Vol. 32, No. 3

JOURNAL OF CLINICAL MICROBIOLOGY, Mar. 1994? p. 783-789 0095-1 137/94/$04.00+0

Identification of Bordetella pertussis Infection by Shared-Primer PCR ZHONGMING LI,'* DEBORAH L. JANSEN,' THERESA M. FINN,' SCOTT A. HALPERIN,2 ALICIA KASINA, STEVEN P. O'CONNOR,3 TATSUO AOYAMA,4 CHARLES R. MANCLARK,' AND MICHAEL J. BRENNAN' Division of Bacterial Prodlucts, Center for Biologics Evalulation and Research, U. S. Food and Drug Administration, Bethesda, Maryland 20892,' Dalholusie University, Halifax, Nova Scotia B3J 3G9, Canada2; Division of Bacterial and Mycotic Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia 303333; and Kawasaki Municipal Hospital, Kawasaki, Japan4 Received 20 August 1993/Returned for modification 12 October 1993/Accepted 9 December 1993

A shared-primer PCR method for the detection of infection was developed by using primers derived from DNA sequences upstream of the structural genes for the porin proteins of Bordetella pertussis and Bordetella parapertussis. This method resulted in a 159-bp PCR product specific for B. pertussis and a 121-bp DNA fragment specific for B. parapertussis and allowed for the simultaneous detection of these pathogens. The PCR procedure was shown to be very specific since no PCR product was obtained from 36 non-Bordetella bacterial DNAs. Nasopharyngeal aspirates (NPAs) from children suspected of having pertussis were evaluated by the PCR method, culture, and the Chinese hamster ovary (CHO) cell assay, which detects pertussis toxin. B. pertussis was cultured from 119 of 205 NPAs assayed, and the presence of pertussis toxin was detected in 69 of the NPAs by the CHO cell assay. When ethidium bromide staining was used to detect PCR products, 100 NPAs gave positive results by shared-primer PCR; 94 of these NPAs were also positive by culture. The result indicated a sensitivity of 79% for PCR when culture was used as the standard. The sensitivity of PCR was increased to 95% when a digoxigenin immunoblot system was used. An additional 20 NPAs from patients with suspected pertussis that were culture negative also gave positive results by PCR. The specific and sensitive PCR method described here should be useful for both the clinical diagnosis of pertussis and case identification in vaccine trials.

genin immunoblot system was developed to increase the sensitivity of the PCR.

The detection of Bordetella pertussis infection in a timely manner by sensitive techniques is very important for the management of pertussis disease. Also, the specific identification of pertussis cases in vaccine trials is essential for an accurate assessment of vaccine efficacy. Current rapid methods of pertussis diagnosis are relatively insensitive or nonspecific. At present, bacterial culture is the standard technique used for the diagnosis of pertussis. However, the organism is slow growing, and in many clinical situations, such as with the concurrent administration of antibiotics, isolation rates are low (19). Reliance on culture alone for the diagnosis of pertussis in a placebo-controlled vaccine trial may lead to falsely high estimates of vaccine efficacy (6). In addition, Bordetella parapertussis can complicate the diagnosis since it is a human pathogen reported to cause approximately 5% of documented Bordetella infections (14). Although B. pertussis can routinely be cultured, rapid diagnostic methods with improved sensitivities and specificities, that are easier to use, and that allow for the simultaneous detection of other pathogenic Bordetella species are needed. In the present study, a shared-primer PCR method was developed by using two primers derived from the unique DNA sequences upstream of the porin genes of B. pertussis and B. parapertussis and a third primer, which is shared by both species, derived from the DNA sequence adjacent to the porin genes. The specificity and sensitivity of this method were compared with those of culture and the Chinese hamster ovary (CHO) cell assay by testing nasopharyngeal aspirates (NPAs) obtained from patients suspected of having pertussis. A digoxi*

MATERIALS AND METHODS Bacterial strains and chromosomal DNA. B. pertlussis Tohama I, Tohama III, 347, 18323, and 10901, B. parapertussis 482, 500, and 23054, and Bordetella bronchiseptica 058, 1 1OH, and 207 were obtained from the Laboratory of Pertussis Culture Collection, the Center for Biologics Evaluation and Research, U.S. Food and Drug Administration. Bacterial culture and chromosomal DNA extraction were performed as described previously (13). Chromosomal DNAs from eight different Bordetella avium species were obtained from Claudia Gentry-Weeks, National Institute of Dental Research, Bethesda, Md. Chromosomal DNAs from 36 different bacteria representing 18 genera were obtained from the Centers for Disease Control and Prevention, Atlanta, Ga. Reagents and synthetic oligonucleotide primers. Rcstriction enzymes and T4 DNA ligase were purchased from Bethesda Research Laboratories (Gaithersburg, Md.). Taq DNA polymerase and the Carry-Over prevention kit were purchased from Perkin-Elmer Cetus (Norwalk, Conn.). Digoxigenin-11dUTP and a digoxigenin detection kit were obtained from Boehringer Mannheim Biochemicals (Indianapolis, Ind.). The TA cloning vector was obtained from Invitrogen (San Diego, Calif.). Oligonucleotide primers were synthesized by Lofstrand Laboratories Ltd. (Gaithersburg, Md.). The Sequenase DNA sequencing kit was purchased from United States Biochemicals

