outbreak (4514,

4515,and4516)or amother-infantpair (4428and4429). Isolates 4397to4432(lanes5 to16)areET 15;isolates 4513, 4405,4582, and 4579

areETs10, 9,9,and8, respectively; andisolates4409to4575 (lanes 21to26) areETs 5, 4, 3, 2, 2, and 1, respectively.

among the other seven isolates, correlating with a

geo-graphicsourceofHouston or Maryland. Thus, visual

inspec-tion suggested three fingerprints, and SimC analysis

sug-gested a fourth strain (Table 4). All of the ET 15 isolates were biotype e except 4432, which was biotype f. Four

different OMPprofileswererepresented bytheseisolates.

SimCs of the unrelated isolates of ETs 16, 11, 9, and 2 were0.909to0.952, 0.909to0.956, 0.909, and1.00,

respec-TABLE 2. Ability of REP-PCRversus MEE andOMPprofiles to discriminateamong C. diversus isolates

No. of strainsidentifiedby: Biotype No.of unrelated

group isolatese ET byMEE OMPprofile REP-PECR type

a 5 3 4 3

c 8 6 4 4

d 5 3 4 3

e 11 4 7 6

Total 29 14c 11 16

aThisanalysisisbasedonlyonisolates for which bothETand OMPprofile areknown.

bREP-PCR typeswerebasedondifferences in fingerprintsapparent on

visualinspection.

cThe ET 3isolate4581was notincluded becauseits OMP

profile

isnot

known.

tively. These SimCs confirmed the impression from visual inspection thatfingerprints within these ETs were similar,

although not always to the same degree as those of the

epidemiologically linked pairs. The members withineach of these ETgroupshad thesamebiotype.

The three ET6 isolates are unrelated epidemiologically

and have different biotypes, although two have the same

OMP profile (Table 1). Visually, their fingerprints were

different(Fig. 2).SimCsamongthesethreeisolatesfailedto reach the level ofsimilarityobserved amongthe

epidemio-logicallyrelated isolates (Table 3).

Analysis of Houston-area isolates and isolates without knownETsorOMPprofiles.FourpairsofunrelatedHouston

isolates with the samebiotype hadfingerprintswhichwere

highlysimilarvisuallyand had SimCs of20.952.Thesewere

4036 and4635,4278 and4635,4397 and4573,and4749 and 5115.Fingerprintsofbiotyped isolates 4632 and4635,which

wereobtained from differentpatientsinthesamehospital10 daysapart,wereslightlydifferentvisuallyand hadaSimCof 0.917. This comparison suggested that these temporally related isolates were not the same strain and arose from differentepidemiologic sources.

Isolate 5595, a biotype a strain (unknown ET) from St. Louis in1991,hadaSimCof 1.0 with isolate4407,anET11 biotypeastrainfromWashington, D.C.,in1969.Theentire fingerprints of these two isolates were highly similar on visualinspection.Isolate5596,abiotypeestrain(unknown ET)fromDelaware in1991,hadaREP-PCR-based SimCof

(¶3

I:S

-L.! C)

o i

v X

6

I

on April 12, 2020 by guest

http://jcm.asm.org/

r- r-H tm r-H o F- %C c £ot- N a, V- Oc tn) o tn

tH

ooooooooooooc

ooc

t t-oo ot ON7 -4

Nd

o oo r-w)r4o

d£

7NC bN

eo

N

ch

bn00

Ot

1-

r0r0

00 w w C~O 10o08%O

§

r- r0 OW) 00 0 0

N

t- en

r-

C1

N

m

W)

