JOURNALOFCLINICALMICROBIOLOGY, Feb. 1994,p.398-402 0095-1137/94/$04.00+0
Copyright C1994, American Societyfor Microbiology
Use
of
Pulsed-Field
Gel
Electrophoresis for Epidemiological
Study of
Bordetella
pertussis
in
a
Whooping Cough Outbreak
YVON R. DEMOISSAC,1 SHERYL L. RONALD,2 ANDMARK S. PEPPLERl*
Departmentsof Medical
Microbiology
and InfectiousDiseases'
and Microbiology,2UniversityofAlberta, Edmonton, Alberta, Canada T6G 2H7
Received 9 August 1993/Returned for modification 18 October 1993/Accepted 9 November1993
We usedpulsed-fieldgel electrophoresis (PFGE) ofchromosomal DNA digested withXbaItodetermine the
distribution of different BordeteUapertussis strains from clinical isolates obtained during a large whooping
cough outbreak that occurred in Alberta, Canada, from December 1989 to May 1991. Our initial study analyzed 28 clinical isolates, 14 fromthe
city
ofEdmonton and 1 from each of 14 northern Albertatowns.Theseclinical isolateswererandomlychosenoverthecourseofthe 18-month outbreak. The DNA profileswere more
heterogeneous thananticipated and causedconcern thatPFGEwastoo sensitiveatechniquetocharacterize strains. Furtheranalysisshowed that thiswasnotthecase, asclustersof similar PFGEpatternswereobserved
in strains isolated from the same outlying town. Identical PFGE patterns werealso seen in clinical strains
obtained from differentmembersofthesamefamily.TwoPFGEpattern types,aand b, predominated in the
outbreak, accountingoverallfor 44 of70 B. pertussis strains tested. Results from isolates from outlyingtowns, however, indicated involvement of local strains rather thana single,highlyinfectious strain in the whooping
cough outbreak in Alberta.
Pulsed-field gel electrophoresis (PFGE) has become a
useful tool in the study ofbacterial genomes. Analysis of DNAfragments generated by rarely cutting restriction
en-donucleasescanbe resolvedby PFGE, allowingtheaccurate determination ofgenomesizes aswell asgenome maps for
Escherichia coli K-12 (16), Pseudomonas aeruginosa (13), Campylobacterjejuni (4), and, recently, Bordetellapertussis (19). Resolution of restriction endonuclease-cleaved DNA
by PFGE has also been used as afingerprint for
epidemio-logical differentiation of Campylobacter spp. (21), Le-gionella pneumophila (14), E. coli (1), enteroinvasive E. coli (6), methicillin-resistant Staphylococcusaureus(7),Listeria
monocytogenes (2), Shigella spp.(17), andB.pertussis (8).
The aim of the present study was to test PFGE as an
epidemiologicaltool forclassifyingclinical isolates collected
duringalarge whooping coughoutbreak inAlberta, Canada,
that occurred from 1989to1991(5). Serologicalclassification
of thegram-negative pathogensB.pertussis and Bordetella
parapertussis is too insensitive for epidemiological studies
because of the relative lack ofdiversityofsurface-exposed antigenic markers (agglutinogens). An exceptionwas dem-onstrated inalong-term study by Preston,who observedan
antigenic shift in the general B. pertussis population from serotypes1,2,0and1,2,3in1958toserotype1,0,3in1963 to 1964, a change which had a negative impact on vaccine
efficacy (10, 11). For short-term strain comparisons,
how-ever, agglutinogens do notprovide enough antigenic heter-ogeneity. Moreover, the expression of AGG 2 and AGG 3 appears to vary in vitro (3, 18) and in vivo (12). These phenotypic changesinsurfaceantigensmaybeimportantfor
pathogenesisbutareproblematicwhen serologyis usedfor strain characterization.
Recently, Khattak et al. have shown PFGE to be a far
moreeffectivetechniquefor theepidemiological studyof B.
