Principles of Bacterial Pathogenesis

By Eduardo A. Groisman

Principles of Bacterial Pathogenesis presents a molecular perspective on a select group of bacterial pathogens by having the leaders of the field present their perspective in a clear and authoritative manner. Each chapter contains a comprehensive review devoted to a single pathogen. Several chapters include work from authors outside the pathogenesis field, providing general perspectives on the evolution, regulation, and secretion of virulence and determinants.

• Explains the basic principles of bacterial pathogenesis
• Covers diverse aspects integrating regulation, cellular microbiology and evolution of microbial disease of humans
• Discusses current strategies for the identification of virulence determinants and the methods used by microbes to deliver virulence factors
• Presents authoritative treatises of the major disease microorganisms

• Language: English
• Publisher: Academic Press
• Release date: Jan 9, 2001
• ISBN: 9780080539584

Book preview

Sweden

Preface

Infectious diseases are the leading cause of morbidity and mortality worldwide. While some of these infections are caused by eukaryotic parasites, bacterial pathogens continue to present a threat to the well-being of humans and animals both in the developing and developed worlds. The use of vaccines and antibiotics, together with changes in sanitary practices, has contributed to an important increase in the life span of humans in the last century. However, these significant improvements are now challenged by the appearance of microbes resistant to multiple antibiotics, the emergence of new bacterial pathogens, and the use of health care treatments that, while prolonging life, render individuals susceptible to opportunistic pathogens. Then, what can be done to sustain the improvements in health care developed during the last 100 years?

Novel strategies are currently being tested to prevent and/or treat bacterial infections. An increasing number of these strategies are based on our understanding of the mechanisms by which pathogenic microorganisms cause disease. This is possible due to exciting developments in the field of bacterial pathogenesis, which started some 20 years ago with the use of molecular genetics to investigate the microorganisms responsible for causing disease, and are now complemented with cell biological and biochemical approaches aimed at unraveling the consequences that infection by such microorganisms has on their hosts. We now have a basic understanding not only of the varied nature of virulence determinants but also of their origin and acquisition by pathogenic microbes. We appreciate that expression of virulence determinants is most often regulated in response to host signals and that microbes use different devices to deliver toxic products to host cells. These studies have revealed a set of principles that govern bacterial pathogenesis and, as the title indicates, constitutes the subject matter of this book.

The motivation for Principles in Bacterial Pathogenesis was twofold: first, to provide in-depth coverage of the best-characterized bacterial pathogens, with the goal of uncovering the salient features that these microbes have in common which allow them to conquer new niches and to circumvent host defense mechanisms, and, second, to group contributions by the world experts in bacterial pathogenesis in which they present a general discussion of the subject beyond the work performed in their own laboratories.

This book is divided in two parts that comprise a total of 16 chapters, each of which can be read independent of the rest. The first part consists of five chapters, three of which discuss aspects of bacterial pathogenesis that are common to all pathogens: evolution, secretion, and regulation of virulence determinants. The fourth chapter presents a thorough description of the strategies currently used to identify virulence determinants. The fifth chapter discusses bacterial pathogens of plants, highlighting the similar mechanisms that bacterial pathogens of animal and plants employ when interacting with their respective hosts. These first 5 chapters serve as a general introduction to the 11 pathogen-based chapters that comprise the second part of the book. Each of the latter chapters provides a broad discussion of the best-understood human pathogens.

In sum, while novel aspects of pathogenic organisms will continue to be discovered, a basic understanding of the principles governing bacterial pathogenesis will not only allow us to appreciate the sophisticated mechanisms used by microbes in their pathogenic lifestyle, but will also be essential in beginning to understand the plethora of information emerging from genomics, and to develop new rational approaches to the treatment and prevention of infectious diseases.

Eduardo A. Groisman,     Department of Molecular Microbiology, Howard Hughes Medical Institute, Washington University School of Medicine, St. Louis. Missouri

CHAPTER 1

Evolution of Bacterial Pathogens

HOWARD OCHMAN

I. Introduction

II. The Genetic Basis of Virulence

III. Identification of Sequences Involved in Bacterial Pathogenesis

IV. Recovery of Genes Contributing to Virulence

V. The Population Genetics of Pathogens

VI. Studying Bacterial Population Genetics

A. Multilocus Enzyme Electrophoresis

B. DNA Sequencing

C. Multilocus Sequence Typing

VII. The Organization of Genetic Diversity in Pathogenic Microorganisms

VIII. Population Genetics of Representative Bacterial Pathogens

A. Bordetella

B. Borrelia

C. Escherichia coli and Shigella

D. Haemophilus

E. Helicobacter

F. Listeria

G. Mycobacterium

H. Neisseria

I. Salmonella

J. Staphylococcus

K. Streptococcus

L. Vibrio

IX. Conclusions

References

I.

Introduction

In what ways do pathogenic microorganisms differ from nonpathogenic forms? For an organism to be considered a pathogen, it must, during some phase of its life-cycle, advance disease and alter the health or behavior of another organism, that is, its host. Every organism can serve as a host for pathogens, and pathogens come into contact with a very large number of species; however, most pathogenic microorganisms are virulent to relatively few, and often only one, host species, and infection may cause disease in only a limited segment of the host population. So despite the wide range of mechanisms deployed by pathogens to disable their hosts (and to promote their own replication and transmission), there is one common theme: virulence depends upon the susceptibility of a host. Therefore, the identification of pathogens, the differences between pathogenic and nonpathogenic microorganisms, and the specific factors required for virulence must each be defined with regard to its relevance to the host.

This chapter addresses two general issues about the evolution of microbial pathogenesis. First, we consider the differences between the genomes of pathogenic and nonpathogenic bacteria, the specific types of genes that contribute to the virulence phenotype, and the evolutionary history of these sequences in the genome of pathogens. Next, we discuss the molecular population genetics of microbial pathogens and the factors that govern the organization of genetic diversity within these populations. Because the origins and genetic bases of virulence influence the population structure of pathogens, these topics are interconnected and broadly relevant to the emergence, outbreak, control, and prevention of the diseases caused by microbial pathogens.

II. The Genetic Basis of Virulence

One way to gain insight into the specific factors contributing to virulence has been to identify the genetic differences between pathogens and closely related nonpathogenic bacteria. In this regard, there are potentially four types of genetic events that might be responsible for the differences in pathogenic potential among related bacteria: (1) virulence as a result of genes specific to the pathogen; (2) virulence as a result of the absence of a suppressor locus in the pathogen; (3) virulence as a result of allelic differences between genes shared by the pathogen and nonpathogen; and (4) virulence as a result of the differential regulation of the same complement of genes in the pathogen and nonpathogen.

1. VIRULENCE AS A RESULT OF SPECIES- OR STRAIN-SPECIFIC GENES

The most common approach to investigate virulence genes—on both analytical and technical grounds—has been to search for sequences that are restricted to pathogenic organisms. Virulence determinants are often acquired through horizontal transfer, which may explain why virulence traits are distributed sporadically among bacterial taxa (Fig. 1) or within some bacterial species (Fig. 2).

