ABSTRACT
Nocardia identification required laborious and time-consuming phenotypic and chemotaxonomic methods until molecular methods were developed in the mid-1990s. Here we reassessed the capacity of PCR-restriction enzyme pattern analysis (PRA) of the hsp65 gene to differentiate Nocardia species, including 36 new species. Our results confirm that hsp65 PRA must no longer be used for Nocardia species identification, as many species have the same restriction pattern. We then compared sequencing-based strategies using an hsp65 database and a 16S rRNA database and found that the hsp65 region contained sufficient polymorphisms for comprehensive Nocardia species identification.
Nocardia species are gram-positive, weakly acid-fast, strictly aerobic bacteria that form filamentous branched cells which fragment into pleomorphic rod-shaped or coccoid elements. Nocardia species are essentially soil saprophytes involved in the decomposition of plant material (12, 16, 18). However, some species can infect both immunocompromised and immunocompetent individuals (18). Genus and species identification is necessary to predict antimicrobial susceptibility and for epidemiological purposes and also for environmental investigations (biodiversity, ecological niches, etc.).
Nocardia identification used to be based on laborious and time-consuming phenotypic and chemotaxonomic methods. Molecular methods were developed in the 1990s, including a 16S rRNA gene PCR-based method capable of distinguishing the genus Nocardia among aerobic actinomycetes (15). PCR-restriction enzyme pattern analysis (PRA) of a 441-bp fragment of the 65-kDa heat shock protein (hsp65) gene was developed to identify individual Nocardia species (28, 29). Sequential use of the two techniques provided rapid and simplified identification of Nocardia isolates from molecular dichotomous decision trees based on amplification/no amplification and the number and size of restriction fragments.
The genus Nocardia has undergone a taxonomic revolution during the last 10 years. Only 12 species were described between 1888, when the genus was first isolated by Nocard (20), and 1996, whereas more than 40 species are now recognized to exist. Some have been collected from clinical specimens; other have been isolated only from environmental specimens. PCR methods developed during the last decade have not yet been tested on the full range of known Nocardia species. For example, no data are available on the hsp65-PRA patterns of the 36 new species. Moreover, sequencing methods are increasingly important identification tools, and those based on 16S rRNA gene polymorphism have been applied to Nocardia (7, 19, 21). The MicroSeq 500 16S rRNA gene kit (PE Applied Biosystems) and the RIDOM database and BIBI database based on this methodology have recently been applied to the species identification of Mycobacterium and Nocardia isolates (6, 7, 9, 19, 21, 32). These approaches proved to be as efficient as conventional methods (biochemical tests, high-pressure liquid chromatography, and molecular probes) for many but not all Nocardia species (7). The latter authors underlined that public databases which are not monitored (no standard annotation, no control of strain identification, etc.) should be used with caution. Moreover, in order to overcome the strong similarity of 16S rRNA gene sequences within the genus Mycobacterium (e.g., M. gastri and M. kansasii) and microheterogeneity within a given species (e.g., M. gordonae), the study of other genes such as hsp65 (23), rpoB (13), sod (35), recA (1), and 16S-23S ITS (24) may be used. Similar problems arise with Nocardia (19, 21).
Here we reevaluated the accuracy of the hsp65-PRA method and compared it with an alternative strategy based on partial hsp65 gene sequences for Nocardia species identification.
MATERIALS AND METHODS
Type and reference strains.Forty-four strains corresponding to 44 species of Nocardia were studied (Table 1). N. asteroides ATCC 49872, representative of “N. asteroides type IV,” was also included, as it corresponded to clearly individualized clusters (which have not yet been named) (4, 17). The strain ATCC 19247T, previously used as a representative of N. asteroides, was included in this study. The taxonomic position of this strain has given rise to much controversy. This strain was in fact representative only of a rare unnamed subgroup of the former N. asteroides complex (21). In the same way, the strain ATCC 14759 was proposed as the reference strain for the type VI drug susceptibility pattern. But some authors indicated that N. cyriacigeorgica may be the same as the major group of isolates (i.e., type VI) within the N. asteroides complex (21, 25). In the absence of information (especially DNA-DNA homology and decision by taxonomic committees) (21) allowing a definitive conclusion, we decided to include the two species in our study and to present separately the data for the two representative strains. Streptomyces somaliensis DSM 41612T was used as the outgroup for phylogenetic analysis. The strains were obtained from international collections and grown on Bennett agar at 37°C for 3 to 15 days.
