Detection and molecular typing ofCampylobacter jejuniin fecal samples by polymerase chain reaction

Molecular and Cellular Probes (1996) 10, 75–80

Detection and molecular typing of Campylobacter jejuni in fecal samples by polymerase chain reaction Alice Waegel1 and Irving Nachamkin2∗ 1

Department of Biology, Neumann College, Aston, Pennsylvania and 2Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA (Received 17 July 1995, Accepted 19 September 1995)

In order to determine whether polymerase chain reaction (PCR) could be used to detect Campylobacter jejuni directly in stool samples, DNA from 66 frozen culture positive and negative fecal samples was purified by column chromatography. The flaA gene was amplified using primers directed against the conserved 5′ and 3′ regions and produced a 1·7 kb amplicon. Fifteen of 18 (83%) C. jejuni culture positive samples were detected by agarose gel electrophoresis and ethidium bromide staining. The test was negative in one sample containing C. coli. Twelve samples containing other enteric pathogens were negative as were 34 of 35 culture negative samples. Flagellin gene typing (see reference14) of the flaA gene product from two stool samples in which the patients’ stool isolate was also available showed the identical flagellin gene types suggesting that molecular typing of Campylobacter could potentially be performed on stool samples without the need for culture.  1996 Academic Press Limited KEYWORDS: Campylobacter, flagellin gene, polymerase chain reaction, molecular epidemiology.

INTRODUCTION Campylobacter jejuni is a leading cause of bacterial gastroenteritis and is estimated to cause over two million cases annually.1 The diagnosis of Campylobacter gastroenteritis is usually made by isolation of the organism from stool cultures and requires several days for isolation and identification.2 Direct methods for detecting Campylobacter in stool samples using microscopic examination and gram-stain or immunofluorescence staining of fresh stool samples have been described with sensitivity ranging from 40–95% with good specificity.3–5 However, the ability to visualize Campylobacter in clinical samples is dependent upon the expertise of the microscopist and may have relatively poor performance in some laboratories.3,5

Molecular approaches to detect Campylobacter in samples in lieu of culture have been described including detection of C. jejuni in chicken6,7 and dairy products.8,9 For clinical purposes, direct detection of Campylobacter in stool samples without culture could be useful if the time to diagnosis was shortened and thus, have an impact on decisions in clinical management of patients with infection. Oyofo et al. reported on the detection of C. jejuni and C. coli in seeded stool samples, infected animals, and some frozen stool samples from military personnel with diarrheal illness.10 Their results suggested that further studies on clinical samples from patients with diarrheal illness were warranted. PCR has the potential to provide a rapid, specific diagnosis of Campylobacter

∗ Correspondence and reprint requests to: Dr Irving Nachamkin, University of Pennsylvania Medical Center, Department of Pathology and Laboratory Medicine, 3400 Spruce Street, Philadelphia, PA 19104-4283, USA.

0890–8508/96/020075+06 $18.00/0

75

 1996 Academic Press Limited

76

A. Waegel and I. Nachamkim

infection considering that the organism is not normally found in healthy individuals and genus/species specific targets have been described. C. jejuni and C. coli produce a single polar unsheathed flagellum and the flaA gene codes for the production of the flagellin subunit protein.11,12 Analysis of flaA sequences from several strains have shown that the 5′ and 3′ regions of the gene are conserved whereas the internal region is variable.13 Using primers derived from the conserved regions of flaA, we used a PCR assay to determine whether Campylobacter jejuni could be detected directly in stool samples. We also recently developed a molecular typing system for Campylobacter jejuni based on RFLP analysis of the flaA gene.14,15 Using this system, we also studied the potential of molecular typing Campylobacter directly from stool samples.

MATERIALS AND METHODS Samples and cultures Stool samples frozen at −70°C had been collected between 1986 and 1992 and were originally submitted to the Clinical Microbiology Laboratory at the Hospital of the University of Pennsylvania for culture and parasitologic examination.

DNA extraction Stool samples (0·2 ml) were diluted in 1·0 ml of 100 m sodium phosphate buffer pH 7·0. After adding 2 ml of a lysing solution containing 8·0  urea, 0·25% sodium dodecyl sulfate, 0·25% sodium lauryl sarcinosate, 50 m disodium EDTA pH 8·0, the samples were incubated at 60°C for 30 min. The material was subjected to centrifugation at 750 g for 5 minutes and the supernatant applied to the DNA ExtractorTM column (Molecular Biosystems, San Diego, CA) and stool DNA purified according to manufacturer’s instructions.

