Analysis of a repetitive DNA sequence from Bordetella pertussis and its application to the diagnosis of pertussis using the polymerase chain reaction

(1)

JOURNALOF CLINICALMICROBIOLOGY, Sept. 1990,p. 1982-1987 0095-1137/90/091982-06$02.00/0

Analysis of

a

Repetitive DNA Sequence from

Bordetella pertussis

and

Its

Application

to

the

Diagnosis of Pertussis Using the

Polymerase Chain Reaction

ERIC M. GLARE,' JAMES C. PATON,1* ROBERT R. PREMIER,2 ANDREW J. LAWRENCE,'

AND IAN T. NISBET2

Microbiology Department, Adelaide Children'sHospital, NorthAdelaide, South Australia 5006,' and Commonwealth Serum Laboratories, Parkville, Victoria 3052,2 Australia

Received 13February1990/Accepted 13 June 1990

A tandemly repeated 1,046-base-pair (bp) ClaI DNA fragment fromBordetellapertussis was cloned into Escherichia coli by using thevectorpUC19. Thisfragment, whenisolated, hybridized stronglytoDNAfrom

all 100 clinical isolates of B. pertussis tested. It was shown to have homology to single-copy sequences in Bordetellabronchiseptica but notBordetellaparapertussis and didnothybridizetolysateblots ofawide range

of otherbacteria,includingmembers of thecloselyrelatedgeneraPasteurella,Alcaligenes,andHaemophilus. The 1,046-bp fragmentwassequenced,andcomplementary synthetic oligonucleotides flankinga153-bp region withinthe repeatedelementwereusedasprimersforspecific amplificationofthisregion usingthepolymerase

chain reaction (PCR). This procedure was then applied to the rapid (5-h) detection of B. pertussis in

nasopharyngeal secretionscollected from 332 children withsuspected pertussis.Thetestyielded positiveresults inatotalof98 samples, comparedwith66 for cultureand33for direct immunofluorescence (IF).Allofthe

IF-positive samples were PCR positive, as were 63 of the samples from which B. pertussis was eventually cultured. Two hundredthirty-one specimenswhichwerenegative byIF and culture werealsonegative inthe PCR assay. However, 33 culture- and IF-negative specimens were positive by PCR assay. Several ofthese

specimens were collected from closecontacts ofculture-proven pertussis patients,were follow-up specimens

from suchpatients, or were frompatientswith serological evidence ofpertussisand thereforemay be

true-rather thanfalse-positives.

In recent years much interest has centered on the appli-cation ofnucleic acid probes to thediagnosis of infectious

diseases, particularlyin view of theavailabilityof nonradio-active labeling techniques. These diagnostic probes will be

particularly useful for organisms which are fastidious or

difficult to culture in vitro. Bordetella pertussis is such an

organism. It takes several days to isolate and identify B.

pertussis from cultures of nasopharyngeal secretions.

Fur-thermore, false-negative culture results are obtained in a

significant proportion of pertussis cases (12).

Recently,thepresenceofspecificreiteratedchromosomal

DNAsequences has beenreported foranumber of bacterial

species (4, 7, 8), includingB.pertussis (1, 14-17).Theseare ideal targets fordiagnostic probes because multiple targets tendtoenhance the sensitivity of probe-based assays. In B.

pertussis the repeat ispresent in 50to 100copies per cell, andconsensus sequences uptoapproximately 1.1 kilobase

pairs (kbp) long have been determined by comparing

se-quence dataobtained from different clones containing por-tions of therepeating element(14, 17).

In the present study, we have shown that the repeated elementisindeed 1,046basepairs (bp) long andthatmanyof

thecopies arearranged in tandem onthechromosome. We have analyzedthe specificity of the repeat and developed a

potentialdiagnostictestfor pertussis basedonamplification

ofa 153-bp region of the repeated element using the

poly-merase chainreaction (PCR).

*Corresponding author.

MATERIALSANDMETHODS

Bacterial strains and cloning vectors. B. pertussis NCTC 10908, NCTC 10911, and NCTC 10739,Bordetella

paraper-tussis NCTC 5952, and Bordetella bronchiseptica NCTC

8344 were obtained from the National Collection ofType

Cultures, London, United Kingdom, and were grown on

charcoal agar (Oxoid Ltd., Basingstoke, United Kingdom) containing10% defibrinated horse blood. Clinical isolatesof B. pertussis and B. parapertussis were collected at the

Adelaide Children's Hospital, Adelaide, South Australia,

Australia. Escherichia coli K-12 strain JM109 (19) has been

described previously and was grown in Luria-Bertani me-dium (13) with or without 1.5% Bacto-Agar (Difco

Labora-tories, Detroit, Mich.). When appropriate, ampicillin was added to media at 50 Fg/ml. Plasmids pUC18 and pUC19

have also beendescribed previously (19).

