|Home | About | Journals | Submit | Contact Us | Français|
We found seven Listeria isolates, initially identified as isolates with the Xyl+ Rha− biotype of Listeria welshimeri by phenotypic tests, which exhibited discrepant genotypic properties in a well-validated Listeria species identification oligonucleotide microarray. The microarray gives results of these seven isolates being atypical hly-negative L. seeligeri isolates, not L. welshimeri isolates. The aberrant L. seeligeri isolates were d-xylose fermentation positive, l-rhamnose fermentation negative (Xyl+ Rha−), and nonhemolytic on blood agar and in the CAMP test with both Staphylococcus aureus (S− reaction) and Rhodococcus equi (R− reaction). All genes of the prfA cluster of L. seeligeri, located in the prs-ldh region, including the orfA2, orfD, prfA, orfE, plcA, hly, orfK, mpl, actA, dplcB, plcB, orfH, orfX, orfI, orfP, orfB, and orfA genes, were checked by PCR and direct sequencing for evidence of their presence in the atypical isolates. The prs-prfA cluster-ldh region of the L. seeligeri isolates was approximately threefold shorter due to the loss of orfD, prfA, orfE, plcA, hly, orfK, mpl, actA, dplcB, plcB, orfH, orfX, and orfI. The genetic map order of the cluster genes of all the atypical L. seeligeri isolates was prs-orfA2-orfP-orfB-orfA-ldh, which was comparable to the similar region in L. welshimeri, with the exception of the presence of orfA2. DNA sequencing and phylogenetic analysis of 17 housekeeping genes indicated an L. seeligeri genomic background in all seven of the atypical hly-negative L. seeligeri isolates. Thus, the novel biotype of Xyl+ Rha− Hly− L. seeligeri strains can only be distinguished from Xyl+ Rha− L. welshimeri strains genotypically, not phenotypically. In contrast, the Rha+ Xyl+ biotype of L. welshimeri would not present an identification issue.
The bacterial genus Listeria is currently taxonomically subdivided into the following six species: Listeria monocytogenes, L. ivanovii, L. innocua, L. seeligeri, L. welshimeri, and L. grayi (33). All Listeria species are ubiquitously distributed in the natural environment and frequently isolated from different biocenoses. Despite the ability of unambiguously distinguishing the pathogenic Listeria species from the other unharmful saprophytic Listeria spp., the identification of all Listeria isolates to the species level is an important taxonomic issue. Listeria spp. identified based on selective cultural enrichments and isolation on selective agar growth media and subsequent isolates can be easily distinguished from each other to the species level by using the following markers: hemolysis (in Christie-Atkins-Munch-Peterson [CAMP] test with Staphylococcus aureus and Rhodococcus equi) and acid production from d-xylose, l-rhamnose, mannitol, and alpha-methyl-d-mannoside. As a result, extra phenotypic or genotypic methods are rarely required (3, 14). The hemolysis and CAMP tests are crucial steps for identification of the hemolytic L. monocytogenes, L. ivanovii, and L. seeligeri species as well as for their differentiation from the nonhemolytic species, L. innocua, L. welshimeri, and L. grayi. The gene hly that encodes hemolysin is present within the prfA virulence gene cluster that is found between prs and ldh in L. monocytogenes, L. ivanovii, and L. seeligeri but is absent from the genomes of the nonhemolytic L. innocua, L. welshimeri, and L. grayi species (7, 8, 33).
The occurrence of atypical or aberrant isolates in the indispensable diagnostic tests (hemolysis and CAMP) can lead to species misidentifications and taxonomic problems (15, 22). The natural and artificially modified Listeria isolates that express weak hemolytic or nonhemolytic phenotypes may be attributable to a constitutive or inducible type of hemolysin expression, inappropriate culturing conditions, or an alteration or deletion of the hly or prfA gene inside the cluster, all of which have been found in L. monocytogenes (19, 20, 29), but no analogous information is available for other hemolytic species such as L. ivanovii and L. seeligeri. Isolates with the natural weak hemolytic or nonhemolytic phenotype of L. monocytogenes may be misclassified as L. innocua strains, but the availability of genomic information for these two species allows accurate identification between them by using multiple-locus sequence typing (16).
Various DNA typing methods have been used to distinguish Listeria isolates at the species or subspecies level (23, 25, 28, 31, 37). Predominantly, DNA typing methods provide better species and even strain definition than conventional phenotypic tests and serotyping, and as a result, comprehensive characterization of prokaryotic species by analysis of diverse chromosomal loci is recommended and can provide bacteriologists with uniform information for species definition as well as phylogenetic and ecological studies (34). This study uses different genetic loci as phylogenetic and identification markers for the characterization of nonhemolytic L. seeligeri isolates and to distinguish them from L. welshimeri isolates.
