Validation of a polymerase chain reaction/restriction enzyme analysis method for species identification of thermophilic campylobacters isolated from domestic and wild animals


E. O. Engvall National Veterinary Institute, SE-751 89 Uppsala, Sweden (e-mail:


Aims: To compare and evaluate a polymerase chain reaction/restriction enzyme analysis (PCR/REA) method with standard phenotypic tests for the identification and differentiation of the thermophilic campylobacters Campylobacter jejuni, C. coli, C. lari and C. upsaliensis.

Methods and Results: One hundred and eighty-two presumptive thermophilic campylobacters from 12 different animal species were tested by a recently published PCR/REA and standard phenotypic tests. By PCR/REA, 95% of the isolates were clearly identified as either one of the four thermophilic Campylobacter species or as not belonging to this group of organisms at all. By standard phenotyping, 174 of the 182 isolates were initially identified as either C. jejuni, C. coli, C. lari or C. upsaliensis. Additional genotypic tests and phenotyping showed that 52 of these identifications were either incorrect or unreliable. Of the C. jejuni isolates, 19% were identified as C. coli by initial phenotyping and 27 sheep isolates phenotyped as C. coli or C. lari were, in fact, arcobacters.

Conclusions: The PCR/REA was more reliable than standard phenotyping for the identification of thermophilic campylobacters from different animals.

Significance and Impact of the Study: Routinely used phenotypic tests often resulted in unreliable identifications, requiring additional testing. The PCR/REA, however, gave unequivocal results and was considered useful for the routine identification of thermophilic campylobacters from different animals.


Campylobacter spp. are recognized as the most common bacterial agents responsible for gastroenteritis in humans in many industrialized countries (Skirrow 1994). Campylobacter jejuni and C. coli are the species most often isolated, but C. upsaliensis and C. lari have also been associated with human disease (Skirrow 1994; Goossens et al. 1995; Lindblom et al. 1995; Bourke et al. 1998). These campylobacters are also referred to as ‘thermophilic’ because of their growth optimum at 42°C. Campylobacter spp. are frequently found in the faeces of healthy animals, especially in birds (Altekruse et al. 1994; Skirrow 1994) and are widely distributed in the environment. Routes of transmission to humans are by consumption of contaminated food, water and unpasteurized milk or by direct contact with infected animals. The handling and consumption of undercooked poultry meat have, since the 1980s, been considered the most important sources of human campylobacter infection (Norkrans and Svedhem 1982; Griffiths and Park 1990; Kapperud et al. 1992). However, an increase in the number of human cases has lately raised the question whether there are other sources which could specifically account for this increase. Consequently, surveys of the prevalence of campylobacters in animals, other than poultry, are carried out in order to identify other potential and significant sources of human infections. Results from these studies show great variation in isolation frequencies and differences in which of the species predominate in certain animals (Kapperud and Rosef 1983; Cabrita et al. 1992; Hald and Madsen 1997; Atabay and Corry 1998; Stanley et al. 1998; Altekruse et al. 1999; Baker et al. 1999; Busato et al. 1999). The discrepancies are partly due to differences in sampling and isolation techniques but also reflect some of the problems with the species identification of campylobacters.

The use of traditional phenotypic tests for the differentiation and species identification of campylobacters is often hampered by the fact that these bacteria are fastidious, asaccharolytic and possess few distinguishing biochemical characteristics (Nachamkin 1995). For epidemiological studies, a correct species identification is essential, especially when further characterization, such as subtyping below species level, is to be carried out. Recently, several genotypic methods have been developed for the identification of Campylobacter species (Jackson et al. 1996; Linton et al. 1997; Van Doorn et al. 1999). Although these methods usually prove more reliable, compared with phenotypic methods, some are laborious and/or offer low discrimination between species. A method based on the amplification of a variable region of the 23S rRNA gene, followed by digestion with two restriction enzymes (polymerase chain reaction/restriction enzyme analysis (PCR/REA)), was found to be easy to perform and seemed suitable for the routine testing of thermophilic campylobacters (Fermér and Olsson Engvall 1999). The purpose of this study was to evaluate the applicability of this PCR/REA for testing a variety of isolates from domestic and wild animals and to compare the results with traditional phenotyping.