Chromosomal DNA extraction from Bordetella species. Chromosomal DNA was extracted from Bordetella species

by a modification of the method ofBrown and Parker (2). Bacteriagrown onfourcharcoalagarplateswere harvested andwashed in 150mMNaCI-10mMTrishydrochloride(pH

7.5) and resuspended in a volume of 20 ml of the same mixture. The suspensionwasfrozenat -20°C. One volume of 1% sodium dodecyl sulfate-100 mM NaCl-100 mM Tris

hydrochloride (pH 8.8)-100 ,ug of pronase (Boehringer

Mannheim, North Ryde, New South Wales, Australia) per mlwasself-digestedat37°C for1h before being usedtothaw the cell suspension by gentle inversion. The lysate was

digested overnightat37°C. Nucleic acids were precipitated with2.5 volumesofchilled ethanol, suspended in TE buffer

(10 mM Tris hydrochloride, 0.1 mM EDTA [pH 8.0]), and

digested again with pronase overnight. The solution was

1982

Vol.28,No. 9

on April 12, 2020 by guest

http://jcm.asm.org/

(2)

extracted with phenol-chloroform (1:1) and then with chlo-roform and was precipitated again. B. pertussis DNA was further purified on a

CsCi

gradient. TE

buffer-CsCI

(735 mg/ml) was added to 8-ml centrifuge tubes and underlaid with 2.4ml of DNA solution to which 4.2 g of

CsCI

and 400

,ul

of ethidium bromide (10 mg/ml) had been added. After centrifugation at 180,000 x g for 42

h,

DNA bands were removed under UV illumination andcleaned as described by Maniatis et al. (13).

Methods for DNA isolation and analysis.Plasmid DNA was isolated by the alkaline lysis method described by Maniatis et al. (13). E. coli cells were treated with CaCl2 and trans-formed with plasmid DNA as described by Brown etal. (3). Digestion of genomic DNA and plasmids with restriction endonucleases was conducted under the conditions recom-mended by the supplier (Boehringer Mannheim). Restricted DNA was electrophoresed in agarose gels with a Tris-borate-EDTA buffer system as described by Maniatis et al. (13). Restriction fragments were purified from agarose gels with Geneclean (Bio 101,Inc., La Jolla, Calif.). Electrophoresed DNA was transferred to nitrocellulose by the method of Southern (18) and then hybridized and autoradiographed as described by Maniatis et al. (13). Hybridization was carried out overnight at 420C in 6x SSC (lx SSC is 0.15 M

NaCI

plus 0.015 M sodium

citrate)-1

x Denhardt

solution-1%

sodium dodecyl sulfate-100 ,ug of herring sperm DNA per ml-100

,ug

of heparin per ml-50% formamide. Filters were washed twice in 2x SSC for5min at65°C and twicein0.2x SSC-0.1% sodium dodecyl sulfate for5 min at 65°C. Probe DNA waslabeled with

[ct-32P]dCTP

(3,000

Ci/mmol;

Amer-sham, North Ryde, New South Wales, Australia) by the method of Feinberg and

Vogelstein

(5).

Lysate blots. A light suspension ofbacterial

cells

from agar plates was made in 200

,ul

of TE buffer.

Cells

werethen

lysed

by the addition of sodium dodecyl sulfateto a final concen-tration of 1%. After the suspension wasvortexed briefly, 0.8 M NaOH was added to a final volume of 1.2ml. Samples of each lysate werediluted 1:100 with 0.8MNaOHandapplied to nitrocellulose by loading 50

,ul

into thewells of aBio-Dot apparatus (Bio-Rad Laboratories, Richmond, Calif.). The samples were washed with 100

,ul

of20x SSC per

well

before the membrane was removed and baked for 2 h in vacuo at 800C.

DNA sequencing. Nested deletions ofcloned DNA were constructed by using an Erase-a-base kit obtained from Promega Corp., Madison,Wis. Bidirectional DNA sequenc-ing was then carried out by the

double-stranded

template technique described by Kraft et

al.

(9).

Preparation and PCR amplification of extracts ofNPA. A total of 332 nasopharyngeal aspirates (NPA) were obtained from children aged 10 days to 16 years (median age, approx-imately 18 months) with suspected pertussis (patients had a persistent cough with or without vomiting or paroxysms). Specimens collected over a12-month period were tested,but most of thesewereobtained duringthelast3monthsof1989, which was a period of high incidence ofpertussis in Ade-laide.

The NPA collection tubing was flushed out by aspiration of 1 ml of sterile saline, and the entire sample was then vortexed. Samples were immediately removed for culture and direct immunofluorescence (IF) analysis as described below. Samples (50 ,ul) ofthe remainder of the NPA were digested by the addition of 1 M Tris hydrochloride, pH 7.6, to 12 mM and proteinase K to 0.2 mg/ml and incubation at

65°C for 1 h. The proteinase K was then inactivated by boiling the extracts for 20 min.