w

c0

Fc

r

0

4tn F o00 00 ON r- I -%C r- t- %O£ 00 ON

C\

00 00000000 00000

_ 0 £cOo en 47 CIONOmF\

:4,t t %O 00 r- 00 t1\, st om^ \ W) tf tn£ F oFooc Fc ro it WI £ oF

_~~~~en0000000000000000,

t-,_ t1 ee£ \0 ON O0N00C

1rh__F tt-^ NN1o°c \2°N

V-t \. o t-£ b o o r'-oo ) W) WIW)%O o

H~~~~~t t-r000000000 00 00 00 0

00 (YONON N O0WIen %ONm

Co tn 'X\OO-O O-O-OO0t-O OOO

o o - co o o- oq oooo o o

o

._

°°

N°

m0 8

°

°e

ON N 0

Z tO F00\ \ 00t- t0000 00 00 00

0 u

caO I--, m mW) Qr- \FO

r

Q~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~r

v~~~~W

~~~~~~~~~~~~;

00000tf) t00 Q =

W)

'o R

"o

\9

CiN

Ch

!St

C)~~~~~~~~~~~~~

Q0\

.0

\J \0 00 \C0 OHNtmmS

C; _c;c;vNc;

m .° mH m oo£ e o o m N^ oo b£ tsovaaJD

-_D F ^ oo X o ^ oo H b o oo b bbNH4t~~~=1r

urtv)etetom et m m s . m m m t m t m t m B Q X~~~r

en te te te te-e ttte 0

2926 WOODS ETAL. J. CLIN. MICROBIOL.

on April 12, 2020 by guest

http://jcm.asm.org/

TABLE 4. REP-PCR-based SimCs amongC. diversusisolateswith ET15a

SimICofb: Strain

4573 4406 4574 4580 4514 4515 4516 4428 4429 2988 4432

4397 1 0.750

f0.96]

0.917 0.917 0.917 0.870 0.917 0.917 0.917 0.870

4573 11 0.750 10.9601 0.917 0.917 0.917 0.870 0.917 0.917 0.917 0.870

4406 0.720 0.667 0.667 0.667 0.696 0.750 0.750 0.750 0.696

4574 0.960 0.960 0.960 0.917 0.960 0.960 0.960 0.917

4580 1.00 1.00 0.956 0.917 0.917 0.917 0.870

4514 1.00 0.956 0.917 0.917 0.917 0.870

4515 0.956 0.917 0.917 0.917 0.870

4516 0.956 0.956 0.9561 0.909

4428 0.956 1.00 0.956

4429 1.00 0.956

2988 0.956

a Visual inspection suggested threefingerprints:4406, 4428 to 4432, and the others. SimC evaluation suggests that 4573 and 4397 may represent a fourth strain inthis group.

b Double boxes indicate closely related groups of strains; single boxes indicate groups of strains with overlappingsimilarities.

1.0 with 4516, an ET 15 biotype e isolate from the same

geographicareain 1983.

DISCUSSION

C. diversus isolates whicharecloselylinked epidemiolog-ically and have the same ET and OMP profiles had

finger-prints generated by REP-PCR whichappeared highly similar

byvisual inspection. This visual impressionwas confirmed

by quantitative analysis using SimCs.Conversely, isolates of C. diversus with different ETs generally had

REP-PCR-based fingerprints which differed significantly by visual inspection and by analysis using SimCs. REP-PCR provided a degree ofdiscrimination among C. diversus isolates with the same biotype similarto thatprovided by MEE (19) and

OMPprofiles (18). REP-PCRtypescorrelatedwell with ETs

withfew exceptions.

Highly similar fingerprints from C. diversus isolates

ob-tained fromdistinct geographicregions and separated in time

by more than 20 years (5595 and 4407) suggest that the locations of the REP elements within the genome can be verystable over time. A similar observation with respect to

the stability of REP-PCR fingerprints over time has been madewithB.subtilisstrains used innewborn screening (32).

DNAfingerprintsbased on arepetitive sequence from

My-cobacterium

tuberculosis, IS986, were identical in strains passagedfor 6 months in culture and in asingle patientover aperiod of1 year,indicating the stability of repetitiveDNA

sequence-basedpatterns (30).

Fingerprints of unrelated C. diversus isolates with the same ETand samebiotype werevery similar in all cases,

althoughoften less sothan

fingerprints

of the

