Campylobacter spp. are a common cause of acute bacterial gastroenteritis in humans 12. The majority of campylobacterioses are caused by C. jejuni and are linked primarily to untreated water or consumption of poultry and raw milk 2. C. jejuni has been divided into two subspecies: C. jejuni subsp. jejuni (Cjj) and C. jejuni subsp. doylei (Cjd). Cjd strains were isolated originally as "gastric campylobacter-like organisms type 2 (GCLO2)" from human gastric biopsies 4 and "nitrate-negative campylobacter-like organisms (NNC)" from Australian pediatric patients with gastroenteritis 5. As the "NNC" designation suggests, the characteristic feature of Cjd, used commonly to distinguish Cjd strains from Cjj strains, is the inability to reduce nitrate. Other phenotypic characteristics, such as variable growth at 42°C 6, high susceptibility to cephalothin 6, and the absence of γ-glutamyl transferase (GGT) and L-arginine arylamidase activity 7 have been associated also with Cjd; however, like Cjj, all Cjd strains are hippuricase positive.

Cjd strains also differ from Cjj in two other clinically-related aspects: first, Cjd strains can be found throughout the gastrointestinal tract, leading to both gastritis 48 and enteritis 691011, and are obtained often from pediatric patients 691112. Second, in South Africa, unlike Cjj strains, Cjd strains are isolated more often from blood cultures than from stool cultures 3; Cjd was isolated from 24% of the Campylobacter-positive blood cultures 3, in contrast to 7.7% of the Campylobacter-positive stool cultures, obtained at Red Cross Children's Hospital, Cape Town, during the years 1990–2005. Additionally, Morey reported that Cjd was isolated from 85.2% of Campylobacter/Helicobacter-related bacteremia cases in Australia during a five-year period 12.

In some parts of the world, notably South Africa, Cjd strains represent a significant proportion of the total campylobacters isolated from human clinical samples; 16% of the non-Cjj/coli Campylobacter strains isolated in Cape Town, South Africa were Cjd. However, despite the unusual clinical symptomatology and relatively high prevalence in certain parts of the world, Cjd is generally isolated infrequently and few strains exist (compared to Cjj) for this subspecies. One possible reason is that many clinical laboratories do not characterize Campylobacter isolates past the genus level, much less subspeciate C. jejuni isolates as Cjj or Cjd. It is also likely that both the susceptibility of Cjd to cephalothin and variable growth at 42°C prevents the isolation of a substantial number of Cjd strains under normal Cjj isolation conditions; the Cjd strains isolated in South Africa were obtained using the Cape Town Protocol 3 which uses passive filtration through a 0.65 μM membrane filter, growth at 37°C and no antibiotic selection.

Another factor in the lack of subspeciation characterization of C. jejuni strains is the absence of a rapid test to distinguish Cjj from Cjd. Phenotypic characterization of Cjd, based on the absence of nitrate reductase activity, remains the primary means of identification; however, such nitrate reductase assays require a large number of cells and can take 24–48 h. Additionally, nitrate reductase assays require pure cultures; mixtures of Cjj and Cjd cells would type as nitrate⁺/Cjj.

Many commonly used molecular-based Campylobacter detection methods cannot be used to subspeciate C. jejuni due to the high similarity between the two subspecies. Molecular-based methods that can subspeciate C. jejuni do exist, based on hybridization 7, amplified fragment length polymorphism fingerprinting 1314, or characterization of the 16S/23S rDNA internal spacer region 15; however, these methods are lengthy or require specialized and costly equipment. In this study we present a simple, novel multiplex PCR method that can be used to unambiguously subspeciate C. jejuni.

The apparent absence of the genes encoding nitrate reductase (napA and napB) in Cjd strain RM2095 correlated well with the nitrate reduction-negative phenotype characteristic of this subspecies. Thus, we reasoned that a potential subspeciation marker for C. jejuni would be the nap locus, specifically napA and napB, and we developed therefore a napA/B multiplex PCR method to subspeciate C. jejuni. Amplification results indicate that this nap multiplex PCR method, using internal and flanking primer sets for napA and napB, can be used to subspeciate unambiguously C. jejuni: Cjd strains (D1-4: Fig. 1A) can be distinguished readily from Cjj strains (J1-2: Fig. 1A). Moreover, different banding patterns were observed with Cjj and Cjd using nap mpx1 or nap mpx2, suggesting that either multiplex primer set is sufficient to subspeciate C. jejuni. The almost total concordance between the subspecies identification of 188 C. jejuni strains and the multiplex PCR results indicates that the new multiplex PCR is robust and can be used successfully to subspeciate C. jejuni. One strain, the Penner HS:63 reference strain, classified originally as Cjj, was identified as Cjd by our assay, suggesting a potential flaw in the PCR method. However, identification of a strain as Cjd by our method requires a deletion in napA and/or napB. Since a deletion in either of these two genes would lead necessarily to a loss of nitrate reductase activity, C. jejuni strains containing such defects would be, by definition Cjd, regardless of the original classification. Therefore, the Penner HS:63 reference strain was either not subspeciated or was subspeciated incorrectly. Finally, combination of the napA/B multiplex method with an lpxA multiplex method designed to speciate thermophilic Campylobacters permits the speciation and subspeciation of thermophilic Campylobacters with a single amplification reaction. Combination of the two methods is especially important when the species identification of a Campylobacter strain is unknown, as C. coli and C. upsaliensis strains may be confused potentially with Cjj (although not Cjd) strains if the nap multiplex method is used solely.