PCR amplification was conducted in

50-,ul

reaction

mix-tures

containing

20

pul

of NPA extract, 200

,uM

deoxynucle-oside

triphosphates,

approximately 1

,uM

eachprimer, and1 U of Taqpolymerase and a buffer(Ten X Activity Grade), both of which were from IBI, New Haven, Conn. The

sequences

of the two oligonucleotide primers are 5'-GAT

TCAATAGGTTGTATGCATGGTT-3' and 5'-AATTGCTG GACCATTTCGAGTCGACG-3'. Samples were subjected to 30 PCR cycles, each consisting of1-min denaturation,

an-nealing,

and

elongation

steps attemperatures of 94, 57,and

72°C,

respectively. The 153-bp amplified product was de-tected by

electrophoresing

20-,ul

aliquots through 2% agar-ose gels in the presence ofethidium bromide for approxi-mately 30 min at 10 V/cm, followed by photography under UV illumination. Positive (B. pertussis DNA) and

negative

(proteinase K-treated saline) control extracts were included in each PCRrun.

Detection of B. pertussis by culture and IF. NPA were

cultured(100to200pil ofsample perplate) immediately after collection on charcoal agar, as well as on charcoal agar supplemented with 40

,ug

ofcephalexin perml andcharcoal agar supplemented with 2.5

ktg

offlucloxacillin per ml, and were incubated for up to 7 days at 37°C. Plates were examined daily, and suspect colonies were subcultured and identified by slide agglutination with B. pertussis and B. parapertussis antisera obtained from WellcomeDiagnostics, Dartford,UnitedKingdom. For direct IF,100-,ul samples of eachaspirate were fixed on slides and stained with

fluores-cein-conjugated

rabbit anti-B. pertussis antibody

(Difco).

RESULTS

Cloningrepetitivesequences in E. coli. B. pertussis NCTC 10908 DNA wasdigestedtocompletion with Sau3AI, ligated into the BamHI site of pUC19, andused totransformE. coli JM109. DNA from recombinant clones was extracted, and the inserts were excised by double digestion with EcoRI and HindIII. After electrophoresis through 2% agarose gels and transfer to nitrocellulose, the filters were probed with ge-nomic B. pertussis DNA. The inserts of four of the recom-binant plasmids produced stronger hybridizationsignals than didthose ofthe other clones, implying that theirsequences were relatively more abundant in the heterogeneous probe DNA.Thesefour inserts wereallthe same size (188 bp) and hybridized strongly to each other.

One of theplasmids, designated pJCP601, was chosen for further study. To confirm that the cloned DNA was reiter-ated, the purified insert wasused toprobe Southern blotsof restricted B. pertussis genomic DNA (Fig. 1). It hybridized to multiple (at least 40) bands in the various digests and particularly highlighted a 1-kbp fragment in the ClaI digest, suggesting the existence of tandem repeats. This fragment was clearly visible as a distinct band in ethidium bromide-stained gels of ClaI-digested DNA and could be easily excised from the gel. The purified ClaI fragment was then ligatedinto the AccIsite ofpUC19, and, after transformation into E. coli, positiveclones were detectedbyprobing colony blots with the 32P-labeled insert of pJCP601; one of these was designated pJCP602. The hybridization pattern of the cloned ClaI fragment (from pJCP602) to various digests of chromosomal DNAis virtually identical tothe hybridization pattern obtained by using the 188-bp Sau3AI fragment in pJCP601 as a probe (Fig. 1),

indicating

that the smaller fragment is containedwithin most of thecopiesofthe

1-kbp

ClaI fragmentin the genome. TheClaI fragmentcontained a single AccI site, and when Southern blots ofgenomic DNA

on April 12, 2020 by guest

http://jcm.asm.org/

(3)

1984 GLARE ET AL.

A_ kb 1 2 3

_

1

f

....w

::

r

.i._.

*`%

e.;,1

kb

B C

1 2 3 1 2 3

FIG. 1. Southern blotanalysisof B.pertussisDNA.B.pertussis NCTC10908 DNAwasdigestedwith BamHI(B),EcoRI(E), PstI (P), or ClaI (C), electrophoresed on a 0.8% agarose gel, and transferred to nitrocellulose. Filters were probed with the 32p_ labeled inserts of pJCP601 (lanes 1) or pJCP602 (lanes 2). The mobilities ofvariousDNA size markersareindicated.

digested with thisenzymeorClaIwereprobedwith theClaI fragment, a comigrating DNA band was strongly labeled

(resultnotshown). This furthersupports thesuggestionthat manyof the copies of the repeated sequencearearrangedin

tandem.