Fig. 1 Phylogenetic relationships among enteric bacteria showing taxa that are normally capable of invading eukaryotic cells (denoted with darkened branches).

Fig. 2 Relationships among commensal and pathogenic strains of Escherichia coli and Shigella spp. based on nucleotide sequences of the gene encoding malate dehydrogenase. Abbreviations are as follows: UTI = urinary tract infection; EHEC = enterohemorrhagic E. coli; EIEC = enteroinvasive E. coli; EPEC = enteropathogenic E. coli; ETEC = enterotoxigenic E. coli. ECOR strains are from the E. coli reference collection [100], and the O:H serotypes of pathogenic E. coli are noted. (adapted with permission from Pupo et al. (1997) [56])

In many bacterial species, there is an association between species-specific genes and virulence. Pathogenicity islands (or Pai)—i.e., segments of the chromosome that encode virulence genes and are absent from related nonpathogenic bacteria [1–3]—have been identified in pathogenic strains of E. coli [4–6], Shigella flexneri [7, 8, 24], Salmonella enterica [9], Yersinia pestis [10–12], Vibrio cholerae [13], Haemophilus influenzae [14, 15], Helicobacter pylori [16], and Staphylococcus aureus [17]. In each of these cases, a specific chromosomally encoded gene, or cluster of genes, has been implicated in the virulence of the microorganism, and the corresponding region was not present in avirulent strains or related species.

Although it has long been known that species-specific regions confer traits that are unique to a bacterial species, the term pathogenicity island was first used to describe a DNA segment harbored by uropathogenic strains of E. coli [18]. Further characterization of pathogenicity islands revealed that many were situated at transfer RNA loci, which are commonly used as integration sites for foreign sequences [19]. For example, certain phages detected in E. coli, such as the retronphage ϕR73 [20] and the bacteriophage P4 [21] insert at, or near, tRNA genes, suggesting that pathogenicity islands are often transferred and acquired through phage-mediated events [1, 22]. The frequent insertion of foreign DNA sequences at tRNA genes is presumably due to the high degree of sequence conservation of tRNA genes across species. In fact, there is recurrent use of the tRNAselC locus, which is targeted by ϕR173 and as the integration site for several pathogenicity islands: Pai-1 of uropathogenic E. coli [4], the LEE island of enteropathogenic E. coli [23], the SHI-2 island of Shigella flexneri [8, 24], and the SPI-3 island of Salmonella enterica [25] each represent independent insertions of different virulence gene clusters into similar chromosomal locations. Flanking many pathogenicity islands there are signature sequences, such as short direct repeats, reminiscent of the integration of mobile elements (or even, in the case of Yersinia pestis, copies of the IS elements themselves) further attesting that these species- or strain-specific regions can be acquired laterally through a variety of transfer mechanisms.

Although research on pathogenicity islands focuses on chromosomally encoded regions, genes involved in bacterial virulence are also carried on extrachromosomal elements that are maintained within the genome of pathogens. For example, many of the genes required for Shigella virulence reside on a 220-kb plasmid [26, 27], and, similarly, all virulent strains of Yersinia harbor a 70- to 75-kb plasmid that encodes proteins necessary for their antihost properties [28, 29]. In this regard, the acquisition of plasmid-borne antibiotic resistance genes will also allow previously sequestered pathogens to exploit new hosts.

Other virulence determinants in these enteric species have been acquired by the organism in phage-mediated events. For example, the cytotoxins first characterized in Shigella are encoded on a bacteriophage that has subsequently been transferred to enterohemorrhagic strains of E. coli [30, 98]. In the case of Vibrio cholerae, two coordinately regulated factors contribute to virulence: cholera toxin, which is encoded by a filamentous bacteriophage (termed CTXϕ) related to the coliphage M13 [31], and the toxin-coregulated pili, which is encoded within a large pathogenicity island. This pathogenicity island of Vibrio cholera is in itself another filamentous bacteriophage [32], and this demonstrates a novel case where one horizontally acquired phage encodes the receptor for the products specified by a second phage, both of which are required for full virulence.

2. VIRULENCE RESULTS FROM THE ABSENCE OF A SUPPRESSOR LOCUS

Similar to the mechanisms described above—whereby a microbe has acquired certain genes that render it virulent—it is also possible that the pathogen has either lost a gene encoding a product capable of diminishing its virulence potential, or that such a determinant was acquired by the related avirulent forms. An early example of a virulence suppressor in enteric bacteria is the surface protease OmpT, which is absent from the genomes of Shigella and enteroinvasive E. coli (EIEC). The presence of ompT results in attenuation of virulence because the encoded protease interferes with expression of the VirG protein, which is required for intercellular spread [33]. The ompT gene is probably not ancestral to enteric bacteria: it is located within the 21-kb cryptic lambdoid phage, suggesting that avirulent strains of E. coli acquired ompT through horizontal gene transfer.

In addition to lacking ompT, Shigellae are also devoid of genes whose products suppress virulence. Representatives of the four species of Shigella, as well as enteroinvasive strains of E. coli (Fig. 2), harbor deletions for the region containing cadA, which encodes lysine decarboxylase. When the cadA gene from an avirulent strain of E. coli was introduced into Shigella flexneri, the resulting strain was still able to invade cells in tissue culture, but did not exhibit the toxic effect that induces the fluid secretion normally associated with infection. In contrast to the situation where a microorganism gains genes that enhance its pathogenic potential (i.e., pathogenicity islands), these regions have been termed black holes to denote deletion of genes that reduce the pathogenic potential of an organism [34].

3. VIRULENCE RESULTS FROM ALLELIC DIFFERENCES BETWEEN HOMOLOGOUS GENES

Because pathogenic and nonpathogenic have often diverged in sequence, it is possible that the differences in pathogenic properties result from allelic variation in homologous genes due to nonsense or missense mutations. For example, point mutations in the fimH gene of E. coli can change the binding of fimbrial adhesins and confer increased virulence in the mouse urinary tract [35, 36].

4. VIRULENCE RESULTS FROM THE DIFFERENTIAL REGULATION OF THE SAME COMPLEMENT OF GENES

In addition to genetic polymorphisms among bacterial strains and species, it is possible that the differences in pathogenic properties are caused by differential regulation of essentially the same set of genes. For example, the invasion gene complexes of S. enterica and S. flexneri are largely homologous, but controlled by very different environmental signals: invasion in S. enterica is regulated by oxygen tension [37], whereas the expression of virulence genes by Shigella is controlled by temperature [38].