In the present study, we describe seven unusual L. seeligeri isolates which were initially identified as being the Xyl+ Rha− biotype of L. welshimeri by bacteriological phenotypic tests. Detailed analysis of the aberrant L. seeligeri isolates has shown a unique organization of the prs-ldh cluster which is different from that described for the species. Multilocus sequence typing and phylogenetic analysis of various housekeeping genes showed an L. seeligeri-specific genome background for all of the isolates. Thus, we show here that these L. seeligeri isolates represent a novel biotype within the taxon. This knowledge provides critical information to allow the definition of L. seeligeri strains that may have evolved into a nonhemolytic biotype and those that may be equally present in the environment together with the typical hemolytic biotype of L. seeligeri but misclassified as Xyl+ Rha− L. welshimeri.
The Listeria strains used in this study were obtained from the Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, College Park, MD. Several L. welshimeri strains were provided by the Institut Pasteur, Paris, France. Working cultures of the strains were maintained on Trypticase soy agar supplemented with yeast extract at 5°C. Bacteria were grown overnight on brain heart infusion plates (Difco, Detroit, Mich.) at 37°C.
Identification of strains was done according to the procedures in the FDA Bacteriological Analytical Manual (14). All of these procedures were done at the Listeria Methods Research Laboratory, Center for Food Safety and Applied Nutrition, FDA.
Freshly grown bacteria were boiled in 1× Tris-EDTA buffer (approximately 108 cells/ml) for 10 min, followed by centrifugation at 14,000 × g for 10 min to remove denatured proteins and bacterial membranes. The presence of genomic DNA in all prepared samples was confirmed by 1% agarose gel electrophoresis followed by staining with ethidium bromide.
The primers used for PCR amplification are listed in Table S1 in the supplemental material. Microarray analysis was performed as described previously (36).
The standard PCR mixture (50 μl) contained 1.5 U of HotStarTaq DNA polymerase, 1× reaction buffer supplemented with 2.5 mM MgCl2 (QIAGEN, Chatsworth, Calif.), 600 nM (each) forward and reverse primers, a 200 μM concentration of each deoxynucleoside triphosphate (dATP, dGTP, dCTP, and dTTP), and 1 to 2 μl of DNA template (approximately 0.2 μg of bacterial DNA). PCR was performed with a Gene Amp PCR system 9700 thermocycler (Applied Biosystems, Foster City, Calif.) under the following conditions: initial activation at 95°C for 15 min, 40 cycles at 94°C for 40 s, 50°C for 40 s, and 72°C for 1 min per 1 kb, and a final extension at 72°C for 10 min. The PCR products were separated by electrophoresis in 1% agarose gels containing 1× Tris-acetate-EDTA buffer and visualized by staining with ethidium bromide.
DNA sequencing was conducted by using an ABI PRISM BigDye Terminator v3.1 cycle sequencing kit. Cycle sequencing reactions were conducted according to the kit's protocol. Reaction samples were then purified with Centrisep spin columns (Princeton Separations, Adelphia, NJ) and dried under a vacuum. Samples were sequenced using an ABI Prism 3100 genetic analyzer system.
Nucleotide diversity (π; average pairwise nucleotide difference/site), numbers of mutations, numbers of synonymous mutations (ds), numbers of nonsynonymous mutations (dn), and dn/ds ratios (ratio of the number of nonsynonymous substitutions/nonsynonymous site [dn] to the number of synonymous substitutions/synonymous site [ds], with a Jukes-Cantor correction of the Nei-Gojobori method ) were calculated using MEGA, version 2.1. The selection pressure on a protein-encoding gene can be measured by comparing the nonsynonymous substitution rate (dn) (amino acid altering) to the synonymous substitution rate (ds) (silent, with no amino acid change) to obtain the dn/ds ratio (ω). An ω value of 1 indicates neutral evolution (relaxed selective constraint; nonsynonymous changes have no associated fitness advantage and are fixed at the same rate as synonymous changes), ω values of <1 indicate purifying selection (strong functional constraint; nonsynonymous changes are deleterious for protein function and are fixed at a lower rate than synonymous changes), and ω values of >1 indicate positive selection (adaptive evolution; nonsynonymous changes are favored because they confer a fitness advantage and are fixed at a higher rate than synonymous changes) (4).