epidemiologi-cally linked

pairs.

Mostof the

highly

similar

pairs

ofisolates

wererecoveredfrom the samegeographicareawithinaspan

of weekstomonths. Isolates fromdifferent

geographic

areas with the same ET which haveSimCsfrom 0.870to0.950may represent the gradual divergence ofprogeny of the parent clone.

Biotypes, while simple to determine for this

species,

providetoolittlediversityforuse as an

independent

system for epidemiologic assessment of C. diversus isolates

(18).

Outbreak-relatedisolates express thesame

biotype (18),

but thedifferences in the REP-PCR

fingerprints

of isolates 4632 and 4635 from Houston support

previous

observations

(9)

thatisolates of thesame

biotype

may berecovered

during

a

short time period in the same institution yet represent

different strains.

Plasmid profiles havebeen usefulinfour outbreaks with small numbers ofisolates(6,9, 20,34). Plasmid profiles may

change within a given strain over time (9, 34) and do not

correlate well withbiotypes ofserotypes orC. diversus (22). Differentstrains from the same locale with different biotypes

alsomay have the same plasmid profile (9). Chromosomal

restriction endonuclease digestionmay correlate with sero-type but appears to be too cumbersome for use with a large

number of C. diversusisolates (22).

Epidemiologically related C. diversus isolates usually have identicalprofilesamong a group of seven OMPs used in theclassificationscheme of Kline et al. (18). In our experi-encedetectionof thethreeminorOMPs used in this scheme

is variable, whichmay affect reproducibility of this method. Anexampleofthismay be the respiratory isolates 4277 and

4278, which were recovered from the same patient on the same day and have OMP profiles of B- and B+,

respec-tively. The difference between these OMP profiles is the presence of the 32-kDa minor OMP which is associated with

isolatesrecovered from the central nervous system (17).

MEE showed that C. diversus exists as a polyclonal

population with atleast 16 distinct strains (19). MEE

con-firmedthe clonal relatedness of the sets ofepidemiologically

linked isolates used in this study to which it was applied. MEEis apowerfulresearch tool but because of its complex-ity isnotlikelytobecomewidelyavailable forstudyof local

outbreaks ofbacterial disease in atimelymanner.

An ideal method of subtyping bacterial isolates from a

givenspecies wouldbesimple, rapid, sensitive, andhighly discriminatory

(4, 24).

These featuresrarelyareencountered in a single method. The DNA-based

fingerprints generated

by REP-PCR, especiallyas PCR technology

improves

and

sample

preparation

becomes

simpler,

may be abletofulfill these tenets.PCR-based

techniques

also

forgo

the

necessity

for cultivation whichmaybe crucial with fastidious

patho-gens andobligateintracellular bacteria.

Ribotyping also appears useful to

distinguish

strains within a species (14, 23,

35)

and shows greater discrimina-torypower among isolates of the same

species

than MEE whencompareddirectly

(2, 35),

although

the

epidemiologic

relevance of this observation is unknown.

Ribotyping

re-quiresrestriction endonuclease

digestion,

Southern

blotting,

and

hybridization

oflabeledribosomal DNA

(12),

whichmay

on April 12, 2020 by guest

http://jcm.asm.org/

2928 WOODS ET AL.

make this method less amenable than

rep-PCR

for routine clinical use in