Bacterial isolates

One hundred and eighty-two isolates originating from cattle (n=36), sheep (n=60), swine (n=19), chicken (n=6), dog (n=12), cat (n=1), Canada goose (Branta canadensis;n=16), gull (Larus spp.; n=17), roe-deer (Capreolus capreolus;n=6), moose (Alces alces;n=1), wild boar (Sus scrofa;n=7) and brown hare (Lepus europaeus;n=1) were included in the study. Except for the feline and eight of the canine isolates, all isolates were obtained from culture of faecal samples in Preston selective enrichment broth with Campylobacter selective supplement (SR117; Oxoid) and growth supplement (84; Oxoid), followed by subculture on Preston Campylobacter selective agar with selective supplement (SR117; Oxoid). Broths and plates were incubated at 41·5–42·0°C for 24 and 48 h, respectively, in a microaerophilic atmosphere generated by using CampyPak Plus (BBL, Cockeysville, MD, USA). The one feline and eight canine isolates were obtained by using a filtration technique where a suspension of faecal sample was passed through a 0·80-μm cellulose acetate filter (Advantec MFS, CA, USA) and 1–3 drops of filtrate spread on non-selective fastidious anaerobe agar (LAB M, Bury, UK) with 10% defibrinated horse blood. Plates were incubated in a microaerophilic environment at 37°C for 72–120 h (Sandstedt et al. 1983). All isolates were initially identified as presumptive campylobacters by colony appearance, microscopic morphology, motility and positive oxidase and catalase reactions.

In addition to the above-mentioned 182 isolates, three urease-positive thermophilic Campylobacter isolates were tested by the PCR/REA and standard phenotyping. Two of the isolates were obtained from the Culture Collection University of Göteborg (Sweden; CCUG 20707 and 20708) and were originally isolated from sea-gulls. The third isolate, a C. jejuni originating from a human case of diarrhoea, was kindly provided by Dr Mikael Rhen (Swedish Institute for Infectious Disease Control, Stockholm, Sweden).

Species identification


Standard tests were used for species identification of the suspected Campylobacter isolates (Nachamkin 1995): catalase production, hippurate and indoxyl acetate hydrolysis, H2S production in triple sugar iron agar and susceptibility to nalidixic acid and cephalothin discs on 5% horse blood agar. In addition, isolates which were untypable by the PCR/REA were tested for growth at 15, 25 and 30°C in microaerophilic and aerobic atmospheres at 30 and 37°C (Table 1).

Table 1.   Phenotypic tests used for differentiation between campylobacters in this study* Thumbnail image of

Genotyping by polymerase chain reaction/restriction enzyme analysis.

DNA extracts from all isolates were subjected to PCR followed by REA essentially as previously described for the specific identification and differentiation of C. jejuni, C. coli, C. lari and C. upsaliensis (Fermér and Olsson Engvall 1999). For this PCR/REA, a fragment of approximately 500 bp of the 23S rRNA gene is first amplified in a PCR reaction, which is specific for C. jejuni, C. coli, C. lari and C. upsaliensis (therm-PCR). The amplified fragment is subsequently digested by AluI and Tsp509I in two separate reactions. The restriction fragments are separated by agarose gel electrophoresis, giving banding patterns specific for each one of the four Campylobacter species. For optimal results with all of the different animal isolates in this study, modifications to the original protocol were made. Briefly, the PCR mixtures contained 0·20 μmol l–1 of each primer THERM1 and THERM4, 50 mmol l–1 KCl, 10 mmol l–1 Tris-HCl, 2·0 mmol l–1 MgCl2, 0·2 mmol l–1 of each deoxynucleotide and 0·02 U AmpliTaq (Applied Biosystems, Branchburg, NJ, USA) per μl. The temperature cycling was performed in a thermal cycler (Progene Techne, Cambridge, UK) at 94°C for 1 min, 60°C for 1 min and 72°C for 1 min for 30 cycles. For digestion, 7·5 μl PCR products were treated with 1 U AluI and TspI, respectively, and electrophoresed in 2·0% agarose gels.

