JCM Figure table search 04
Home Help [Feedback] [For Subscribers] [Archive] [Search] [Contents]
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nightingale, K. K.
Right arrow Articles by Wiedmann, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nightingale, K. K.
Right arrow Articles by Wiedmann, M.

 Previous Article  |  Next Article 

Journal of Clinical Microbiology, October 2006, p. 3742-3751, Vol. 44, No. 10
0095-1137/06/$08.00+0     doi:10.1128/JCM.00618-06
Copyright © 2006, American Society for Microbiology. All Rights Reserved.

Novel Method To Identify Source-Associated Phylogenetic Clustering Shows that Listeria monocytogenes Includes Niche-Adapted Clonal Groups with Distinct Ecological Preferences

K. K. Nightingale,1,2* K. Lyles,1 M. Ayodele,1 P. Jalan,3 R. Nielsen,4 and M. Wiedmann1

Department of Food Science, Cornell University, Ithaca, New York,1 Department of Animal Sciences, Colorado State University, Fort Collins, Colorado,2 Department of Biostatistics and Computational Biology, Cornell University, Ithaca, New York,3 Center for Bioinformatics and Institute of Biology, University of Copenhagen, Copenhagen, Denmark4

Received 22 March 2006/ Returned for modification 30 April 2006/ Accepted 27 May 2006


   ABSTRACT
Top
 Abstract
Introduction
 Materials and Methods
 Results and Discussion
 References
 
While phylogenetic and cluster analyses are often used to define clonal groups within bacterial species, the identification of clonal groups that are associated with specific ecological niches or host species remains a challenge. We used Listeria monocytogenes, which causes invasive disease in humans and different animal species and which can be isolated from a number of environments including food, as a model organism to develop and implement a two-step statistical approach to the identification of phylogenetic clades that are significantly associated with different source populations, including humans, animals, and food. If the null hypothesis that the genetic distances for isolates within and between source populations are identical can be rejected (SourceCluster test), then particular clades in the phylogenetic tree with significant overrepresentation of sequences from a given source population are identified (TreeStats test). Analysis of sequence data for 120 L. monocytogenes isolates revealed evidence of clustering between isolates from the same source, based on the phylogenies inferred from actA and inlA (P = 0.02 and P = 0.07, respectively; SourceCluster test). Overall, the TreeStats test identified 10 clades with significant (P < 0.05) or marginally significant (P < 0.10) associations with defined sources, including human-, animal-, and food-associated clusters. Epidemiological and virulence phenotype data supported the fact that the source-associated clonal groups identified here are biologically valid. Overall, our data show that (i) the SourceCluster and TreeStats tests can identify biologically meaningful source-associated phylogenetic clusters and (ii) L. monocytogenes includes clonal groups that have adapted to infect specific host species or colonize nonhost environments.


   INTRODUCTION
Top
 Abstract
 Introduction
Materials and Methods
 Results and Discussion
 References
 
Recent advances in DNA sequencing technologies have made DNA sequence-based subtyping methods such as multilocus sequence typing (MLST) widely available and affordable (2, 4), and as a result, a rich source of data for evolutionary and population genetics analyses of pathogenic and nonpathogenic bacteria is available (30). Traditional phylogenetic methods are frequently used to re-create the evolutionary history of a group of bacterial isolates and identify clonal groups of isolates that share a recent common ancestor. These methods also allow preliminary identification of clonal groups that show obvious associations with specific ecological niches, hosts, or disease symptoms in affected hosts. For example, MLST studies with Campylobacter jejuni showed that some clonal complexes within this species are exclusively or predominantly associated with specific host populations, including clonal complexes that were associated with human hosts and specific animal host species, such as sheep or poultry (5, 6). Similarly, Luan et al. (13) used MLST analysis to describe a specific clonal complex among invasive group B Streptococcus isolates which was associated with neonatal infections but which rarely caused invasive disease in adults. A recent MLST study of Candida albicans also showed that isolates from different infection sites varied in their distribution among phylogenetic clades (29), indicating that clonal complexes in this fungal pathogen may have evolved to demonstrate tissue specificity in infected hosts. While previous studies have used traditional statistical and phylogenetic methods to make inferences about the distribution of isolates within a phylogeny, these studies usually evaluated only specific clades which showed exclusive or almost exclusive associations with isolation from a particular niche. There is thus a clear need for the development of an unbiased statistical approach to assess the significance of phylogenetic clustering between isolates from the same source as a first step for the identification of niche-adapted clonal groups.

Listeria monocytogenes, a facultative intracellular pathogen that may cause severe invasive infections in humans and more than 40 species of animal hosts (25, 32), was chosen as a model organism for the development and implementation of a novel two-step statistical approach for the unbiased identification of phylogenetic clades that are significantly associated with different source populations, i.e., humans, animals, and food. L. monocytogenes was chosen as a model since it not only causes human and animal food-borne infections but also is commonly isolated from different environmental sources (e.g., soil, surface water, vegetation, and manure) and from food (7, 8). Consequently, a number of epidemiological studies have previously been performed to identify the L. monocytogenes subtypes and clonal groups that may differ in their ability to cause disease (9, 19, 20, 33). In general, the results of these previous studies supported the finding that L. monocytogenes represents two major genetic lineages (termed lineages I and II) and a third minor lineage (lineage III) that appear to differ in their abilities to cause human disease (3, 9, 21, 33). L. monocytogenes isolates representing lineage I, which includes serotypes 1/2b, 4b, and 3b (17), are more prevalent among human clinical isolates (9, 11, 16, 31), while lineage II, which includes isolates belonging to serotypes 1/2a, 1/2c, and 3a (17), even though they are regularly isolated from human clinical cases, are overrepresented among L. monocytogenes strains isolated from food (9, 31). Lineage III includes isolates belonging to serotypes 4a, 4b, and 4c; and animal clinical isolates are overrepresented among this third rare lineage (11, 17, 31, 33). Modeling of the dose-response relationships for these L. monocytogenes lineages also supported the finding that lineage I strains show higher levels of virulence for humans than lineage II strains (3). Previous studies (9, 20, 33) performed by use of a tissue culture plaque assay also showed that lineage I isolates, on average, formed larger plaques than lineage II isolates, indicative of an enhanced ability of lineage I isolates to spread intracellularly between host cells.