epidemiologic

evaluations.

Biotyping

remains an

appropriate

first step for evaluation of apparent outbreaks of C. diversus with the caveat that different strains with the same

biotype

may exist

concur-rently

within the same locale. REP-PCR appears to be a

useful

adjunct

in terms of its

ability

to discriminate among unrelated isolates with the same

biotype

while

demonstrat-ing

a

high degree

of

similarity

between strains which are

epidemiologically

linked. REP-PCR may be an appropriate

nextlevel of

analysis

toevaluate relatedness ofC. diversus isolates when a series of isolates with the same

biotype

is encountered. ERIC-PCR has also been useful for

distin-guishing

strains ofC. diversus

(data

not

shown).

REP- and

ERIC-PCR methods which use whole cells

directly

from brothcultures orcolonies on

plates

allow discrimination of C. diversus isolates within several hours

(data

not

shown).

The

quantitative analysis

using

SimCwas undertaken to

provide

a more

objective

analysis

than visual inspection

alone. These results confirmed that a

high

degree

of visual

similarity

of the DNA

fingerprints

oftwoisolatesgenerated

by REP-PCR, especially

when these have thesamebiotype,

is sufficient to consider them the same strain. The visual distinctions of

fingerprints

of different strains also were

confirmed

by

SimC.

Therefore,

visual

inspection

alone,

without themore

rigorous

mathematical

comparisons,

may

be

adequate

for routine

epidemiologic

use of

rep-PCR,

at

least for C. diversus.

rep-PCR,

based on REP and ERIC sequences, has been

applied successfully

to assess relatedness of isolates within

the

species

B. subtilis

(32),

Rhizobium meliloti

(3),

and

Enterobacteraerogenes

(7).

To

date,

these evaluations have involved a limited number of isolatesor lackeda

compara-tive standard. Results of this

study

withC. diversus isolates addfurther evidence that

rep-PCR

maybe

broadly

applica-bleto

fingerprinting

bacterialgenomesand for

epidemiologic

evaluations of outbreaks of bacterial

species

which possess these

extragenic repetitive

elements. This method should be

especially

useful among members of theEnterobacteriaceae

as both ERIC- and REP-based

oligonucleotide

primers

are able to generate

sufficiently

complex

DNA band patterns frommostgenera of this

family

(31).

ACKNOWLEDGMENTS

We thankSheldon Kaplan,MarkKline, and Edward Masonfor criticalreviews.

This researchwassupportedin partbyanAdvancedTechnology

Program grant from the Texas Higher Education Department to J.R.L. J.R.L. acknowledges supportfrom thePEWScholars Pro-gram inBiomedical Sciences. J.V. is arecipientofNational Insti-tutesof HealthpredoctoralfellowshiplF31GM14601-01 SRC-7and also issupported bythe Medical ScientistTrainingProgram at the

Baylor College of Medicine. C.R.W. is a Daland Fellow of the AmericanPhilosophicalSociety.

REFERENCES

1. Blake, P. A., D. T. Allegra, J. D. Snyder, T. J. Barrett, L. McFarland,C. T.Caraway,J.C.Feeley, J. P. Craig, J. V. Lee, N. D. Puhr, and R. A. Feldman. 1980. Cholera-a possible endemic focus in the United States. N. Engl. J. Med. 302:305-309.

2. Brenner, D.J., L. W. Mayer, G. M.Carlone,L. H.Harrison, W. F.Bibb,M.C. de CuntoBrandileone,F.0.Sottnek, K. Irino, M. W.Reeves,J. M.Swenson, K. A. Birkness, R. S. Weyant, S. F.Berkley,T.C.Woods,A.G.Steigerwalt,P. A. D. Grimont, R. M. McKinney, D. W. Fleming, L. L. Gheesling, R. C.