* Correspondingauthor.Mailingaddress:Departmentof Medical
Microbiology and Infectious Diseases, UniversityofAlberta,
Ed-monton, Alberta T6G 2H6, Canada. Phone: (403) 492-2304. Fax:
(403)492-7521.
pertussis (8) than serology or multilocus enzyme
electro-phoresis (9). These authors showed the existence of
geno-typic diversity in isolates from Germany and the United Kingdom. Not surprisingly, they couldnot showany
geno-type-serotypecorrelations. Theirtyping scheme wasbased onXbaIdigestionof B.pertussis DNA,whichgenerated 8to 11 DNA fragments perisolate. The authors suggested that
PFGE typing could be used to trace the movement of B. pertussisstrains withinand/orbetweentowns,communities,
andcities(8).
During the period of December 1989 to May 1991, the
provinceof Albertaexperienceditslargestrecorded whoop-ing cough outbreak since the beginning of vaccination in 1943 (5). This event provided a unique opportunity to use
PFGEto gather epidemiological data pertinent to this out-break. We conducted a preliminary study by randomly
selectingisolates fromtowns,communities,and cities span-ning northern Alberta to get an impression of the genetic
diversity shown bythe clinical isolates. Following the pre-liminary survey, we examined two localized outbreaks. In
total, 70 strains (of 552 collected during this pericd)were
examined, and contact between certain towns appears to haveoccurred.Ourdatasuggestthat theoutbre:.ik in
north-ern Albertawas probablya result of immune status in the
population rather than a result of a particularily virulent
strain.
MATERIALS AND METHODS
Bacterial strains and enzymes. The clinical isolates were
provided bythe Provincial LaboratoryofNorthern Alberta
(Edmonton, Alberta, Canada). Organismswereculturedon
Bordet-Gengouagar (Difco, Detroit, Mich.)with 15% defi-brinated sheep'sblood (Gibmar Labs, Ardrossan, Alberta, Canada)and incubatedat37°Cin98%humidity.The restric-tionenzymeXbaIwasobtainedfromBoehringer Mannheim
Canada,Laval, Quebec, Canada.
Preparation of agarose-embedded DNA. Chromosomal DNAwasembedded inlow-melting-pointagarosetoprepare
398
Vol. 32, No. 2
on May 15, 2020 by guest
http://jcm.asm.org/
DNA plugs for subsequent restriction endonuclease diges-tion and analysis by PFGE as described by Smith et al. (15). Three- tofour-day-old cultures were suspended in 5 ml of 50 mMTris-HCl-5 mM EDTA, pH 8.0 (1 x TE buffer). Suspen-sions werestandardized to an
A540
of0.80 in screw-capped tubes (13 by 100 mm) and transferred to a 25-ml Corex centrifuge tube. An extra 5 ml of lx TE was added to the suspension, and the cells were centrifuged for 10 min (8,000 xg) in 1 x TE. Pellets were washed twice more in 1x TE, resuspended in 2.5 ml of 10 mM MgCl2, and mixed with 2.5 ml of low-melting-point agarose (Chromorose; Clontech, Mississauga, Ontario, Canada). Two percentlow-melting-pointagarose wasprepared in 0.5x Tris-borate-EDTA (0.45 M Tris-borate, 0.01 M EDTA [5x TBE]). The bacterium-agarose mixture was gently but quickly mixed, carefully
poured onto a clean, level, glass microscope slide, and allowed tosolidifyatroom temperaturefor about 15 min. A gel lane cutter (Bio-Rad, Mississauga, Ontario, Canada; catalog no. 170-4120) or a large razor blade was used to cut the solidified mixture into plugs of approximately 6 by 4 by 2 mm. The plugs were transferred to a 15-ml tube and exposed to 10 ml of alysis buffer consisting of 0.1 M EDTA
(pH 8.0), 0.01 M Tris-HCl (pH 7.6), 0.5% sodium dodecyl sulfate, 0.5% lauroyl sarcosine, 0.2% deoxycholic acid, 200 ,ugof RNase per ml, and 1 mg of lysozyme per ml and was incubated at 37°C for 16 to 18 h. After incubation, the lysis bufferwasremoved andreplaced with 10 ml of a proteinase K solution (1% lauroyl sarcosine, 1 mg of proteinase K
[Boehringer Mannheim] per ml, 0.5 M EDTA), and this mixture was incubated at 50°C for 8 h and then at 37°C overnight. Following incubation, theplugs were washed for
two1-hintervalsat370Cin 1x TE,fortwo0.5-h intervalsat
roomtemperature in lx TE with 1 mM phenylmethylsulfo-nylfluoride,andfortwo0.5-h intervals at room temperature in lx TE. The plugswere stored in lx TE at 40C.