While previous studies clearly indicate that the genetic lineages within L. monocytogenes differ in their virulence characteristics and their association with different source populations (e.g., human, animal, and food), each L. monocytogenes lineage contains considerable subtype diversity and subtypes that are associated with isolation from specific source populations (9). Further studies are thus needed to probe the evolution of niche adaptation and to define biologically meaningful clonal groups within the major L. monocytogenes genetic lineages. Specifically, there is a clear need to differentiate L. monocytogenes clonal groups that are adapted to infect humans and that may have virulence enhanced over that of clonal groups that are adapted to colonize the environment and that may have limited virulence. We thus used L. monocytogenes as a model organism to (i) develop and implement a novel statistical approach for the unbiased identification of phylogenetic clades that are significantly associated with a particular source population (i.e., humans, animals, or food) and (ii) validate the biological significance of clades associated with specific source populations using tissue culture pathogenicity data and epidemiological information.


   MATERIALS AND METHODS
Top
 Abstract
 Introduction
 Materials and Methods
Results and Discussion
 References
 
Bacterial isolates. A set of 120 geographically and temporally matched L. monocytogenes isolates from human (n = 60) and animal (n = 30) clinical cases as well as food samples (n = 30) was used in this study. These isolates were selected from a larger set of L. monocytogenes isolates collected in New York State between January 1999 and December 2001 (n = 354), as described in detail in a previous study (18). All isolates were previously characterized by EcoRI ribotyping (24) and were assigned to one of three genetic lineages as described by Wiedmann et al. (33). In addition, all 120 isolates were previously characterized by an MLST scheme that included partial sequencing of actA, inlA, gap, prs, purM, ribC, and sigB (18). A detailed description from a previous study (18) of the isolate set investigated here (including isolate source, serotype, EcoRI ribotype, allelic types for all seven genes, and MLST type) is available elsewhere (http://www.foodscience.cornell.edu/wiedmann/Nightingale%20Supplementary.txt). DNA sequence data for these 120 L. monocytogenes isolates are available for download at www.pathogentracker.com. From the seven genes sequenced in our previous MLST study (18), we selected partial sequences for the two virulence genes (actA and inlA) and a concatenated sequence composed of partial sequences for the housekeeping genes gap and prs and stress response gene sigB to evaluate niche adaptation in L. monocytogenes. These two housekeeping genes and the stress response gene were previously shown to have evolved primarily by point mutation and thus provide a data set suitable as a reliable probe for L. monocytogenes niche adaptation at the core genetic level (18). A concatenated housekeeping and stress response gene sequence rather than individual gene sequences were used here, as the concatenated sequence provided a more robust phylogeny due to the limited number of polymorphisms found in each individual gene (18). The virulence genes actA and inlA were selected because we expected that these genes would be the most likely to show evidence of adaptive evolution, as these genes play key roles in host-pathogen interactions (12).

Frozen stocks of L. monocytogenes isolates were maintained at –80°C in brain heart infusion (Difco, Detroit, MI) broth with 15% (wt/vol) glycerol.

Mouse cell plaque assay. The in vitro virulence phenotype of all 120 L. monocytogenes isolates was evaluated by a plaque assay with mouse L cells, which was performed essentially as described previously (9, 26, 27). Briefly, L. monocytogenes isolates were grown in brain heart infusion broth overnight (18 h) at 30°C without shaking. Overnight bacterial cultures (1 ml) were pelleted and resuspended in phosphate-buffered saline (PBS; pH 7.4), and 1:10 serial dilutions were performed in PBS. Mouse L cells were grown to confluence in treated flat-bottom tissue culture six-well plates (Corning; Acton, MA), and approximately 1.5 x 105 or 4.0 x 106 CFU of a given L. monocytogenes isolate was inoculated into one well containing an L-cell monolayer. The lineage II, EcoRI ribotype DUP-1030A, standard laboratory control strain 10403S (1) was included as an internal control in each plaque assay. Plaques were visualized by staining with neutral red (Sigma Chemical, St. Louis, MO), as described previously (9, 26, 27), and images of infected L cells were captured with a digital scanner (Perfection 1650; Epson, Long Beach, CA). The area of approximately 25 plaques, selected to represent each L. monocytogenes isolate, was measured with SigmaScan Pro software (version 5.0; Statistical Solutions, Saugus, MA). The ability of each L. monocytogenes isolate to spread from cell to cell was expressed as the average plaque size for a given L. monocytogenes isolate relative to the average plaque size for 10403S, which was set equal to 100%. Two independent plaque assays were performed for all 120 L. monocytogenes isolates studied here.

