The isolates used in this study were divided into three collections. All were collected from the Rance estuary (Brittany, France) and stored at -80 °C in brain heart infusion with 10% glycerol added.

Collection A: Collection A comprised 93 iso­lates from three samples (two freshwater speci­mens and one aquatic sediment specimen) col­lected in August 2000. These specimens were incorporated (10% w/w) in saline alkaline pep­tone water (0.3% yeast extract, 1% peptone, 2% NaCl, and pH 8.6) (SAPW) and incubated at 41 °C±1 °C for 16-18 hours (3,4,5). Volumes of 0.1 mL of 10-fold dilutions of the enrichments were spread onto TCBS agar (Difco). All the yellow colonies present on TCBS agar were kept for the screening procedure, and all the presumptive iso­lates were biochemically identified with the API 20E system.

Collection B: Collection B was composed of 774 isolates from 115 samples (56 water samples, 46 sediment samples, and 13 cockle samples) col­lected during June-November 2000. These speci-mens were incorporated (10% w/w) in saline alkaline peptone water (0.3% yeast extract, 1% peptone, 2% NaCl, and pH 8.6) (SAPW) and in­cubated at 41 °C±1 °C for 16-18 hours (3,4,5). Only those isolates which were recognized by the screening procedure as presumptive V. cholerae were kept for definite identification.

Collection C: Collection C comprised isolates col­lected from a water sample in September 2003: 10 volumes of 0.1 mL of the water sample were di­rectly (without any enrichment step) plated onto TCBS (DIFCO). One hundred presumptive isolates of V. cholerae, according to the screening proce­dure, were randomly selected and kept for identi­fication by the three methods tested in this study.

Identification

Biochemical identification

For each isolate, one colony from an overnight culture on nutrient agar with 2% NaCl (NA₂)was suspended in sterile saline water (0.85% NaCl) for inoculation of an API 20E strip (Biomerieux Industries, France) according to the instructions of the manufacturer. This was then incubated at 37 °C for 24 hours. The series of 21 miniatur­ized biochemical tests was interpreted with the API 20E analytical profile index (version 5). The quality of the identification was based on two statistical values (% of identification [id] and t value). In this study, the identification results were classified as: (a) excellent identification when % of id was ≥99.9 and t value ≥0.75; (b) good iden­tification when 90.0 ≤% of id <99.9%, and t was ≥0.25; and (c) low discrimination, when V. cholerae was not the only identification proposed.

Molecular identification

DNA extraction: DNA extraction was done on overnight subculture on NA₂ using the chloro­form-phenol procedure (8). DNA extracts were treated with RNAse (1%) and dialyzed.

PCR assays: PCR amplification of the target DNA was carried out in a thermal cycler (Hybaid-PCR express) using 200-µL PCR tubes with a re­action mixture volume of 25 µL containing 3 µL of template DNA, 0.2 µL of each primer (100 µM), 2.5 µL of 2 mM dNTP, 0.125 µL (5 U/µL) of Taq polymerase (Eurobio), 2.5 µL of 10X reaction buffer, 1.25 µL of MgCl2 50 M (Eurobio), and ul­tra-pure water. The sequences of the primers and the conditions of amplification used in this study are shown in Table 1.

PCR products were separated by agarose gel electro­phoresis, followed by ethidium bromide staining. For each PCR reaction, a reagent blank was run, where the DNA template was replaced by ultra-pure water. DNA of a strain of V. cholerae O1, Classical, Inaba (Institut Pasteur—CNRVC 940147) was run as a positive control. When an isolate gave a nega­tive result, the PCR reaction was repeated once.

As the authors had already tested specificity, only the absence of cross-reaction of the two sets of primers with V. mimicus was checked on 14 envi­ronmental strains of V. mimicus (provided by the LERES collection). The quality of DNA extraction for each strain was verified by amplification of 16S rDNA with the 27F and 1492R primers (9).

Screening procedure

The screening procedure was based on pheno­typic traits. The sucrose-fermenting colonies on TCBS were transferred onto nutrient agar with­out the addition of NaCl (NA₀) for growth testing and were then submitted to an oxidase test (Bac­tident strip). All sucrose-fermenting isolates, able to grow on NA₀ and oxidase-positive, were con­sidered to be presumptive isolates of V. cholerae and were then tested for culture purity on NA₂.

Experimental plan

Routine screening of isolates on any selective agar relies on the capacity of selection of the target bac­teria and on the choice of a few relevant taxonom­ic traits for presumptive identification. TCBS agar is commonly recommended for the isolation of Vibrio spp. (11,12). It eliminates non-bile salt tolerant species and remains the best agar for its ability to isolate vibrios from their natural estua­rine environment (13).