Additional genotyping.

Biochemically hippurate-negative C. jejuni, as defined by PCR/REA, were also analysed by a hippuricase gene PCR (hipp-PCR) (Linton et al. 1997). A positive hipp-PCR was considered sufficient for verifying the PCR/REA identification. For confirmation of Arcobacter species identification, a selection of 10 therm-PCR-negative aerotolerant isolates were tested by PCR/restriction fragment length polymorphism (RFLP) analysis of the 16S rRNA gene according to a method by Marshall et al. (1999).

Final species identification.

Concordant identification by standard phenotyping and PCR/REA was accepted as final species identification. Isolates with discrepant identifications by standard phenotyping and PCR/REA or inconclusive test results were subjected to additional tests and, based on these test results, an identification, designated as final species identification, was made.


In total, 182 presumptive Campylobacter isolates from different domestic and wild animals were examined by phenotypic and genotypic methods for species identification. One hundred and forty-four isolates which were amplified by the therm-PCR showed electrophoretic patterns of the digested PCR products corresponding with either C. jejuni, C. coli, C. lari or C. upsaliensis (Fig. 1 and Tables 2 and 3). By standard phenotypic tests, 174 of the 182 isolates were initially identified as one of the four thermophilic Campylobacter species (Table 3). The PCR/REA and standard phenotyping equally identified 122 isolates (Table 2) and these species identifications were considered final. In an attempt to conclusively identify the remaining isolates, additional phenotyping and genotyping were performed. After additional testing, 172 of all 182 isolates received a final identification and for only 10 isolates could no final species (NFS) identification be made (Table 4). For 52 of the 174 species identifications made by standard phenotyping, additional tests showed that the first identification was incorrect (n=46) or could not be considered conclusive (n=6).

Table 2.   Species identification by polymerase chain reaction/restriction enzyme analysis of 144 animal Campylobacter isolates Thumbnail image of
Table 3.   Species identification of 182 presumptive thermophilic campylobacters by genotyping and phenotyping Thumbnail image of
Figure 1.

 Polymerase chain reaction (PCR)/restriction enzyme analysis banding patterns of thermophilic campylobacters. PCR products digested with AluI (lanes 1–4) and Tsp509I (lanes 5–8), respectively. Lanes: 1 and 5, Campylobacter coli; 2 and 6, C. jejuni; 3 and 7, C. lari; 4 and 8, C. upsaliensis; M, 100 bp molecular size marker

Table 4.   Characteristics of 10 isolates with no final species identification Thumbnail image of

Campylobacter jejuni

Eighty isolates were identified by both standard phenotypic tests and PCR/REA as C. jejuni. Another 20 isolates were speciated as C. jejuni by PCR/REA but gave a negative hippurate hydrolysis reaction or a weak reaction which was interpreted as negative (Table 3). All 20 isolates were, in addition, tested by a hipp-PCR (Linton et al. 1997). Nineteen isolates were positive by this PCR and considered to be verified as C. jejuni. One isolate was negative by the hipp-PCR and was designated NFS (Table 4). The C. jejuni isolates originated from all animals except cat and brown hare. Eight isolates were resistant to nalidixic acid. By PCR/REA, the human isolate of urease-positive C. jejuni resulted in typical banding patterns when the therm-PCR product was digested by AluI and TspI (data not shown).

Campylobacter coli

Twenty-nine isolates were equally identified as C. coli by standard phenotyping and PCR/REA (Table 2). Initially, another 45 isolates were speciated as C. coli when only standard phenotypic tests were used (Table 3). Of these, 20 were the above-mentioned hippurate-negative isolates identified as C. jejuni by PCR/REA and 25 were identified as arcobacters by additional tests (see below). Two isolates, identified as C. coli only by PCR/REA, could not be identified by phenotyping, due to conflicting test results and were designated NFS (Table 4). The C. coli isolates came from cattle, sheep, swine, wild boar and hare. Twelve were resistant to nalidixic acid.

