Rapid and Accurate Identification of Human Isolates of Pasteurella and Related Species by Sequencing the sodA Gene

Bacterial strains and culture conditions.The main characteristics of the 60 isolates used in this work are listed in Tables 1 and 2. The 27 type strains were obtained from the Collection de l'Institut Pasteur (Paris, France). Actinobacillus ureae, Haemophilus influenzae, and Haemophilus parainfluenzae type strains were chosen as representatives of the other main genera of the family Pasteurellaceae. Pasteurella lymphangitidis and Pasteurella piscicida were not included in our investigations because of their relatedness to the families Enterobacteriaceae and Vibrionaceae, respectively (10).

The 33 epidemiologically unrelated human isolates were obtained from clinically relevant specimens and were selected to encompass a variety of infections. Among them, only Pasteurella caballi isolate CNP 1019 had been previously reported (13). Throughout this paper, names refer to type strains unless an isolate number is mentioned. All strains were grown aerobically at 37°C on chocolate agar supplemented with Polyvitex (AES Laboratories, Combourg, France), except for Pasteurella skyensis, which was grown anaerobically at 20°C on 10% sheep blood agar. In selected cases, complementary study of acid production from carbohydrates was performed by using Taxo disks and CTA medium (Becton Dickinson, Le Pont-De-Claix, France), and p-nitrophenyl-β-d-glucoside tests were performed in API 20 NE strips (bioMérieux, Marcy-l'Étoile, France).

DNA manipulations.Based on the published sodA sequence of P. multocida strain Pm70 (23), we assumed that degenerate primers d1 (5′-CCITAYICITAYGAYGCIYTIGARCC-3′) and d2 (5′-ARRTARTAIGCRTGYTCCCAIACRTC-3′), designed by Poyart et al. (29), could be used for sodA amplification from Pasteurella and related species. An internal fragment (sodA_int) representing approximately 80% of the sodA gene was amplified by colony PCR with d1 and d2. PCR conditions were as described by Poyart et al. for gram-positive cocci (30), and the bacterial lysis was carried out during the initial denaturing step. Briefly, reactions were performed in a final volume of 50 μl containing ca. 10³ bacteria from a single colony, 0.5 μM (each) primer (Proligo, Paris, France), 200 μM (each) deoxynucleoside triphosphate (Invitrogen, Oxon, United Kingdom), and 1 U of AmpliTaq Gold DNA polymerase (Applied Biosystems, Courtabeuf, France) in a 1× amplification buffer (10 mM Tris-HCl [pH 8.3], 50 mM KCl, 1.5 mM MgCl₂). The PCR mixtures were denatured (7 min at 95°C) and then subjected to 30 cycles of amplification (60 s of annealing at 37°C, 45 s of elongation at 72°C, and 30 s of denaturation at 95°C) and to a final elongation cycle of 72°C for 10 min. The PCR products were resolved by electrophoresis on a 1% agarose gel stained with ethidium bromide. Amplicons were purified on S-400 Sephadex columns (Pharmacia Biotech, Orsay, France).

Cloning and sequencing.Purified amplicons obtained from the type strains were cloned into the pUC18-SmaI dephosphorylated vector by using the Sure-clone ligation kit (Pharmacia Biotech). Recombinant plasmids were analyzed by colony PCR on 12 randomly chosen clones with the universal −21 (5′-GTAAAACGACGGCCAGT-3′) and reverse (5′-AACAGCTATGACCATG-3′) primers in a final volume of 50 μl containing 10³ bacteria, 0.1 μM (each) primer, 200 μM (each) deoxynucleoside triphosphate, and 1 U of AmpliTaq Gold DNA polymerase in a 1× amplification buffer (10 mM Tris-HCl [pH 8.3], 50 mM KCl, 1.5 mM MgCl₂). The PCR mixtures were denatured (10 min at 95°C) and were then subjected to 30 cycles of amplification (90 s of annealing at 45°C, 1 min of elongation at 72°C, and 1 min of denaturation at 95°C). PCR products were directly sequenced after purification on a Sephadex S-400 column. The entire nucleotide sequences of both strands of two cloned amplicons obtained from independent PCRs were determined with the BigDye terminator cycle sequencing kit (version 3.1; Applied Biosystems) per the manufacturer's instructions. Sequencing products were resolved with an ABI 3100 automated sequencer (Applied Biosystems). Purified sodA_int amplicons obtained from the clinical isolates were directly sequenced on both strands with the degenerate primers d1 and d2 by using the BigDye terminator kit and the ABI 3100 sequencer.