In this study, the biochemical traits were chosen among 13 readily-determinable ones proposed by Baumann et al. (15). Three ordered traits (suc-rose fermentation, non-requirement of added Na+ for growth, and presence of oxidase) are sufficient to distinguish V. cholerae from the other spe­cies of Vibrio. Sucrose non-fermenting spe­cies are not taken into consideration, and, thus, the related V. mimicus species is eliminated at the beginning. Growth on nutrient agar without added NaCl eliminated most sucrose-positive halotolerant or halophilic vibrios that may be important in differential diagnoses (V. algino­lyticus, V. metschnikovii, and V. fluvialis and to a lesser extent V. furnissii, V. cincinnatiensis, V. an­guillarum,and V. carchariae). Moreover, a oxidase-positive reaction eliminated V. metschnikovii.

While these three taxonomic traits are insufficient to definitively identify the isolates as V. cholerae, they do represent minimal traits which all mem­bers of the species must have. Thus, any isolates which do not meet these minimal criteria are deemed not to be V. cholerae.

Our results, which are based on a very large num­ber of environmental isolates (collection A and collection B) emphasize the fact that the efficiency of the proposed screening procedure is reliable for environmental monitoring. Even if one no­tices that some rare variants (sucrose-negative or inability to grow whithout added NaCl) of V. cholerae may be lost by this screening scheme, the option remains relevant for routine screening.

Choopun et al. proposed another selection of traits for the rapid presumptive identification of V. chole-rae from aquatic environments (arginine dihydro­lase activity-negative and esculin hydrolysis-nega­tive—two expensive and relatively difficult tests) (16). Although esculin hydrolysis has been pro­posed by Baumann et al. as a key taxonomic trait (15), it is not easy to interpret the test without a fluorometer, which is absent in most laboratories.

Choice of a confirmative identification method

The API 20E system is indeed considered an ac­ceptable method for the identification of the more commonly-occurring members of the fami-ly Vibrionaceae (17,18), even if there are very few reports expressly concerned with the ability of commercial systems to identify members of the genus Vibrio (19).

Besides conventional biochemical identification, alternative molecular methods, such as PCR or hybridation of colonies by labelled probes, are available for the confirmation of presumptive identifications.

Colony-blot probing on selective agar, such as alkaline nutrient agar without added NaCl, has been proposed for the isolation of V. cholerae (20). This alternative process does not need presump­tive biochemical identification, but cannot use TCBS agar selectivity. Moreover, its application is limited to the case of brackish water monitor­ing. Another common alternative method is PCR, which is now used by reference laboratories as the most reliable means of routine identification.

Only two of four pairs of primers which were available for the identification of V. cholerae spe­cies were compared. The one developed by Chun et al. which targets a highly-conserved intergenic spacer region of the 16S-23S rRNA gene sequences (6), and one of the two developed by Nandi et al. which targets the ompW gene (7). The primers tar­geting the haemolysin genes of V. cholerae were not included in our comparison (21) because, in our collection of environmental strains, haemo­lysin genes were often absent (unpublished data). The numerous other primers, described in the lite-rature, target cholera virulence-associated genes and do not allow the identification of V. cholerae species. They are used for distinguishing between toxigenic and non-toxigenic V. cholerae. Moreover, in most V. cholerae non-O1/non-O139 strains, cholera virulence-associated genes are absent.

In our work, the use of three identification meth­ods in parallel on a large number of isolates (col­lection C) was able to illustrate the advantages of PCR for the confirmation of presumptive identi­fications. Based on our data (very few API-, PCR+ results and no API+, PCR- results), we can say that there is a little difference between the tested bio­chemical and the molecular methods used for confirming the presumptive identifications born of the screening procedure. Previous observation on this had never been carried out using such a large number of environmental isolates. Howev­er, identification by PCR should be the preferred method as it is known to be quicker to perform and more accurate. Leroux et al. obtained different results in a similar study, with no API-, PCR+ and a significant proportion of API+, PCR- results, but they used another pair of primers targeting the ompW gene (22).

In this study, PCR reactions targeting ISR and the ompW gene were compared on a large col­lection of environmental isolates of V. cholerae and a significant number of V. mimicus isolates, which had never been done before. Based on our results, it can be concluded that ompW-based PCR should be preferred to ISR-based PCR and that there was no evidence of cross-reaction with the closely-related V. mimicus species.

In conclusion, the proposed simple procedure for the identification of V. cholerae after isolation on TCBS agar is based upon a combination of phe­notypic and genotypic testing methods, which is recommended for the identification of any taxon (23). This is suited to microbiological monitoring of the aquatic environment which requires tech­niques with higher resolution.

The screening of presumptive isolates on TCBS agar is quick, cheap, and easy to perform, and the read­ing of results are clear-cut and reliable. Also, the associated identification method by ompW-based PCR is quicker and slightly more sensitive than the API 20E system.

As a whole, the proposed procedure is also rea­sonable with respect to time consumption and expense. It is also adapted to the capabilities of a routine microbiology laboratory (clinical, environ­mental and food-testing) and is easily adaptable to the workflow in such laboratories. All these charac­teristics are beneficial to public health.