The origin of virulence properties in many pathogenic species is likely to result from a combination of the factors presented above. For Shigella and enteroinvasive E. coli, it is clear that virulence is the result of the incorporation of a large virulence plasmid into a strain lacking the ompT gene and the deletion of cadA from their genomes. And while the virulence genes on the Shigella and EIEC plasmids are 99% identical, these species exhibit large differences in their median infective doses, which could be due to allelic variation or to differential regulation of homologous genes or to species-specific chromosomal genes. Also note that these types of genetic changes do not pertain to the analysis of most opportunistic or newly emerging pathogens. Because these microbes are displaced to nonstandard hosts, tissues or environments, the genes contributing to virulence are not preadapted to the host and are not likely to differ from the repertoire required for growth in their customary environments.

III. Identification of Sequences Involved in Bacterial Pathogenesis

From the previous discussion, it is obvious that most of the differences in pathogenic potential among related bacteria are due to changes in gene content (mechanisms 1 and 2) rather than to changes in the ancestral genes themselves (mechanisms 3 and 4). Although the specific approach, as well as technical considerations, bias the identification of the particular genetic events and the recovery of genes involved in virulence (see below), the vast majority of traits that are unique to a species are encoded on segments of the genome that arose through horizontal transfer [39]. Stepwise mutational changes in existing genes only rarely confer novel functions, whereas traits encoded by acquired DNA will occasionally confer the ability to explore new hosts or environments and can have a large impact on bacterial evolution [40]. As a result, none of the phenotypic characteristics that distinguish E. coli from S. enterica are attributable to the divergence of homologous genes by mutation; instead, all of the species-specific traits derive from functions encoded by horizontally transferred genes (e.g., lactose utilization, citrate utilization, indole production) or from the loss of ancestral DNA (e.g., alkaline phosphatase) [39].

The broad association of species-specific traits with unique portions of the chromosome does not imply that all (or even the majority of) genes contributing to the virulence phenotype are restricted to pathogens and absent from the related nonpathogenic bacteria. The classification of genes as being involved in pathogenesis depends largely on the particular approach used to define and identify these factors [41].

The traditional approach to recognizing virulence determinants was purification of microbial products, which, upon introduction into a susceptible host, produced some of the symptoms advanced by the whole organism. This biochemical approach resulted in identification of a variety of factors, usually toxins, produced by several pathogens such as Vibrio cholerae and Clostridium botulinum.

The classical bacterial genetics approach defines virulence genes as those that, on mutation, give rise to strains with median lethal doses (LD50) higher than that corresponding to the wild-type parent [42]. This interpretation of virulence includes all relevant loci—apart from those essential for growth under laboratory conditions—without assumptions about the precise role that particular virulence determinants play in the pathogenicity of the microorganism.

The molecular genetic approach is used to isolate virulence genes based on their capacity to confer certain pathogenic properties on normally benign strains, such as E. coli. A prime example of this approach was the recovery of DNA segments contributing to the invasive character of Yersinia by selecting for clones that could render an E. coli laboratory strain capable of eliciting its uptake by epithelial cells [43]. Similarly, the introduction of a plasmid containing the LEE island into a laboratory isolate of E. coli creates strains that produce attachment and effacing lesions in host cells [23].

It is not surprising that these three strategies have led to the recovery of somewhat different subsets of genes involved in pathogenesis. However, if virulence is to be characterized in terms of the consequences of bacterial infection on the health of the host, the classical bacterial genetics approach—whose aim is to identify all genes that affect host fitness—provides the most comprehensive definition of virulence genes.

Given this perspective on bacterial pathogenicity, many of the virulence genes required for propagation of pathogens within a host would be identical to those required in commensal or benign interactions with hosts. In fact, the molecular genetic experiments, which attempt to convert nonpathogenic E. coli into pathogens, suggest that E. coli, as normal constituents of the human intestinal flora, already contain genes necessary for interaction with animal cells and are, thus, predisposed to become pathogens on acquisition of a particular virulence gene cluster. In addition, many of the genes implicated in Salmonella virulence are also present in nonpathogenic strains of E. coli [9]. These genes encode enzymes responsible for the biosynthesis of nutrients that are scarce within host tissues, transcriptional and posttranscriptional regulatory factors, proteins necessary for the repair of damaged DNA, and products necessary for defense against host microbicidal mechanisms. The presence of these genes in nonpathogenic species suggests that they promote survival within the nutritionally deprived and/or potentially lethal environments that microbes encounter inside and outside animal hosts.

IV. Recovery of Genes Contributing to Virulence

Although the identification and isolation of virulence genes largely depends on how these genes are defined, many pathogens require genes that are absent from related nonpathogenic bacteria. Therefore, several researchers have applied molecular and genetic techniques to recover segments of the genome that are specific to particular bacterial lineages. These procedures yield anonymous DNA fragments and have been typically employed to obtain diagnostic probes for the identification of particular bacterial strains or species [44–49]. However, in a few cases, these techniques have been exploited to find new pathogen-specific genes having a potential role in virulence.

A subtractive hybridization procedure [50] was used to recover DNA sequences present in an avian pathogenic strain of Escherichia coli but absent from a nonpathogenic laboratory strain of Escherichia coli K-12 [51]. The pathogen-specific sequences recovered by this method mapped to at least 12 positions in the chromosome. Subsequently, the phenotype of mutant strains harboring deletions for each of these unique fragments was tested, and two were found to contain genes required for virulence in avian hosts. Other studies have examined the unique DNA in the genome of pathogens, but have yet to directly assess the function of these sequences. For example, a subtractive hybridization procedure used to examine the differences in gene content among strains of Helicobacter pylori yielded 18 clones, several of which were presumed to have a role in the specific virulence characteristics of H. pylori strains [52]. Although genome subtraction and physical mapping techniques allow one to identify and clone the differences between two genomes, there is usually no rapid way to determine if these pathogen-specific sequences are indeed relevant to pathogenesis. A thorough review of the strategies used to identify virulence determinants is presented in the chapter by Camilli et al. in this book (see Chapter 4).

V. The Population Genetics of Pathogens

What is the genetic structure of pathogen populations, and how much genetic variation is present in these populations compared with that in related nonpathogenic microorganisms? Moreover, what is the apportionment of genetic diversity among pathogenic strains in relation to that in species at large?

Although there are several genetic mechanisms by which microorganisms change their pathogenic potential, it has become evident that the bacteria responsible for disease outbreaks are distinct clones that are frequently characterized by unique combinations of virulence genes or of alleles at virulence genes. The situation is not as clear for pathogenic species, principally because the classification of a pathogenic species is somewhat arbitrary, and historically reflects the ability of epidemiologists to classify strains. On one hand, pathogens have been overclassified—that is, they are typed into a multitude of genetically narrow groups or species—compared to nonpathogens. This is probably judicious, given the importance of assigning an identity of each isolate implicated in human disease. For example, based on serological characteristics, the Salmonellae were assorted into thousands of distinct species (now termed serovars), but a recent taxonomic revision based on DNA–DNA hybridization reclassified these strains into a single species, Salmonella enterica [53, 54]. Similarly, the Shigellae have traditionally been subdivided into four species—Shigella boydii, S. flexneri, S. dysenteriae, and S. sonnei—although the total amount of genetic variation within this genus is contained within E. coli [55–58]. The classification of Shigellae based on serologic and metabolic characteristics illustrates two additional points. First, the amount of diversity can vary widely among species: while Shigella sonnei consists a single genetically uniform clone, each of the other three species comprises a heterogeneous array of clones. Second, the classification of strains does not always reflect their true genetic similarities or relationships. As shown in Figure 2, certain strains of S. flexneri can be more closely related to S. boydii than to any other strains typed S. flexneri, and evidence from other studies have shown that Shigella boydii and S. dysenteriae have multiple origins from within E. coli [55–58].