Cooksey,

R.J.Arko,C.V.Broome,andthe BrazilianPurpuric

FeverStudy Group. 1988. Biochemical, genetic, and epidemio-logic characterization ofHaenophilus influenzae biogroup ae-gyptius(Haemophilusaegyptius) strains associatedwith Brazil-ianpurpuricfever. J. Clin.Microbiol. 26:1524-1534.

3. deBruin, F.J. 1992. Use ofrepetitive (repetitive extragenic palindromicandenterobacterialrepetitiveintergenicconsensus) sequences and thepolymerasechain reactiontofingerprint the genomes ofRhizobium meliloti isolates and other soil bacteria. Appl.Environ. Microbiol. 58:2180-2187.

4. Falkiner,F. R. 1988. Epidemiological typing: auser'sview.J. Hosp. Infect. 11:303-309.

5. Farrar,W.E. 1983. Molecularanalysisofplasmidsin epidemi-ologic investigations. J. Infect.Dis. 148:1-6.

6. Finn, A., G. H. Talbot, E. Anday, M.Skros, M. Provencher, and C. Hoegg. 1988. Vertical transmission of Citrobacter diversus from mothertoinfant. Pediatr. Infect. Dis.J.7:293-294. 7. Georghiou, P., J. Versalovic, andJ. R. Lupski. Unpublished

data.

8. Gilson, E.,J.-M. Clement, D.Brutlag, and M.Hofnung. 1984.A family of dispersed repetitive extragenic palindromic DNA sequencesin E. coli. EMBO J. 3:1417-1421.

9. Goering,R.V., N. J. Ehrenkranz, C. C. Sanders, and W. E. Sanders. 1992. Longtermepidemiologicalanalysisof Citrobac-terdiversus in a neonatal intensive care unit. Pediatr. Infect. Dis. J. 11:99-104.

10. Graham, D. R., R. L. Anderson, F. E.Ariel, N. J.Ehrenkranz, B. Rowe, H. R. Boer, and R. E. Dixon. 1981. Epidemic noso-comialmeningitis due to Citrobacter diversus in neonates. J. Infect. Dis. 144:203-209.

11. Graham,D. R., andJ. D. Band. 1981. Citrobacterdiversus brain abscess andmeningitis inneonates.JAMA245:1923-1925. 12. Grimont, F.,and P. A. D. Grimont. 1986. Ribosomalribonucleic

acidgenerestriction patternsaspotential taxonomictools. Ann. Inst. Pasteur/Microbiol. 137B:165-175.

13. Hulton, C. S.J.,C. F.Higgins,and P. M. Sharp. 1991. ERIC sequences: anovelfamily of repetitiveelements in thegenomes of Escherichiacoli, Salmonella typhimurium and other entero-bacteria. Mol.Microbiol.5:825-834.

14. Irino, K., F. Grimont, I. Casin, P. A. D. Grimont, and the BrazilianPurpuric Fever Study Group. 1988. rRNA gene restric-tion patterns of Haemophilus influenzae biogroup aegyptius strains associatedwith Brazilianpurpuricfever.J. Clin. Micro-biol. 26:1535-1538.

15. John, J. F., andJ. A.Twitty. 1986. Plasmidsas epidemiologic markers in nosocomial gram-negative bacilli: experience at a universityandreview ofthe literature. Rev. Infect. Dis. 8:693-704.

16. Kline,M. W.1988.Citrobactermeningitis and brain abscessin infancy:epidemiology, pathogenesisandtreatment. J. Pediatr. 113:430-434.

17. Kline,M.W.,E.0.Mason,andS. L. Kaplan. 1988. Character-izationofCitrobacterdiversus strains causing neonatal menin-gitis.J. Infect.Dis.157:101-105.

18. Kline,M.W.,E.0. Mason,andS. L. Kaplan. 1989. Epidemi-ologic marker system for Citrobacter diversus using outer membraneproteinprofiles. J. Clin. Microbiol.27:1793-1796. 19. Li, J., J. M. Musser, P. Beltran, M. W. Kline, and R. K.

Selander. 1990.Genotypic heterogeneityofstrains of Citrobac-terdiversus expressinga32-kilodalton outer membraneprotein associated with neonatal meningitis. J. Clin. Microbiol. 28: 1760-1765.