Digestion by restriction endonuclease. One DNA plug pre-pared as outlined above represents an approximate volume of 40,ul. A digestion mixture with a final volume of 200 ,ul
consisted ofone DNAplug, 136 ,u of sterile distilled water, 20
RI
of the supplied lOx enzyme buffer, 2RI
of 100-mg/ml bovineserumalbumin, and 30 U of restriction endonuclease. The reaction mixture was incubated overnight at 370C and stored in 1x TEat4°Cuntil theplugcould be loaded onto a gel.PFGE. The PFGE apparatus used in this study used con-tour-clamped homogeneous field electrophoresis (CHEF). It
consistedofa2015PulsaphorPlus Control Unit powered by
anelectrophoresis power supply EPS 500/400 and cooled by
a 2219 Multitemp II Thermostatic Circulator (Pharmacia Biotech, Inc., Baie d'Urfe, Quebec, Canada). Digested DNA plugs for CHEF resolution were loaded onto a gel consisting of 1% agarose in 0.5 x TBE. Approximately one-thirdtoone-half of thedigested plug volume was loaded. Pulse times were programmed into the CHEF apparatus to
giveatotal run time of 24 h. The pulse times are ramped over sixphases beginningat15sfor thefirst phase and increasing by 5-s increments for each new phase. Each of the first three phases wererun for 3 h, and each of the last three phases
were run for 5 h, at 175 V at 80C. These conditions consistently resolved DNA fragments in the range of 50- to 534-kb as determined from the DNA standard (Lambda Ladder PFG Marker; New England Biolabs, Mississauga,
Ontario, Canada). The gels were stained in 0.5 ,ug of ethid-ium bromide per ml for 20 min and visualized by a UV transilluminator. TheDNAbands were sized by comparison ofmigration distances with those of the DNA standard.
RESULTS
Analysis
ofclinical isolates by PFGE. Our surveyof clinical isolates was based onthe analysis of 28 strains, 14 from the city of Edmonton and 1 from each of 14 northern Alberta towns. Between eight and nine fragments in the molecular size range of 130 to 534kb were generatedbyXbaIdigestion, which resulted in a diverse array of DNA profiles. Bands of less than 130 kb were not resolved adequately enough to provide useful information. Among the 14 Edmonton iso-lates, 11 unique PFGE types were observed. Differences in the 11 unique DNA profiles ranged from a low of one band between types a and1, f and g, and b and o to a high offive bands between types h and f. In addition, six common restriction fragments were observed in the Edmonton iso-lates. Fragments with molecular sizes of 300, 190, and 130 kb were seen in all 14isolates, while fragments with molecular sizes of 260 and 200 kb wereseen in 13 of the 14 Edmonton isolates. The 150-kb fragment was seen in 12 of the 14 Edmonton isolates.XbaI digestion of the 14 northern Alberta isolates resulted in 7 to 10fragments in the 130- to 534-kb range. In contrast to the Edmonton isolates, the northern Alberta isolates had only sixdifferent PFGE types amongthem (a, b, d, e, f, and g). Of these, four (a, b, e, andf) were identical to four of the 11 Edmonton PFGE types. The remaining two unique pat-ternsincreased the total number of clinical PFGE types from 11 to 13. Differences in the DNA profiles of the northern Alberta isolates were essentially the same as those observed with the Edmonton isolates. The same common fragments were also observed.