Phylogenetic analysis. In a previous study (18), the 120 L. monocytogenes isolates described above were assigned sequence types (STs) on the basis of MLST analyses that included partial gene sequences for two key virulence genes (actA and inlA), a stress response gene (sigB), two hypervariable (15) housekeeping genes (purM and ribC), and two slowly evolving (18) housekeeping genes (gap and prs). While we previously reported phylogenetic trees inferred from a single isolate selected to represent each of the 52 unique STs (18), phylogenetic trees based on DNA sequence data for all 120 L. monocytogenes isolates were constructed here to allow the implementation of the novel statistical approach to detection of the same-source clustering described below.

MODELTEST (22) was used to optimize the input parameters to infer maximum-likelihood phylogenetic trees in PAUP* (28), based on partial sequences of two key virulence genes (i.e., actA and inlA) and a concatenated sequence representing two housekeeping genes (gap and prs) and the stress response gene sigB. Heuristic searches were performed by using equal weights for all sites, and the tree-bisection-reconnection branch-swapping algorithm was used. While unrooted maximum-likelihood phylogenies were used as inputs for the TreeStats test (described below), phylogenies were displayed here as rooted phylograms (see Fig. 1A to C). A homologous concatenated gene sequence from Bacillus subtilis (http://genolist.pasteur.fr/SubtiList/) was used as the outgroup for the concatenated housekeeping and stress response gene phylogeny (Fig. 1C), while the three lineage III sequences (i.e., FSL F2-525, FSL F2-655, and FSL-F2-695) were defined as the outgroups for the inlA and actA phylogenies (Fig. 1A and B), as described previously (18), because homologous B. subtilis sequences are not available for these sequences.


Figure 1
Figure 1
Figure 1
View larger version (78K):
[in this window]
[in a new window]
  		FIG.1. Maximum-likelihood phylogenetic trees inferred from (A) actA, (B) inlA, and (C) concatenated gap, prs, and sigB sequences for a sample set containing 60 human clinical (red), 30 animal clinical (blue), and 30 food (green) L. monocytogenes isolates. Taxon labels include the Cornell Food Safety Laboratory Culture Collection isolate name (e.g., F2655 represents isolate FSL F2-655), EcoRI ribotype (e.g., 44A represents ribotype DUP-1044A), and source (e.g., human isolate from New York State Department of Health [HS], human isolate from New York City Department of Health and Mental Hygiene [HC], animal isolate [AN], and food isolate [FD]). Isolates in clades that show significant or marginally significant associations with source populations (i.e., humans, animals, and food), as identified by the TreeStats test, are indicated by italics and are marked with a vertical bar; significance levels are indicated by * and ** (which represent P < 0.05 and P < 0.01, respectively) and # (which indicates marginally significant associations; P < 0.10). Clades that are significantly associated with source populations in at least one phylogeny and that contain the same or overlapping isolates are designated with the same letter (A to J) across all three phylogenies (Table 3); the predominant source population for each clade (i.e., human, animal, or food) is also indicated. The mean plaque size (PS) for isolates in each cluster as well as classification into ECs (as described by Kathariou [12]) is also indicated at each significant source-associated clade.

 
Identification of phylogenetic clades that are significantly associated with different source populations. To test if isolates from different source populations, i.e., humans, animals, and food samples, were evolutionarily related, we used a systematic two-step approach. We first tested the null hypothesis that the genetic distances among isolates within and between source populations were identical (SourceCluster test). If this null hypothesis can be rejected, then there is significant evidence for overall clustering between isolates from the same source population within a phylogeny, and specific clades with a significant overrepresentation of sequences from a given source population can be reliably identified (TreeStats test).

The SourceCluster test was performed by using uncorrected pairwise distance matrices (raw distance) generated with PAUP* software (28) from (i) actA sequences, (ii) inlA sequences, and (iii) a concatenated sequence of two housekeeping and a stress response gene (i.e., gap, prs, and sigB) for the 120 L. monocytogenes isolates described above. The SourceCluster test used the average genetic distance (raw distance) between sequences from the same source (i.e., human [H], animal [A], and food [F] samples) as a test statistic (T) calculated as follows:

Formula 1(1)
where dij is the raw genetic distance between isolates i and j, Si is the source of isolate i, and I(Si = k) is an indicator function that returns a value of 1 if isolate i is from source population k. Significance is determined by a procedure that permutes the labeling of the source population among isolates to generate 1,000 new data sets. For each resampled data set, a new value of T, T*, is calculated; and if T falls in the 5th percentile of the distribution of T*, the null hypothesis of no clustering between isolates from the same source population is rejected. The relative position of T in the distribution of T* was expressed as a P value, which was calculated as the number of T* values that exceeded T divided by the number of resamplings. For example, if 10 of 1,000 T* estimates were greater than T, the P value would be 0.01. The program SourceCluster is publicly available at http://www.foodscience.cornell.edu/wiedmann/SourceCluster.txt.

The TreeStats test was performed by using unrooted maximum-likelihood phylogenetic trees for the 120 L. monocytogenes isolates inferred from (i) actA sequences, (ii) inlA sequences, and (iii) a concatenated sequence of two housekeeping and a stress response gene (i.e., gap, prs, and sigB). In the TreeStats test, for each clade on a given phylogenetic tree, the expected number of isolates belonging to group A, F, and H, under the null hypothesis of no clustering between isolates from the same source population, was calculated by permuting the labeling of the leaf clades 10,000 times. A chi-square goodness-of-fit test was then used to compare the observed to the expected numbers for each clade on the tree. This procedure is implemented in the program TreeStats, which is publicly available at http://www.foodscience.cornell.edu/wiedmann/TreeStats.htm. Clades with P values <0.05 and <0.10 were considered to show statistically significant and marginally significant evidence for clustering among isolates belonging to the same source population, respectively. No correction for multiple testing was performed, since the SourceCluster test was used to first test the null hypothesis of no clustering between isolates from the same source population. Significant and marginally significant P values from TreeStats analysis were superimposed on the maximum-likelihood trees that were constructed and rooted with an appropriate outgroup as described above.