Strains of Nocardia studied
Clinical isolates.We also studied 21 clinical isolates sent for identification to the Observatoire Français des Nocardioses (Lyon, France). We confirmed that they belonged to the genus Nocardia by analyzing basic phenotypic characteristics such as culture morphology, mesodiaminopimelic acid, lysozyme resistance, substrate use (2), and also PCR (15).
DNA extraction.DNA was extracted with achromopeptidase. Colonies were picked off with a loop, and one loopful was suspended in 250 μl of sterile pyrolyzed water and vortexed for 1 minute. The bacterial filaments were crushed manually with conical plastic crushers. The mixture was then incubated for 15 min at 70°C. Fifty microliters of the suspension plus 1.5 μl of achromopeptidase (10 U/ml; Sigma, Steinheim, Germany) was incubated at 55°C for 15 min. The suspensions were then centrifuged for 3 min at 13,000 rpm. The supernatants were stored at −20°C until use.
hsp65 amplification and PRA identification of Nocardia species.A 441-bp fragment of the hsp65 gene encoding the 65-kDa heat shock protein was amplified with primers described by Telenti et al. (TB11, 5′-ACCAACGATGGTGTGTCCAT-3′; TB12, 5′-CTTGTCGAACCGCATACCCT-3′) (30). Amplification was carried out in packaged PCR tubes (Ready-to-Go PCR Beads; Amersham Biosciences, Piscataway, N.J.) in a final volume of 25 μl (2.5 U of Taq polymerase puRe Taq, 10 mM Tris-HCl [pH 9], 50 mM KCl, 1.5 mM MgCl2, 200 μM each deoxynucleoside triphosphate) with 10 μl of DNA extract. Amplification was carried out in a thermal cycler (PTC-100; MJ Research, Boston, Mass.). Amplification runs included a 5-min initial denaturation step at 94°C, followed by 35 cycles (94°C for 60 s, 55°C for 60 s, and 72°C for 60 s), and a 10-min final extension step at 72°C.
Ten microliters of amplification product was digested with BstEII, MspI, and HinfI according to the manufacturer's instructions (New England Biolabs, Saint Quentin en Yvelines, France) and then electrophoresed in a 3% agarose gel (Agarose 3:1; Eurobio, Les Ulis, France) containing 0.5 μg/ml ethidium bromide (Sigma) at 90 V for 6 h. A 50-bp DNA ladder (Sigma) was used to interpret restriction patterns. The results were analyzed and included in a dichotomous decision tree based on the number and size of restriction fragments obtained with each enzyme, in the following order: BstEII, MspI, and HinfI.
hsp65 and 16S rRNA gene sequencing and analysis.A 606-bp fragment of the 16S rRNA gene was amplified with primers Noc1 (5′-GCTTAACACATGCAAGTCG-3′) (positions 46 to 64, Escherichia coli numbering system) and Noc2 (5′-GAATTCCAGTCTCCCCTG-3′) (positions 663 to 680, E. coli numbering system). After 40 cycles of denaturation at 94°C for 60 s, primer annealing at 58°C for 60 s, and primer extension at 72°C for 60 s, followed by postamplification extension at 72°C for 5 min, PCR products were purified with the EZNA gel extraction kit (EAZY Nucleic Acid Purification; Omega Bio-Tek, Vaulx-en-Velin, France). For hsp65 gene sequencing, the amplification protocol was the same as that used for hsp65 PRA. Sequencing was performed with fluorescence-labeled dideoxynucleotide terminators implemented with the ABI PRISM Big Dye terminator cycle sequencing reaction kit (PE Applied Biosystems, Foster City, CA) and primers Noc1 and Noc2. The hsp65 gene was amplified and sequenced with primers TB11 and TB12 as described below. Nucleotide sequences were determined with an ABI 377 automated sequencer according to the manufacturer's instructions (PE Applied Biosystems). Each sequence was manually aligned and analyzed to ensure a high quality of sequence data.
According to the 441-bp fragment of the hsp65 gene, we achieved an in silico restriction analysis by using DNAStrider software. We tried to select new enzymes allowing a better differentiation of Nocardia species.
Moreover, the target 16S rRNA gene and hsp65 nucleotide sequences were aligned with the Clustal W program (31). Phylo_win software (10) was used to infer evolutionary trees according to neighbor-joining methods (26) using the Kimura two-parameter model (14). Tree robustness was assessed by bootstrap resampling (1,000 replicates each).