Specificity of the repeatedsequence.Toassessthe

suitabil-ity of the cloned ClaI fragment as adiagnostic DNA probe specific forB.pertussis, theClaIfragmentwashybridizedto

lysate blots of 96 different bacteria isolated from clinical

specimens. These bacteria included 45 different species representing 24genera. While theprobe hybridized strongly

to B. pertussis strains, it did not hybridize to any other

bacteria, including members of the most closely related

genera (Pasteurella, Alcaligenes, and Haemophilus).

Fur-thermore, no hybridization was observed with B.

paraper-tussis lysates, even when autoradiography at -80°C was

extended considerably. However, under these conditions

somehybridization could be detectedtothelysate blot ofB.

bronchiseptica.

The nature of the homology between the 1-kbp ClaI fragment and the genomic DNA of the other Bordetella species was determined by Southern transfer of various

restriction digests of DNA (Fig. 2). After prolonged

autora-diography, hybridizationtoone ortwobands in each restric-tiondigest of B. bronchiseptica DNAwas detected butwas notdetected withB. parapertussisDNA. Thestrengthof the hybridization suggeststhatsingle-copy sequences that have

at least partial homology to the repetitive ClaI fragment

found inB. pertussisexist inB. bronchiseptica.

To determine whether the ClaI fragment in pJCP602 is present inall strains ofB. pertussis, the insertwas

hybrid-ized to lysate blots of 100 clinical isolates ofB. pertussis.

Hybridization to all B. pertussis isolates was detected but

wasnotdetected withanyof the controlsamples(DNAfrom

otherbacteriaorhuman DNA) (result notshown).

DNAsequence analysis of repetitiveclones. The insertsof

pJCP601 and pJCP602 were subcloned into pUC18 by

iso-lating the EcoRI-HindIII fragmentandligatingthisfragment

FIG. 2. Southern blot analysis of DNAfrom otherBordetella species. DNA from B.pertussis (A),B. bronchiseptica (B), orB. parapertussis(C)wasdigestedwith BamHI(lanes 1),ClaI (lanes 2), orEcoRI (lanes 3). Southernblots ofelectrophoresed DNAwere

then probed with the insert ofpJCP602. Filter Awas autoradio-graphedfor 4h, while filters B and Cwereautoradiographedfor4 days.

into EcoRI-HindlII-cut pUC18, producing pJCP603 and pJCP604, which had the respective inserts in the reverse

orientation withrespecttotheuniversalsequencing primer. Bidirectional double-stranded DNA sequencing was then conducted with theseplasmidsand, whennecessary,nested deletion derivatives, as templates. The sequence of the

insertofpJCP602 is comparedwith twopreviouslypublished

sequencesfor theB. pertussisrepeated element(14, 17) in Fig. 3. Thesequenceofthe tandemrepeat has beenarranged (i.e., "wrapped around") so that it lines upwith that ofa

nontandemcopyof thesequencepublished byMcLaffertyet

al. (14), and this places the ClaI site at position 620. The

sequenceofthe insert ofpJCP601 wasidentical tothe first

168 bases ofourrepeatingunit butincluded 20

nonhomolo-gous bases at the 5' end, terminating in a Sau3AI site.

Presumably, the insert of pJCP601 wasderivedfromoneof

the nontandem copies of the repeated element, which in-cluded 20 bases offlanking chromosomalDNAendingina

Sau3AI site (the 1,046-bp tandemly repeated element does

nothaveinternal Sau3AI sites thatwouldgeneratea188-bp

fragment). There are 12 base differences (five substitutions

andsevendeletions) betweenoursequencefor the repeated

element and that of McLafferty et al. (14). For all the substitutions and five of the deletions, our dataare consis-tent withthe(incomplete) sequencedata of Parketal. (17). Their sequence, however, included a further five base

dif-ferences(three deletions andtwosubstitutions) withrespect toboth thepresentstudyand that of McLaffertyetal. (14). Thus, it appears that the DNA sequence of the repeated

element ofB. pertussis is not absolutely conserved. The existence of this sequence variation, particularly the

dele-tions, makes it extremely unlikely that the repeatedelement

encodes aprotein product.

Detection ofB.pertussisbyPCRamplificationofpartof the

repeatedelement. Oligonucleotides flankinga153-bpregion

ofthe repeated element (Fig. 3)weresynthesizedand used as primers for PCRamplificationas described in Materials B

1 2

E 1 2

p 12

C 1 2

J.CLIN. MICROBIOL.

Ai"

v VI_q7-Ë".

4

-:=

-P.i. iw..:

on April 12, 2020 by guest

http://jcm.asm.org/

(4)

PCR DIAGNOSIS OF PERTUSSIS 1985

A

B

A B

A e C

CAGCTGTGAAGATTCAATA GGTTGTATGCATGGTTCATCCGAACCGGATTTGAGAAACTGGAAATCGCC AACCCCCCAG TTCACTCAAGGAGCCCGGCC

150 200

...