The genes in the prs-ldh region and other housekeeping gene sequences were compared to the GenBank nucleotide and protein databases using the BLASTN and BLASTP algorithms at GenBank (6). Nucleotide and deduced amino acid sequences for each gene were initially aligned with CLUSTALX (2). Inter- and intraspecies similarity score matrixes for each gene were generated using MEGA, version 2.1, and BioEdit software (http://www.megasoftware.net and http://www.mbio.ncsu.edu/BioEdit/bioedit.html, respectively). Phylogenetic and molecular evolutionary analyses were conducted using MEGA, version 2.1. Genetic distances were calculated by the Kimura two-parameter and Tamura-Nei models. The method of Nei and Gojobori was applied to the various sequences to obtain synonymous and nonsynonymous distances (multiple substitutions adjusted by the Jukes-Cantor formula). Phylogenetic trees were constructed and compared using neighbor-joining, maximum parsimony, and minimum evolution algorithms (18). The same gene sequences of other bacteria were used as outgroups for phylogenetic comparisons of some housekeeping genes, and bootstrap analyses were performed with 1,000 replicates.
The GenBank accession numbers of the sequences determined for this study are DQ153172 to DQ153182, DQ154289, DQ154290, DQ151663 to DQ151667, DQ093572 to DQ093578, DQ091845 to DQ091849, DQ091851, DQ092637 to DQ092642, DQ091852, DQ091853, DQ091833 to DQ091844, DQ086415 to DQ086421, DQ083394 to DQ083400, DQ065839 to DQ065846, DQ060335 to DQ060361, DQ016504 to DQ016510, AY994168 to AY994175, AY994176 to AY994187, AY878348, AY822470 to AY822475, AY785381, AY785382, AY748441 to AY748446, AY753217 to AY753221, AY729917 to AY729926, AY521653, AY521654, AY553865 to AY553869, and AY352074 to AY352076.
Seven isolates (LS159, LS165, LS160, SE107, SE116, LS166, and 2436KA) of L. seeligeri were originally obtained from food (raw milk and crabmeat) and environmental (salt marsh/estuarine water or sediment) samples by Listeria selective enrichment and selective isolation. Each isolate had a typical Listeria colonial appearance on esculin-containing PALCAM and Oxford agars. Each isolate was a gram-positive bacillus. Each isolate was motile, exhibiting tumbling motility in wet mounts and producing an umbrella pattern in motility-stab agar. Each isolate was positive in catalase, methyl red, and Voges-Proskauer tests as well as in esculin, maltose, and glucose fermentation tests. The isolates were negative in oxidase and indole tests. Triple sugar iron agar reactions were an acid slant and butt with no gas or hydrogen sulfide. Therefore, it was concluded that each isolate is a Listeria species. All seven isolates were nonhemolytic by sheep blood agar stab testing and by CAMP hemolysis enhancement testing (S− and R− reactions with Staphylococcus aureus and Rhodococcus equi, respectively). The negative hemolysis reactions suggested that the isolates were not L. ivanovii, L. monocytogenes, or L. seeligeri. The isolates did not produce acid from l-rhamnose and mannitol but did from d-xylose. Together, the hemolysis and sugar reactions suggested that the strains were L. welshimeri Rha− Xyl+ strains. Serotyping of the original isolate of 2436KA with polyclonal antibodies had shown that it was serotype 4, but no serotyping had been performed on the rest of the original isolates.
Two primer sets, LisF-LisR and IsoF-IsoR, were used for duplex PCR amplification of the Listeria species-specific alleles of the iap and hly genes, respectively (36). No synthesis of detectable amounts of amplicons was observed for the hly gene with any of these isolates, whereas these isolates were positive in iap-specific PCR amplifications. All seven isolates, previously identified as L. welshimeri, were reidentified as L. seeligeri on the basis of microarray hybridization with the iap-specific oligonucleotides. Thus, by the microarray assay and PCR results, these seven putative L. welshimeri strains were reidentified as atypical hly-negative L. seeligeri isolates. Partial sequencing of the iap, prfA, and 16S rRNA genes of these strains confirmed that the strains were L. seeligeri but had lost their hly genes due to a substantial deletion in the central virulence gene cluster (LIPI-1). Therefore, the absence of other LIPI-1 genes was suspected (36).