Analysis of sequences from clinical strains.The hsp65 and 16S rRNA sequences of the 21 clinical strains were determined as described above. They were then compared to the Nocardia entries in the hsp65 and 16S rRNA databases that included only the sequences obtained in this study. The database comparison, using Bibi software (9), generated a list of the closest matches with pairwise distance scores indicating the percent difference between the unknown sequence and the database sequences.
RESULTS
Reference strains.Primers TB11 and TB12 amplified the expected 441-bp fragment of the hsp65 gene for all the reference strains. The restriction patterns obtained for the species previously included in decision trees (Fig. 1, species indicated by arrows) conformed to those described by Steingrube et al. (28, 29). The restriction patterns of the 36 species not previously tested with this method (Fig. 1) were added to the previously published patterns. The new decision tree was much more complex. Thirteen species (N. arthritidis, N. beijingensis, N. cerradoensis, N. crassostreae, N. farcinica, N. ignorata, N. mexicana, N. pseudobrasiliensis, N. pseudovaccinii, N. seriolae, N. tenerifensis, N. testacea, and N. transvalensis) each had a unique restriction pattern. Conversely, a significant number of species had identical patterns that formed six clusters (Fig. 1), as follows: cluster 1, N. asteroides type IV (ATCC 49872) and N. vaccinii; cluster 2, N. fluminea and N. salmonicida; cluster 3, N. carnea, N. flavorosea, and N. uniformis; cluster 4, N. abscessus, N. araoensis, N. asiatica, N. asteroides type VI (ATCC 14759), N. asteroides ATCC 19247T, N. brevicatena, N. cyriacigeorgica, N. higoensis, N. neocaledoniensis, N. paucivorans, N. puris, N. vinacea; cluster 5, N. brasiliensis, N. inohanensis, N. niigatensis, N. otitidiscaviarum, N. shimofusensis, and N. yamanashiensis; cluster 6, N. africana, N. cummidelens, N. nova, N. pneumoniae, N. soli, and N. veterana.
Steingrube's decision tree for the identification of Nocardia species by hsp65 PRA (29), based on reference strains. The previously tested species are indicated by arrows. a, reclassified as N. abscessus by Roth et al. (25); b, reclassified as a Nocardia sp. by Roth et al. (25); *, results from the work of Steingrube et al. (29).
Surprisingly, the N. cerradoensis and N. ignorata amplification products had a BstEII restriction site, which is characteristic of mycobacteria (29).
According to these first results, we decided to generate a new decisional dichotomous tree with additional discriminating enzymes. More than 550 enzymes were tested in silico on the hsp65 sequences of the 44 reference strains. Although some combinations of restriction enzymes provided better differentiation within a few clusters, none discriminated among all the species (data not shown). Importantly, the undifferentiated clusters included the species most commonly identified in clinical samples.
We then created two databases, one based on the hsp65 gene sequence (401 bp) and the other based on the 16S rRNA gene sequence (569 bp) (excluding primer sequence regions), and compared their discriminatory power. Alignment of the hsp65 sequences revealed a large number of variable nucleotide positions bearing at least one substitution in all species. The hsp65 interspecies dissimilarity ranged from 12% between N. fluminea and N. transvalensis to 0% between N. soli and N. cummidelens and between N. flavorosea and N. uniformis (Table 2). By comparison, the interspecies dissimilarity based on the 16S rRNA fragment ranged from 9.5% between N. fluminea and N. testacea to 0% between N. soli and N. cummidelens and between N. cyriacigeorgica and N. asteroides type VI (Table 2). The number of variable sites was higher in the hsp65 gene than in the 16S rRNA gene (26.4% versus 14.6%). Moreover, when we analyzed each pair of species individually (Table 2), 71% of pairwise distances (672/946) were higher with hsp65 than with 16S rRNA.
16S rRNA and hsp65 gene sequence dissimilarities, with evolutionary distances, of 44 species of Nocardia, determined with ClustalW and Mega software
Two phylogenetic trees were then generated from the hsp65 and 16S rRNA gene sequences of the 44 species studied (Fig. 2 and 3). The S. somaliensis DSM 41612T sequence was used to root both trees.