(Cleveland, Ohio). Collection and preparation of samples. NPAs were obtained by syringe aspiration with a fine flexible catheter from children suspected of having pertussis and from family members of patients diagnosed with pertussis. Of 239 NPAs, 212 specimens

Corresponding author. Mailing address: Laboratory of Pertussis,

Division of Bacterial Products, Centcr for Biologics Evaluation and Research, U.S. Food and Drug Administration, 880() Rockville Pike, Bethesda, MD 2(1892. Phone: (301) 496-4288. Fax: (301) 402-2776.

783

784

LI ET AL.

including 7 samples collected from patients without pertussis who were confirmed to have respiratory syncytial virus infections were obtained from the Izaak Walton Killam Children's Hospital, Halifax, Nova Scotia, Canada. Culture and the CHO cell assay were performed at the Izaak Walton Killam Children's Hospital. Another 27 NPAs including 8 sequential samples from two children collected at intervals of between 2 and 5 days were from the Kawasaki Municipal Hospital, Kawasaki, Japan. Following aspiration, mucus was expelled and rinsed from the catheter with 0.8 ml of phosphate-buffered saline-I % Casamino Acids. Samples of diluted NPAs were plated onto media for bacterial culture before storage at - 20°C and were sent to the Center for Biologics Evaluation and Research, U.S. Food and Drug Administration, for analysis by PCR. Bacterial culture. For culture, 50 jil of diluted NPAs from each clinical sample was plated onto charcoal blood agar containing cephalexin (40 Kg/ml), as described by Regan and Lowe (20), and the plate was incubated at 37°C. The plate was examined at 72 h and then daily for 7 days. Suspect colonies were identified by Gram stain morphology, immunofluorescense, and agglutination by using B. pertussis- and B. parapertussis-specific antisera (8). Paired serology assay. Serum samples taken from patients at their initial visit and during convalescence were used in a paired serology assay. Antibodies were measured by immunoglobulin G (IgG) and IgA enzyme immunoassay (8) by using pertussis toxin and filamentous hemagglutinin as antigens. The criterion for positivity was a fourfold rise in either IgG or IgA antibody titer to either antigen in the convalescent-phase serum sample compared with that in the initial serum sample. CHO cell assay. The CHO cell assay was performed as described previously (7). CHO cells (3 x 104) and diluted NPAs (50 [l) were added to each well of a 96-well microtiter plate containing 250 [l of medium (Ham's F-12 medium containing 10% fetal calf serum) per well. After incubation for 20 h in a CO2 incubator at 37°C, the cells were examined under a microscope to determine the extent of clustering, which is a measure of the presence of pertussis toxin. PCR amplification of porin genes. Several pairs of synthetic oligonucleotide primers derived from the porin gene sequence of B. pertussis (13) were used for amplification of DNA fragments of the B. pertussis, B. parapertussis, B. bronchiseptica, and B. avium porin genes. PCR amplification was conducted in a 30-[il reaction mixture containing 3 [l of template of each chromosomal DNA (3 ,ug/ml), 20 pmol of each primer, 125 p.M (each) dATP, dCTP, dGTP, and dTTP, 2.5 U of Taq polymerase, and 3 pl of 10x reaction buffer (100 mM Tris [pH 8.4], 500 mM KCl, 0.1% Triton X-100, 1.0% gelatin, 15 mM MgCl2). The reaction mixture was covered with mineral oil and was amplified for 30 cycles in a Perkin-Elmer Cetus thermal cycler. Each cycle consisted of 1 min each of denaturation, annealing, and elongation steps at temperatures of 94, 50, and 72°C, respectively. The amplified PCR products were detected after electrophoresis through 2% agarose gels in the presence of 0.5 Kig of ethidium bromide per ml for approximately I h at 10 V/cm; this was followed by photography under UV illumination. Inverse PCR and DNA sequencing. Chromosomal DNAs (0.5 [ig) of B. pertussis Tohama I, B. parapertussis 23054, and B. bronchiseptica 058 were each digested with Sall. The enzyme was inactivated by phenol extraction, and the Sall DNA fragments were self-ligated with T4 DNA ligase in a dilute DNA concentration (1 p.g/ml) that favors the formation of monomeric circles (2). Inverse PCR was performed on these circular DNA templates from B. pertussis, B. parapertussis, and B. bronchiseptica by using primers derived from the N-terminal