Sequence analysis.The nucleotide sequences were analyzed with Applied Biosystems software (Sequencing Analysis, SeqScape). The sequences corresponding to the oligonucleotides d1 and d2 were excluded for subsequent analysis because of their degeneracy. Multiple alignment was carried out with the CLUSTAL X program (19). Alignment gaps were not taken into consideration for calculations. The construction of the sodA_int and 16S rRNA gene phylogenetic trees was performed by the neighbor-joining method (37). The topology of the phylogenetic tree was evaluated by bootstrap analysis to give the degree of confidence intervals for each node on the phylogenetic tree. The reproducibility of tree nodes was evaluated by generating 1,000 bootstraps trees. The sodA and 16S rRNA gene sequences of Escherichia coli K-12 (complete genome, accession no. U00096 ) were used as outgroups. The 16S rRNA gene phylogenetic tree was based on a 1,342-bp partial sequence corresponding to nucleotide positions 49 to 1391 (E. coli numbering) which was obtained from GenBank for each of the type strains (accession numbers are as follows: P. aerogenes, U66491 ; P. avium, AY362916; P. bettyae AY362917; P. caballi, AY362918; P. canis, AY362919; P. dagmatis, AY362920; P. gallinarum, AY362921; P. langaaensis, AY362922; P. mairii, AY362923; P. multocida subsp. multocida, AF294410; P. multocida subsp. gallicida, AF294412; P. multocida subsp. septica, AF294411; P. pneumotropica, AY362924; P. stomatis, AY362925; P. skyensis, AJ243202; P. testudinis, AY362926; P. trehalosi, AY362927; P. volantium, AY362928; M. glucosida, AY362912; M. granulomatis, AY362913; M. haemolytica, AF060699; M. ruminalis, AY362912; M. varigena, AY362912; A. ureae, AY362900; G. anatis, AF228001; H. influenzae, M35019; and H. parainfluenzae, AY362908). Ambiguities or gaps present in some of those sequences were solved by partial 16S rRNA gene sequencing of the corresponding type strain before performing alignments.

Nucleotide sequence accession numbers.The 60 sodA_int sequences determined in this work were submitted to GenBank and assigned the accession numbers listed in Tables 1 and 2.

A sodA gene is present in all Pasteurella and related species.The degenerate primers d1 and d2 were designed on the basis of a reverse translation of two conserved amino acid domains that are characteristic of the sodA gene products of low-GC gram-positive eubacteria (29). Therefore, they made it possible to amplify a DNA fragment internal to the sodA gene (designated sodA_int) of enterococcal, streptococcal, and staphylococcal type strains and clinical isolates (30, 31). Sequence analysis of the genome of P. multocida strain Pm70 revealed the presence of sodA and sodC genes, which are thought to encode Mn-SOD and Cu/Zn-SOD, respectively (23). Interestingly, a computer-assisted analysis of the Pm70 genome indicated that the 642-bp sodA gene was the sole target for primers d1 (no mismatches) and d2 (two mismatches) and that a 506-bp sodA_int fragment should be amplified by PCR if d2 misprimability is tolerated. Consistently, by using primers d1 and d2 in a colony PCR assay, we amplified a single DNA fragment of the expected size with all 18 type strains of Pasteurella used in this study. A similar result was obtained with Gallibacterium and Mannheimia species and with the type strains of A. ureae, H. influenzae, and H. parainfluenzae (Fig. 1). The nucleotide sequences of the sodA_int fragments from these type strains were determined following cloning into pUC18. Analysis of the deduced amino acid sequences (data not shown) revealed that they all shared one aspartyl residue and three histidyl residues that are thought to be involved in Fe²⁺ or Mn²⁺ binding by SOD enzymes and four other residues that are characteristic of Mn-SODs (28). We therefore concluded that the PCR products cloned and sequenced were actual internal fragments of sodA genes. In the family Pasteurellaceae, an Mn-SOD activity has been demonstrated so far only in H. influenzae (22) and Haemophilus ducreyi (38), while a sodA gene is also present in the genomes of four species, Actinobacillus pleuropneumoniae (27), P. multocida (23), M. haemolytica (unpublished, accession no. L47537 ), and Histophilus (Haemophilus) somnus (unpublished genome shotgun sequence, accession no. NZAACJ01000012 ). Here we demonstrate that all the species of the Pasteurella, Gallibacterium, and Mannheimia genera also possess a sodA gene.