In contrast to these enteric pathogens, the classification of strains into other pathogenic species is more liberal and diffuse, due, in part, to the inability to rapidly differentiate among isolates that display a diagnostic phenotype or produce a specific pathology in hosts. However, the subsequent analysis of such species using additional genetic and/or phenotypic markers often reveals the true nature of the diversity and the relationships among strains.

VI. Studying Bacterial Population Genetics

Studies addressing the genetic structure of bacterial populations began relatively recently [59, 60] by evolutionary geneticists who had originally examined the amount and distribution of genetic variation among natural populations of eukaryotes. Bacteria, particularly E. coli and Salmonella, were an attractive group of organisms on account of their phenotypic diversity, haploid chromosomes, large populations sizes, short generation times, and ease of propagation and experimental manipulation. Hence, bacterial population genetic research was originated by population geneticists interested in bacteria, rather than by bacterial geneticists or medical microbiologists interested in population genetics. The information gathered by population geneticists has broad implications for medicine and epidemiology, and these studies typically go beyond the simple identification of strains and address questions pertaining to their genetic relationships and their levels of allelic diversity [61, 62].

Numerous molecular techniques—such as RAPDs, IS (and other repetitive element) fingerprinting, ribotyping, phage typing, macrorestriction mapping by pulsed-field gel electrophoresis, and plasmid profile analysis—have been applied to establish the identity of strains for the epidemiological purposes. But these methods typically do not supply the information necessary to establish the relationships among strains, infer the genetic structure of natural populations, or assess the relative roles of natural selection, random drift (i.e., the change in gene frequencies caused by the chance event of random sampling in small populations), new mutations, and horizontal gene transfer (including intragenic and intergenic recombination) on the organization of allelic diversity [63]. Once the evolutionary relationships of clones of a species is available, it is possible to examine the manner in which the total genetic diversity is apportioned with respect to host species, geographic populations, and the specific disease pathology. The following methods have been applied to establish the evolutionary relationships and genetic structure of bacterial populations.

A. Multilocus Enzyme Electrophoresis

Multilocus locus enzyme electrophoresis (or MLEE) has been the primary method used to assess genetic variation in bacterial populations [64]. The main advantage of this technique is that many genes can be readily examined in hundreds, if not thousands, of isolates. However, this method relies on the discrimination of alleles of distinct electrophoretic mobilities—also called allozymes or electromorphs in this context—and, therefore, can detect only a portion of the sequence variation at a locus (Fig. 3). The key concept underlying the use of MLEE in population genetics is that the electromorphs can be directly equated with alleles of the corresponding structural gene and that electromorph profiles over the sample of different enzymes (frequently termed electrophoretic types or ETs) correspond to multilocus chromosomal genotypes [65]. The proteins assayed by this method are usually metabolic enzymes, such as those involved in glycolysis, which are expressed in all isolates of strains, and the allelic variation is unaffected by environmental conditions, including host, culture medium, or laboratory storage. Moreover, the allelic variation detected at these enzyme loci is selectively neutral, or nearly so, such that there is minimal convergence to the same allele through adaptive evolution [66–68]. Hence, this technique provides a rapid way to index the overall levels of genetic diversity at numerous loci throughout the chromosome and to infer genetic relationships among strains. Because the number of alleles at a locus is fairly large in bacterial populations, we expect that recombination would, by chance, only rarely generate strains of identical multilocus chromosomal genotypes. Therefore, strains of the same ET are considered to be similar by descent and shared ancestry. Over the past two decades, MLEE has become established as the standard procedure for assessing genetic variation in bacterial populations and the one against which the discriminatory power of all other techniques is measured.

Fig. 3 Multilocus enzyme electrophoresis (MLEE). (A) Supernatants extracted from individual bacterial cultures (A–I) are extracted, electrophoresed through a support matrix, and selectively stained to detect the product of a specific enzyme locus (e.g., PGI). (B) Allelic variants (a.k.a. electromorphs) at each locus are numbered and cumulatively form the basis of the electrophoretic type (ET) of the isolate. (C) Relationships among isolates can be visualized as a dendrogram, constructed from a matrix of pairwise differences between the allelic profiles of the isolates. Reprinted with permission from and through the courtesy of Dr. Thomas Whittam.

B. DNA Sequencing

Although MLEE offers a rapid and inexpensive way to appraise the genetic variation in bacterial populations, many laboratories have turned to directly sequencing the genes specifying several of the enzymes originally indexed by MLEE or the genes encoding proteins involved in virulence. In contrast to MLEE, nucleotide sequencing offers a means of uncovering all of the allelic variation at a locus and detecting events of intragenic recombination. Moreover, nucleotide sequences are composed of discrete characters—i.e., the four bases—as opposed to MLEE, which can only provide the relative mobilities of electromorphs. This allows for the unambiguous identification of alleles, and for comparison and portability of data, from different studies and laboratories. Furthermore, the use of PCR to generate sequencing templates permits the analysis of noncultivable organisms. Although nucleotide sequencing provides the most complete information about the genetic variation and relationships among strains, it is still relatively costly and time consuming, especially for the analysis of variation at several loci in a large number of strains. And in many applications, the level of variation detected by MLEE has been sufficient to answer all but the subtlest questions about the genetic diversity and structure of natural populations.

C. Multilocus Sequence Typing

Maiden et al. [69] have devised a method for the identification and typing of bacterial clones based on the determination of sequences of several gene fragments (Fig. 4). Multilocus sequence typing (MLST) exploits the advantages of nucleotide sequence data, but also constructs chromosomal genotypes, which can be used to detect intergenic recombination in the manner of MLEE, through an analysis of multiple chromosomally encoded loci [70]. In the initial application of MLST, the nucleotide sequences of PCR-amplified fragments from 11 housekeeping genes were obtained for more than 100 isolates of Neisseria meningitis. For MLST, the gene fragments (i.e., alleles) are 400 to 500 nucleotides in length—a convenient size for the direct automated sequencing of a DNA fragment with a single primer—and each unique combination of alleles over loci is referred to as a sequence type (ST). As the number of sequencing facilities increase, and the costs of DNA sequencing fall, MSLT is certain to become the method of choice for assessing variation in bacterial populations.

Fig. 4 Multilocus sequence typing (MLST). The method for allocation of the allelic profile, or sequence type (ST), of a bacterial isolate is shown. As in MLEE, the relationships among isolates can be visualized as a dendrogram, constructed from a matrix of pairwise differences between the allelic profiles of the isolates. Reprinted with permission from Spratt (1999) [70].