Since these seven aberrant L. seeligeri isolates were genetically and phenotypically hly negative, other structural alterations in the prfA-regulated virulence gene cluster of L. seeligeri were suspected and then found. All genes of the prfA gene cluster of L. seeligeri, located in the prs-ldh region, including the orfA2, orfD, prfA, orfE, plcA, hly, orfK, mpl, actA, dplcB, plcB, orfH, orfX, orfI, orfP, orfB, and orfA genes, were checked by PCR for molecular evidence of their presence in the atypical isolates (see Table S1 in the supplemental material). All PCR primers were based on GenBank data for the genetic organization of the cluster in reference strain L. seeligeri ATCC 35967. The PCR results and subsequent sequencing of the entire prs-ldh region suggested that the gene complement of the virulence gene cluster of the aberrant isolates was different from that in the reference strain L. seeligeri ATCC 35967 (Fig. (Fig.1).1). The prs-LIPI-1 cluster-ldh region of the L. seeligeri isolates was approximately threefold shorter due to the loss of the following 13 genes: orfD, prfA, orfE, plcA, hly, orfK, mpl, actA, dplcB, plcB, orfH, orfX, and orfI. The genetic map order of the prs-prfA cluster-ldh region genes of all the atypical L. seeligeri isolates was prs-orfA2-orfP-orfB-orfA-ldh, and direct sequencing showed that all of these genes were L. seeligeri specific. However, the cluster organization in the isolates was comparable to that in the nonhemolytic species L. welshimeri (AJ249808), except for one gene, orfA2, which was located between prs and orfP. The 3′ end of the cluster had the same organization (orfB-orfA-ldh) as all Listeria species, with the exception of L. grayi. These two putative open reading frames (ORFs) (orfB and orfA) are conceptually transcribed in the same direction as ldh. Remarkably, the three Listeria species, L. seeligeri, L. welshimeri, and L. ivanovii, which belong to the same cluster based on iap gene phylogeny have the same organization (orfP-orfB-orfA-ldh) in the 3′ region of the cluster (7, 33). The 5′ end has the order prs-orfA2-orfP, where orfA2 appears to be specific for the L. seeligeri LIPI-1 cluster only, and these two putative ORFs are conceptually transcribed in the same direction as prs. The orfA2-orfP ORFs are separated by a 157-bp intergenic noncoding spacer. A BLASTN search of the intergenic noncoding spacer did not reveal any homology to sequences in other bacteria, nor did it provide evidence for any transposon insertion site or transposon-related repeat structures in the region. The LIPI-1 gene cluster organization of the isolates suggests an L. seeligeri genomic background and confirms the exceptional nature of the cluster organization of the isolates relative to that of typical L. seeligeri isolates. However, no clues about the horizontal transfer/deletion of LIPI-1 in the isolates were found.
Five isolates (LS159, LS165, LS160, SE107, and SE116) have identical cluster sequences, and two, LS166 and 2436KA, have 98% and 96% homology to these five isolates, respectively. Detailed analysis of the orfA2-orfP-orfB-orfA cluster showed that the orfA gene of L. seeligeri has homologues in L. welshimeri, L. innocua, L. monocytogenes, and L. ivanovii (Fig. (Fig.2A).2A). The nucleotide homology for the gene was 91.3 to 100% between the analyzed L. seeligeri isolates and the reference strain L. seeligeri ATCC 35967. On the other hand, the gene homology was only 82% for L. ivanovii (AJ249805), 76% for L. innocua and L. monocytogenes EGD (AJ249804, AF322004, NC_003212, NC_003210, and M82881), and 72 to 73% for L. welshimeri and L. monocytogenes F2365 serovar 4b (AJ249808, AF322005, and NC_002973).
The orfB genes in these L. seeligeri isolates were 96 to 100% homologous among all atypical isolates and the reference strain L. seeligeri ATCC 35967 (Fig. (Fig.2B),2B), whereas the homology was 88% for L. ivanovii (AJ249805), 86% for L. innocua (AL596164, AF322004, and AJ249804), L. monocytogenes F2365 serovar 4b, and L. monocytogenes EGD (AE017322 and AL591974), 84% for L. welshimeri (AJ249808 and AF322005), and 74% for L. grayi (AJ249739). A BLASTP search for the deduced protein, OrfB, found it to be present, with a broadly conserved 55 to 71% amino acid similarity to hypothetical proteins in a wide selection of bacterial genera, including Bifidobacterium longum (NP_695750), Clostridium thermocellum (ZP_00504811), Vibrio vulnificus (AAO08334), Geobacter sulfurreducens (NP_953835), Nitrosomonas europaea (NP_841105), Vibrio cholerae (AAF96847), and Pasteurella multocida (AAK03752).