... ...

GGATGAACACCCATAAGCAT GCCCGATTGA CCTTCCTACGTCGACTCGAAATGGTCCAGCAATTGATCGCCCATCAAGTTTGTGTGCCTGAAGCGGCCCG

250 300

...

... ...

CGCCTATGGGGTCACCGCGCCGACTGTGCÇ CAAATGGCTGGGCCGCTTCCTGGCTCAGGGCCAGGCGGGCTTGGCCGATGCGTCCTCGCGCCCGACGGTC

...

. AG..-..

350 400

A ... G.. CC...c...

B TCGCCCCGAGCGATTGCGCC GGCCAAGGCG CTGGCTATCG TGGAGCTCCG --G-AAGCGGCTGACCCAAGCGCGCAT-GCCCAGGCGCTGGGCGTGTCAG

C ... C

450 500

A ...

B CCAGCACC6TCAGCCGCGTC CTGGCCCGCG CCGGTCTGTC GCACCTGGCCGACCTGGAGC -GGCCGAGCC GGTGGTGCGCTACGAGCATCAGGCCCCCGG

C ... ... ... ...

550 600

A ... 8 CGATCTGCTGCACATCGACA TCAAGAAGCT GGGACGTATCCAGCGCCCTG GCCACCGGGT CACGGGCAAC CGACGCGATA CCGTTGAGGG GGCCGGCTGG C ... ... ... ... ... ...

650 700

A ...

B GACTTCGTCTTCGTGGCCAT_AT6GACCAC GCCCGCGT-GCCTTCACCGA CATCCACCCC GACGAGCGCTTCCCCAGCGCCGTCCAGTTC CTCAAGGACG

C ... G .6. ... ...

Clal

750 800

A ... ... ... ....

B CAGTGGCCTA CTACCAGCGC CTGGGCGTGACCATCCAGCGCTTGCTCACC6ACAATGGCT CGGCCTTTCGCAGCCG-GCC TTCGCCGCGCTGTGCCATGA

C ... ... ... ...

850 900

A ... ... ...r... ...

B GCTGGGCATC AAGCACCGCTTTACCCGACC TTACCGCCCACAGACCA'ATGGCAAGGCCGA ACGCTTCATC CAGTCGGCCTTGCGTGAGTGGGCTTACGCT

C ... ... ... ... ...

950 1000

A ...

B CACACCTACC AGAACTCCCAACACCGAGCC GATGCCATGA AATCCTGGCT ACACCACTAC AACTGGCATCGACCCCACCAAGGCATCG66CGCGCTGTAC

C ... ... ... ... ...

1050 A ...ATC.

B CCATCTCCAGACTCAACCTGGACGAATACA ACCTATTGACAGTTéCACAC

TAG-C ...C

FIG. 3. DNA sequence of theB. pertussis repeated element. The complete DNA sequence shown (B) is that of the insert ofpJCP602,

aligned with data of McLafferty etal. (14) (A) and data of Park etal. (17) (C). Forsequences AandC, a dot indicates base identitywith

sequence

B,

whereas a dash in any of the sequencesindicates a basedeletion. The location of the regionamplified byPCR is indicated by the arrows, which underscorethe position of annealing of the two primer oligonucleotides. The ClaI recognition site at position 620 is underlined.

and Methods. In initial experiments, proteinase K-treated suspensions of various bacteria were tested. PCR product was detected by agarose gel electrophoresis and ethidium bromide staining. This isarapid technique which is suitable for analyzing large numbers of samples, and photography under UV illumination provides a permanent record. Fur-thermore, assessment of the size of the PCR product is an added check on specificity. PCR-amplified extracts of B. pertussis showed the presence of an intensely stained DNA band of the expected size (153 bp). A very weak band of approximately the same size was seen in amplified extracts of B. bronchiseptica, but no bands were seen inamplified extracts of B. parapertussis. No DNA bands were detected in the amplified extracts of any of the other bacteriatested earlier by hybridization or with human DNA as the template (results not shown). Thus, the specificity of the PCR ampli-fication when the chosen oligonucleotide primers wereused was the same as that shown for direct hybridization to the insert ofpJCP602.

The sensitivity of the PCR assay for detection of B. pertussis was then examined. A fresh (24-h) culture ofB. pertussis was serially diluted inTrypticase soy broth (BBL Microbiology Systems, Cockeysville, Md.), and

50-ktl

ali-quots were immediately plated on charcoal agarfor deter-mination of the number of viable bacteria. Samples(50,ul) from each dilution were also treated with proteinase K, amplified by PCR, and electrophoresed, as described in Materials and Methods (Fig. 4, lanes1 to 7). A153-bpDNA band was still detectable in the reaction mixturethatinitially contained approximately 3 viable bacteria (lane5) and was faintly discernible in the reaction mixture supposedly con-taining 1/10 of this amount (lane 6). No bandwas seenwhen

proteinase K-treated diluentwas

amplified (lane

7)

orwhen further dilutions of the bacterial

suspension

were

similarly

treated (resultnot

shown).