Statistical analysis. Chi-square tests of independence or Fisher's exact tests (if appropriate; i.e., if more than 25% of expected values in a table were less than five) were used to probe associations between L. monocytogenes source populations (i.e., human, animal, or food) and L. monocytogenes genotypes (i.e., genetic lineage and EcoRI ribotype). Human and animal clinical isolate were also pooled to create a categorical variable termed "host" to evaluate associations between this source population and L. monocytogenes genotypes. Categorical analyses were performed only for the EcoRI ribotypes that were observed at least four times among the 120 L. monocytogenes isolates studied here.

One-way analysis of variance was used to determine the relationship between the measure "plaque size" and the categorical variables "lineage," "source," and "ribotype." Analyses were performed only for ribotypes that were observed at least four times among the 120 isolates. All analyses were performed with Statistical Analysis Systems software (SAS Institute, Cary, NC). While P values of <0.05 were considered statistically significant, P values of <0.10 are also reported and were considered marginally significant.


   RESULTS AND DISCUSSION
Top
 Abstract
 Introduction
 Materials and Methods
 Results and Discussion
References
 
We combined MLST, epidemiology, and virulence phenotype data for 120 L. monocytogenes isolates from human and animal clinical cases as well as food samples to develop, implement, and biologically validate a novel two-step statistical approach for the unbiased identification of phylogenetic clades that are significantly associated with different source populations. Our results indicate (i) that the combined SourceCluster and TreeStats approach identifies biologically meaningful source-associated clonal groups, providing evidence for niche adaptation within the major L. monocytogenes genetic lineages, and (ii) that L. monocytogenes contains both host- and non-host-adapted clonal groups. The power of the combined SourceCluster and TreeStats approach for the identification of niche-adapted clonal groups is further supported by the fact that most of the source-associated clades identified in our study represent previously described subtypes with unique epidemiological or virulence characteristics, including (i) the major human epidemic clones (ECs) and (ii) subtypes with mutations that result in premature stop codons in the key L. monocytogenes virulence gene inlA, which are further associated with a significantly reduced ability to invade human intestinal epithelial cells (19). While it is well known that L. monocytogenes is an environmental pathogen (8), our data support the finding that clonal groups within this species may differ in their ability to cause human disease. Significant associations between clades in a given gene tree and a source population do not necessarily indicate that specific polymorphisms in a given gene are responsible for the source association. In fact, the evolution of linked genetic traits by different mechanisms (e.g., gene acquisition, gene loss, and polymorphisms within linked genes) may allow host tropism and niche adaptation.

The combined SourceCluster and TreeStats approach identifies biologically relevant phylogenetic clades, which provide evidence for niche adaptation within L. monocytogenes lineages. Initial categorical analyses showed that the distribution of the 120 L. monocytogenes isolates studied here among source populations, genetic lineages, and EcoRI ribotypes was consistent with that found in previous studies, which were often based on larger L. monocytogenes isolate sets (9, 11, 16, 31). Specifically, L. monocytogenes lineage I was marginally overrepresented (P = 0.07) among isolates from human listeriosis cases, while lineage II was significantly (P = 0.04) overrepresented among food isolates (Table 1). Determination and analyses of mean plaque sizes, a measure for the ability of a given L. monocytogenes strain to spread intracellularly between mammalian host cells, indicated that isolates representing lineage II produced significantly (P < 0.0001) smaller plaques than isolates belonging to lineages I and III and, similarly, that isolates belonging to lineage I formed significantly (P < 0.01) larger plaques than lineage II and III isolates (Table 2), also consistent with previous studies (9, 20, 33). Interestingly, L. monocytogenes isolated from food samples also formed significantly (P = 0.03) smaller plaques than human and animal clinical isolates (Table 2). Since isolates that form larger plaques have accelerated rates of intracellular spread (26), these data further suggest that lineage I isolates are more virulent than those belonging to lineage II and, similarly, that isolates from clinical cases of listeriosis (i.e., both human and animal cases) are typically more virulent than isolates from food, providing preliminary evidence for niche adaptation within L. monocytogenes. Our analyses also support the finding that the collection of 120 L. monocytogenes isolates studied here is representative of the previously reported genetic and phenotypic diversity for this pathogen and is thus appropriate for use for the development and biological validation of a new statistical approach to the identification of the phylogenetic clades associated with specific source populations.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Distribution of L. monocytogenes molecular subtypes among isolate sources

 

View this table:
[in this window]
[in a new window]
  		TABLE 2. Summary of associations between plaque size and categorical variables