VII. The Organization of Genetic Diversity in Pathogenic Microorganisms

Early studies on microbial pathogens, particularly enteropathogenic E. coli [71, 72], suggested that very few clones, as identified by serotyping, were associated with disease outbreaks. Subsequent analysis using MLEE provided the first indication that the species E. coli as a whole was clonal, as evident from the repeated recovery of strains of the same chromosomal genotypes from different times and geographic locations [60, 73]. Three generalizations have emerged from the broad-scale application of MLEE to the study of common human pathogens [61, 63]. First, most species of bacteria are highly polymorphic for electrophoretically detectable alleles at each enzyme locus, such that a typical locus may have 10 to 20 electromorphs. Second, despite harboring large amounts of genetic variability, most bacterial species are clonal and consist of a relatively small number of genotypes. This implies that rates of recombination between genetically distinct clones, which would serve to generate new combinations of alleles over loci, must be very low. Finally, in most pathogenic species, only a very small proportion of clones promotes most of the disease worldwide (which indicates large differences in virulence among strains). Furthermore, the same clones that cause disease can be recovered over long periods of time.

VIII. Population Genetics of Representative Bacterial Pathogens

Having considered the connections among the molecular, genetic, and evolutionary perspectives on bacterial virulence, we can now turn our attention to the population genetic analysis of specific pathogens. Most of the information summarized below is based on data obtained through the application of MLEE; however, for several organisms, we can also integrate information on the relationships and diversity among strains, as achieved through the analysis of DNA sequences. The species are discussed in alphabetical order (with the exception of Shigella, which, based on its proper taxonomic position, is included within E. coli), and the key characteristics of their genetic variation and population structure are summarized in Table I.

Table I

Population Structure and Genetic Diversity of Representative Bacterial Pathogens as Assayed by MLEE

aET = electrophoretic type; a unique combination of alleles over all loci studied.

bMajor genetic divisions within a sample (often designated with roman numerals) are defined on the basis of a dendrogram (or tree) representing the relationships of ETs, and are also referred to as phylogenetic subgroups, clusters or subdivisions, primary lineages, or subspecies.

cFor a sample of isolates or ETs, genetic diversity at a locus is expressed as

where xi is the frequency of the ith allele at the locus, and n is the number of isolates or electrophoretic types in the sample. Mean genetic diversity per locus (H) is the arithmetic mean of h over all loci studied.

dThe genetic variation in Mycobacterium tuberculosis has been assessed by direct nucleotide sequencing of several loci, but not by MLEE.

eData on Streptococcus pneumoniae was obtained through MSLT, not MLEE.

A. Bordetella

Bacteria belonging to the genus Bordetella are of primary importance in pediatric and veterinary medicine because of their ability to colonize the epithelium of the respiratory tract of a variety of vertebrate hosts, there causing bronchial and pulmonary pathology (see Chapter 13 herein, by Cotter and Miller). Based on phenotypic characteristics, the genus Bordetella nominally consists of four species: B. pertussis, an obligate human pathogen causing whooping cough, B. parapertussis, which has been isolated from humans and sheep, B. bronchiseptica, which is the etiologic agent of canine kennel cough, and B. avium, which causes respiratory disease in fowl. These four species have been further subdivided on the basis of serology and biotyping; however, evidence from MLEE indicates that from an evolutionary genetic standpoint B. pertussis and human isolates of B. parapertussis should be considered clones of B. bronchiseptica that adapted to the human host relatively recently [74]. Furthermore, many electrophoretic types (ETs) within each species are strongly host adapted. For example, ETs 22 through 26 of B. parapertussis are confined to sheep, while B. parapertussis ET 28 and the four ETs of B. pertussis are only recovered from humans [75].

Although laboratory experiments have shown the Bordetellae to be naturally competent and capable of recombination, their population structure is predominantly clonal, with several of the same ETs recovered from different localities and at different times. While there is no evidence of gene exchange between distantly related strains, recombination occurs among some of the more closely related strains with broader host ranges [75].

B. Borrelia

The spirochete Borrelia burgdorferi is the causal agent of Lyme disease and is transmitted to humans and animals by the bite of infective Ixodes ticks [76]. In humans, the clinical features of Lyme disease progress from a rash and flu-like symptoms to arthritis, carditis, and neurologic disorders; this pathology was first described in Europe several decades ago. Based on genetic criteria, strains from Europe, Asia, and the United States that were originally classified as Borrelia burgdorferi (sensu lato) were reassorted into a complex of at least four highly divergent species—B. burgdorferi (sensu stricto), B. garinii, B. afzelii, and B. japonica—and there are likely to be additional species in other geographic regions [77–79].

Among the ETs defined by MLEE, there were highly nonrandom associations of alleles over the chromosome (i.e., linkage disequilibrium), indicating a clonal population structure. All isolates from the United States (including the type strain from Shelter Island, New York isolated in 1982) and many from west-central Europe formed a monophyletic clade of lineages descended from a single common ancestor, which is presently considered B. burgdorferi sensu stricto [77]. In addition, comparisons of the branching orders of phylogenetic trees based on the DNA sequences of two chromosomally encoded genes (fla and p93) and a plasmid-borne gene (ospA) provide no evidence of lateral gene transfer—again indicating that Borrelia burgdorferi (sensu lato) is strictly clonal—and that plasmid transfer between clones is very rare [80].

Recent analysis of the two outer surface lipoproteins encoded by ospA and ospC within Borrelia burgdorferi (sensu stricto) revealed a strong association between ospA and ospC alleles [81], despite the fact that these genes are encoded on separate plasmids. The ospC gene is highly variable, and even in local populations there are large numbers and high frequencies of divergent alleles, as expected if the diversity is maintained through some form of selection acting to maintain variation in the population. Furthermore, the geographic distribution of opaC alleles confirmed that Borrelia burgdorferi (sensu stricto) originated in the United States and only recently spread to Europe [81].

C. Escherichia coli and Shigella

Although typically a benign constituent of the mammalian intestinal flora, E. coli is occasionally an infectious agent responsible for several human diseases. Unlike many of the pathogenic microbes discussed so far, most information on the evolutionary and population genetics of E. coli was obtained from studies of wild strains from healthy hosts. And for the past 20 years, the genetic structure and level of variation observed in these commensal populations of E. coli have served as the basis for interpreting results from pathogenic and nonpathogenic bacteria alike [82].

Given that the organization of genetic diversity is known for natural populations of E. coli isolated from a wide range of sources [83], we can begin to ask questions about the emergence of pathogenic strains and the extent of genetic variation in pathogenic strains of E. coli. Strains of E. coli have been traditionally typed on the basis of polymorphism in their O (somatic), K (capsular) and H (flagellar) antigens [72, 84]. Despite more than 1000 combinations of antigens of the O, K, and H groups, certain serotypes were repeatedly associated with certain diseases, providing the first indication that E. coli populations were clonal [85, 86].