J. CLIN. MICROBIOL.

DNA and C-terminal DNA of the B. pertussis porin gene (see Fig. IA). Amplification was performed as described above. For nucleotide sequence analysis, the B. parapertussis inverse PCR product was then inserted into the TA cloning vector. This insert was sequenced with the Sequenase DNA sequencing kit by using the dideoxy chain-termination method, and the sequence result was confirmed by Lark Sequencing Technology Inc., Houston, Tex. Shared-primer PCR. A shared-primer PCR protocol was developed by using two primers derived from unique DNA sequences upstream of the porin structural genes of B. pertussis and B. parapertussis. A third primer was derived from the DNA sequence common to both species (see Fig. 2). This PCR method amplifies a 159-bp DNA fragment specific for B. pertussis and a 121-bp DNA fragment unique to B. parapertussis and allows for the simultaneous detection of B. pertussis and B. parapertussis. For PCR, NPAs were initially prepared by liquefying a 35-p,l sample with an equal volume of a 1:9 solution of 20% N-acetylcysteine and 0.5 N NaOH (11). The mixture was vortexed for 30 s and was incubated at room temperature for 15 min. The samples were then centrifuged at 14,000 x g for 10 min in a Microfuge, and the pellet was resuspended in 30 pl of the PCR mixture. The PCR mixture contained 1 x reaction buffer, the three primers at 20 pmol each (Fig. 2), 2.5 U of Taq polymerase, 125 jiM dATP, 125 jiM dCTP, and 125 jiM dGTP. To prevent carryover contamination (16), 0.5 U of uracil N-glycosylase (UNG) was added and 250 jiM dUTP was substituted for 125 FiM dTTP (UNG and dUTP were supplied in the Carry-Over prevention kit). Samples were incubated at 37°C for 10 min and then for 10 min at 94°C to inactivate the UNG and lyse the bacteria. Two-temperature DNA amplification (3) was used for 40 cycles. Denaturation and annealing were carried out at 94 and 65°C for 30 s each. Elongation occurred during the 1°C/s change in temperature from 65 to 94°C. Positive controls of B. pertussis and B. parapertussis DNAs and a negative control from a pool of NPAs from five healthy adults were included in each PCR run. A positive PCR result was detected by ethidium bromide staining of agarose gels after electrophoresis. Digoxigenin immunoblot system. In the detection system described here, sample treatment and DNA amplification were completed as described above, except that 1 nmol of digoxigenin-1 1-dUTP was included in the PCR mixture (15). The digoxigenin-labeled PCR products were first analyzed by ethidium bromide staining after electrophoresis and were then transferred onto nitrocellulose sheets (Schleicher & Schuell, Keene, N.H.) without DNA denaturation. Digoxigenin detection was performed according to the manufacturer's protocol. Briefly, filter membranes containing the immobilized PCR products were first incubated in 10 ml of blocking solution provided by the digoxigenin detection kit at room temperature for 30 min. Following a brief wash with 50 mM Tris-buffered saline (TBS; pH 7.5), membranes were incubated for 1 h with a 1:1,000 dilution of alkaline phosphatase-conjugated antibody against digoxigenin in TBS. Membranes were washed three times for 10 min each time with TBS and were then developed in 10 ml of developing buffer (100 mM Tris [pH 9.5], 100 mM NaCl, 50 mM MgCl2) containing 50 ,ul of Nitro Blue Tetrazolium and 37.5 [l of 5-bromo-4-chloro-3-indolylphosphate solutions. Color development was terminated after 10 min by a brief washing in TE buffer and air drying.

VOL. 32, 1994

IDENTIFICATION OF B. PERTUSSIS BY PCR A

785

B PrimeF

Brgmer

ORF of B. perrussis porin gene

Sal I

1 5 Kb Sal

fragmenlt

SaI

---.

723 bp

---- 359 bp

t

123 bp

_

1 2 3 4 FIG. 1. Inverse PCR and amplified products. (A) Primers derived from B. pertussis N-terminal and C-terminal DNAs are indicated with arrows. The restriction enzyme Sall site is shown. ORF, open reading frame. (B) Inverse PCR products amplified from circularized Sall DNA fragments of B. pertussis, B. parapertussis, and B. bronchiseptica. The positions of the 123-bp DNA ladder marker (lane 1), the 359-bp PCR product found in B. pertussis Tohama I (lane 2), and the 723-bp PCR products found in B. parapertussis 23054 (lane 3) and B. bronchiseptica 058 (lane 4) are shown.