Phylogenetic tree based on 16S rRNA sequencing of collection strains belonging to the genus Nocardia. The tree was constructed by using the neighbor-joining method, based on a 569-nucleotide stretch. Bootstrap values are expressed as a percentage of 1,000 replications. The scale bar represents 0.02 substitutions per nucleotide position.
Phylogenetic tree based upon hsp65 sequencing of collection strains belonging to the genus Nocardia. The tree was constructed by using the neighbor-joining method, based on a 401-nucleotide stretch. Bootstrap values are expressed as a percentage of 1,000 replications. The scale bar represents 0.02 substitutions per nucleotide position.
Clinical strains.Twenty-one isolates were submitted to 16S rRNA and hsp65 sequence analysis using our own nocardial databases and Bibi software (http://pbil.univ-lyon1.fr/bibi/ ). For each clinical isolate and each fragment (16S rRNA and hsp65), we searched for the three closest species in the databases, e.g., the three species showing the lowest dissimilarity (Table 3). For 19 of the 21 isolates, the closest species were the same whether the hsp65 or 16S rRNA sequences were used (Table 2). For the two remaining isolates (04.17 and 04.18), the hsp65-based analysis was inconclusive: the pairwise distances from N. nova and N. africana were equal, while those based on the 16S rRNA sequences were identical (isolate 04.18) or very close (isolate 04.17, 0.2% dissimilarity) to N. nova. In contrast, the two species were much more remote from N. africana, which was only the third most closely related species. The 16S rRNA sequence was thus more discriminatory.
List of the three closest species obtained by using Bibi software applied to the 16S and hsp65 databases for 21 clinical Nocardia isolates
For the other isolates, the hsp65 gene was more discriminatory than the 16S rRNA gene. Indeed, the closest species were more distant (higher dissimilarity differences with hsp65 sequences than with 16S rRNA sequences). For instance, the 16S rRNA sequence of isolate 04.07 showed 0.2% dissimilarity to N. cerradoensis, 0.3% dissimilarity to N. veterana, and 0.9% dissimilarity to N. africana. On the basis of the hsp65 sequences, N. cerradoensis was the closest species (0.5% dissimilarity) and N. veterana did not appear among the three closest species.
As it was the first description of clinical isolates belonging to N. cerradoensis and N. ignorata (03.14, N21, 27.3837, 00.18, and 04.07), we have confirmed species identification by an extensive phenotypic and chemotaxonomic analysis for all isolates (data not shown).
DISCUSSION
In the mid-1990s, Wallace et al. (33), Steingrube et al. (28, 29), and Laurent et al. (15) described new approaches to Nocardia genus and species identification. Here, when we included the restriction profiles of the 36 new Nocardia species inthe decision tree described in 1997 by Steingrube et al., a significant number of species had identical profiles (Fig. 1). Thus, the “N. asteroides type VI” pattern (corresponding to drug susceptibility types) (34) according to the work of Steingrube et al. (28, 29) was identical to the pattern obtained for N. abscessus, N. asteroides (ATCC 19247T), N. brevicatena, N. cyriacigeorgica, N. paucivorans, and N. vinacea. This pattern is also obtained for 60% of French clinical isolates (F. Laurent, personal data). On the other hand, it was not possible to distinguish N. africana, N. cummidelens, N. nova, N. soli, and N. veterana. These results confirmed that PRA is unsuitable for delineating the new nocardial taxonomic background (8, 22). Moreover, we found that the PRA pattern for N. abscessus, previously included in the N. asteroides type I subspecies, differed from the profile reported by Steingrube (Fig. 1). This was confirmed according to 21 clinical isolates of N. abscessus/N. asiatica (unpublished data).
Thus, the use of PRA can lead to erroneous species identification of both clinical and environmental Nocardia isolates. Pottumarthy et al. (22) recently reported that the N. veterana hsp65 PRA profile was identical to that of N. nova (22). The absence of the newly described Nocardia species in the PRA tree inevitably leads to an overestimation of the prevalence of infections due to the species included in the tree and to an underestimation of infections due to species that are missing from the tree. For instance, in our large collection of clinical strains, for which hsp65 PRA was used for species identification, a recent reexamination based on 16S sequence analysis demonstrated that 34% and 26% of strains with an N. asteroides type VI PRA pattern in fact belonged to the species N. cyriacigeorgica and N. abscessus, respectively (the other isolates belonged to N. transvalensis, N. paucivorans, N. brevicatena, or N. asteroides or were identified as Nocardia sp.) (F. Laurent, personal data).