Additional clinical isolates were obtained from members of the same family to test the ability of PFGE to determine whether intrafamilial transmission was occurring. Subjects ofthis testing included a 30-year-old mother and her 13-year-old daughter from Edmonton, both type m; a pair of 7-year-old twins from Edmonton, both type k; a pair of siblings from Berwyn and a pair of siblings from Manning, all four type b; and two pairs of siblings from High Level, all four type c. Of the six pairs of intrafamilial isolates, each indi-vidual of a particular pair had a PFGE type identical to that of his or her sibling or parent, suggesting that transmission of some fashion occurred between them.
From these 40 strains (18 from Edmonton and 22 from northern Alberta) 14 distinct PFGE types were obtained, 12 ofwhich can be found in Edmonton. A 15th type (type o) was identified in samples from Whitehorse, The Yukon. Figure 1 allows a direct comparison of all the PFGE types generated by this study.
To further test the usefulness ofPFGE for the epidemiol-ogy ofpertussis, isolates from two smaller, more isolated communities were examined. One was Fort Smith (popula-tion, 2,480) (see Fig. 4), where 18 culture-positive cases occurred within a 3-month period during the outbreak. The other community was Whitehorse, The Yukon (population, 17,925), where 12 culture-positive cases occurred within a 12-month period. Whitehorse is approximately 2,000 km northwest of the city of Edmonton and is not shown on the map in Fig. 4. The PFGE profiles from all 18 Fort Smith isolates are identical (Fig. 2), suggesting that a single strain (type a) was responsible for the outbreak in Fort Smith. Type a strains were also isolated from Fort McMurray, Lloydminster, Athabasca, Ponoka, and Edmonton and were the most prevalent strains overall, representing 24(34%) of the 70 strains analyzed. Eleven of the 12 Whitehorse isolates (Fig. 3) share a DNA profile (type b) that is also seen
on May 15, 2020 by guest
http://jcm.asm.org/
400 DE MOISSAC ET AL.
X 1 2 3 4 5 6 7 8 9 1011 12 1314 15 16 A 1 2 3 4 5 6 7 8 9 10 11 9Y12
kb 533.5 388.C 291.C
a b c d e f g h i j k mno A FIG. 1. Clinical isolates digested by XbaI and resolved by PFGE. The isolates shownrepresent each of the 15 PFGE types generated by this study. Lanes 1 to 15 are clinical isolates; each
represents a unique DNA profile or PFGE type, and the letters
below the lanes correspond with the letters shownonthemapinFig. 4. Lane 16 is the referencelaboratory strain,B.pertussis 338. Lane
X is the X ladder molecular size marker, a successively larger
concatamerof 48.5-kb DNA fragments.
distributedacross northern Alberta and isseenasfar south
asRed Deer(see Fig. 4). Typebstrainswerealsoprevalent
in the outbreak, as they represented 20 (29%) of the 70 strains tested.Thus, the twopertussis PFGEtypes aand b were observed in 44 (63%) of the 70 isolates testedso far. Figure4 illustrates the distribution of PFGEtypesacrossthe
northernpartof theprovinceof Alberta.Typesaand b and
A 1 2 3 4 5 6 7 8 9 Ak
kb 533.E 388.C 291.C
145.-48C.
a a a a a a a a a
FIG. 2. Clinical isolates from FortSmith, Alberta,Canada(no.1
onthemapinFig. 4).Nine of 18 strains isolated withina3-month period areshown, althoughall 18possessedthesame PFGEtype (datanotshown). All isolatesaretypea.X, molecular sizemarker
(see Fig. 1legend).