RESULTS PCR amplification of porin genes and inverse PCR. The porin gene of B. pet-tussis has recently been cloned and sequenced (13). When chromosomal DNAs from various strains of B. pertussis, B. parapertussis, and B. bronchiseptica were digested with the restriction enzyme Sall and probed with a DNA fragment from the N-terminal portion of the porin gene, a 1.5-kb band from B. pertussis DNA and a 1.7-kb band from B. parapertussis or B. bronchiseptica DNA were detected by Southern blotting (13). Several pairs of primers derived from the B. pertussis porin gene were used for PCR with DNA templates from different Bordetella species. None of these primers amplified a PCR product from B. avium DNA. Similar PCR products were obtained from B. parapertussis and B. bronchiseptica DNAs when the primers derived from the DNA sequence within the B. pertussis porin structural gene were used, but when a pair of primers derived from the region upstream of the porin gene was used, PCR products were made only from B. pertussis DNA (data not shown). This suggests that both B. parapertussis and B. bronchiseptica have similar porin structural genes compared with that of B. pertussis, but that the DNA sequences upstream of the porin gene are different. To characterize the upstream DNA sequences, inverse PCR was performed and the amplified PCR products of both B. parapertussis and B. bronchiseptica were determined to be 364 bp longer than the B. pertussis PCR product (Fig. I B). The inverse PCR product from B. parapertussis was inserted into the TA cloning vector, and the DNA upstream of the B. parapertussis porin gene was sequenced (Fig. 2). The DNA sequences of B. pertussis and B. parapertussis are identical up to nucleotide 210 upstream of the start codon of the porin structural gene (Fig. 2; note arrow). At this site, a 1,053-bp repetitive, transposon-like DNA sequence is inserted into the B. pertussis chromosomal DNA (13, 18). Three 24-mer oligonucleotide primers from this region were designed; P1 (5'-TGCAACATCCTGTCCCCTTAATCC-3') is derived from the DNA sequence which is identical between B. pertussis and B. parapertussis, P2 (5'-ATGCTTATGGGTGTTCATC CGGCC-3') is specific for B. pertussis, and P3 (5'-CGTCCAC CAGGGGTGGTAGGAGAT-3') is specific for B. parapertussis (Fig. 2). Specificity and sensitivity of shared-primer PCR. By using the three primers mentioned above, shared-primer PCR was

evaluated by amplifying DNAs from Bordetella and nonBordetella species. The results showed that a 159-bp PCR product was detected on 2% agarose gels only from B. pertussis DNA (Fig. 3, lanes 2 and 3). A 121-bp PCR product was detected when DNA from either B. parapertussis (Fig. 3, lanes 2 and 4) or B. bronchiseptica (Fig. 3, lane 5) was used as the PCR template. A doublet (159- and 121-bp fragments) was obtained when DNAs from both B. pertussis and B. parapertussis were present (Fig. 3, lane 2). No PCR product was identified on the ethidium bromide-stained agarose gel after amplification of DNA from B. avium (Fig. 3, lane 6). Similar results were obtained when DNAs from 12 B. pertussis, 6 B. parapertussis, 8 B. bronchiseptica, and 8 B. avium strains were tested. In addition, no PCR products were obtained when human genomic DNA or DNAs from 36 non-Bordetella species representing 17 genera including Streptococcus, Neisseria, Escherichia, Haemophilus, Mycobacterium, Pseudomonas, Brucella, Mycoplasma, Staphylococcus, Salmonella, Candida, Alcaligenes, Chlamydia, Proteus, Legionella, Klebsiella, and Moraxella were tested. The sensitivity of shared-primer PCR was evaluated by using purified chromosomal DNA. The results showed that as little as 500 fg of B. pertussis or B. parapertussis DNA allowed amplification of the expected 159- or 121-bp DNA fragment, respectively. However, by using a digoxigenin immunoblot system for the detection of PCR products, as little as 5 fg of DNA as starting material could be detected with no apparent loss of specificity (data not shown). Comparison of culture, CHO cell assay, and PCR. NPAs were obtained from children suspected of harboring pertussis infections and from family members of patients diagnosed with pertussis. Of 205 NPAs tested, B. pertussis was cultured from 119 NPA samples by standard techniques (Table 1), but no B. parapertussis infections were identified. The presence of pertussis toxin in NPAs, an indication of pertussis infection, was investigated by the CHO cell assay, and 69 positive NPAs were identified when the 205 NPAs were assayed. The CHO cell assay detected 65 of the 119 culturepositive samples, while 4 culture-negative samples gave positive CHO cell assay results. These four samples were, however, positive by PCR (Table 1). This indicates that although the CHO cell assay can detect pertussis toxin in some patients from whom B. pertussis cannot be cultured, it is unlikely that