The orfP genes of the L. seeligeri isolates were 95 to 100% homologous among the atypical isolates and the L. seeligeri reference strain, ATCC 35967 (Fig. (Fig.2C),2C), 81 to 82% homologous to the orfP gene of L. ivanovii (AJ249805, AY510073, and AY510072), and 75% homologous to the orfP gene of L. welshimeri (AJ249808). No orfP homologue was found in L. innocua and L. monocytogenes. A BLASTP search of the conceptually translated protein OrfP found a low amino acid similarity to serine/threonine protein phosphatases and metallophosphoesterases in a variety of different bacteria.
The orfA2 genes in these L. seeligeri isolates were similar among all the analyzed isolates. However, in all the isolates, the gene was found to be about 8% shorter (2,508 bp) than the orfA2 gene of the reference strain L. seeligeri ATCC 35967 (2,709 bp). Consequently, the deduced protein OrfA2 was found to have 80 amino acids truncated in the C-terminal region and to also have a five-amino-acid insertion (EKPKN) in the same region. The insertion was not present in the reference strain. In general, the orfA2 gene was found to be present only in L. seeligeri. However, orfA2 homologues with 57 to 63% similarity were found in other Listeria species. They were the lin0665 gene (LPXTG motif) of L. innocua Clip11262 (AI1515) and the cell wall surface anchor family protein genes of L. monocytogenes F2365 serovar 4b (YP_013297) and L. monocytogenes H7858 serovar 4b (ZP_00229416). Remarkably, in comparison to the reference strain of L. seeligeri, the 80-amino-acid deletion also apparently occurred in the C-terminal regions of the OrfA2 homologues of L. innocua and L. monocytogenes. However, no orfA2 gene homologues were found within the prs-ldh cluster regions of L. innocua and L. welshimeri. BLASTP searches for OrfA2 revealed low-level homologies to conserved hypothetical proteins of L. monocytogenes F6854 serovar 1/2a (EAL05820) and L. monocytogenes H7858 serovar 4b (ZP_00232016), the hypothetical protein Lmo0333 of L. monocytogenes EGD (NP_463863), the internalin-like protein of L. monocytogenes strain 1/2a F6854 (ZP_00234811), and the internalin-like protein (LPXTG motif) of L. innocua Clip11262 (NP_470664).
Although the prfA cluster genes of L. seeligeri were found to have homologues in various other Listeria species, sequencing and analysis of the target genes within the prs-ldh region showed that all genes were L. seeligeri specific (Fig. (Fig.2).2). Thus, they form a novel unique region that has not been previously described for L. seeligeri (Fig. (Fig.1).1). Also, these findings genetically confirmed the initially observed nonhemolytic phenotype of the atypical L. seeligeri isolates.
The sequences of the complete 16S rRNA genes and the large 16S-23S rRNA intergenic spacer (ITS) regions (13) of the atypical nonhemolytic L. seeligeri isolates were determined in order to classify them phylogenetically. Dendrograms representing the nucleotide sequence homologies for the partial 16S rRNA gene for 53 analyzed sequences available in GenBank (Fig. (Fig.3A)3A) and 20 analyzed sequences of the large 16S-23S rRNA ITS regions (Fig. (Fig.3B)3B) of the test strains and of strains of other Listeria species from the GenBank database were constructed. These analyses of the 16S rRNA gene and the ITS consistently grouped the atypical strains, with 85 and 99% bootstrap values, respectively, into the L. seeligeri branch of the L. seeligeri-L. ivanovii phylogenetic cluster. The average percentage of similarity of the 16S rRNA gene partial sequences (950 bp) for all the Listeria species examined was 98%. The gene sequences from all the L. seeligeri isolates analyzed were 99.3% similar, on average; they were 97.8% similar to L. ivanovii, L. welshimeri, L. innocua, L. monocytogenes, and L. grayi. BLAST alignments of the complete 16S rRNA genes (1,546 bp) of all seven isolates showed 100 and 99% homologies with the corresponding genes from L. seeligeri (X98531, AJ535699, and AJ535698), L. welshimeri (DQ065846, X56149, X98532, and AJ535694), L. innocua (AL596173, AJ535695, AJ535693, and X98527), L. monocytogenes EGD (AL591983), L. monocytogenes F2365 serovar 4b (AE017330), and L. ivanovii (X98528 and X98529). However, despite the high percentage of 16S rRNA gene interspecies homology, the species-specific character of nucleotide substitutions and comparable trees constructed using different phylogenetic algorithms allows the clustering of all analyzed sequences in a species branch. The large 16S-23S rRNA ITS regions of the L. seeligeri isolates were 583 bp (DQ065839 to DQ065843) and 586 bp (DQ065844, DQ065845, and U57916), with the average percentage of similarity of the ITS regions in all the Listeria species examined being 93.1%, while the gene sequences for all the L. seeligeri isolates analyzed had an average similarity of 99%, with a 96.1% similarity to L. ivanovii (U57913), 92.6% similarity to L. welshimeri (U57917 and DQ065846), and 91.6% similarity to L. innocua (NC_003212 and U57915) and L. monocytogenes (U44061, U57912, and S83265). The L. seeligeri sequences exhibited only a 78.2% match with the L. grayi sequence (U57918). By analysis with tRNAscan-SE, version 1.21, software (http://www.genetics.wustl.edu/eddy/tRNAscan-SE), the large 16S-23S ITS regions of all the L. seeligeri isolates examined also contained regions with nucleotide sequence homology to genes for the tRNA molecules for isoleucine and alanine, a feature common to all Listeria species. The isoleucine tRNA gene was present between positions 180 and 253, while the alanine tRNA gene was present between positions 300 and 372 of the large L. seeligeri 16S-23S ITS. The phylogenetic results based on sequences of the 16S rRNA gene and the large 16S-23S rRNA ITS region showed that all seven isolates were L. seeligeri.