Reverse Primer (1728–1705) 5′-CTG TAG TAA TCT TAA AAC ATT TG-3′ In addition we used an internal forward primer located within the conserved, C1 region of flaA13 to confirm the presence of a Campylobacter gene product in selected samples: Internal Forward Primer (433–456) 5′ CAA GAA TTC CAA ATC GGC GCA AGT TCA-3′ Reverse Primer (1728–1705) 5′-CTG TAG TAA TCT TAA AAC ATT TG-3′

PCR amplification Each 100 ll PCR reaction mixture contained 10 ll Taq polymerase reaction buffer (Promega), 5 ll each of the forward and reverse flaA primers, 6 ll of 25 m MgCl2 (Promega), 8 ll deoxyribonucleotide mix, 0·5 ll Taq polymerase and 0·75 lg stool DNA template. PCR was performed using a MJ Research MiniCycler (Model PTC-150, Watertown, MA) with the following cycle conditions: 94°C for 1 min, 35 cycles at 92°C for 30 s, 55°C for 1·5 min, 72°C for 2 min, then a final 72°C for 5 min and hold samples at 4°C until analysis. The presence of amplified product was detected by agarose gel electrophoresis using 0·7% agarose. Gels were stained with ethidium bromide and photographed on an UV transilluminator. HindIII digests of bacteriophage lambda DNA were included as a molecular size standard.

Flagellin gene typing Flagellin gene typing was performed essentially as previously described with slight modification.14,15 Rather than using 1X TBE buffer for electrophoresis, 1X TAE buffer was used. Restriction fragment patterns were analyzed using ProRFLPTM analysis software (DNA ProScan, Nashville, TN) and compared with patterns contained in the typing database (Campylobacter RFLP Database v1.0 Trustees of the University of Pennsylvania, 1994).15

Primers Synthetic Oligonucleotide primers were prepared using an automated DNA synthesizer (PCR Mate, Applied Biosystems, Foster City, CA) and adjusted to 20 m for PCR reactions. The primers were as follows:14 Forward Primer (1–26) 5′-GGA TTT CGT ATT AAC ACA AAT GGT GC-3′

RESULTS Detection of Campylobacter The primer set used in this study (forward 1–26; reverse 1728–1705) has been used to amplify the flaA gene from both C. jejuni and C. coli and a limited number of other Campylobacter species. The primers

Detection of Campylobacter

77

of the assay was 97·8% if the latter sample was considered to be a false positive.

Flagellin gene typing

Fig. 1. Agarose gel electrophoresis analysis of PCR products obtained from representative stool samples containing Campylobacter jejuni (#146, 197, 147) and other enteric pathogens (#466-(Giardia lamblia; 219Salmonella; 246-Salmonella; 14-7-Aeromonas; 261negative; 262-negative). Lane 1 contains molecular size markers, lane 2-chromosomal DNA from C. jejuni IN1 and lane 3-negative control (no DNA).

have amplified the flaA gene from C. jejuni (n=322), hippurate-negative C. jejuni/C. coli (n=21), C. coli (n=14), and C. lari (n=3).15 We have recently tested some other Campylobacter species and the primers also amplify C. upsaliensis (n=3) and C. sputorum (n=2) (data not shown). Sixty-six frozen stool samples were used in this retrospective study. Nineteen samples contained C. jejuni (n=18) or C. coli (n=1), 12 samples contained other enteric pathogens and 35 were culture negative. Of 18 samples containing Campylobacter jejuni, 15 (83%) were positive in the PCR as detected by the presence of a single band (c. 1·7 kb) by visualization on agarose gel electrophoresis (Fig. 1). None of the 12 samples containing other enteric pathogens were positive. Thirty-four of 35 culture negative samples were negative in the assay. One of the culture negative samples, however, consistently produced a 1·7 kb fragment suggesting that Campylobacter was present but could not be recovered by culture. Repeat testing of the sample using a primer internal to the flaA gene within the C1 conserved region13 and the same reverse primer used above was also positive. The specificity