145.5
48.5
b b b b b b b o b b b b
FIG. 3. Clinical isolates fromWhitehorse, The Yukon. Shown are 12strains isolated withina 12-month period.Eleven of the 12 strains share the same PFGE type (shown below the lanes). X, molecular size marker(see Fig.1legend).
others,suchastypese and c, are seen inmultiple locations
across theprovince.
DISCUSSION
Our results support evidence thatPFGEofXbaI-digested chromosomal DNAprovidesa sensitivemeansfor
discrim-inating between B. pertussis isolates (8). Our preliminary
epidemiologicaldatawerebasedontheanalysis of 28 clinical
isolates of B. pertussis, 14 from Edmonton and 14 from
townsspanningnorthern Alberta. Weweresurprised by the
heterogeneityof the DNAprofilesexhibited by the isolates: 11differentDNAprofilesfrom 14 Edmonton isolates and a total of 13 different DNA profiles for all 28 strains tested. Because of this, we considered the possibility that PFGE may generate toomuchheterogeneityin the DNA patterns,
complicatingtheanalysisofprevalence andspreadof
com-monly occuring strains of B. pertussis. On the basis of the
similarity ofprofiles from intrafamilial isolates as well as
from isolates from the small communities of FortSmithand
Whitehorse,we nolongerconsiderthis aproblem.
Khattaketal. (8)havealready establishedaPFGE-based
genotypicclassification system for B.pertussis strains,but
we were unable toalign ourtypes with theirs. We believe thatthis is due todifferences in the
electrophoretic
condi-tions used by us andby
Khattak et al. In contrast, ourparameters closely resemble those used by Stibitz and Garletts (19) to physically map the chromosome of B.
pertussis Tohama I.
Accordingly,
our results with XbaI digestsof Tohama I(Fig.
1,lane16)
areverysimilartothose obtainedbyStibitz and Garletts(Fig.
la,lane1,
of reference19). Thus,we areconfidentthatourresultscanbeachieved inother laboratories andare animprovementbecauseofthe increased resolution in the 200-to450-kbrange.Wefeel that
ourparametersaresufficient fortypingB.pertussisoutbreak strains.
The PFGE results thatweobtained with clinical isolates
canbeusedtoaddress twotheoriesas tothereasonfor the
year-and-a-half-long
whooping cough
outbreak:(i)
the emer-gence of a highly infectious strain of B.pertussis
in the J. CLIN. MICROBIOL.on May 15, 2020 by guest
http://jcm.asm.org/
a,b,C,e,\1ea\
f, h, I, j,
k,
1,
m, n100km
FIG. 4. Simplified geographic representation of the distribution
ofPFGEtypesidentifiedacrossthe province ofAlberta(land area
equivalent to 93% of that of the state of Texas). The numbers representthefollowingtownsorcities:1,FortSmith; 2,HighLevel; 3, Manning; 4, Fort McMurray; 5, Hines Creek; 6, Berwyn; 7, Wabasca; 8,Hythe; 9, Athabasca; 10,St. Paul; 11,Edmonton; 12, Hobbema; 13, Viking; 14,Lloydminster; 15, Edson; 16, Ponoka; 17, Lacombe; 18, RockyMountain House;and 19,RedDeer. The city
ofEdmonton'sclinicalisolatesareshownatthelowerleft.
Individ-ual isolatesfrom northern Alberta towns (e.g., Edsonand Hythe)
wererandomlychosenandmaynotnecessarily reflectthe prevalent PFGEtypesoftheirrespective areas.