The iap gene, which is regarded as a virulence-associated gene in L. monocytogenes, codes for a gene product that has murein-lytic activity and is involved in cell division. Direct sequencing and phylogenetic analysis of the entire iap genes from all the atypical nonhemolytic L. seeligeri strains were carried out to identify their species of origin by comparison with the complete iap sequences of 80 different Listeria strains available in GenBank. The results clearly showed that the iap genes from these strains were of L. seeligeri origin (Fig. (Fig.3C)3C) and belonged to a branch of the phylogenetic cluster L. seeligeri-L. welshimeri-L. ivanovii. The average percentage of similarity of the iap genes in all the Listeria species examined was 82.5%, while the gene sequences from all the L. seeligeri isolates analyzed had an average similarity of 97.7%, with 82.8% and 71% similarity to L. ivanovii-L. welshimeri and L. innocua-L. monocytogenes, respectively, and only 59% similarity to L. grayi. The conceptually translated IAP60 proteins from all L. seeligeri isolates were homologous and had 525 amino acids. The IAP60 protein of L. seeligeri 2436KA contained only 522 amino acids due to a deletion of 276Asn, 277Thr, and 289Thr and had 8 amino acid substitutions. The average nucleotide diversity, π (i.e., nucleotide divergence per site), in the iap gene ranged from 0.008 for L. welshimeri and L. innocua to 0.017, 0.025, and 0.028 for L. seeligeri, L. ivanovii, and L. monocytogenes, respectively. The average dn/ds ratios (ω) ranged from 0.008 for L. innocua to 0.010, 0.012, 0.025, and 0.041 for L. ivanovii, L. welshimeri, L. seeligeri, and L. monocytogenes, respectively. Therefore, the dn/ds ratios for the iap genes of all the Listeria isolates were <1, which shows a purifying gene selection. In all pairwise comparisons of the iap genes from different species, the mean divergence between species (0.154) was higher than the mean divergence within species, which supported their phylogenetic distinctness. A species that diverges from a common ancestor and subsequently evolves is expected to accumulate fixed polymorphisms (sites at which all sequences in one species population differ from all sequences in a second species population), shared polymorphisms, and unique polymorphisms (i.e., sites polymorphic in one species population but monomorphic in a second population) in its DNA sequences (9, 12, 17). Based on this theory, the nonhemolytic biotype of L. seeligeri may represent one subpopulation diverging from L. seeligeri.
The putative internalin protein-like gene has been shown to be present in L. seeligeri and is very useful for the identification of L. seeligeri species (21). Direct sequencing and phylogenetic analysis of the putative internalin protein-like genes (1,473 bp) from all the atypical nonhemolytic L. seeligeri strains were carried out to identify their species of origin by comparison with the putative internalin protein-like gene sequences of nine different L. seeligeri strains available in GenBank. Five isolates (LS159, LS165, LS160, SE107, and SE116) had identical putative internalin protein-like gene sequences, and two, LS166 and 2436KA, had 98.5% and 93.1% homology to these five isolates, respectively. In general, the putative internalin protein-like gene was found to be present only in the L. seeligeri strains. However, putative internalin protein homologues with 63 to 65% similarity were found. They were the lmo0333 gene product (LPXTG motif, probable peptidoglycan-bound protein) of L. monocytogenes EGD (AF1116) and the cell wall surface anchor family proteins of L. monocytogenes F2365 serovar 4b (YP_012959) and L. monocytogenes F6854 serovar 1/2b (ZP_00234784).