Enteropathogenic E. coli (EPEC) (see Chapter 9 herein, by Puente and Finlay) were first identified from outbreaks of infantile diarrhea in Great Britain nearly 50 years ago [87]. The EPEC strains constitute very few serotypes, many of which are represented by a single temporally stable, and geographically widespread, clone [88]. Based on MLEE, the EPEC strains form two clonal complexes, each possessing the plasmid-borne adherence phenotype and distantly related to other pathogenic E. coli [89].

In the early 1980s, there were several outbreaks of hemorrhagic colitis associated to the rare E. coli serotype O157:H7, the so-called Jack-in-the-Box strain [90, 91]. These isolates are serologically distinct from the EPEC strains, and over the past decade there have been numerous outbreaks of hemorrhagic colitis and hemolytic uremic syndrome in North America and Asia linked to a variety of foodstuffs [92–94]. The virulence characteristics of enterohemorrhagic E. coli O157:H7 derive from several sources: strains produce one or more forms of Shiga cytotoxin, contain a 92-kb virulence plasmid (plus a 3.3-kb plasmid in Japanese strains) and harbor a 43-kb pathogenicity island conferring the ability to evoke attachment and effacing (A/E) lesions [95–98].

MLEE was used to determine the relationships of enterohemorrhagic E. coli O157:H7 isolates to one another and to other strains causing enteric infections. A majority of O157:H7 strains belong to a clone complex that is only distantly related to other Shiga-toxin–producing strains of E. coli or to other ETs of the O157 group causing enteric infections in animals [99]. The O157:H7 clone is most closely related to strains O55:H7, also capable of producing A/E lesions. The most likely hypothesis for the origin of O157:H7 is that it emerged from an O55:H7 progenitor after acquiring the phage-encoding Shiga toxins and plasmid-encoded adhesins through horizontal transfer [89].

An alternative approach to studying the evolutionary genetics of virulence in E. coli has been to examine the phylogenetic distribution of pathogenic strains in relation to the species as a whole. To index the degree of genetic diversity and relationships among strains from natural populations of E. coli, most researchers have relied on the ECOR collection, which includes 72 strains originally selected to encompass the range of genetic variation in the entire species [100]. Based on MLEE [101] and on the nucleotide sequences of several genes [102], the phylogeny of ECOR strains consists of four major subgroups (A, B1, B2, and D), and the characterization of the ECOR collection by most other techniques has yielded essentially the same groupings [83, 103].

From such studies, it is clear that pathogenic strains do not form a phylogenetically distinct group, nor do they have a single evolutionary origin within E. coli [56, 104, 105]. In addition, strains of E. coli are predisposed to become human pathogens on the disruption of chromosomal genes and/or the acquisition of additional plasmids, phages, or pathogenicity islands. Yet, some generalizations can be made: commensal strains of E. coli are principally within subgroups A and B1 (and these strains typically lack virulence determinants), whereas many of the genes associated with pathogenic isolates of E. coli are prevalent among ECOR strains in subgroups B2 and D [106, 107].

Despite the tendency for pathogens to be related to commensal strains in certain subgroups and the phylogenetic clustering of certain virulence-associated genes, there are pathogenic isolates in each of the E. coli subgroups. In a comprehensive analysis of the relationships between nonpathogenic (as represented by ECOR) and a variety of pathogenic strains of E. coli, analysis by MLEE and mdh (malate dehydrogenase) sequences shows that enteropathogenic (EPEC), enterotoxigenic (ETEC), enteroinvasive (EIEC), and enterohemorrhagic (EHEC) strains were distributed among the ECOR subgroups [56].

Shigella is the etiologic agent of bacillary dysentery and has long been known to be closely related to E. coli (see Chapter 8 herein, by Sansonetti et al). Although classified into a separate genus, the four subspecies of Shigella constitute lineages that actually fall within the range of variation detected in E. coli [55–58]. Strains of Shigella sonnei form a single clone that has remained virtually unchanged for at least the past 40 years [108], but each of the other Shigella species have multiple independent origins from within E. coli (Fig. 2). Based on genetic criteria, the Shigellae should not be considered species separate from E. coli, and, like other agents of enteric disease, they appear to be secondarily derived from commensal strains of E. coli.

D. Haemophilus

Isolates of Haemophilus influenzae have traditionally been serotyped on the basis of six chemically distinct polysaccharide capsule types, but many isolates are unencapsulated and, hence, untypable (sec Chapter 14 herein, by Tang et al.). These unencapsulated forms are generally noninvasive, but encapsulated strains are invasive; and, in particular, those expressing serotype b are the major agents of meningitis in children. There were originally few useful epidemiological markers for serotype b strains of H. influenzae because most isolates were of identical biotype. However, the application of MLEE to Haemophilus [109–111] provided the first genetic framework for studying the population structure and relationships among strains and species.

Like several other pathogens, the population structure of H. influenzae is clonal, with certain ETs distributed worldwide and persisting over long periods of time [109, 112]. Among the encapsulated strains (serotypes a–f). there were two major phylogenetic divisions. Strains producing a and b capsules occur within each of these divisions, most probably as a result of transfer of the serotype-specific capsule genes between lineages [113, 114]. Although serotype a strains are rarely virulent, some have been implicated in cases of septicemia and meningitis. These isolates harbor a virulence-enhancing mutation usually detected in type b strains, providing additional evidence of horizontal transfer [115].

Although it was originally thought that unencapsulated forms of H. influenzae would simply represent a subset of the encapsulated strains that recently lost the polysaccharide capsule, these nontypeable strains belonged to distinct clusters and are, as a group, much more genetically diverse than clones expressing capsule-type b [74, 116, 117]. In general, there is little relationship between ET and particular disease conditions; however, the nonencapsulated strains isolated from healthy carriers were distinct from those causing various diseases [118].

E. Helicobacter

Nearly half the human population is colonized by Helicobacter pylori, but clinical symptoms, ranging in type from peptic ulcer to gastric carcinoma and lymphoma, manifest in only a very small proportion of those infected ([119]; see Chapter 11 herein, by Cover et al.). Although host and environmental factors—such as blood type, cigarette smoking, and gender—affect the risk of gastric ailments [120, 121], strains of H. pylori causing disease contain genetic determinants not found in avirulent strains [16, 122]. As a result, strains have been classified into two groups (type I and type II) that differ in the presence of the cytotoxin-associated gene cagA, which is encoded on a pathogenicity island, and of the vacuolating cytotoxin tox, whose phenotype is controlled by expression of the vacA gene [123]. Type I strains, which represent 60% of all isolates examined, are cagA+/tox+ and are associated with duodenal and gastric ulceration, while type II strains are cagA−/tox− and are largely asymptomatic [124–126].