 
In order to allow the unbiased identification of phylogenetic clades that are associated with different sources populations, rather than using clades defined a priori (e.g., the major L. monocytogenes genetic lineages), we developed a two-step statistical approach to the analysis of phylogenetic trees constructed for the 120 L. monocytogenes isolates included in our study for evidence of clustering between L. monocytogenes isolates from a given source population (i.e., humans, animals, and food). The SourceCluster test was first used to test the null hypothesis that the genetic distances for isolates within and between source populations were identical. Results from SourceCluster analysis indicated that the genetic distances for isolates within and between source populations were not identical, based on actA sequences (P = 0.02), providing overall evidence of clustering between isolates from the same source. While marginally significant evidence for overall clustering of L. monocytogenes isolates from the same source population (P = 0.07) was observed for inlA sequences, no evidence for overall clustering between isolates from same source population (P = 0.29) was found on the basis of the concatenated housekeeping and stress response gene sequence data. Subsequent analyses by use of the TreeStats test identified a number of clades within each phylogeny that showed significant (P < 0.05) clustering of isolates from a specific source population. Importantly, the results from the TreeStats analyses found that L. monocytogenes lineages I and II represent clades that show significant associations with specific source populations (P < 0.05 for inlA and housekeeping gene-derived phylogenies and P < 0.10 for the actA phylogeny; Fig. 1A to C). These findings not only are consistent with our plaque assay data and the results of the categorical analyses, which show that lineages I and II differ in their distributions among source populations and in their ability to spread from cell to cell, but are also consistent with the findings of previous prevalence studies (9, 11, 16, 31) that showed that human clinical isolates are overrepresented among lineage I isolates, while lineage II isolates are more commonly isolated from food. While one might expect a stronger association between source population and lineages in the actA phylogeny, particularly since the lineages differ in their ability to spread from cell to cell, as indicated by the average plaque sizes (9), variation in virulence genes other than actA (e.g., prfA) or ActA amino acid residues outside the region sequenced here may also have contributed to cell-to-cell-spread phenotype. The strong association between the source population and lineage in the inlA tree suggests that variation in InlA-mediated attachment and invasion is fundamental for niche adaptation in L. monocytogenes and is consistent with the fact that naturally occurring polymorphisms in inlA (e.g., nonsense mutations that lead to premature stop codons) have a significant effect on the invasiveness of L. monocytogenes for human intestinal epithelial cells (19). It is, however, important to recognize that significant associations of clades in a given gene tree with a specific source population do not necessarily indicate that the given gene is solely responsible for the source association. Analyses of additional virulence gene phylogenies (e.g., phylogenies based on other internalin genes, such as inlB) as well as analysis of potentially linked gene indels or islands (e.g., through microarray analysis) are likely to provide further insight into the genetic mechanisms underlying niche adaptation in L. monocytogenes.

Implementation of SourceCluster and TreeStats tests can identify biologically meaningful same-source clades in phylogenetic trees. Because the SourceCluster and TreeStats tests identified the same, biologically validated, source-associated lineages as previous studies (9, 11, 16, 31), this approach has the potential to identify specific clades representing clonal groups that may have common biologically relevant characteristics, consistent with niche adaptation within the major genetic lineages. Specifically, the TreeStats test identified eight, six, and five clades with significant (P < 0.05) source associations in the actA (Fig. 1A), inlA (Fig. 1B), and concatenated housekeeping and stress response gene (Fig. 1C) phylogenies, respectively. In addition, one clade each in the actA and the concatenated housekeeping and stress response gene phylogeny as well as three clades in the inlA phylogeny showed marginal (P < 0.10) evidence for associations with specific source populations. Overall, a total of 10 clades covering the same or overlapping isolates showed significant associations with specific source populations across the three phylogenies analyzed here, and all 10 of these clades were identified independently in at least two phylogenies (Table 3). For example, clade F (Table 3; Fig. 1) was identified as a significant (P < 0.05) source-associated clade in all three phylogenies. Additionally, each of the 10 clades listed in Table 3 was identified as a significant (P < 0.05) source-associated clade in at least one phylogeny. For example, while clade A was identified only as a marginally significant (P < 0.10) source-associated clade in the inlA phylogeny, this clade was significantly (P < 0.05) overrepresented by isolates obtained from human clinical cases in the actA phylogeny (Fig. 1). While we appreciate that the source-associated clades identified by TreeStats analysis of the phylogeny inferred from the concatenated housekeeping and stress response gene sequence should be interpreted carefully, since the overall SourceCluster statistic was not significant, the overlap between significant clades identified in the different trees supports the robustness of the TreeStats approach, particularly when it is applied to pathogens with a highly clonal population structure, such as L. monocytogenes (18).


View this table:
[in this window]
[in a new window]
  		TABLE 3. Description of significant same-source isolate clusters in L. monocytogenes phylogenies inferred from key virulence genes (actA and inlA) and a concatenated housekeeping gene sequence composed of gap, prs, and sigB

 
Of the 10 clades that were identified as being significantly associated with different source populations, three were characterized by an overrepresentation of isolates from human clinical cases, including two clades within lineage I (clades A and C) and one clade within lineage II (clade G) (Fig. 1A and B and Table 3). Interestingly, all three of these clades not only correlate with previously identified human ECs ( 12), including ECI (clade A and C) and ECIII (clade G), but also correlate with EcoRI ribotypes previously identified to be associated with human listeriosis epidemics (9, 11), including ribotypes DUP-1038B (clade A), DUP-1042B (clade C), and DUP-1053A (clade G). Our results also suggest that ECI represents two distinct clones (clades A and C), as supported by previous reports that ECI represents two distinct ribotypes (9). Interestingly, L. monocytogenes isolates classified into clades A and C showed some of the largest mean plaque sizes observed among all 10 of the source-associated clades identified here (Fig. 1A and B and Table 3), consistent with the observation that isolates of the ribotypes associated with these clades (DUP-1038B and DUP-1042B) also formed significantly (P < 0.05) larger plaques than isolates belonging to other ribotypes (Table 2). Our findings not only further support the biological significance of the human-associated clades identified here, but they also indicate that different clades associated with human clinical cases may show distinct phenotypic characteristics, as isolates in the human-associated clade G (which belongs to lineage II and ribotype DUP-1053A) produced relatively smaller (91.0%) plaques (Table 3). Interestingly, lineage II ribotype DUP-1053A isolates appear to be characterized by enhanced invasiveness for human intestinal epithelial Caco-2 cells (K. Nightingale and M. Wiedmann, unpublished data) compared to the invasiveness of the standard lineage II laboratory control strain 10403S, providing a potential biological explanation for their association with human hosts. These findings support the importance of future studies on the host-associated clades identified here and serve as a reminder that while in vitro assays such as the plaque assay used here may provide useful information on the virulence characteristics of L. monocytogenes isolates, virulence for humans cannot be fully measured by an in vitro assay.