To accurately identify the genomic background of the nonhemolytic L. seeligeri isolates and distinguish it from L. welshimeri, we sequenced a set of different genetic loci, including some genes which have been proposed as suitable tools for studying Listeria phylogeny (23, 28, 31, 38). The prs, ldh, recA, comK, cpn-60, kat, sigB, gyrB, ribC, hisC, addB, lisR, bglA, and betL genes were sequenced in all the hly-negative L. seeligeri isolates and the reference strain (see details in the supplemental material). Their homologues in several L. welshimeri strains, including the reference strain, were also sequenced. In order to improve the accuracy of our phylogenetic relationship predictions, we traced trees using different methods. There were slight differences between phylogenetic structures for L. seeligeri predicted by analyses of nucleotide and amino acid sequences of the housekeeping genes mentioned above. Most of the intraspecific base substitutions in the housekeeping genes analyzed were synonymous (ω < 1), i.e., they did not result in amino acid changes. This is not surprising and indicates a purifying selection of most essential housekeeping genes, which encode proteins that have functions which are vital for survival during evolution and adaptation to an ecological niche. The L. seeligeri topologies in different trees were comparable. According to analyses of some genes (sigB, recA, the 16S rRNA gene, and the ITS region), L. seeligeri forms part of an L. seeligeri-L. ivanovii branch, but in analyses of other genes (iap, prs, and lisR), it forms part of an L. seeligeri-L. ivanovii-L. welshimeri cluster. This analysis allowed us to precisely identify all the nonhemolytic L. seeligeri isolates and to distinguish them from L. welshimeri, which has analogous phenotypic features. This is taxonomically important because, as previously recorded, about 13% of natural L. welshimeri strains do not acidify rhamnose and give CAMP-negative reactions (5). If these phenotypic characteristics are taken alone without genomic background analysis, one is unable to differentiate naturally nonhemolytic L. welshimeri strains from the nonhemolytic L. seeligeri biotype.
Only a few classical tests are needed to identify members of the genus Listeria to the species level, and although aberrant strains may require extra tests, in general, species-level identification is not an onerous task. However, the strains studied here did require considerable extra effort in order to identify them to the species level. Thus, definitive species-level identification of these nonhemolytic L. seeligeri isolates was very comprehensive, being based on the combined results of the following different tests: DNA microarray hybridization and the detection of genes indicating an L. seeligeri genetic background, e.g., sequencing of the 16S and 23S intergenic regions, the 16S rRNA gene, 15 other housekeeping genes, and the LIPI-1 region with its neighboring genes. Nevertheless, in the absence of any information on the frequency of occurrence of such strains, it is not yet clear how widespread the taxonomic problem they potentially pose is with respect to complicating the species-level identification methodology for Listeria isolates. Pragmatically, if such isolates are more common than is apparent so far, they may justify a separate taxon. Otherwise, they can be regarded as taxon-transitional strains. In general, aberrant Listeria strains are not that rare in nature, and the further addition of the unique strains studied here emphasizes the importance of not casually dismissing aberrant strains as being unrepresentative of nature in terms of their frequency of occurrence.
