On the identification of clinical Aeromonas by a new restriction fragment length polymorphism of 16S rDNA method

Sir—Ghatak et al. (2007) published an identification method for Aeromonas species, based on a 16S rDNA‐RFLP technique, claiming that this method offered more easily recognizable patterns than a previous published one (Borrell et al. 1997; Figueras et al. 2000a). In fact the only advantage recorded by Ghatak et al. on their method is that it shows simple patterns comprising only from two to five bands. However, the method requires four independent sequential digestions (BstSNI, MboI, PvuII and NruI) for the differentiation of only a very limited number of Aeromonas species (A. hydrophila, A. caviae, A. veronii biovars sobria and veronii and A. jandaei), claimed by the authors to be the only ones with some clinical relevance. Conversely, using the previously described method (Borrell et al. 1997) all those species can be differentiated with only a single digestion of the 16S rRNA gene with the enzymes AluI and MboI, which produce species‐specific and clear patterns. In addition, this digestion enables to discriminate also A. media, A. sobria, A. trota, A. schubertii, A. eucrenophila and A. allosaccharophila, all of them, species isolated with different frequencies from clinical specimens too (Janda and Abbott 1998; Figueras et al. 2000b; Figueras 2005; Saavedra et al. 2007). Apart from the limited species discrimination, another shortcoming of the Ghatak’s et al. method is that the patterns they describe for those four species are identical to those obtained for other Aeromonas spp. This was observed when we performed a simulation on the type strains of all accepted species, using the same procedure of Ghatak’s et al. For instance, the pattern of A. hydrophila is shared with two other species, A. encheleia and A. molluscorum; the pattern of A. caviae is shared with A. trota and A. simiae; the pattern of A. jandaei is shared with other four species (A. bestiarum, A. salmonicida, A. popoffii and A. bivalvium) and the one of A. veronii is identical to those of A. sobria and A. allosaccharophila. Several of the species which patterns can be mistaken had previously been involved in clinical cases (Janda and Abbott 1998; Hua et al. 2004; Figueras et al. 2000b; Figueras 2005; Saavedra et al. 2007). In our view it is very important that this less frequently isolated clinical species can be properly characterized in order to know their epidemiology and be able to establish their true pathogenic role. Furthermore, Ghatak et al., made a mistake concerning the separation of A. veronii bv. sobria and A. veronii bv. veronii using the endonuclease NruI. The simulated digestion with this enzyme on the 16S rRNA gene of the reference strain of A. veronii bv. sobria and of the type strain of A. veronii bv. veronii (GenBank accession numbers AF410949 and X60414 respectively) showed that this enzyme did not cut any of those genes in disagreement with Ghatak et al. (2007). However, it cuts the 16S rRNA gene of the type strain of A. sobria, a species unrelated to A. veronii bv. sobria. The term Aeromonas sobria, had classically been a misnomer used by clinicians to refer to A. veronii bv. sobria (Figueras et al. 2000b; Figueras 2005), but in fact the former species belongs to hybridization group 7, and had rarely been isolated from clinical cases (Janda and Abbott 1998; Figueras 2005). Probably the authors had wrongly chosen the GenBank sequence of A. sobria (X60412) to simulate the digestion they commented at the end of their paper. The authors claimed that the identification method of Borrell et al. (1997) laid an emphasis on the taxonomic aspect of the species identification, without any attention to clinical relevant aeromonads. We have to totally disagree with their opinion, because that study demonstrated the prevalence on clinical samples of A. veronii (46%), A. caviae (29%) and A. hydrophila (16%) and that accounted for ca 90% of the clinical strains, which is in agreement with other authors (Janda and Abbott 1998). Furthermore, since the aim of our study was to provide an identification tool for clinical microbiologists, the Borrell et al. (1997) RFLP method was published in a clinical journal.