province affected both the immunized and nonimmunized populations,and(ii)adecreaseintheimmunizedpopulation resulted in an increase in the population susceptible to pertussisdisease.Thefirsttheorystemmedfromthe thought thatanespeciallyvirulentstrainofB.pertussiswasbrought
intoCanadafromoutsidethecountrybysomeoneattending
an InternationalStoryteller's Convention that took place in Whitehorse, TheYukon(20). Thisiswhyisolates from this particularsiteoutsidetheprovince,andover2,000 kmfrom Edmonton,were includedinourstudy. Themostprevalent
strainofpertussisinWhitehorsewas typeb. Althoughtype
bstrainswerefoundasfarsouthasRedDeer, Alberta,type
bstrainsaccounted foronly 29% of all theoutbreakstrains
tested.Typeastrainswerealsowidelydistributed duringthe
outbreak, representing 34% of tested strains. Moreover,
local strainsof other PFGEtypes besides a and bmay be
responsible fordisease inoutlyingtowns duringthe course
oftheoutbreak. Forexample, it ispossiblethattype c may
bethepredominanttype in HighLevel (population, 2,849),
as two different pairs of siblings shared one PFGE type.
MoreisolatesfromHighLevelwouldneedtobeanalyzedto
confirm this. Regardless, no onestrain oforganisms is
fully
responsible for the outbreak.Regarding the second theory, the outbreak occurred dur-ing the peak of the periodic 4-year cycle ofpertussis
infec-tion, when disease would be expected tobe highest.
How-ever, 57% of greater than 5,600 patients for whom records
wereavailable had full immunization (at least four doses of vaccine) (6a). This suggests that perhaps vaccination was less protective than in previous periods of peak pertussis
activity.
Anotable observation from our data was the
heterogene-ity of PFGE patterns of Edmonton isolates compared with
thosefromtherestoftheprovince.Onepossible
explanation
may be that the city of Edmonton (population, 616,741) is visited bymanytravellerswhomaybring in different strains. This would allow many different strains to become estab-lished and spread throughout the city and perhaps the province. Despite this heterogeneity, there is evidence of
close-contact transmission, as shown by analysis ofsibling
or parent-child isolates. As more isolates from different
areas of the city are tested, it would be interesting to see whethercertain PFGEtypespredominated in different local-ities withinthe city.
These typeofanalysesillustrate the potentialof PFGE of
XbaI-digested Bordetella DNA for studying the
epidemiol-ogyof strains involved inwhoopingcough outbreaks. ACKNOWLEDGMENTS
This study was funded by the Canadian Bacterial Diseases Net-work (CBDN).
Thanks to William Albritton of the Provincial Laboratory of Northern Alberta for providing the clinical isolates and the data regarding their isolation. We thank our colleagues in the Department of Medical Microbiology and Infectious Diseases: Diane Taylor for the use of her CHEF apparatus, Sam Salama for his advice regarding PFGE, and San Vinh and Richard Sherburne for photo-graphic assistance. Thanks also to John Waters and Anita Hanra-han, of Alberta Social Services and Community Health (Communi-cable Disease Control and Epidemiology), for information regarding immune status.
REFERENCES
1. Arbeit, R. D., M. Arthur, R. Dunn, C. Kim, R. K. Selander, and R. Goldstein. 1990. Resolution of recent evolutionary diver-gence among Escherichia coli from related lineages: the appli-cation of pulsed-field gel electrophoresis to molecular epidemi-ology. J. Infect. Dis. 161:230-235.
2. Brosch, R., C. Buchrieser, and J. Rocurt. 1991. Subtyping of
Listeria monocytogenes serovar 4b by use of low-frequency cleavage restriction endonucleases and pulsed-field gel electro-phoresis. Res. Microbiol. 142:667-675.
3. Cameron, J. 1993. Variation in Bordetella pertussis. J. Pathol. Bacteriol. 94:367-374.
4. Chang, N., and D. E. Taylor. 1990. Use of pulsed-field gel electrophoresis to size genomes of Campylobacter species and to construct a Sall map of Campylobacter jejuni UA580. J. Bacteriol. 172:5211-5217.
5. Ewanowich, C. A., L. W.Chui, G. Paranchych, M. S.Peppler, R. G. Marusyk, and W. L. Albritton. 1993. Major outbreak of pertussis in northern Alberta, Canada: analysis of discrepant direct fluorescent-antibody and culture results by using poly-merase chain reaction methodology. J. Clin. Microbiol. 31: 1715-1725.