Campylobacter lari

Five isolates were identified as C. lari by both standard phenotyping and PCR/REA. All were isolated from gulls (Table 2). Digestion of the PCR product with TspI gave an atypical banding pattern in one case. The same pattern has been observed previously for other gull isolates of C. lari (unpublished observations). Another six isolates were identified by standard phenotypic tests as C. lari but were negative by the therm-PCR. Additional testing showed that two were aerotolerant and identified as arcobacters (see below). The four remaining isolates could not be conclusively species identified and were therefore designated NFS (Table 4). The two urease-positive CCUG isolates, designated C. lari and Campylobacter UPTC group, respectively, were both identified as C. lari by standard phenotyping. They were amplified by the therm-PCR but digestion of the PCR products gave identical and atypical banding patterns for both isolates. The atypical TspI pattern was the same as mentioned above. The AluI pattern also differed from the expected C. lari pattern (data not shown).

Campylobacter upsaliensis

Eight isolates were identified as C. upsaliensis by standard phenotyping and PCR/REA (Table 2). One more isolate was identified as C. upsaliensis by standard phenotyping (Table 3), but was negative by the therm-PCR. Since the isolate grew very poorly the phenotypic test results could not be considered reliable and the isolate was finally designated NFS (Table 4). All nine isolates came from dogs. Two of the C. upsaliensis isolates were resistant to nalidixic acid.

Non-thermophilic Campylobacter species

Two isolates from cattle, negative by the therm-PCR, grew under microaerophilic conditions at 30 and 25°C and were identified as C. fetus. Another two isolates, one from sheep and one from roe-deer, were negative by the therm-PCR and untypable by standard phenotyping. They were identified as C. hyointestinalis by additional testing (Table 3).

Arcobacter species

Twenty-seven isolates, all from sheep, were identified by standard phenotyping as C. coli (n=25) and C. lari (n=2) but were negative by the therm-PCR. The isolates could be grown in a microaerophilic environment at 15 and 25°C and also aerobically. These aerotolerant isolates were presumed to be Arcobacter species. Ten isolates were verified as arcobacters by PCR/RFLP analysis (Fig. 2) (Marshall et al. 1999) and by tests performed at the Danish Veterinary Laboratory (Copenhagen, Denmark).

Figure 2.

 Polymerase chain reaction (PCR)/restriction fragment length polymorphism banding patterns of arcobacters and campylobacters. PCR products digested with DdeI. Lanes: 1–10, Arcobacter sp.; 11, Campylobacter jejuni; 12, C. coli; 13, C. upsaliensis; M, 100 bp molecular size marker

Isolates with no or no final species identification

By standard phenotyping, eight isolates received no species (NS) identification because of negative, intermediate or conflicting test results. Two of the isolates were identified by additional tests as C. fetus, two as C. hyointestinalis (see above) and four received no final identification (Table 4).

For 38 of the 182 isolates, no species identification could be made by PCR/REA (Table 3) due to either a negative therm-PCR (n=37) or illegible banding pattern in the REA (n=1). The negative therm-PCR isolates consisted of the above-mentioned 27 arcobacters, two C. fetus, two C. hyointestinalis and six other isolates (Table 4). One wild boar isolate was amplified by the therm-PCR but digestion of the PCR product gave illegible banding patterns (NT). No conclusive identification could be made by phenotyping. The isolate ceased to grow upon subcultivation and could, therefore, not be investigated any further (NFS; Table 4).

Ten isolates could not be identified to species by the combination of genotypic and phenotypic tests used in this study (NFS; Table 4). For eight isolates, there was a contradiction between phenotypic and genotypic identifications and, for two isolates, no species identification could be made by either of the methods.


The species identification of Campylobacter isolates by traditional phenotypic methods is problematic due to many conflicting and intermediate test results. For the identification of the thermophilic campylobacters of public health significance, many clinical laboratories choose to identify only to genus level or simply as C. jejuni/coli (Nachamkin 1995; On 1996). In epidemiological surveys, where samples from different sources are analysed, it is particularly important that the campylobacters are correctly identified, as conclusions are drawn about the prevalence of different Campylobacter species in, for instance, different animals. It should be noted that the animal campylobacters tested in the present study were not chosen in a way that allows any conclusions about prevalence. In general, the isolates tested were of Campylobacter species often found in the animal species (Cabrita et al. 1992; Skirrow 1994; Atabay and Corry 1998; Baker et al. 1999).