We recently developed a molecular typing system for Campylobacter isolates based on restriction fragment length polymorphism (RFLP) analysis of flaA.14,15 We examined the RFLP pattern of the amplicons from two positive stool samples and their corresponding patterns derived from the patient’s stool isolate (Fig. 2). The RFLP type of the amplicon derived from PCR of the stool sample and the patient’s clinical isolate showed identical patterns for both samples analyzed. Isolate 309 was determined to be flaA type-80 and isolate 14-8 was flaA type-1. Compared with a random distribution of strains isolated from sporadic cases of Campylobacter infection, flaA-1 is one of the most common types observed (21·7%) and flaA80 one of the more uncommon types observed (<1%).15

DISCUSSION The results of this study show that PCR has potential as a direct procedure for detecting Campylobacter in stool samples. PCR for detecting Campylobacter directly in human stool samples was originally described by Oyofo and colleagues,10 however, no other studies have been published to date. PCR has been used in other applications to directly detect Campylobacter in chicken products7,9 and milk and dairy products.6,7 Oyofo et al. developed a PCR assay that detected a 450 bp internal region of flaA.10 Using 13 frozen culture positive stool samples, the assay had a sensitivity of 69% (9 of 13 detected). Our assay appeared to have slightly better sensitivity than the former study, even when detecting a larger gene product. We only were able to test one stool sample containing C. coli and the PCR analysis was negative with this sample. Thus, we could not determine the sensitivity of the assay for C. coli. Further testing of stools containing C. coli is necessary. The assay appeared to be quite specific with only a single sample giving a positive result in the PCR assay but was culture negative. Additional studies using an internal flaA primer suggested that this sample contained Campylobacter, however, reculturing of the frozen sample was not performed. Since the primers used in this study could react with other Campylobacter species, such as C. upsaliensis, this organism is infrequently isolated on routine cultures for Campylobacter jejuni and requires

78

A. Waegel and I. Nachamkim Table 1. Retrospective Analysis of Frozen Stool Samples for the Detection of Campylobacter by PCR. Stool sample containinga

No. tested

Campylobacter jejuni C. coli Salmonella serogroup D Salmonella serogroup B Salmonella species Aeromonas sp. Giardia lamblia No Pathogens

18 1 5 1 1 4 1 35

a b c

No. positive by PCR 15 0 0 0 0 0 0 1

Percent positiveb,c 83 0 0 0 0 0 0 2·9

Based on culture isolation. Sensitivity for C. jejuni: 83%. Specificity: 97·8%.

Fig. 2. Restriction fragment length polymorphism analysis of PCR product obtained from two clinical isolates of C. jejuni 309 and C. jejuni 14-8 and the PCR product obtained from direct amplification of the patient’s frozen stool sample. Lanes 1 and 7 contain 123 molecular size standard.

filtration methods for isolation.2 We did not use filtration for routine culture of the fresh stool sample so it is possible that other Campylobacter species were present in the sample. In order to keep the assay as practical as possible, we used ethidium bromide staining for the presence of the 1·7 kb flaA gene fragment. The minimum amount of Campylobacter DNA that could be amplified and detected by visualization on agarose gel

electrophoresis was 1×10−5 lg DNA (c. 104 Campylobacter genome equivalents). We did not use Southern blot analysis, but suspect that higher sensitivity might have been achieved using hybridization methods. Analysis of ethidium bromide staining was also found to be approximately 10 fold less sensitive than hybridization by Oyofo et al.10 Despite the fact that the PCR assay by Oyofo et al.10 was quite sensitive using seeded stool samples, the clinical sensitivity of the assay did not appear to approach the theoretical threshold. Enhanced sensitivity of the assay might also be achieved using nested PCR strategies as well. Previous studies have shown that highly purified stool DNA was necessary for good performance of the assay.16,17 One major problem is the presence of inhibitors in the stool sample that could result in false-negative test results. We did not test the culture positive/PCR negative samples for the presence of inhibitors, so this may be an additional explanation for the sensitivity of our PCR assay. We used a column chromatographic method to obtain highly purified DNA samples, but the presence of inhibitors in these samples cannot be ruled out. The commercial column method used in our study, the DNA ExtractorTM, is unfortunately no longer available from the supplier. A similar method is available using a genomic DNA isolation kit (ASAPTM, Boehringer Mannheim, Indianapolis, IN) and Qiagen Genomic-Tip DNA purification system (Qiagen Inc., Chatsworth, CA). We have processed four stool samples (two positive and two negative) using the Qiagen column method and this appears to be equivalent in terms of amount of stool sample needed and time for purifying the DNA. Amplification of C. jejuni flaA from these samples correlated with our previous PCR results. Other stool DNA extraction methods have been described including column purification,18 multiple phenol/chloroform extractions,19