• Open in new tab
• Download powerpoint

FIG. 1.

Amplification of type strains of Pasteurella and related species with the primers d1 and d2 and separation of the sodA_int amplicons (arrowheads) by 1% agarose gel electrophoresis. Lanes: 1 and 21, 50-bp DNA ladder (Fermentas, Lithuania); 2, P. avium; 3, P. gallinarum; 4, P. volantium; 5, P. skyensis; 6, P. testudinis; 7, A. ureae; 8, Mannheimia sp.; 9, P. trehalosi; 10, P. canis; 11, P. dagmatis; 12, P. multocida; 13, P. stomatis; 14, P. aerogenes; 15, P. mairii; 16, G. anatis; 17, H. parainfluenzae; 18, P. bettyae; 19, P. langaaensis, 20, P. caballi. The size of the amplicons is either 503 bp (lanes 2 to 9), 506 bp (lanes 10 to 17), 515 bp (lane 18), 518 bp (lane 19), or 527 bp (lane 20).

Comparison of the sodA_int sequences of type strains.Depending on the species, the length of most sodA_int sequences (primers excluded) was either 449 bp (P. avium, P. gallinarum, P. volantium, P. skyensis, P. testudinis, A. ureae, Mannheimia sp., and P. trehalosi), 452 bp (P. canis, P. dagmatis, P. multocida, P. stomatis, P. aerogenes, P. mairii, G. anatis, and H. parainfluenzae), or 455 bp (P. pneumotropica and H. influenzae). However, P. bettyae, P. langaaensis, and P. caballi sodA_int fragments displayed longer sequences of 461, 464, and 473 bp, respectively. The 9, 12, and 21 extra nucleotides correspond to supplemental blocks of three codons (P. bettyae), four codons (P. langaaensis), and four plus three codons (P. caballi) in the deduced amino acid sequences. Interestingly, these blocks are located in a short and otherwise conserved region (Fig. 2) and were not found in other bacterial SodA sequences determined in this work or obtained from protein data banks (data not shown). The authenticity of these additional codons was confirmed by sequencing the sodA_int fragments of unrelated clinical isolates of P. caballi (GenBank accession number AY702524 ) and P. bettyae (GenBank accession numbers AY702521 , AY702522 , and AY702523 ).

• Open in new tab
• Download powerpoint

FIG. 2.

Additional amino acids (boxed) in the deduced SodA sequence of Pasteurella bettyae, Pasteurella caballi, and Pasteurella langaaensis are located in the same region. The corresponding codons are absent from sodA sequences of Pasteurella multocida and other bacterial species. The authenticity of these additional codons was confirmed by sequencing the sodA_int fragments of independent P. caballi and P. bettyae isolates. Amino acid numbering refers to the SodA sequence of P. multocida Pm70 (accession no. AAK02085 ).