Currently, Cjd strains are subspeciated using the nitrate reductase assay, an assay used initially also in this study to identify Cjd strains. Although the nitrate reductase assay is relatively easy to perform, the multiplex PCR methods described here have several advantages over the nitrate reductase assay. First, subspeciation of C. jejuni strains, as well as the speciation/subspeciation of thermophilic Campylobacters, can be accomplished with this assay in a matter of hours, compared to the 2–3 days necessary for the nitrate reductase assay, and in a single amplification reaction. Second, the nitrate reductase assay requires a pure culture. Mixtures of Campylobacter strains of the same species or strains representing different species occur often during isolation 31819 and would be problematic with regards to phenotypic assays; mixtures of Cjj and Cjd cells would type as Cjj. In contrast, the multiplex PCR method does not require a pure culture and would identify the species/subspecies comprising such a mixture. In fact, the multiplex speciation method identified successfully each species in a mixture of four thermophilic Campylobacter genomic DNAs (data not shown). Third, the nitrate reductase assay requires often more than a single colony for accurate identification. However, the multiplex PCR method requires at most only a single colony. Indeed, 19 of the Cjd genomic DNAs used in this study were obtained by boiling a single storage bead in 10 mM Tris pH 8.0, indicating that strains in storage at -80°C do not even have to be grown out to be characterized. Finally, the nitrate reductase assay does lead occasionally to false identification. One of the strains in this study, identified initially as Cjd, was determined by the multiplex PCR method to be Cjj. A repeat of the nitrate reductase assay on this strain indicated that it was, in fact, Cjj and that the original nitrate reductase identification was in error. Taken together, these results indicate that the napA/B multiplex PCR method described here is a valuable tool for clinical and research laboratories and can be used successfully to subspeciate rapidly C. jejuni strains and speciate thermophilic Campylobacters when combined with the lpxA multiplex PCR method.

An unexpected outcome of this study arose from the sequencing of the Cjd napA and napB amplicons. Some Cjd strains contained deletions in napB (Cjd2) and some contained apparently full-length napB genes (Cjd1), although point mutations in many of the Cjd1 napB genes would result in non-functional proteins; however, all 27 characterized Cjd strains contained identical napA deletions of 2761 bp. Based on the presence of these identical napA deletions, it appears that the Cjd strains, isolated on four continents over at least two decades, share a common ancestor. Furthermore, as the defining phenotype of Cjd is the absence of nitrate reduction, the evidence suggests strongly that Cjd arose from a single evolutionary event, i.e. the napA deletion, with the divergence at napB occurring somewhat later; deletion of the gene encoding the large subunit of nitrate reductase would remove the constraint maintaining a functional small subunit-encoding gene. Obviously, further experiments will be necessary to validate this hypothesis. Such experiments would entail the isolation and characterization of additional Cjd isolates, facilitated by the multiplex PCRs described above. Although the results presented here will need to be investigated further, they nevertheless provide intriguing insights into the origin of this subspecies.

Campylobacter jejuni subsp. doylei strains are isolated infrequently. They are isolated primarily from human clinical patients and often from blood cultures, suggesting that they may be more pathogenic than Cjj strains. However, little is known about the doylei subspecies, especially with regards to the environmental reservoirs for this organism; no animal host has been yet identified for Cjd 20. Cjj strains are isolated often from avian hosts. The absence of Cjd from avian sources, such as poultry, may be due to the lower maximum growth temperature (37°C) for this subspecies. This lower growth temperature may restrict Cjd strains to other hosts with lower normal body temperatures, such as swine and domestic pets. It may influence also the isolation of Cjd at the clinical level since enrichment and selection of C. jejuni as a whole is normally performed at 42°C, precluding the isolation of most Cjd strains. Also, antibiotics used in typical C. jejuni isolation media may restrict the growth of some Cjd strains.

The multiplex PCR method presented in this study will enhance greatly the identification of Cjd strains from human clinical, veterinary and environmental samples. Since this PCR does not require large numbers of cells or pure cultures, Cjd strains can be identified even when they comprise only a small subset of the overall population. Removing the need for enrichment or selection would reduce the isolation bias that might minimize or eliminate detection of Cjd strains. Finally, we have demonstrated here that PCR identification of Cjd strains can be performed using only the cells coating bacterial storage beads from -80°C freezer stocks. Thus, strains identified only as C. jejuni can be re-assessed rapidly without the necessity of growing the strains out to purity. This was instrumental in identifying, for example, an HS:63 Penner serotype reference strain, identified initially as Cjj, as a Cjd strain; several other strains identified only as C. jejuni may be, in fact, C. jejuni subsp. doylei.

WGM and AJL designed the research project. WGM designed the nap mpx1 and mpx2 multiplex primer sets and was the principal author of the manuscript. CTP constructed the C. jejuni DNA microarray and performed the microarray experiments and analysis. SH performed all of the multiplex amplifications and sequenced all of the napA and napB amplicons. AJL collected the 21 South African Cjd strains and performed the initial nitrate reductase assays on these isolates. All authors approved and read the final manuscript.