The PCR assay was then used to detect B.

pertussis

in NPA collected from children with

suspected

pertussis,

as

described in Materials and Methods

(Fig. 4,

lanes 8 to

12).

The resultsofthePCR assaywerethen

compared

with those obtained by IF or culture. In

general,

aspirates

yielding

a

moderate to

heavy

growth

of B.

pertussis

were

strongly

positive

by PCR,

whilethose

yielding

scanty

growth

(which

were usually negative

by

IF) resulted in a less intense but nevertheless

clearly

positive

PCR result. A subset ofNPA samples (6

positive

and 14

negative)

wasretestedinanother laboratory with identical

results,

demonstrating

the

repro-ducibility

of the PCR assay.

The results obtained for a total of332 NPA are

summa-rized in Table 1. ThePCR test

yielded

positive

results in a

totalof98

samples,

compared

with 66 for culture and 33for direct IF. Allofthe

IF-positive

samples

were PCR

positive,

as were 63 of the

samples

from which B.

pertussis

was eventually cultured. The three

culture-positive,

PCR-nega-tive samples grew fewer than five colonies each. Two hundred thirty-one

specimens

which were

negative by

IF andculture werealso

negative

in thePCR assay.

However,

33 culture- and

IF-negative

specimens

were

positive

by

PCR

assay;atleast threeofthese

specimens

were collectedfrom

close contacts of

culture-proven

pertussis

patients

andtwo

were follow-up

specimens

from recent

culture-confirmed

patients. Sera were available from 10 of the

remaining

patients, and when these were tested

by

enzyme-linked

immunosorbent assay

(11)

for

immunoglobulin

A

(IgA),

IgG,

VOL. 28,1990

on April 12, 2020 by guest

http://jcm.asm.org/

(5)

1986 GLARE ET AL.

FIG. 4. PCRamplification ofrepetitiveDNAsequencesfor directdetection ofB.pertussis.SamplesweretreatedwithproteinaseK,PCR

amplified, electrophoresed, and stained with ethidium bromide, as described in thetext. Lanes: 1, amplified purified B.pertussis DNA

(positive control); 2to6,amplifiedB.pertussis cultureextractscontaining(approximately) 3,000, 300, 30, 3,and0.3CFU,respectively,in

the total PCRmix; 7, amplified diluent(negative control); 8 and 9, culture- andIF-negative NPA;10and11,culture- andIF-positiveextracts; 12,culture-positive, IF-negative extract; 13,DNAsize markers(500,404, 331, 242,190, 147,and110bp,respectively,from toptobottom).

and IgM antibodies to B. pertussis, 5 showed serological evidence ofrecent infection(results not shown).

DISCUSSION

ThecompleterepetitiveDNAsequence inB.pertussis has been shown in this study tobe 1,046bp long, with unique ClaIand AccIrestrictionendonucleasesites. Whengenomic DNA was digested with either ClaI or AccI, a prominent 1-kbp repetitive band was easilyobserved in ethidium bro-mide-stainedgels. Hence,many ofthe repeatsarearranged intandem on thechromosome.We haveclonedthecomplete repetitive unitas a ClaI fragment

(pJCP602)

afterisolating therepetitiveDNAbandfrom

electrophoresed ClaI-digested

genomicDNA.

Thechromosomalarrangementofthe repeat appearstobe conserved between strains ofB. pertussis. When genomic DNAfromsevenclinical isolatesand threeNational Collec-tion ofTypeCultures strainsofB.pertussis wassubjectedto restriction fragment length polymorphism analysis (result not presented), only minor differences in restriction frag-mentlengthsand bandintensities wereobserved. In partic-ular, all strains had repetitive 1-kbp bands of consistent intensities in ClaIdigests. Thehigh degree of conservation offragments bearingtherepetitivesequenceimpliesthat the repeatis notinvolved in DNArearrangements as has been suggested elsewhere (1, 15).