786

J. CLIN. MICROBIOL.

LI ET AL. ATG

Sal

Sal

I

ORF of B. Pertussis Porin Gene P2

P1

Sal

ATG

Sal I P3 ---o

ORF of B. parapertussis Porin Gene

P1 ~-,

Primer P2 Pert. ATGCTTATGGGTGTTCATCCGGCCGGGCTCCTTGACTGAACTGGGGGGTCGGCGA TTTCCAG Para.

cgggcgggcacgggcacacttggtgaggtcgggcgaatcgtccaccaggggtggtaggagat

Pert. Para.

TTTCTCAAATCCGGTTCGGATGAACCATGCATACAACCTATTGAATCTTCACAGTTAGCCCG

Primer P3 V

cggcgcggggcaggcgggcaggagcttgttgcattgcgatgcgccgccctagggttagcccg

Pert. CGCGCGATTCCGGATTAAGGGGACAGGATGTTGCAACTTACCAACAATGGGGCGGG:::::: Para. cgcgcgattccggattaaggggacaggatgttgcaacttaccaacaatggggcggg::::::

Primer P1 Pert. Para.

:

:::::::

:ATG

:::::::::::::::::atg

ORF of B.pertussis Porin Gene ORF of B.parapertussis Porin Gene

FIG. 2. Primers used for shared-primer PCR. Regions where B. pertussis (Pert.) and B. parapertussis (Para.) DNA sequences differ are indicated in italics. The B. pertussis-specific 1,053-bp repetitive DNA sequence insertion site is designated with an arrowhead. The three primers used for shared-primer PCR (P1, P2, and P3) are underlined with arrows. P1 is derived from the DNA sequence which is identical between B. pertussis and B. parapertussis. ORF, open reading frame.

the test will detect additional positive NPAs not identified by PCR. By using the ethidium bromide staining method to detect PCR products, 100 positive NPAs among 205 specimens tested were identified by shared-primer PCR. Among them, 94 were culture positive, indicating a sensitivity of 79%. PCR did not detect 25 of the culture-positive patients (Table 1). However, when the more sensitive digoxigenin immunoblot system was used, 95% of the 119 culture-positive samples were then detected by PCR (Table 1). Seven culture-negative NPAs from

TABLE 1. Detection of B. pertussis in NPAs by PCR and the CHO cell assay compared with that by culture No. of NPAs PCR" Culture result

1 2 3 4 5 6 FIG. 3. Specificity of shared-primer PCR. The 123-bp DNA ladder marker (lane 1) and PCR products from B. pertussis Tohama I and B. parapertussis 23054 DNA (lane 2), B. pertussis Tohama I DNA (lane 3), B. parapertussis 23054 DNA (lane 4), B. bronchiseptica 058 DNA (lane 5), and B. avium GOBLI24 DNA (lane 6). A 159-bp PCR product was amplified from B. pertussis DNA, a 121-bp PCR product was amplified from B. parapertussis or B. bronchiseptica DNA, and no PCR product was amplified from B. avium DNA.

+

+ -

65 4

Ethidium bromide or staining immunoblotting"

Ethidium

CHO cell assay

bromide

bromiden staining

-

+

_

54 82

94 6

25 80

+_

113 20

6 66

An additional seven NPAs (not shown here) collected from patients without pertussis and which were confirmed as being infected with respiratory virus were all negative by PCR. * Because of the lack of sample, PCR results were not detected by both ethidium bromide staining and immunoblotting in all cases.

VOL. 32, 1994

IDENTIFICATION OF B. PERTUSSIS BY PCR

787

TABLE! 2. Comparison of characteristics of PCR-positive, culturenegative patients with PCR-negative, culture-negative patients

A

PCR result

Culture result

+

-

-

B

-

Corroborating evidence for pertussis

CHO cell assay Paired serology

No. of positive patients

Epidemiologic link" Clinically compatiblec No supportive evidencee

4 (20)" 1(6) 4 (13) 9 (1t)d 1 (16Y

CHO cell assay Paired serology Epidemiologic link Clinically compatible No supportive evidence

0 (66) 14 (38) 6 (35) 15 (28)C 13 (29k

aValues in parentheses are the total number of samples examined. Diagnosis was made by exposure history without laboratory confirmation. Epidemiologic link includes family member with a positive culture for B. pertussis or a patient having prior or subsequent positive culture of NPAs. c Diagnosis made by the presence of clinically compatible symptoms without laboratory confirmation. Clinically compatible symptoms include paroxysmal cough, cough with whoop, or cough ending with vomiting. d p = 0.04 (chi-square test). ' No laboratory or epidemiologic supporting evidence in which both additional laboratory tests (CHO and/or serology) and historical and epidemiologic information were available. f P = 0.008 (chi-square test). b