Overall, a total of three clades were significantly associated with isolation from animal clinical cases, including one clade within lineage I (clade B) and two clades within lineage II (clades E and H) (Table 3). Interestingly, isolates in clade B predominantly represent ribotype DUP-1042B, which, on average, shows an enhanced ability to spread intracellularly between host cells (Table 2). Our data suggest that DUP-1042B isolates represent two distinct clonal groups that may have adapted to infect specific host species. This observation is further supported by the fact that DUP-1042B isolates in the human- and animal-associated clades (clades C and B, respectively) represent distinct serotypes, with isolates in clades C and B belonging to serotypes 4b and 1/2b, respectively, consistent with the fact that most human listeriosis epidemics are caused by serotype 4b strains (14). While isolates in clades E and H do not show any apparent phenotypic or epidemiological characteristics that explain their association with animal clinical cases, further characterization of isolates belonging to these clades may provide new insights into the evolution of host specificity in L. monocytogenes.

Overall, a total of four clades were significantly associated with isolation from contaminated food, including two clades within lineage I (clades D and J) and two clades belonging to lineage II (clades F and I) (Table 3). Interestingly, isolates in clade F belong to three L. monocytogenes ribotypes (i.e., DUP-1062A, DUP-1039C, and DUP-1046B) previously shown to carry nonsense mutations that lead to premature stop codons in the virulence gene inlA, express a truncated and secreted form of InlA, and demonstrate attenuated invasiveness in Caco-2 cells (19; R. Orsi and M. Wiedmann, unpublished data). We recently identified two similar additional mutations that lead to premature stop codons in inlA in isolates belonging to ribotypes DUP-1045B, DUP-1062D, and DUP-1048B, which also demonstrated attenuated invasion of Caco-2 cells (Nightingale and Wiedmann, unpublished), supporting the finding that the second lineage II food-associated clade identified here, clade I, may also represent a human virulence-attenuated clonal group. These findings not only provide a clear biological explanation for the association of clades F and I isolates with food but also are consistent with the findings of a previous study (19), which found that isolates with premature stop codons in inlA are significantly underrepresented among human clinical isolates compared to their prevalence in food. On very rare occasions, L. monocytogenes isolates containing premature inlA stop codons have been linked to human disease (<2% of more than 1,000 human listeriosis cases), suggesting that these strains may have the potential to cause disease in extremely immunosuppressed individuals (19). Isolates in clades I and F also showed, on average, a smaller plaque size (Table 3), indicative of a reduced ability to spread intracellularly in mammalian hosts cells (Table 3), further supporting the hypothesis that isolates in these clades may represent virulence-attenuated clonal groups. Interestingly, ribotype DUP-1042C isolates belonging to food-associated clade D also formed smaller (92.9%) plaques (Table 3). The classification of clade D as a food-associated clade is also consistent with the results from a previous study (9), which included a larger number of DUP-1042C isolates (n = 14) and which showed that DUP-1042C isolates were significantly more common among food isolates and were not observed among nearly 500 human clinical isolates. Some DUP-1042C isolates from food were also recently found to be characterized by attenuated invasiveness in Caco-2 cells (Nightingale and Wiedmann, unpublished), providing a possible biological explanation for their classification into a food-associated clade. Interestingly, isolates in food-associated clade J, on average, formed large (120%) plaques (Table 3), further indicating that different clades associated with food may show distinct phenotypic characteristics. Future studies are thus required to fully understand the biological underpinnings for the observed sources associations of specific clades.

Conclusions. While a number of previous studies have provided preliminary evidence that the main L. monocytogenes lineages (i.e., lineages I and II) differ in their ability to cause human disease, only a few clonal groups with specific host or niche preferences (e.g., epidemic clones and virulence-attenuated strains with inlA mutations) have been described (10, 12, 19, 23). Our findings not only support the finding that previously described epidemic clones and virulence-attenuated strains represent distinct clades within L. monocytogenes but also identified a number of additional clades that are significantly associated with specific source populations and may thus represent host- or niche-adapted ecotypes. Most importantly, our data, obtained by using L. monocytogenes as a model system, show that a novel two-step statistical approach (SourceCluster and TreeStats analyses) for the unbiased identification of phylogenetic clades that are significantly associated with particular source populations can reliably identify biologically relevant clonal groups that may represent niche-adapted L. monocytogenes ecotypes. This approach will thus provide a valuable set of tools for the characterization of other bacterial pathogens as well as nonpathogenic bacteria, allowing the rapid identification of meaningful subtypes and clades that differ in their relevant phenotypic characteristics, without requiring considerable information a priori on the biology of the organism being characterized. As larger sequence data sets for L. monocytogenes and other pathogens become available, this approach may also provide an opportunity to identify associations between clades and more narrowly defined hosts and niches (e.g., specific animal species and environments).