All isolates of B. pertussis tested in the present study contained therepeated DNAsequence, which (apart from a very weak

hybridization

with B. bronchiseptica) wasabsent from allother species of bacteria tested. B. bronchiseptica,

TABLE 1. Detection of B. pertussis in NPA by PCR assay comparedwithcultureand IF'

No. of results

PCR

result IF+, IF-, IF +, IF- Total

culture+ culture+ culture- culture

-+ 31 32 2 33 98

- O 3 0 231 234

aNPA weretreatedwith proteinaseKandamplifiedbyPCR aftersamples

hadbeenwithdrawnforcultureand IFanalysis. The numberofspecimens yielding a positive (+) or negative (-) result for the various assays is

indicated.

however,is a very rare (0.1% ofallisolated bordetellae[10]) and clinically unimportant bacterium in humans (6). This suggested that this sequence would be an ideal target for diagnostic probes for B. pertussis. The limit of detection when the 32P-labeled repeated sequence was used toprobe dot blots of B. pertussis suspensions was approximately 1,000 organisms after overnight autoradiography (resultnot

shown), and this limit is clearly inadequate for clinical testing. However, the use of syntheticoligonucleotide prim-ers for PCR amplification of part of the repeated element, although vastlyincreasing the sensitivity, couldconceivably becomplicated by variations in the precise DNA sequence from one strain of B. pertussis to another. Indeed, we detected a totalof 12 base differences (five substitutions and seven deletions) between the sequence of our clone (pJCP602) and that of McLafferty et al. (14). In the present study, the region of the repeated element from bases 12 to 164(inclusive) was chosen for amplification by PCR because it is a region of absolute sequence identity between the insertsofpJCP601 and pJCP602 and that ofMcLaffertyetal. (14). Furthermore, our Southern blot experiments using the inserts of pJCP601 and pJCP602 as probes indicated that virtuallyallcopies (completeorotherwise) of thereiterated sequences in the B. pertussis genome included thisregion.

Under ourconditions, the PCR assay for direct detection of B.pertussis had an apparent sensitivity limitof less than oneviableorganism (i.e., CFU).Itispossible, however,that the diluted cell suspensions tested included nonviable B. pertussiscells,andthereforeitisuncertainwhetherless than one B. pertussis genome (which would contain approxi-mately 100 copies ofthe target sequence) could have been detected. The theoretical limitofdetectionby PCR assay is onecopy of thetarget sequence, and therefore it should be possible to increase the sensitivityofour assay. This could beachievedbyincreasingthenumber of amplification cycles or by increasing the sensitivity of PCR product detection (comparedwithethidiumbromide staining) by hybridization with a labeledoligonucleotide probe specific for a section of theamplifiedsequence which did not overlap with the primer sequences. It wasdecided, however, that these steps would increasethe time ofthe assay to a point at which a same-day resultwould not beachievable in the clinical setting.

Formalassessmentof thespecificity and sensitivity of new methods for diagnosis of pertussis is complicated by the absence of asatisfactory "gold standard" with which

com-J. CLIN. MICROBIOL.

on April 12, 2020 by guest

http://jcm.asm.org/

(6)

parsons can be made. Culture, for example, is highly specific, but its sensitivity is very poor, with positive results obtained in at most 50% ofcases (12). In the present study,

IF,while specific, was even less sensitive than culture.

Also,

serological studies may yield false-negative results early in thecourse of infection and in children under 4 months of age (11). Furthermore, it is not acceptable to use clinical criteria alone as a gold standard, because some pertussis cases presentatypically and the classical pertussis symptoms can also be produced by other infectious agents, such as adeno-virus or Mycoplasma pneumoniae. Consideration should be given to developing gold standard criteria based on an appropriate combinationofthe methods described above as well as otherdiagnostic methods.

Notwithstanding the limitations described above, the PCR assayappeared to perform well in comparison with the other diagnostic techniques currently in use in our clinical labora-tory. Under optimum conditions, the PCR result was avail-able within 5 h ofcollection of the NPA, whereas culture tookatleast 3 or 4 days. The PCR results were negative in only 3 of 66 samples from which B. pertussis was cultured, andthese samples grew fewer than five colonies each. NPA weresubmitted for PCR assay only after samples had been removed for culture and IF analysis. For both these tech-niques, parts ofthe specimen containing mucus were con-sidered more likely to contain the organism and were sam-pledpreferentially. Also, a much larger inoculum was used for culture than for PCR (100 to 200 ,ul for each of three cultureplates,comparedwith20 ,ulfor PCR). Thus, the PCR assays were conducted on much smaller and probably infe-rior-qualitysamples. Giventhe small number of B. pertussis colonieswhich grew from the three aspiratesyielding appar-ently false-negative PCRresults, it seems probable that the samples assayed by PCR did not actually contain any B. pertussis atall. ThePCRassay was positive, however, ina further 35sampleswhichwerenegative by culture;2of these were positive by IF. Classification of the remaining 33 culture- and IF-negative, PCR-positive samples as either true-positiveorfalse-positive isdifficult.For severalofthese specimens, there is strong circumstantial evidence that the PCR result is genuine, as they were collected either from symptomatic family contacts of culture-proven pertussis patients or were follow-up specimens from patients who werepreviously culture positive (and PCR positive)but were culture negative at the time of follow-up sampling. The maximum durationbetweencollection ofthe twospecimens was6weeks. Inaddition, appropriate serumspecimensfor assessment of serological response to B. pertussis by

en-zyme-linked immunosorbent assay wereavailable for 10 of the remaining

PCR-positive,

culture- and IF-negative pa-tients, and five ofthese

yielded positive

results. Unfortu-nately, sera werenotavailable fromthe

remaining patients.