1

2 3 4

5

6

7

8

9 10 11 12 13

FIG. 4. Use of the digoxigenin immunoblot system for the detection of B. pertussis DNA in clinical samples. PCR results were detected by ethidium bromide staining (A) or by the digoxigenin immunoblot system (B). Lane 1, 123-bp DNA ladder marker; lanes 2 through 11, NPAs from patients suspected of having pertussis; lane 12, NPA pool from normal adults as a negative control; lane 13, 1 pg of B. pertussis DNA as a positive control.

patients without pertussis who were confirmed to have respiratory virus infections were included in this experiment, and all were negative by PCR. Examples of the PCR results obtained by ethidium bromide staining and with the digoxigenin immunoblot system are compared in Fig. 4. Ten clinical samples were evaluated; seven were culture positive. Four culturepositive samples were PCR positive (Fig. 4, lanes 5 to 8) by ethidium bromide staining. However, all seven culture-positive samples were identified by PCR when the digoxigenin immunoblot system was used (Fig. 4, lanes 4 to 9 and 11). Twenty PCR-positive samples, including 6 samples that were positive by ethidium bromide staining and 14 samples that were positive by immunoblotting, were detected in those samples that were negative by culture (Table 1). The characteristics of PCR-positive, culture-negative patients compared with those of PCR-negative, culture-negative patients are given in Table 2. Four of the PCR-positive samples were also positive in the CHO cell assay (one was positive by paired serology and the patient presented with a clinically compatible illness), and four had strong epidemiologic links to culturepositive patients, including an individual whose NPA was previously positive on culture. An additional eight individuals had clinical symptoms, including paroxysmal cough, cough with whoop, or cough ending with vomiting, which were compatible with pertussis. Only one patient with respiratory illness had no laboratory, clinical, or epidemiologic evidence of pertussis (Table 2). No clinical or epidemiologic information was available for three patients. A clinically compatible illness was less

in culture-negative, PCR-negative samples (P = 0.04). In addition, more samples from this group of patients had no laboratory, clinical, or epidemiologic evidence supportive of pertussis (P = 0.008). In an additional study, sequential NPAs taken from two Japanese children who were treated with antibiotics were evaluated by the shared-primer PCR. A total of eight samples from the two patients were obtained. Four samples were collected from each patient at intervals of 2 to 5 days following the initial visit. While only three specimens from the first and second samplings were positive by culture for B. pertussis, all eight NPAs were determined to be positive by PCR by using the digoxigenin immunoblot system. This result indicates that PCR may detect B. pertussis in patients who have been treated with antibiotics and whose NPAs give negative results by culture techniques. common

DISCUSSION A rapid diagnostic method for the detection of B. pertussis will allow for the early detection and treatment of patients with active pertussis. Prompt prophylactic treatment of household members and other contacts with antibiotics may help interrupt disease transmission and prevent further illness. In addition, a specific and sensitive diagnostic method for pertussis is essential for the accurate assessment of the efficacies of vaccine trials. At present, confirmation by bacterial culture is the most specific test for the diagnosis of pertussis, but it takes investigators or clinicians at least 4 or 5 days to obtain results. Moreover, the sensitivity of the culture method decreases with disease duration and antibiotic treatment. Additional time is also needed for the confirmation of the presence of B. pertussis in cultures by using fluorescent-antibody or bacterial agglutination assays. The serological response to specific pertussis antigens is used to verify infection (17), but this method suffers from the lack of an antibody response during the early course of infection and in children under 4 months of age. It appears that no single test will be appropriate for the diagnosis of all suspected cases of pertussis. In the present study, the specificity and sensitivity of a

788

LI ET AL.