Multiple alignment of the sodA_int DNA sequences was carried out with the CLUSTAL X program. Pairwise comparison of these sequences showed that their mean identity (83.5%) is less than that calculated from a comparison of their 16S rRNA gene sequences (mean identity, 98.2%). The identity matrix of the sodA_int sequences of Pasteurella, Gallibacterium, and Mannheimia type strains is shown in Table 3. These data indicate that sodA might be a more discriminative target than the 16S rRNA gene to differentiate closely related species of the Pasteurellaceae family, as was previously demonstrated for differentiation of streptococci, enterococci, and coagulase-negative staphylococci (30-32). The phylogenetic positions of P. testudinis and P. skyensis were the most distant from those of other species (Fig. 3A; Table 3), an observation consistent with the phenotypic characteristics and host specificities of both species (2, 39). The topology of the sodA_int phylogenetic tree obtained (Fig. 3A) was in general agreement with that inferred from an analysis of partial 16S rRNA (Fig. 3B) or rpoB (21) gene sequences, although some differences could be observed. Only three species clusters were supported by a significant bootstrap value of >95% (14). The Pasteurella sensu stricto core group (P. canis, P. dagmatis, P. multocida, and P. stomatis), which encompasses the species most frequently associated with human infections (17), constituted a monophyletic clade defined in 98% of the bootstrap trees (Fig. 3A), in agreement with DNA-DNA hybridization studies (25) and 16S rRNA phylogeny (11). However, the P. canis, P. dagmatis, and P. stomatis cluster defined by 16S rRNA gene analysis (bootstrap value, 97%) was not supported by sodA_int analysis, which implied only the possible association of P. canis and P. stomatis (bootstrap value, 91%) (Fig. 3).

• Open in new tab
• Download powerpoint

FIG. 3.

Phylogenetic trees showing relationships among sodA_int (A) and rrs (16S rRNA gene) (B) sequences from type strains of Pasteurella, Gallibacterium, and Mannheimia species. The trees were established by using the neighbor-joining method from an analysis of the sodA_int sequences obtained in this work (listed in Table 1) (A) and rrs sequences obtained from GenBank (B). The sodA_int and rrs sequences of A. ureae, H. influenzae, and H. parainfluenzae type strains are also included in order to show the relationships between representatives of the main genera of the family Pasteurellaceae. The sodA_int and rrs sequences of E. coli strain K-12 were used as outgroups to root the trees. The value on each branch is the estimated confidence limit (expressed as a percentage) for the position of the branch as determined by bootstrap analysis. Only bootstrap values of greater than 75% are indicated. The scale bar represents 10% differences in nucleotide sequences.

An avian Pasteurella group consisting of P. avium, P. gallinarum, and P. volantium appeared clearly independent from the Pasteurella sensu stricto core group, as it was recovered in 100% of the bootstrap trees (Fig. 3A). Gallibacterium anatis did not cluster with this group, in agreement with infB-based phylogeny (8). This finding is at odds with results from phylogenetic analysis of 16S rRNA gene sequences (Fig. 3B), which indicate that G. anatis is related to the P. avium, P. gallinarum, and P. volantium group (7). Our results confirm the recognition of Gallibacterium as an independent monophyletic genus (6).

The third monophyletic clade encompassed all Mannheimia species, with a branching order identical to that of the 16S rRNA gene sequence-derived tree (Fig. 2), in agreement with the findings of Angen et al. (1).

The sodA_int-based phylogenetic positions of P. pneumotropica and A. ureae (formerly Pasteurella ureae) indicated that these species are not related to the aforementioned Pasteurella and Mannheimia clusters (Fig. 2). Moreover, a bootstrap value of 92% suggests that P. pneumotropica might be related to the Haemophilus influenzae group. Preliminary analysis of sodA_int sequences of other Haemophilus and Actinobacillus species is in progress in our laboratory and seems to favor that hypothesis (V. Cattoir and O. Gaillot, unpublished data).