Detection of Campylobacter

adsorption to glass beads or silica particles,20 and immunomagnetic separation.17 Further development of a rapid stool DNA extraction procedure that reduces hands on time is needed to make PCR detection of enteropathogens more practical for routine laboratory use. A particularly noteworthy finding from our results was the application of molecular typing of Campylobacter directly in the stool sample without first isolating it on solid media. This is the first report, to our knowledge, of molecular typing of Campylobacter directly from a stool sample. Using RFLP analysis of the flagellin gene, flaA,14,15 we could determine the flagellin gene type of Campylobacter present in two stool samples and the patterns exactly matched those obtained from typing the isolated organism. With regard to RFLP analysis, several very faint bands were noted that were apparent on either the isolate or stool sample gel (Fig. 2). Our laboratory has performed this method on over 400 strains of Campylobacter15 and such faint bands are irregularly seen and probably represent artifacts inherit in most amplification systems. RFLP analysis of the flaA gene in C. jejuni has been shown to correlate very well with conventional serotyping but can discriminate among strains within a particular serotype.14,15 Other molecular methods such as pulse field gel electrophoresis and ribotyping have been used to subtype Campylobacter.21 It is difficult to estimate whether flagellin gene typing has better discrimination power than these other methods since they have not been compared. However, in a study by Patton et al.21 on different typing methods, molecular typing methods such as restriction endonuclease analysis and ribotyping were able to distinguish outbreak from non-outbreak strains, and like flagellin gene typing, were able to distinguish strains within a particular serotype. However, as pointed out by Patton et al.21, these methods are complex and limited to specialized laboratories. Flagellin gene typing appears to be an easier method for most laboratories to perform and might be useful for initial epidemiologic investigations. More complex method might be reserved for specialized subtyping studies. The molecular typing studies are limited since only a few stool samples were analyzed, however, the results clearly show the utility of such a method. Obviously, further studies are needed to verify the utility of such an approach and, if verified, could greatly enhance epidemiological investigations on Campylobacter infections. By improving the sensitivity of PCR detection combined with flagellin gene typing, this method would be particularly well suited to field studies where cultures might be difficult to perform.

79

ACKNOWLEDGEMENTS Development of the flagellin gene typing database was supported in part by a grant from the National Research Initiative Competitive Grants Program/US Department of Agriculture, 93-37201-9192. We thank Rebecca Hong and Huong Ung for technical assistance for some of the experiments. We thank Mabel Ann Nicholson from the Centers for Disease Control for providing some of the Campylobacter species for analysis.