BstEII enzyme restriction provides rapid and simple separation of Mycobacterium with PRA (28, 29). Other actinomycetes, including Nocardia, were reported to have no BstEII restriction site. However, we found that the Nocardia species ignorata and cerradoensis had a BstEII restriction site. Moreover, Brunello et al. (3) showed that several Mycobacterium species (M. confluentis, M. gilvum, M. tusciae, M. brumae, M. pulveris, M. duvalii, M. szulgai, and M. gadium) had no BstEII restriction site and had a BstEII restriction profile identical to that of Nocardia strains. This underlines the importance of the Nocardia-specific 16S rRNA PCR method (15), which is the only approach currently capable of offering rapid and accurate identification of the genus Nocardia (15).
hsp65 PRA has been shown to be unreliable for identification of Nocardia species before (8, 22). Our results confirm and extend these findings. The use of new restriction enzymes together with a new dichotomous decision tree failed to improve Nocardia species identification, and the inseparable clusters included the most common clinical species. PRA explores only the polymorphism of a few bases present at the restriction site of each enzyme, and it would be difficult to differentiate more than 40 species on the basis of such a small region of an amplified fragment.
Advances in DNA sequencing and the increasing number of sequences available in databases have greatly improved the molecular identification of various bacteria. For Nocardia, tools such as the Ribosomal Differentiation of Medical Microorganisms (RIDOM) database (19) and the MicroSeq kit (Perkin-Elmer Applied Biosystems) (7) are based on the 16S rRNA gene. The latter authors highlighted the value of such an approach but stressed that (i) interspecies heterogeneity of closely related nocardial species is very low, (ii) results for a given isolate are not always congruent and conclusive when using different databases of a given gene (7, 19, 25), and (iii) the most important component for successful identification of bacterial isolates is an accurate and complete database (7). But at present, validated databases for the molecular identification of Nocardia isolates are available with a commercial kit (MicroSeq; Applied Biosystems, Foster City, Calif.) or are not freely accessible (RIDOM). Moreover, these databases comprise fewer than 30 validated Nocardia species. This is why we decided to develop specific databases of hsp65 and 16S rRNA sequences for all validated and putative Nocardia species (as of 31 October 2004, except for N. alba, N. pigrifrangens, and N. caishijiensis) and compared their performance.
The sequenced 16S rRNA region corresponds to bases 64 to 663 of the E. coli 16S rRNA gene. It is very close to the region used in the RIDOM database (bases 54 to 510) and the MicroSeq 500 database (bases 4 to 532) and includes the most variable regions of the nocardial 16S rRNA gene. We observed a higher average dissimilarity among the different species with the hsp65 sequence (mean, 6.3%; range, 12 to 0%) than with the 16S rRNA sequence (mean, 5.2%; range, 9.5 to 0%). In addition, the distances between each pair of species were three to four times larger with hsp65 than with 16S rRNA. Likewise, the number of variable sites was larger in the hsp65 gene sequence (26.4% versus 16.5%). Whereas variable sites are confined to certain areas of the 16S rRNA gene, they are dispersed throughout the hsp65 gene and show a higher frequency in two regions (positions 624 to 664 and positions 683 to 725), as previously reported for mycobacteria by Ringuet et al. (23). Most substitutions in the hsp65 gene are confined to the third position of each codon (the wobble position), ensuring that the resulting amino acid sequence is preserved across the genus (data not shown), together with the crucial function of Hsp65 protein in resistance to environmental stress (23).
The relative positions of each species were similar in the hsp65 and 16S rRNA phylogenetic trees when bootstrap values higher than 80% were examined. However, the tree based on hsp65 was more discriminatory and more robust. Respectively, 23 nodes (55%) and 17 nodes (41%) were supported by bootstrap values greater than 50% and 80% in the hsp65 tree, compared to only 21 (50%) and 11 (26%) nodes in the 16S rRNA tree.
Some species that were not clearly distinguished by their 16S rRNA sequences were clearly distinguished by their hsp65 sequences; this was the case for N. shimofusensis versus N. higoensis and N. abscessus versus N. asiatica, for example. In contrast, N. flavorosea and N. uniformis had identical hsp65 sequences but very different 16S rRNA sequences. N. soli and N. cummidelens could not be separated by using either database. For N. cyriacigeorgica and N. asteroides type VI (ATCC 14759), 16S rRNA sequences and hsp65 sequences revealed 0% and 0.5% dissimilarity, respectively. These data reinforced the idea that these two species form in fact a single taxon as recently suggested by Roth et al. and Patel et al. (21, 25).