Species identification of human clinical isolates.Direct sequencing of the amplicons obtained from the 33 clinical strains yielded electropherograms devoid of overlapping peaks, confirming that the strains contained a single type of sodA gene. Both primers d1 and d2 produced unambiguous sequences of the whole length of the sodA_int amplicons. In most cases, the sodA_int sequences displayed less than 2.5% divergence with those of the corresponding type strains (Table 2). The only major discrepancy between initial species assignment and sodA_int-based identification concerned isolate CNP 531, tentatively identified as “Pasteurella haemolytica.” Its sodA_int sequence was found to be 74.8% and 99.1% identical to those of M. haemolytica and G. anatis, respectively. Acid production from d-(+)-mannose and p-nitrophenyl-β-d-glucoside hydrolysis, which allow for differentiation of G. anatis from M. haemolytica (6), subsequently confirmed our molecular identification of G. anatis. To our knowledge, this hemolytic isolate, from a patient with chronic bronchitis, constitutes the first G. anatis human strain ever reported. Conversely, the sodA_int sequence of the other “Pasteurella haemolytica” isolate used in this work was 99.8% identical to that of M. haemolytica.

All but one of the P. multocida isolates from various clinical origins were unambiguously identified to the subspecies level, including P. multocida subsp. gallicida CNP 982, which displayed 100% sodA_int identity with the corresponding type strain. Only P. multocida subsp. multocida isolate CNP 978 (Table 2) was identified as P. multocida subsp. septica based on sodA_int sequence identity (100% and 96% identity with P. multocida subsp. septica and P. multocida subsp. multocida, respectively), although it fermented sorbitol, a key characteristic of P. multocida subsp. multocida which is absent in P. multocida subsp. septica (25). This result confirms that the differentiation of the two dulcitol-negative P. multocida subspecies based solely on sorbitol fermentation might not be reliable and that sorbitol-positive P. multocida subsp. septica might exist, as previously suggested (18). Other members of the Pasteurella stricto sensu core group were unambiguously identified, as were isolates of P. bettyae, P. aerogenes, P. caballi, and P. trehalosi (Table 2). Some heterogeneity is seen among the five P. canis isolates, with 97.6% to 99.1% sodA_int identity with the type strain, although they formed a lineage clearly distinct from that of P. stomatis, P. dagmatis, and P. multocida.

Lastly, we compared the sodA_int sequences of four unrelated clinical isolates of P. pneumotropica and found out that they were distributed in a single cluster with 98.1 to 99.2% identity but shared only 91% identity with the sodA_int sequence of the type strain (Table 2). As type strain CIP 66.16 (ATCC 35149) is a murine isolate not associated with human infection, the possibility arises that P. pneumotropica human clinical isolates may belong to a different lineage than murine colonizing isolates, although they share identical phenotypic features (biotype Jawetz). However, clinical isolates CNP 728 and RSP 877 are most likely of rodent origin, since they were isolated from rat and guinea pig bite wounds, respectively. The analysis of the genetic relatedness of these and other isolates of various animal or human origins is in progress in our laboratory.

Conclusion.Conventional identification of Pasteurella and related bacteria remains a challenge to many laboratories. Most commercial identification systems are likely to overlook species other than P. multocida, and further phenotypic characterization is long, fastidious, and sometimes inconclusive or misleading. In spite of its cost, sequence-based identification is a convenient alternative which should be recommended when dealing with unusual or atypical isolates and when performing epidemiological studies. The method described here provides a rapid and accurate tool for species identification when access to a sequencing facility is available. Heat lysis was adequate for extracting DNA for amplification, and although sodA_int fragments were sequenced on both strands in this work, single-strand sequencing with either primer d1 or d2 was accurate enough for routine identification. The high discriminative power related to the sodA gene variability is particularly helpful in recognizing closely related species or subspecies, which cannot be achieved with the same confidence through 16S rRNA gene analysis. The database generated from the type strains allowed unambiguous identification of all human isolates tested, including those belonging to the unusually encountered species G. anatis, P. trehalosi, and P. caballi. The sequencing of sodA in Haemophilus and Actinobacillus species is in progress in our facility and should provide new insights on the phylogeny of the family Pasteurellaceae. This convenient genetic approach might help to investigate the distribution of Pasteurellaceae species in human infections, a prerequisite for the study of their pathogenesis.