The genus Arcobacter, included in the family Campylobacteraceae, has expanded rapidly since it was first recognised in 1991 [1], and currently includes 17 species. Some of these species are considered enteropathogenic to humans and animals, as well as important zoonotic agents. Arcobacter species negatively impact the food industry, as many meat products are frequently contaminated with these bacteria, and multiple species have been described from shellfish [2–6]. In addition, the International Commission on Microbiological Specification for Foods classified A. butzleri as a serious hazard to human health [7]. However, the true incidence of Arcobacter species in environmental and clinical samples is thought to be underestimated because specific detection and identification methods are not normally applied and can be inaccurate [2, 8].

A 16S rRNA restriction fragment length polymorphism (RFLP) method for the identification of Arcobacter species has previously been described [9]. The method involved a single digestion with the MseI endonuclease and discriminated all Arcobacter species that had been described up to 2008, i.e. A. butzleri, A. cryaerophilus, A. cibarius, A. skirrowii, A. nitrofigilis and A. halophilus[9]. Further molecular methods for the identification of Arcobacter species have been reviewed elsewhere [2, 9]. Most of the methods described target only the most common species i.e. A. butzleri[10, 11], A. cryaerophilus[12] and/or A. skirrowii[13, 14]. Even the most recently proposed identification method, m-PCR, described by Douidah et al. [15] in 2010, only targeted five species: A. butzleri, A. cryaerophilus, A. skirrowii, A. cibarius and A. thereius. Furthermore, using this method, the species A. defluvii, A. ellisii, A. venerupis and A. butzleri produced an identical and therefore uninformative amplicon [2, 5, 6].

The limitations of the current methods have arisen because of the limited testing of certain species, as well as the identification of novel species [2, 4–6]. Douidah et al.[15] suggested that the reliance of the currently-available 16S rRNA-RFLP method on polyacrylamide gel electrophoresis was a major disadvantage for its routine use. Furthermore, the recently described species A. thereius, isolated from aborted pig foetuses [16], and A. trophiarum, which was recovered from porcine faecal matter [17], produce the same RFLP pattern as A. butzleri[2]. Additionally, the new species A. venerupis, from clams, produces a pattern that is very similar to A. marinus[6, 18].

The aim of the present study was to update the 16S rRNA-RFLP identification method to include all the currently characterised species of Arcobacter, and to provide protocols for both polyacrylamide and agarose gel electrophoresis so that the method can easily be adapted.

The proposed 16S rRNA-RFLP method described here used an initial digestion with MseI endonuclease, as in the original method [9], which enabled 10 of the 17 accepted species, including the recently described species A. cloacae, to be identified. Further digestion was necessary to resolve species that showed the MseI digestion pattern of A. butzleri (also common to A. thereius, A. trophiarum and to the atypical strains of A. cryaerophilus with 16S rRNA gene microheterogeneities). Computer simulation revealed that two endonucleases, MnlI and TasI, produced discriminative patterns between the species A. butzleri and A. thereius (Figure 2 and Additional file 5: Figure S2). Furthermore, these two enzymes also produced discriminative patterns between A. marinus and A. venerupis (Figure 2), which showed distinctive but very similar patterns following MseI digestion (Figure 4 and Additional file 1: Table S1). MnlI was selected because it generated more distinctive banding patterns, enabling easier discrimination than TasI (Additional file 5: Figure S2). Considering that A. butzleri is a very common species [2, 8, 19–21], the identification of the majority of strains will normally be obtained with this second (MnlI) endonuclease reaction (Figures 1, 2, 4). In fact, 79.3% of the strains (96/121) included in the current study were correctly identified with this second digestion step.

However, a third digestion, using the enzyme BfaI, was required to distinguish between A. defluvii and the recently described species A. suis and for distinguishing A. trophiarum from the atypical A. cryaerophilus strains following MnlI digestion (Figures 3,4 and Additional file 3: Table S3). The proposed method enables reliable and fast species identification for a large collection of isolates, requiring, at most, digestion of the PCR-amplified 16S rRNA gene (1026 bp) with three restriction endonucleases (MseI, MnlI and/or BfaI).

The original 16S rRNA-RFLP method [9] has been used to identify more than 800 Arcobacter strains recovered from meat products, shellfish and water in various studies [3–6, 19–22]. The existing method has also helped to discover new species on the basis of novel RFLP patterns, including A. mytili[3], A. molluscorum[4], A. ellisii[5], A. bivalviorum, A. venerupis[6] and A. cloacae[23]. Furthermore, as well as identifying the more common Arcobacter species, this technique has confirmed the presence of other rare species in atypical habitats, such A. nitrofigilis in mussels and A. thereius in pork meat [20]. The updated technique described here is likely to supersede the current method in all of these areas.

The use of the 16S rRNA-RFLP method in parallel with the more commonly used molecular identification method, m-PCR [13], as well as the fact that strains with incongruent results were sequenced (rpoB and/or 16S rRNA gene sequencing), ensured accurate species identification, and highlighted the limitations of both identification methods [2, 4–6, 23]. The presence of microheterogeneities in the 16S rRNA gene, as in the case of the 11 atypical A. cryaerophilus strains, had not previously been observed. These strains produced the m-PCR amplicon expected for A. cryaerophilus, which targets the 23S rRNA gene [13], but showed the A. butzleri 16S rRNA-RFLP pattern [9]. However, rpoB and 16S rRNA gene sequencing results confirmed these strains as A. cryaerophilus. 16S rRNA-RFLP patterns that differ from those described here can be expected for any newly discovered Arcobacter species [3–6, 9, 23]. Nevertheless, intra-species nucleotide diversity (i.e. mutations or microheterogeneities in the operon copies of the 16S rRNA gene) at the endonuclease cleavage sites can also generate a novel RFLP pattern for a given isolate, or result in a pattern identical to another species [9, 24, 25]. In the latter situation, misidentifications may occur, as described here.

In conclusion, the 16S rRNA-RFLP protocols described here for the identification of Arcobacter spp. can be carried out using either agarose or polyacrylamide gel electrophoresis (Figures 1–3, Additional file 1: Table S1, Additional file 2: Table S2, Additional file 3: Table S3), depending on the requirements of an individual laboratory. It is important, however, to carry out the 16S rRNA gene digestions in the order illustrated in the flow chart (Figure 4).

The method provided in this study is reproducible, reliable, simple, fast, and reasonably inexpensive, and can be carried out efficiently in any laboratory. The technique is highly applicable for investigations of the prevalence of arcobacters in a variety of food products, water, wastewater or other environmental samples. It will enable investigators to determine the true incidence of the recently described species A. mytili, A. marinus, A. trophiarum, A. molluscorum, A. defluvii, A. ellisii, A. bivalviorum, A. venerupis, A. cloacae and A. suis clarifying their prevalence and epidemiology.