The simple classification of H. pylori as type I or type II does not impart the true extent of genetic diversity within the species as a whole, or the degree of genetic differentiation among strains. Most typing methods can discriminate among isolates from different individuals [127], and minor genetic changes, arising from rearrangements occurring within a lineage, have been detected in samples from a single host [128]. The mean genetic diversity of H. pylori, as assessed by MLEE, exceeded that detected for any other bacterial species examined by this technique [129]. Moreover, the index of association (IA) value between loci [130] did not differ significantly from zero, implying free recombination within natural populations of H. pylori. Such high levels of genetic exchange obscure the clonal descent and genealogical relationships among strains, and can also impede the use of certain therapies against this pathogen.

As a step in the control and treatment of H. pylori infections, the sequence diversity at several virulence-associated loci has been investigated, often in an attempt to identify the association between sequence variants and disease pathology [131–133]. All loci examined display unusually high levels of genetic variability, and in some cases unique alleles have been formed through intragenic recombination [132, 134–137]. In an analysis of the partial nucleotide sequences from two flagellins, flaA and flaB, and from vacA, it was noted that the vast majority of H. pylori strains had unique sequences at all three loci and that approximately 20% of the nucleotide positions were polymorphic in different strains, due principally to changes at synonymous sites [138]. This pattern of substitutions, characterized by high levels of divergence at silent sites but very low levels at nonsynonymous sites, suggests that these genes are under strong selection. Moreover, analysis of these data confirms that recombination is so frequent within H. pylori that the alleles at different loci, as well as the polymorphisms within loci, are effectively at linkage equilibrium [138].

F. Listeria

Among the species of this genus, only Listeria monocytogenes is commonly pathogenic for humans, causing several invasive diseases including septicemia, meningitis and meningoencephalitis (see Chapter 16 herein, by Fsihi et al.). L. monocytogenes have been isolated from environmental sources and a wide variety of raw and processed foodstuffs, particularly dairy products, which has led to the application of numerous techniques for the detection, identification, and epidemiological tracing of strains [139–145]. Based of MLEE, strains of L. monocytogenes belong to two major clusters [146–148], and the existence of two primary lineages has been confirmed by ribotyping [149], restriction mapping [150], and DNA sequencing [151]. One cluster contained strains serotyped as 1/2a and 1/2c. and the other included strains of serotypes 1/2b. 3b. and 4b; and, except in a very few cases, each ET contained strains typed to only one serotype. Strains recovered from animals and humans were distributed into each of these phylogenetic groups, indicating that there is no genetic differentiation among strains infecting a variety of mammalian hosts. Subsequently, on the basis of the DNA sequences of several genes, a third evolutionary lineage of L. monocytogenes has been defined [152, 153], and this group comprises strains recovered from animals, but not from humans.

Several lines of evidence are consistent with the hypothesis that recombination of chromosomal genes is an infrequent event in natural populations of L. monocytogenes. First, there is no overlap in the serotypes of isolates assigned to the two phylogenetic clusters. Second is the nonrandom association of alleles among ETs for many pairs of loci (i.e., linkage disequilibrium). Third is the finding that ETs are very stable, such that genetically indistinguishable isolates have been recovered in widely separated geographic locations. Despite the large number of clones detected within L. monocytogenes, clones typed as ET1 have been implicated in disease outbreaks in California and Switzerland, suggesting that this ET is either very abundant in the environment (or in foodstuffs), or highly pathogenic to humans [147].

G. Mycobacterium

Despite a steady decline in the incidence of tuberculosis in the United States beginning in the 1950s, there has been an alarming increase in the number of individuals infected by Mycobacterium tuberculosis, and perhaps one-third of the world’s population harbor this pathogen ([154]). In spite of this very large population size, the species as a whole displays almost no sequence variation, suggesting that all strains of M. tuberculosis shared a common ancestor an estimated 15,000 years ago [155]. Due to the close relationships among strains, the epidemiological tracing of M. tuberculosis is typically based on RFLP analysis of a mobile genetic element, IS6110. which is variable in its copy number and genomic location among strains [156, 157].

Analysis of more than two megabases of sequence data from a total of 26 loci has provided the first comprehensive view of the origin, ancestry, and genetic population structure of M. tuberculosis [158]. Unlike most bacterial species that gain antibiotic resistance genes through the acquisition of plasmids or other elements, the vast majority (>95%) of sequence variation within this species is attributable to nonsynonymous substitutions or other mutations in chromosomal loci that confer resistance to antibiotics. Despite very low levels of neutral sequence variation—in fact, the M. tuberculosis species has the lowest level of nucleotide diversity of any bacterial pathogen—strains could be assigned to three genotypic groups based on combinations of polymorphisms at two sites: codon 463 in katG, and codon 95 in gyrA. (These two sites are not involved in antibiotic resistance and, hence, serve as useful genetic markers to examine the evolutionary history and relationships of strains.) All isolates of the predominantly nonhuman pathogens, M. microti and M. bovis, and the human pathogen, M. africanum, have the same combination of polymorphisms characteristic of M. tuberculosis genotypic group 1, suggesting that this genotypic group is ancestral to groups 2 and 3. Because the most ancient group is expected to contain the most variation, this view of the relationships among the genotypic groups is further supported by the fact the genotypic group 1 contains the most variation in IS6100 copy numbers and in the nucleotide sequences at other loci [158].

Despite the low rate amount of nucleotide sequence diversity, the three genotypic groups of M. tuberculosis have diverged with respect to IS6110 copy numbers. IS6110 profiles can change relatively rapidly, and in many cases isolates resampled after 90 days from the same patients displayed changes in IS6100 genotype, particularly among strains with greater numbers of these IS elements. And due to relatively minor changes in the IS6110 fingerprint patterns of strains from single individuals, this variation could not have been produced by reinfection by a different strain [159].

H. Neisseria

Due to the widespread and recurrent epidemics of meningococcal disease, there has been extensive work on the population biology of Neisseria meningitidis (see [160–163] for comprehensive reviews and Chapter 12 herein, by Meyer et al.). In general, the variation among strains of Neisseria is generated through horizontal exchange, but epidemics are often caused by the spread of specific genetic variants, which results in clonal replacement. After their descent for a common ancestor, strains rapidly diversify through mutation and recombination [162].

N. meningitidis are conventionally typed on the basis of capsular polysaccharides, and serogroups A, B, and C account for more than 90% of the cases of meningococcal disease worldwide ([164]). In an early study, it was demonstrated that the European epidemic of serogroup B disease that began in the 1970s was caused by a group of 22 very closely related clones, designated as the ET-5 complex, that have no close genetic relationship to other groups of clones [165]. Clones of the ET-5 complex were also the causative agents of later outbreaks in Africa, South America, Cuba, and the United States (where it was likely to have been introduced by Cuban immigrants). N. meningitidis is carried asymptomatically in the upper respiratory tract by about 15% of the human population, and the clones isolated from carriers are only rarely represented among those causing disease, showing that certain complexes of clones have a low virulence potential [166].