REFERENCES 1. Skirrow, M. B. & Blaser, M. J. (1992). Clinical and epidemiologic considerations. In: Campylobacter jejuni: Current Status and Future Trends, Nachamkin, I., Blaser, M.J. and Tompkins, L.S. eds, pp. 3–8. Washington: American Society for Microbiology. 2. Nachamkin, I. (1995). Campylobacter and Arcobacter. In: Manual of Clinical Microbiology, 6th edition, Murray, P.R., Baron, E.J., Pfaller, M.A., Tenover, F.C. & Yolken, R.H. eds, 6th edition, chapter 36. Washington: ASM Press. 3. Sazie, E. S. & Titus, A. E. (1982). Rapid diagnosis of campylobacter enteritis. Annals of Internal Medicine 96, 62–3. 4. Park, C. H., Hixon, D. L., Polhemus, A. S., Ferguson, C. B., Hall, S. L., Risheim, C. C. & Cook, C. B. (1983). A rapid diagnosis of Campylobacter enteritis by direct smear examination. American Journal of Clinical Pathology 80, 388–90. 5. Hodge, D. S., Prescott, J. F. & Shewen, P. E. (1986). Direct immunofluorescence microscopy for rapid screening of Campylobacter enteritis. Journal of Clinical Microbiology 24, 863–5. 6. Giesendorf, B. A. J., Quint, W. G. V., Henkens, M. H. C., Stegeman, H., Huf, F. A. & Niesters, H. G. M. (1992). Rapid and sensitive detection of Campylobacter spp. in chicken products by using the polymerase chain reaction. Applied Environmental Microbiology 58, 3804–8. 7. Itoh, R., Saitoh, S. & Yatsuyanagi, J. (1995). Specific detection of Campylobacter jejuni by means of polymerase chain reaction in chicken litter. Journal of Veterinary Medicine & Science 57, 125–7. 8. Wegmuller, B., Luthy, J. & Candrian, U. (1993). Direct polymerase chain reaction detection of Campylobacter jejuni and Campylobacter coli in raw milk and dairy products. Applied Environmental Microbiology 59, 2161–5. 9. Allmann, M., Hofelein, C., Koppel, E., Luthy, J., Meyer, R., Niederhauser, C., Wegmuller, B. & Candrian, U. (1995). Polymerase chain reaction (PCR) for detection of pathogenic microorganisms in bacteriological monitoring of dairy products. Research in Microbiology 146, 85–97. 10. Oyofo, B. A., Thornton, S., Burr, D. H., Trust, T. J., Pavlovskis, O. R. & Guerry, P. (1992). Specific detection of Campylobacter jejuni amd Campylobacter coli by using polymerase chain reaction. Clinical Microbiology 30, 2613–19. 11. Guerry, P., Logan, S. M., Thornton, S. & Trust, T. J. (1990). Genomic organization and expression of

80

12.

13.

14.

15.

16.

17.

A. Waegel and I. Nachamkim Campylobacter flagellin genes. Journal of Bacteriology 172, 1853–60. Nuijten, P. J. M., VanAsten, F. J., Gaastra, W. & Van Der Zeijst, B. A. M. (1990). Structural and functional analysis of two Campylobacter jejuni flagellin genes. Journal of Biology & Chemistry 265, 17798–804. Fischer, S. & Nachamkin, I. (1991). Common and variable domains of the flagellin gene, flaA, in Campylobacter jejuni. Molecular Microbiology 5, 1151–8. Nachamkin, I., Bohachick, K. & Patton, C. M. (1993). Flagellin gene typing of Campylobacter jejuni by restriction fragment length polymorphism analysis. Journal of Clinical Microbiology 31, 1531–6. Nachamkin, I., Ung, N., Patton, C. M. (1996). Analysis of HL and O Serotypes of Campylobacter Strains by the Flagellin gene typing system. Journal of Clinical Microbiology 34, 277–81. Khan, G., Kangro, H. O., Coates, P. J. & Heath, R. B. (1991). Inhibitory effects of urine on the polymerase chain reaction for cytomegalovirus DNA. Journal of Clinical Pathology 44, 360–5. Widjojoatmodjo, M. N., Fluit, A. C., Torenama, R., Verdonk, G. P. H. T. & Verhoef, J. (1992). The magnetic

18.

19.

20.

21.

immuno polymerase chain reaction assay for direct detection of Salmonellae in fecal samples. Journal of Clinical Microbiology 30, 3195–9. Coll, P., Phillips, K. & Tenover, F. C. (1989). Evaluation of a rapid method for extracting DNA from stool samples for use in hybridization assays. Journal of Clinical Microbiology 27, 2245–8. Gumerlock, P. H., Tang, Y. J. & Silva, J. (1993). PCR detection of toxigenic Clostridium difficile. In: Diagnostic Molecular Microbiology (Persing, D.H., Smith, T.F., Tenover, F.C. & White, T.J. eds). Washington: American Society for Microbiology. Brian, M. J., Frosolono, M., Murran, B. E., Miranda, A., Lopez, E. L., Gomez, H. F. & Cleary, T. G. (1992). Polymerase chain reaction for diagnosis of enterohemorrhagic Escherichia coli infection and hemolytic uremic syndrome. Journal of Clinical Microbiology 30, 1801–6. Patton, C. M., Wachsmuth, I. K., Evins, G. M., Kiehlbauch, J. A., Plikaytis, B. D., Troup, N., Tompkin, L., Lior, H. (1991). Evaluation of 10 methods to distinguish epidemic-associated Campylobacter strains. Journal of Clinical Microbiology 29, 680–8.