6. Gordillo, M. E., G. R. Reeve, J. J. Pappas, J. J. Mathewson, H.L. DuPont, and B. E. Murray. 1993. Molecular characteriza-tion of strains of enteroinvasive Escherichia coli 0143, includ-ing isolates from a large outbreak in Houston, Texas. J. Clin. Microbiol.30:889-893.
6a.Hanrahan, A. (Alberta Health, Communicable Disease Control
on May 15, 2020 by guest
http://jcm.asm.org/
402 DE MOISSAC ET AL.
and Epidemiology). Personalcommunication.
7. Ichiyama, S., M. Ohta, K. Shimokate, N. Kato, and J. Takeuchi. 1991. Genomic DNAfingerprintingby pulsed-field gel electro-phoresisas anepidemiological marker for study of nosocomial infections caused by methicillin-resistant Staphylococcus au-reus. J. Clin. Microbiol.29:2690-2695.
8. Khattak, M., R. C. Matthews, and J. P. Burnie. 1992. Is Bordetella pertussis clonal? Br. Med. J. 304:813-815.
9. Khattak, M. N., and R. C. Matthews.1993. Geneticrelatedness of Bordetellaspeciesasdeterminedby macrorestriction digests resolved by pulsed-fieldgel electrophoresis. Int. J. Syst. Bac-teriol. 43:659-664.
10. Preston, N. W.1963.Typespecific immunity against whooping cough. Br.Med.J.2:724-726.
11. Preston, N. W. 1965. Effectiveness ofpertussisvaccines. Br. Med. J. 2:11-13.
12. Preston, N. W., and T. N. Stanbridge.1972. Efficacy of pertussis vaccines:abrighter horizon. Br.Med. J. 3:448-451.
13. Romling, U., D. Grothues, W. Bautsch, and B.Tummier. 1989. A physical genome map ofPseudomonas aeruginosa PAO. EMBO J. 8:4081-4089.
14. Schoonmaker, D., T. Heimberger, and G. Birkhead. 1992. Com-parison of ribotyping and restriction enzyme analysis using pulsed-field gel electrophoresis for distinguishing Legionella pneumophilaisolatesobtainedduringanosocomialoutbreak. J.
Clin. Microbiol.30:1491-1498.
15. Smith, C. L., S. R. KIco, and C. R. Cantor. 1988. Pulsed-field electrophoresis and the technology of largeDNAmolecules, p. 41-72. In K. E. Davies (ed.), Genomic analysis: a practical approach. IRL Press, Washington, D.C.
16. Smith, C. L., S. R. Klco, and C. R. Cantor. 1989. Aphysical genome map of the Escherichia coli K-12 genome. Science 236:1448-1453.
17. Soldati, L., and J. C. Piffatetti. 1991. Molecular typing of Shigella strains using pulsed-field gel electrophoresis and ge-nome hybridization with insertion sequences. Res.Microbiol. 5:489-498.
18. Stanbridge, T. N., and N. W. Preston. 1974. Variation of serotype in strains of Bordetella pertussis. J. Hyg. Camb. 73:305-310.
19. Stibitz, S., and T. L. Garletts. 1992. Derivation of a physical map of the chromosomeofBordetellapertussisTohamaI. J. Bacteriol. 174:7770-7777.
20. Varughese,P. V. 1990.Pertussis outbreakin theYukon Terri-toryin 1989. Can.Dis.WeeklyRep. 16:63-68.
21. Yan, W., N. Chang, and D. E. Taylor. 1991. Pulsed-field gel electrophoresis of Campylobacter jejuni and Campylobacter coli genomicDNAand itsepidemiologicapplication. J. Infect. Dis. 163:1068-1072.
J. CLIN. MICROBIOL.