Unlike most other serogroups of N. meningitidis, which are generally associated with endemic disease, isolates of serogroup A are unusual in that they may cause very large epidemics. Among the serogroup A organisms responsible for 23 epidemics between 1915 and 1983, there were 34 distinctive ETs constituting four clone complexes, each representing a group of related clones [167]. Most epidemics were caused by a single clone, and the same clone was responsible for concurrent epidemics in different countries. Clonal analysis has also demonstrated that serogroup A isolates are a restricted phylogenetic subpopulation of the species [168], which probably arose no more than a few hundred years ago [169]. Although recombination produces genetic variation within clone complexes, these results would indicate that only limited amounts of genetic exchange occur between phylogenetically unrelated strains of N. meningitidis.

Sequence analysis of meningococcal genes provides clear evidence that the evolution of N. meningitidis has been characterized by high levels of intra- and intergenic recombination, which is perhaps not surprising for a naturally transformable species. For example, allelic variants of the gene encoding adenylate kinase (adk), whose variation is regularly assayed by MLEE, have a mosaic structure produced by recombination between genes from different strains [170]. Moreover, within and among species of Neisseria, patterns of sequence divergence for adk, recA, aroE (shikimate dehydrogenase), and glnA (glutamine synthetase) are very different [171, 172], and the phylogenetic trees based on each of these genes are not congruent, as expected for species that undergo frequent recombination. Note that even in recombining species, such as N. meningitidis, the epidemic increase of certain ETs can give the appearance of a clonal population structure despite high levels of gene exchange [130, 173].

I. Salmonella

Under the original serotyping schemes of White [174] and Kauffman [175], the Salmonellae were assigned to nearly 3000 serotypes or serovars, with each considered a distinct species. However, based on the biotyping and molecular genetic evidence, all strains were subsequently typed as single species, S. enterica, which comprises eight subspecific groups (designated I, II, IIIa, IIIb, IV, V, VI, and VII) [176, 177]. Hence, the nomenclature has changed such that Salmonella typhimurium would now be referred to as Salmonella enterica serovar Typhimurium or, simply, Typhimurium. Over 60% of the serotypes belong to subspecies I, including those strains causing >99% of the cases of human salmonellosis (see Chapter 7 herein, by Scherer and Miller). Due to its genetic relationships to the other subspecific groups, subspecies V has recently been reclassified as a separate species, Salmonella bongori [54], and MLEE as well as nucleotide sequence analysis of several genes have confirmed its divergent phylogenetic position [178, 179, 63, 180–182].

MLEE was originally used to investigate the allelic variation in genes in large samples of Salmonella [183–189]. The total genetic diversity in Salmonella, as assessed by MLEE, is among the highest reported for any bacterial species (Table I)—nearly twice that observed in E. coli. Clonal aspects of the genetic structure of S. enterica are well illustrated by the serovars Typhi and Paratyphi A, B, and C, all of which are agents of enteric fever in humans. MLEE analysis has demonstrated that there are no close relationships among these serovars, implying independent evolutionary derivation [187, 188]. Typhi is an unusually distinctive and homogenous serovar, and over 80% of the worldwide isolates are of a single ET (with a second ET comprising 16% of strains, all from West Africa). Paratyphi B consists of a large and heterogenous group of lineages that are closely related to Typhimurium. However, the ability of Paratyphi B to cause human enteric fever arose in a single globally distributed clone, and only recently, since it is only weakly differentiated.

The genetic variation and relationships among strains have also been assessed by the nucleotide sequencing of several housekeeping genes, including proline permease (putP) [178], glyceraldehyde-3-phosphate dehydrogenase (gapA) [190], malate dehydrogenase (mdh) [180], 6-phosphogluconate dehydrogenase (gnd) [179], isocitrate dehydrogenase (icd) [191], and isocitrate dehydrogenase kinase/phosphatase (aceK) [192]. For these five housekeeping genes, on average, about 16% of nucleotides and 5% of amino acids are polymorphic. With the exception of gnd, the level of sequence diversity is greater in S. enterica than in E. coli, which is attributable to the unusually high divergence of subspecies V (S. bongori) from the other subspecies. Comparisons of the individual trees based on the nucleotide sequences have revealed several cases in which the branching orders of lineages are not congruent. These disparities are due both to intragenic recombination events, which can involve regions ranging from six basepairs to more than 1 kb, and to the exchange of entire genes [63].

Notwithstanding low levels of recombination at some loci, the relationships among strains based on nucleotide sequences matched those established by DNA hybridization and MLEE. Based on these phylogenetic relationships, serovars that are exclusively or predominantly diphasic (subspecies I, II, IIIb, and VI) cluster apart from the monophasic subspecies (IIIa, IV, V, and VII). This suggests that, following the divergence of S. enterica and E. coli from a common ancestor, E. coli evolved as an commensal of mammals while Salmonella remained associated with reptiles, which are still the primary hosts of the monophasic subspecies. Salmonella serovars are typically classified as either monophasic or diphasic based on their ability to produce one or two forms of flagellin. Subsequently, Salmonella evolved as an intracellular pathogen through the acquisition of several pathogenicity islands, which conferred the ability to invade host epithelial cells and circumvent host defenses. The diphasic condition originated in the lineage ancestral to subspecies I, II, IIIb, and VI is a mechanism to further evade the host immune system and is likely to have assisted in the exploitation of birds and mammals of potential hosts [181]. This scenario is supported by the phylogenetic distribution of pathogenicity islands in Salmonella [193]. The SPI-1 island, which confers the ability to invade nonphagocytic host cells, was acquired very early in the evolution of Salmonella and is present in all subspecies, whereas the SPI-2 island, which is necessary for intracellular proliferation, is absent from Salmonella bongori, which were originally recovered from nonmammalian hosts. This suggests that the evolution of Salmonella as a pathogen has been marked by the acquisition and/or generation of several genes that facilitate interactions with the host [9, 194].

J. Staphylococcus

The are two notable cases where MLEE has been applied to uncover the evolutionary history of infective strains of Staphylococcus aureus. The first involves strains of S. aureus causing toxic shock syndrome (TSS) in young, healthy menstruating women. Almost all strains of S. aureus recovered from TSS patients express a toxin (designated TSST-1) [195, 196], which is now known to be encoded as part of a 15-kb pathogenicity island present only in TSST-1-positive strains [17]. The analysis of genetic variation in 315 isolates of S. aureus expressing TSST-1 revealed that toxin production occurs in association with chromosomal backgrounds representing the full breadth of genotypic diversity in the species as a whole, as might be expected for genes encoded on a mobile element [197]. But despite the diversity among strains expressing TSST-1, a single distinctive clone causes the majority of cases of toxic shock syndrome. It is not known whether the present-day distribution of the TSST-1 gene in S. aureus reflects an evolutionarily old association or the independent acquisition of the TSST-1 pathogenicity island by multiple strains. However, these results suggest that the particular clone causing TSS has properties conferring strong affinity for human cervicovaginal surfaces [197].

A second case where MLEE has