shared-primer PCR technique were assessed by culture and the CHO cell assay, which measures the presence of pertussis toxin, a protein secreted by viable B. pertussis cells. Ethidium bromide staining of PCR products in agarose gels yielded 94 PCR-positive samples in the set of 119 samples (79%) from which B. pertussis could be cultured. When the digoxigenin immunoblot system was applied, 95% (113 of 119 samples) of the culture-positive samples were PCR positive. In the digoxigenin immunoblot system described here, digoxigenin-11dUTP was added to the PCR mixture and the PCR products were detected with conjugated digoxigenin-specific antibody. The digoxigenin immunoblot system increased the sensitivity of PCR for detecting B. pertussis DNA by approximately 100-fold over that by ethidium bromide staining and had no apparent effect on specificity (data not shown). An additional 20 PCR-positive NPAs were detected among culture-negative NPAs. The PCR technique described here appears to be more sensitive than culture for detecting B. pertussis in patients with mild disease or those treated with antibiotics, as indicated by reviewing the clinical histories of these patients. PCR may be able to detect B. pertussis DNA in samples from which organisms cannot be cultured, perhaps because of the fastidious nature of the organism, the timing of the specimen collection, or prior antimicrobial use. A paired serologic assay was performed on samples from 79 individuals. Samples from 14 patients were positive by serologic testing but negative by culture and PCR. The frequency of negative PCR results for patients with positive serologic results suggests that, like culture, diagnosis by PCR may be most useful early in the course of illness (Table 2). Detection of B. pertussis DNA in clinical samples by PCR was improved in the present studies by combining the annealing and extension steps under stringent temperature conditions (3). By using a two-temperature (65 and 94°C) PCR protocol and modifying the DNA transfer procedure, the PCR immunoblotting technique was conducted within 8 h and the results were obtained on the same day. Before DNA amplification, clinical samples need to be treated to remove Taq polymerase inhibitory factors (21) and to release the DNA from cells. Many different approaches can be used to prepare clinical samples for PCR (10). In our hands, treatment of the NPAs from patients suspected of having pertussis with the mucolytic agent N-acetyl-L-cysteine at high pH (11) facilitated the centrifugation of the organisms and eliminated Taq polymerase inhibitory factors. In other studies, samples were treated, before PCR amplification, with proteinase K at 65°C for 1 h and were heated for 10 min at 100°C in order to release the DNA from cells (5, 9). However, in the present study, incubation of the sample for 10 min at 94°C was sufficient to release the DNA from the bacteria. In addition, the use of PCR for sensitive detection is complicated by the fact that amplified molecules can potentially contaminate subsequent amplifications of the same target sequence (12). In the present study, dUTP was substituted for dTTP in the PCR mixture, resulting in incorporation of dU in place of dT in the amplified PCR products. Since digoxigenin is linked to dUTP, carryover contamination can also be prevented by using UNG in the digoxigenin immunoblot system (16), which can be universally applied for the diagnosis of other infectious diseases by PCR. B. parapertussis is capable of producing mild upper respiratory tract disease, and patients with severe cases of infection may present with dual infections with both B. pertussis and B. parapertussis (14). The shared-primer PCR described here is also capable of detecting B. pertussis and B. parapertussis in NPAs simultaneously without adjusting the relative concentrations of different primers or staggering the amplifications (1,

J. CLIN. MICROBIOL.

4). In our analysis of 27 samples obtained from Japan, one B. parapertussis infection was detected (data not shown) by the shared-primer PCR method. B. parapertussis was not found in the 205 NPAs collected in Canada. Although it is impossible to distinguish B. parapertussis from B. bronchiseptica on the basis of an evaluation of PCR products on agarose gels, human infection with B. bronchiseptica is very rare and the clinical symptoms of lower respiratory tract infections can be distinguished from those of upper respiratory tract infections caused by B. parapertussis (22). The simultaneous detection of different Bordetella pathogens in clinical specimens by sharedprimer PCR may aid in the epidemiologic analysis of the respiratory illnesses caused by these organisms. A number of laboratory methods used to diagnose pertussis were compared in the present study. The CHO cell assay appeared to be very specific, but it was somewhat less sensitive than culture. Negative results may be obtained at the early stages of the disease, when the concentration of pertussis toxin is too low to be detected. Bacterial culture has been used as the standard for the diagnosis of pertussis, but many factors may affect the laboratory's ability to culture B. pertussis, particularly the time of sample collection and the use of antibiotics. The molecular diagnostic techniques described here provide rapid, sensitive, and reliable methods for the diagnosis of pertussis and should be considered for use in the laboratory diagnosis of pertussis in clinical trials. ACKNOWLEDGMENTS We thank Claudia Gentry-Weeks, National Institute of Dental Research, for B. avium DNA, Annette Morris, Dalhousie University, Halifax, Nova Scotia, Canada, for compiling the clinical data on the patients; and Sheldon Morris and Bruce Meade, Center for Biologics Evaluation and Research, U.S. Food and Drug Administration, for helpful comments. REFERENCES 1. Bej, A. K., M. H. Mahbubani, R. Miller, J. L. DiCesare, L. Haff, and R. M. Atlas. 1990. Multiplex PCR amplification and immobilized capture probes for detection of bacterial pathogens and indicators in water. Mol. Cell. Probes 4:353-365. 2. Collins, F. S., and S. M. Weissman. 1984. Directional cloning of DNA fragments at a large distance from an initial probe: a circularization method. Proc. Natl. Acad. Sci. USA 81:6812-6816. 3. Dodson, L. A., and J. A. Kant. 1991. Two-temperature PCR and heteroduplex detection: application to rapid cystic fibrosis screening. Mol. Cell. Probes 5:21-25. 4. Frankel, G., J. A. Giron, J. Valmassoi, and G. K. Schoolnik. 1989. Multi-gene amplification: simultaneous detection of three virulence genes in diarrhoeal stool. Mol. Microbiol. 3:1729-1734. 5. Glare, E. M., J. C. Paton, R. R. Premier, A. J. Lawrence, and I. T. Nisbet. 1990. Analysis of a repetitive DNA sequence from Borde-