   ACKNOWLEDGMENTS
 
This work was supported by National Institutes of Health (NIH) grant R01GM63259 awarded to M. Wiedmann.

We thank Qi Sun (Computational Biology Service Unit, Cornell University) for his expertise in setting up the computing system to perform evolutionary analyses. We also thank Wan-Lin Su for helpful discussions. We are also indebted to Esther Fortes and Alphina Ho for their technical assistance with DNA sequencing.


   FOOTNOTES
 
* Corresponding author. Mailing address: Colorado State University, Department of Animal Sciences, 108B Animal Science Building, Fort Collins, CO 80523-1171. Phone: (970) 491-1556. Fax: (970) 491-5326. E-mail: kendra.nightingale{at}colostate.edu. Back


   REFERENCES
Top
 Abstract
 Introduction
 Materials and Methods
 Results and Discussion
 References
 

  1. Bishop, D. K., and D. J. Hinrichs. 1987. Adoptive transfer of immunity to Listeria monocytogenes: the influence of in vitro stimulation on lymphocyte subset requirements. J. Immunol. 139:2005-2009.[Abstract]
  2. Cai, S., D. Y. Kabuki, A. Y. Kuaye, T. G. Cargioli, M. S. Chung, R. Nielsen, and M. Wiedmann. 2002. Rational design of DNA sequence-based strategies for subtyping Listeria monocytogenes. J. Clin. Microbiol. 40:3319-3325.[Abstract/Free Full Text]
  3. Chen, Y., W. H. Ross, M. J. Gray, M. Wiedmann, R. C. Whiting, and V. N. Scott. 2006. Attributing risk to L. monocytogenes subgroups: dose response in relation to genetic lineages. J. Food Prot. 69:335-344.[Medline]
  4. Clarke, S. C. 2002. Nucleotide sequence-based typing of bacteria and the impact of automation. BioEssays 24:858-862.[CrossRef][Medline]
  5. Colles, F. M., K. Jones, R. M. Harding, and M. C. J. Maiden. 2003. Genetic diversity of Campylobacter jejuni isolates from farm animals and the farm environment. Appl. Environ. Microbiol. 69:7409-7413.[Abstract/Free Full Text]
  6. Dingle, K. E., F. M. Colles, R. Ure, J. A. Wagenaar, B. Duim, F. J. Bolton, A. J. Fox, D. R. A. Wareing, and M. C. J. Maiden. 2002. Molecular characterization of Campylobacter jejuni clones: a basis for epidemiological investigation. Emerg. Infect. Dis. 8:949-955.[Medline]
  7. Farber, J. M., and P. I. Peterkin. 1991. Listeria monocytogenes, a food-borne pathogen. Microbiol. Rev. 55:476-511.[Abstract/Free Full Text]
  8. Fenlon, D. R. 1999. Listeria monocytogenes in the natural environment, p. 21-39. In E. T. Ryser and E. H. Marth (ed.), Listeria listeriosis and food safety, 2nd ed. rev. and expanded. Marcel Decker, Inc., New York, N.Y.
  9. Gray, M. J., R. N. Zadoks, E. D. Fortes, B. Dogan, S. Cai, Y. Chen, V. N. Scott, D. E. Gombas, K. J. Boor, and M. Wiedmann. 2004. Food and human isolates of Listeria monocytogenes form distinct but overlapping populations. Appl. Environ. Microbiol. 70:5833-5841.[Abstract/Free Full Text]
  10. Jacquet, C., M. Doumith, J. I. Gordon, P. M. V. Martin, P. Cossart, and M. Lecuit. 2004. A molecular marker for evaluating the pathogenic potential of foodborne Listeria monocytogenes. J. Infect. Dis. 189:2094-2100.[CrossRef][Medline]
  11. Jeffers, G. T., J. L. Bruce, P. L. McDonough, J. Scarlett, K. J. Boor, and M. Wiedmann. 2001. Comparative genetic characterization of Listeria monocytogenes isolates from human and animal listeriosis cases. Microbiology 147:1095-1104.[Abstract/Free Full Text]
  12. Kathariou, S. 2002. Listeria monocytogenes virulence and pathogenicity, a food safety perspective. J. Food Prot. 11:1811-1829.
  13. Luan, S., M. Granlund, M. Sellin, T. Lagergard, B. G. Spratt, and M. Norgren. 2005. Multilocus sequence typing of Swedish invasive group B Streptococcus isolates indicates a neonatally associated genetic lineage and capsule switching. J. Clin. Microbiol. 43:3727-3733.[Abstract/Free Full Text]
  14. McLauchlin, J. 1990. Distribution of serovars of Listeria monocytogenes isolated from different categories of patients with listeriosis. Eur. J. Clin. Microbiol. Infect. Dis. 9:210-213.[CrossRef][Medline]
  15. Meinersmann, R. J., R. W. Phillips, M. Wiedmann, and M. E. Berrang. 2004. Multilocus sequence typing of Listeria monocytogenes by use of hypervariable genes reveals clonal and recombination histories of three lineages. Appl. Environ. Microbiol. 70:2193-2203.[Abstract/Free Full Text]
  16. Mereghetti, L., P. Lanotte, V. Savoye-Marczuk, N. Marquet-Van Der Mee, A. Audurier, and R. Quentin. 2002. Combined ribotyping and random multiprimer DNA analysis to probe the population structure of Listeria monocytogenes. Appl. Environ. Microbiol. 68:2849-2857.[Abstract/Free Full Text]