Thus, PCR

amplification

ofreiterated B.

pertussis-specific

DNA sequences has the

potential

to be a rapid, highly sensitive, and

specific

method for

laboratory

diagnosis of

pertussis.

ACKNOWLEDGMENTS

We aregratefulto AndreaMcAdam,Michael Summerford, and AndrewMooreforassistance withthePCR assays,IFanalyses,and

culture analyses, respectively. We also thank John Cox and Anne Berryfor helpful discussions and Peter Griffiths for synthesis of the oligonucleotides.

This work wassupported by a grant from the National Health and Medical ResearchCouncil of Australia.

LITERATURE CITED

1. Alsheikhly, A.-R.,and S. Lôfdahl. 1989. Identification of a DNA fragment in the genome ofBordetella pertussis carrying re-peatedDNAsequences also present in other Bordetella species. Microb.Pathog. 6:193-201.

2. Brown, D. R., and C. D. Parker. 1987.Cloning of the filamen-toushemagglutinin of Bordetella pertussis and its expression in Escherichia coli. Infect. Immun. 55:154-161.

3. Brown, M. C. M., A. Weston, J. R. Saunders, and G. O. Humphreys. 1979.Transformation ofE. coliC600 by plasmid DNA at different phases of growth. FEMS Microbiol. Lett. 5:219-222.

4. Clark-Curtiss, J. E., and M. A. Docherty. 1989. A species-specific repetitive sequence in Mycobacterium lepraeDNA. J. Infect. Dis. 159:7-15.

5. Feinberg, A. P., and B. Vogelstein. 1983. A technique for radiolabelingDNA restriction endonucleasefragmentsto high specific activity. Anal. Biochem. 132:6-13.

6. Goodnow, R. A. 1980. Biology of Bordetella bronchiseptica. Microbiol. Rev.44:722-738.

7. Higgins, C.F., G. F.-L.Ames, W. M.Barnes, J. M. Clement, and M. Hofnung. 1982. A novel intercistronic regulatory ele-mentofprokaryotic operons. Nature (London)298:760-762. 8. Kleckner, N. 1981. Transposable elements in prokaryotes.

Annu. Rev.Genet. 15:341-404.

9. Kraft, R., J.Tardif,K. S.Krauter, and L. A. Leinwand. 1988. Using miniprepplasmid DNAforsequencing double stranded templateswithSequenase. Biotechniques6:544-546.

10. Lautrop,H. 1960. Laboratorydiagnosis of whooping coughor Bordetella infections. Bull. W.H.0.23:15-35.

11. Lawrence, A. J., and J. C. Paton. 1987. Efficacy of enzyme-linked immunosorbent assayfor rapid diagnosis of Bordetella pertussisinfection. J. Clin. Microbiol. 25:2102-2104.

12. Linneman, C. C. 1978.Host-parasite interactions inpertussis, p. 3-18. In C. R. Manclark and J. C. Hill (ed.), International symposium onpertussis. U.S. Departmentof Health, Educa-tion, and Welfare, Bethesda, Md.

13. Maniatis, T.,E. F. Fritsch,andJ. Sambrook. 1982. Molecular cloning: alaboratory manual. ColdSpring Harbor Laboratory, ColdSpring Harbor,N.Y.

14. McLafferty, M. A., D. R. Harcus, and E. L. Hewlett. 1988. Nucleotide sequence and characterization ofarepetitive DNA elementfrom the genome of Bordetellapertussiswith charac-teristicsofaninsertion sequence. J. Gen. Microbiol. 134:2297-2306.

15. McPheat,W. L.,andT. McNally. 1987. Phase IandphaseIV strains ofBordetella pertussis carryarepeatedDNAsequence notfound in other Bordetella species. FEMS Microbiol. Lett. 41:357-360.

16. McPheat,W.L.,and T.McNally. 1987.Isolation ofarepeated

DNA sequencefrom Bordetella pertussis. J. Gen. Microbiol. 133:323-330.

17. Park,I.,W.Saurin,and A. Ullmann.1988.Ahighlyconserved 530base-pair repeatedDNA sequence specific for Bordetella pertussis. FEMSMicrobiol. Lett. 52:19-24.

18. Southern, E. M. 1975. Detection ofspecific sequences among DNAfragmentsseparated bygelelectrophoresis. J. Mol. Biol. 98:503-517.

19. Yanisch-Perron, C., J. Vieira,andJ. Messing. 1985.

Improved

M13 phage cloning vectors and host strains: nucleotide

se-quences oftheM13mpl8andpUC19vectors.Gene33:103-119.