What can be the cause of this difference between nonhemolytic and typical hemolytic L. seeligeri isolates? Two possibilities can explain how these unique strains of L. seeligeri contain a threefold shorter analog of the LIPI-1 cluster, whose gene organization (prs-orfA2-orfP-orfB-orfA-ldh) is more comparable with that of naturally nonhemolytic L. welshimeri. One is that a free-living ancestor of L. seeligeri that does not contain virulence-associated genes gave rise to the currently prevailing typical hemolytic strains of L. seeligeri and that the nonhemolytic L. seeligeri strains (phenotypically L. welshimeri-like) described here represent an atypical biotype. The other possibility is that nonhemolytic biotype strains of L. seeligeri arose by secondary loss of the LIPI-1 virulence cluster genes by some type of genetic transfer. LIPI-1 is flanked by the chromosomal genes prs and ldh, which are within about 21 kb of each other in the recently sequenced region of L. seeligeri (GenBank accession no. AY878348 and AY510074). The virulence gene cluster may have been lost by phage-mediated transfer or by processes such as genetic drift, natural selection, and recombination. However, during this study, no traces of such events have been found. Theoretically, this secondary loss of the LIPI-1 region might have been possible through “genetic garbage” removal or adaptive gene loss (11, 24, 26, 35) because L. seeligeri is typically an avirulent free-living bacterium, and missing such a large useless gene cluster might not have been vital for the survival of the atypical strains. In the case of bacteria, such types of genome transformation are particularly important processes since they may involve transfer from widely divergent backgrounds of groups of genes conferring adaptive traits on the bacterium, including those involved in adaptation to an ecological niche or in pathogenicity (11, 24, 26). Normally, L. seeligeri is nonpathogenic, although it was reported that a case of human infection in a previously healthy adult presenting with acute purulent meningitis was due to L. seeligeri infection (30). This published result showed that L. seeligeri, previously assumed to be experimentally nonpathogenic for mice, may in fact be a heterogeneous species regarding its pathogenicity and thus may include strains that could cause life-threatening diseases in humans. Recently, a case of fatal bacteremia was reported to be caused by L. innocua, which had never been described in association with human disease (27). Hypothetically, the natural heterogeneity of avirulent Listeria species regarding their capability of acquiring virulence-associated genes may reflect their potential ability to be causative agents of diseases, especially in immunocompromised mammals.
How can these two species, L. seeligeri and L. welshimeri, which share a common nonhemolytic Xyl+ Rha− phenotype, be distinguished from each other? The 16S rRNA gene has been used extensively as a phylogenetic marker in bacterial taxonomy for evolutionary relationships due to its extremely slow rate of evolution, and the 16S-23S ITS has been selected for interspecies comparisons in the genus Listeria because it is not as conserved as the 16S rRNA gene (10, 13, 32, 33). However, it must be noted that rRNA phylogeny provides only low resolution among the six-species cluster of the genus Listeria since the 16S rRNA or 23S rRNA sequences of the respective species are almost identical (13, 33).
Multilocus sequence typing of housekeeping genes has recently become a fashionable way to characterize the population genetic structures of various bacterial species. Protein-encoding genes such as gyrB, prs, cpn-60, sigB, recA, ribC, hisC, addB, lisR, and ldh have been reported to evolve much faster than rrn operons, thus providing higher resolution and allowing comparisons between closely related species (13, 25, 28, 33). At the same time, the amino acid sequences are conservative enough to allow comparisons of taxa that are not closely related. In this study, the aforementioned housekeeping genes were included in the scheme of genetic identification of the atypically nonhemolytic L. seeligeri isolates. As a result, sequence profiles indicated that all seven atypically nonhemolytic L. seeligeri isolates should be considered a novel biotype of L. seeligeri. The occurrence of Listeria spp. with atypical phenotypic features is not surprising, but it could cause taxonomic misclassification of strains. Recently, a few strains of hemolysis-positive L. innocua were described which were surprisingly difficult to identify to the species level due to contradictory results in standard confirmatory tests (16). In contrast to the atypical L. seeligeri isolates described here, those atypical L. innocua strains contained all the members of the PrfA-regulated virulence gene cluster of L. monocytogenes.
This study provides the first substantial documentation of naturally occurring nonhemolytic strains of L. seeligeri, which can only be distinguished from Xyl+ Rha− L. welshimeri genotypically, not phenotypically. In contrast to this case, the Rha+ Xyl+ biotype of L. welshimeri would not present a potential identification issue unless it were a nonhemolytic L. seeligeri strain that had gained the capacity to ferment l-rhamnose. The strains of the nonhemolytic biotype of L. seeligeri are examples of natural species intermediates at the chromosomal level that seem to be virtually undocumented, in contrast to artificial strains among bacteria (1, 16). Strains like these nonhemolytic L. seeligeri isolates seem to be examples of intermediates in bacterial evolution. In conclusion, the current study, coupled with a previous study (16), illustrates how a taxonomic phenotypic property delineating an atypical strain of a Listeria sp. may not be just a simple reflection of structural changes concerning a single gene. The observed phenotypic property may represent just the tip of the iceberg of changes that also involve topologically and/or functionally related genes. Such atypical phenotypes, although complicating taxonomic identification, should contribute fruitful insights into the evolutionary events underlying the phylogeny of the genus Listeria.
We express our appreciation to Michael Klutch for his assistance in sequencing and to Paul Martin for providing cultures of Listeria welshimeri. We thank Jocelyne Rocourt for the helpful advice that initiated this study.
We thank Konstantin M. Chumakov and Vladimir E. Chizhikov for financial support of this study.
†Supplemental material for this article may be found at http://aem.asm.org/.