The standard phenotypic tests used for species identification in this study were clearly inadequate, since 52 (30%) of the identifications were either incorrect or unreliable. Adding more tests increased the possibility of a reliable species identification, but this approach would not be feasible for a routine laboratory. The hippurate hydrolysis test, which is the phenotypic test most commonly used in routine diagnostics, was unable to identify many C. jejuni in this study. Nineteen (19%) of the isolates, genotyped by PCR/REA and verified by hipp-PCR as C. jejuni, were, in fact, either negative or gave weak or indeterminate hippurate test reactions, which could not be securely interpreted. However, a clearly positive hippurate reaction always correlated to an isolate genotyped as C. jejuni. In agreement with the findings reported by Steinbrueckner et al. (1999), hippurate-positive C. jejuni were correctly identified in the present study by phenotyping, whereas hippurate-negative isolates needed additional genotyping in order to obtain definitive species identification.

Antibiotic susceptibility tests were found to be very unreliable, since several of the isolates of C. jejuni, C. coli and C. upsaliensis, which are supposed to be susceptible to nalidixic acid, were resistant. This test was, therefore, considered to be not reliable and never decisive. If the indoxyl test had not been included in the battery of phenotypic tests, many nalidixic acid-resistant C. coli would have been incorrectly identified as C. lari.

In this study, 27 sheep isolates initially identified as thermophilic campylobacters were, in fact, aerotolerant organisms. So-called aerotolerant campylobacters, with uncertain pathogenicity, are sometimes isolated from clinical samples from animals (Skirrow 1994). This group of organisms has been recognized as belonging to a new genus, Arcobacter (Vandamme and Goossens 1992). Although arcobacters are phenotypically similar to campylobacters, they differ significantly by two features according to the classification schemes of Vandamme and Goossens (1992): aerotolerance and the ability to grow at low temperatures (15–25°C). However, in a recent study by Chynoweth et al. (1998), isolates of C. jejuni were shown to be capable of adapting to growth under aerobic conditions. This information makes it difficult to evaluate aerotolerance as a distinguishing feature between campylobacters and arcobacters. Genotyping of the aerotolerant sheep isolates in the present study was, therefore, very helpful. The possibility of the isolates being aerotolerant C. coli or C. lari was ruled out by the negative therm-PCR. Thus, the PCR/REA proved valuable for avoiding the false interpretation of biochemical test reactions given by these organisms.

With three C. lari isolates, inconclusive results were obtained with the PCR/REA, due to atypical electrophoretic banding patterns of the digested PCR products. One of the isolates was a urease-negative routine sample from a gull, whereas the other two were urease-positive CCUG type strains. The C. lari species comprises a more heterogeneous group of isolates compared with the other three thermophilic campylobacters, a fact that has been confirmed by others (Van Doorn et al. 1998). The problem with typing the three C. lari strains in this study with the PCR/REA may well be a reflection of this heterogeneity. Further analysis of C. lari isolates will confirm whether the new banding patterns are common within the species and whether the method can be used for differentiation of isolates below the species level in addition to species determination.

Phenotypic methods, other than the traditional biochemical tests, along with a variety of genotypic methods have been reported for the speciation of campylobacters (On 1996). However, the usefulness for routine testing may vary. A good method should be easy to perform, give unequivocal test results and be applicable for all types of isolates. In the present study, the PCR/REA was found to be both robust and practical to use with isolates of varying origin. For the majority of isolates (67%), the PCR/REA and standard phenotypic tests resulted in concordant species identifications. Furthermore, for the majority of isolates which received no or discrepant species identifications with PCR/REA and standard phenotyping, additional tests could explain the discrepancies and prove the accuracy of the PCR/REA. It was concluded that the PCR/REA method was useful, either as a complement or alone, for the routine identification of thermophilic campylobacters from different animals.


The authors thank Dr Stephen On at the Danish Veterinary Laboratory for help with the identification of the Arcobacter isolates. Financial support by the Swedish Board of Agriculture is acknowledged.