Analysis of the clinical isolates indicated that (i) species discrimination was better with hsp65 than with 16S rRNA (larger distances between the closest species pairs) and (ii) combined analysis of the two databases resolved some inconclusive results obtained with only one database. How “similar” or “close” a strain must be to the reference strain before it can be considered a species is controversial, as is the choice between similarity (or dissimilarity) and phylogenetic analysis for species designation. Some authors answer this question by statistical analysis. For instance, Cloud et al. used a reporting criterion of ≥99% similarity (or ≤1% dissimilarity) and a statistical error probability of 3% to separate two species with the MicroSeq 500 kit, according to the normal distribution of pairwise distances (7). Meanwhile, Mellmann et al. chose a reporting criterion of ≥99.12% similarity (or ≤0.88% dissimilarity) to define a distinct species and calculated an error probability of 1% with the RIDOM database (19). Whatever the criterion and the database, the choice is arbitrary and will likely change as new species are added to the database. It is interesting that, according to the chosen statistical approach, it is possible to define, from a threshold, a residual error to attribute a strain to a given species although it does not belong to this species. However, in the absence of data on intraspecies variability, Mellmann et al. and Cloud et al. cannot quantify the probability of not rightfully attributing a strain to a given species. In a recent review (5), Clarridge stated that no similarity or dissimilarity value could be assigned to a defined species on the basis of its 16S rRNA sequence. This is partly because different values are generated by the use of different databases and different methods (5) and emphasizes the need for clean, updated, and controlled reference databases.
Nocardia taxonomic studies must allow new bona fide species to be identified on the basis of morphological, biochemical, physiological, and chemotaxonomic properties coupled with genomic and phylogenetic analyses (27), while species identification of clinical isolates must be based on conformity to the phylogenically closest known species contained in updated and comprehensive databases. The latter task is greatly facilitated by new bioinformatics tools based on simplified analysis of sequence similarities (or dissimilarities) within an accepted and updated bacterial framework. One such tool, named Bibi, was recently described by Devulder et al. (9), but it requires complete and accurate sequence data on relevant genes if it is to be used for Nocardia identification.
The choice of genes is indeed crucial. According to the work of Goebel and Stackebrandt (11), 16S rRNA dissimilarity of ≥3% always clearly separates isolates belonging to different species; in other words, a strain does not belong to a given species if its 16S rRNA sequence differs by more than 3%. However, some species of Nocardia have identical 16S sequences (Fig. 2), as is the case for many other taxa. As expected for a protein-encoding gene, the Nocardia hsp65 gene is less conserved than the 16S rRNA gene.
In conclusion, our results suggest that hsp65 sequencing has the potential for providing a reliable means of identifying clinical Nocardia isolates to the genus and species level. The hsp65 sequences presented here can be used to establish a new database, analogous to the 16S rRNA gene database, containing type strain sequences that serve as standards for identification. In addition, combining the hsp65 and 16S rRNA sequences could form the basis for a new Nocardia species identification system, particularly for species with similar 16S rRNA sequences. A multilocus sequencing strategy, including a specific, clean, and controlled database containing all validated type strains belonging to the genus Nocardia and to closely related genera (Mycobacterium, Rhodococcus, Tsukamurella, Gordonia, etc.), would answer the new taxonomic background of the actinomycetes family.
ACKNOWLEDGMENTS
We are grateful to Xavier Nesme for his help during the preparation of the document and for interesting discussions on bacterial phylogeny and bacterial identification. We thank Catherine Pichat for her technical assistance; Brigitte Lamy and Marie-Laure Muller-Delignette for their help with statistical analysis; and C. Friedman, N. Kinoshita, L. Maldonado, and A. Yassin for providing strains.
V. Rodríguez-Nava is grateful to the Consejo Nacional de Ciencia y Tecnologia (CONACYT), Mexico City, Mexico, and the Société Française d'Exportation des Ressources Educatives (SFERE), Paris, France.
This work was supported by a grant from Institut Fédératif de Recherche 41 (IFR41), Université Claude Bernard Lyon 1, CNRS, Lyon, France.
FOOTNOTES
- Received 6 October 2004.
- Returned for modification 30 November 2004.
- Accepted 22 November 2005.
- Copyright © 2006 American Society for Microbiology