tella pertussis and its application to the diagnosis of pertussis using the polymerase chain reaction. J. Clin. Microbiol. 28:1982-1987. 6. Hallander, H. 0., J. Storsaeter, and R. Mollby. 1991. Evaluation of serology and nasopharyngeal cultures for diagnosis of pertussis in a vaccine efficacy trial. J. Infect. Dis. 163:1046-1054. 7. Halperin, S. A., R. Bortolussi, A. Kasina, and A. J. Wort. 1990. Use of a Chinese hamster ovary cell cytotoxicity assay for the rapid diagnosis of pertussis. J. Clin. Microbiol. 28:32-38. 8. Halperin, S. A., R. Bortolussi, and A. J. Wort. 1989. Evaluation of culture, immunofluorescence, and serology for the diagnosis of pertussis. J. Clin. Microbiol. 27:752-757. 9. Houard, S., C. Hackel, A. Herzog, and A. Bollen. 1989. Specific identification of Bordetella pertussis by the polymerase chain reaction. Res. Microbiol. 140:477-487. 10. Kawasaki, E. S. 1990. In M. A. Innis, D. H. Gelfand, J. J. Sninsky, and T. J. White (ed.), PCR protocols: a guide to methods and applications, p. 146. Academic Press, Inc., San Diego, Calif. 11. Kubica, G. P., W. E. Dye, and M. L. Cohn. 1963. Sputum digestion

IDENTIFICATION OF B. PERTUSSIS BY PCR

VOL. 32, 1994

12. 13.

14.

15. 16. 17.

and decontamination with N-acetyl-L-cysteine-sodium hydroxide for culture of mycobacteria. Am. Rev. Respir. Dis. 87:775-779. Kwok, S., and R. Higuchi. 1989. Avoiding false positives with PCR. Nature (London) 339:237-238. Li, Z. M., J. H. Hannah, S. Stibitz, N. Y. Nguyen, C. R. Manclark, and M. J. Brennan. 1991. Cloning and sequencing of the structural gene for the porin protein of Bordetella pertussis. Mol. Microbiol. 5:1649-1656. Linnemann, C. C., Jr., and E. B. Perry. 1977. Bordetella parapertussis. Am. J. Dis. Child. 131:560-563. Lion, T., and 0. A. Haas. 1990. Nonradioactive labeling of probe with digoxigenin by polymerase chain reaction. Anal. Biochem. 188:335-337. Longo, M. C., M. S. Berninger, and J. L. Hartley. 1990. Use of uracil DNA glycosylase to control carry-over contamination in polymerase chain reactions. Gene 93:125-128. Manclark, C. R., B. D. Meade, and D. G. Burstyn. 1986. Serological response to Bordetella pertussis, p. 388-394. In N. R. Rose, H.

18.

19. 20. 21. 22.

789

Friedman, and J. L. Fahey (ed.), Manual of clinical laboratory immunology, 3rd ed. American Society for Microbiology, Washington, D.C. McLafferty, M. A., D. R. Harcus, and E. L. Hewlett. 1988. Nucleotide sequence and characterization of a repetitive DNA element from the genome of Bordetella pertussis with characteristics of an insertion sequence. J. Gen. Microbiol. 134:2297-2306. Onorato, I. M., and S. G. F. Wassilalk 1987. Laboratory diagnosis of pertussis: the state of the art. Pediatr. Infect. Dis. J. 6:145-151. Regan, J., and F. Lowe. 1977. Enrichment medium for the isolation of Bordetella. J. Clin. Microbiol. 6:303-309. Walsh, P. S., D. A. Metzger, and R. Higuchi. 1991. Chelex 100 as medium for simple extraction of DNA for PCR-based typing from forensic material. BioTechniques 10:506-513. Woolfrey, B. F., and J. A. Moody. 1991. Human infections associated with Bordetella bronchiseptica. Clin. Microbiol. Rev. 4:243255.