20. Lin,F.C., W. F. Devoe, C. Morrison, J.Libonati,P.Powers, R. J. Gross, B. Rowe, E. Israel, and J. G. Morris. 1987. OutbreakofneonatalCitrobacter diversusmeningitis in a sub-urbanhospital. Pediatr. Infect. Dis. J.6:50-55.

21. Lupski, J. R., and G. M.Weinstock 1992.Short, interspersed repetitiveDNAsequences in prokaryotic genomes. J.Bacteriol. 174:4525-4529.

22. Morris, J. G., F. C. Lin, C. B. Morrison, R. J. Gross, R. Khabbaz,K.0.Maher, B. Rowe, E. Israel, and J. P. Libonati. 1986. Molecular epidemiology of neonatal meningitis due to Citrobacter diversus: a study of isolates from hospitals in Maryland.J. Infect. Dis. 154:409-414.

J. CLIN. MICROBIOL.

on April 12, 2020 by guest

http://jcm.asm.org/

23. Owen, R. J., A. Beck, P. A. Dayal, and C. Dawson. 1988. Detection of genomic variation in Providencia stuartii clinical isolates by analysis of DNA restriction fragment length poly-morphisms containing rRNAcistrons. J. Clin. Microbiol. 26: 2161-2166.

24. Parker, M. T.1978.Hospital-acquired infections: guidelines to laboratory methods, European series no. 4, p. 35-39. World Health Organization Regional Publications, Copenhagen. 25. Parry,M. F., J. H. Hutchinson, N. A. Brown,C.-H.Wu, and L.

Estreller. 1980. Gram-negative sepsis in neonates: a nursery outbreakdue tohand carriage of Citrobacter diversus. Pediat-rics 65:1105-1109.

26. Penner,J. L., J. N. Hennessy,S.D.Mills, and W.C.Bradbury. 1983. Application of serotyping and chromosomal restriction endonuclease digest analysis in investigating a laboratory-ac-quired case of Campylobacter jejuni enteritis. J. Clin. Micro-biol. 18:1427-1428.

27. Richard,C.,B.Brisou, andJ.Lioult. 1972. Etudetaxonomique de "Levinea"nouveaugenredela famille des enterobacteries. Ann. Inst.Pastuer 122:1137-1146.

28. Sharples, G. J.,and R. G.Lloyd. 1990. Anovelrepeated DNA sequencelocated in theintergenic regions of bacterial chromo-somes.Nucleic Acids Res. 18:6503-6508.

29. Stern,M.J., G.F.L.Ames,N.H.Smith,E.C.Robinson,and C. F. Higgins. 1984. Repetitive extragenic palindromic

se-quences: a major component of the bacterial genome. Cell 37:1015-1026.

30. vanSoolingen, D., P. W. M.Hermans,P. E. W.deHaas,D. R.

Soll, and J. D. A. van Embden.1991.Occurrenceandstability of insertion sequences in Mycobacterium tuberculosis complex strains: evaluation of an insertion sequence-dependent DNA polymorphismas atoolintheepidemiologyoftuberculosis.J. Clin.Microbiol. 29:2578-2586.

31. Versalovic, J., T. Koeuth, and J. R. Lupski. 1991.Distribution of repetitive DNA sequences in eubacteria and application to fingerprinting of bacterial genomes. Nucleic Acids Res. 19: 6823-6831.

32. Versalovic, J., T. Koeuth, Y.-H. Zhang, E. R. B. McCabe, and J. R. Lupski. 1992. Quality control for bacterial inhibition assays: DNA fingerprinting of microorganisms by rep-PCR. Screening 1:175-183.

32a.Versalovic,J., andJ. R.Lupski.Unpublished data.

33. Wachsmuth,K. 1986. Molecularepidemiologyofbacterial in-fections: examples of methodology and of investigations of outbreaks.Rev.Infect. Dis. 8:682-692.

34. Williams,W.W., J. Mariano,M.Spurner,H. D.Donnell,R. L. Breckenridge, R. L. Anderson, I. K. Wachsmuth, C. Thorns-berry, D. R.Graham, D.W.Thibeault,and J. R. Allen. 1984. Nosocomialmeningitis duetoCitrobacter diversus inneonates: newaspectsof theepidemiology. J. Infect. Dis. 150:229-235. 35. 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, andC.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.

on April 12, 2020 by guest

http://jcm.asm.org/