  17. Nadon, C. A., D. L. Woodward, C. Young, F. G. Rodgers, and M. Wiedmann. 2001. Correlations between molecular subtyping and serotyping of Listeria monocytogenes. J. Clin. Microbiol. 39:2704-2707.[Abstract/Free Full Text]
  18. Nightingale, K. K., K. Windham, and M. Wiedmann. 2005. Evolution and molecular phylogeny of Listeria monocytogenes from human and animal cases and food. J. Bacteriol. 187:5537-5551.[Abstract/Free Full Text]
  19. Nightingale, K. K., K. Windham, K. E. Martin, M. Yeung, and M. Wiedmann. 2005. Selected Listeria monocytogenes subtypes commonly found in food show reduced invasion in human intestinal cells due to distinct nonsense mutations in inlA leading to expression of truncated and secreted internalin A. Appl. Environ. Microbiol. 12:8764-8772.
  20. Norton, D. M., J. M. Scarlett, K. Horton, D. Sue, J. Thimothe, K. J. Boor, and M. Wiedmann. 2001. Characterization and pathogenic potential of Listeria monocytogenes isolates from the smoked fish industry. Appl. Environ. Microbiol. 67:646-653.[Abstract/Free Full Text]
  21. Piffaretti, J. C., H. Kressebuch, M. Aeschbacher, J. Bille, E Bannerman J. M. Musser, R. K. Selander, and J. Rocourt. 1989. Genetic characterization of clones of the bacterium Listeria monocytogenes causing epidemic disease. Proc. Natl. Acad. Sci. USA 10:3818-3822.
  22. Posada, D., and K. A. Crandall. 1998. MODELTEST: testing the model of DNA substitution. Bioinformatics 14:817-818.[Abstract/Free Full Text]
  23. Rousseaux, S., M. Olier, J. P. Lamaitre, P. Piveteau, and J. Guzzo. 2004. Use of PCR-restriction fragment polymorphism of inlA for rapid screening of Listeria monocytogenes strains deficient in the ability to invade Caco-2 cells. Appl. Environ. Microbiol. 70:2180-2185.[Abstract/Free Full Text]
  24. Sauders, B. D., K. Mangione, C. Vincent, J. Schermerhorn, C. M. Farchione, N. B. Duman, D. Bopp, L. Kornstein, E. D. Fortes, K. Windham, and M. Wiedmann. 2004. Distribution of Listeria monocytogenes molecular subtypes among human and food isolates from New York State shows persistence of human-disease associated Listeria monocytogenes stubypes in retail environments. J. Food Prot. 67:1417-1428.[Medline]
  25. Schlech, W. F. 2000. Foodborne listeriosis. Clin. Infect. Dis. 31:770-775.[CrossRef][Medline]
  26. Smith, G. A., J. A. Theriot, and D. A. Portnoy. 1996. The tandem repeat domain in the Listeria monocytogenes ActA protein controls the rate of actin-based mobility, the percentage of moving bacteria, and the localization of vasodilator-stimulated phosphoprotein and profilin. J. Cell Biol. 135:647-660.[Abstract/Free Full Text]
  27. Sun, A. N., A Camilli, and D. A. Portnoy. 1990. Isolation of Listeria monocytogenes small-plaque mutants defective for intracellular growth and cell-to-cell spread. Infect. Immun. 58:3770-3778.[Abstract/Free Full Text]
  28. Swofford, D. 1997. Phylogenetic analysis using parsimony (*and other methods). Sinauer Associates, Sunderland, Mass.
  29. Tavanti, A., A. D. Davisdon, M. J. Fordyce, N. A. R. Gow, M. C. J. Maiden, and F. C. Odds. 2005. Population structure and properties of Candida albicans, as determined by multilocus sequence typing. J. Clin. Microbiol. 43:5601-5613.[Abstract/Free Full Text]
  30. Urwin, R., and M. C. J. Maiden. 2003. Multi-locus sequence typing: a tool for global epidemiology. Trends Microbiol. 11:479-487.[CrossRef][Medline]
  31. Ward, T. J., L. Gorski, M. K. Borucki, R. E. Mandrell, J. Hutchins, and K. Pupedis. 2004. Intraspecific phylogeny and lineage group identification based on the prfA virulence gene cluster of Listeria monocytogenes. J. Bacteriol. 15:4994-5002.
  32. Wesley, I. V. 1999. Listeriosis in animals, p. 39-73. In E. T. Ryser and E. H. Marth (ed.), Listeria listeriosis and food safety, 2nd ed. rev. and expanded. Marcel Decker, Inc., New York, N.Y.
  33. Wiedmann, M., J. L. Bruce, C. Keating, A. E. Johnson, P. L. McDonough, and C. A. Batt. 1997. Ribotypes and virulence gene polymorphisms suggest three distinct Listeria monocytogenes lineages with differences in pathogenic potential. Infect. Immun. 65:2707-2716.[Abstract]

Journal of Clinical Microbiology, October 2006, p. 3742-3751, Vol. 44, No. 10
0095-1137/06/$08.00+0     doi:10.1128/JCM.00618-06
Copyright © 2006, American Society for Microbiology. All Rights Reserved.



This article has been cited by other articles:


This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nightingale, K. K.
Right arrow Articles by Wiedmann, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nightingale, K. K.
Right arrow Articles by Wiedmann, M.

Home Help [Feedback] [For Subscribers] [Archive] [Search] [Contents]
Antimicrob. Agents Chemother. Clin. Microbiol. Rev.
Clin. Vaccine Immunol. ALL ASM JOURNALS