 

 

 

This is to certify that the
dissertation entitled

POPULATION GENETIC ANALYSES OF SALMONELLA
ENTER/CA SEROVAR ENTERITIDIS AND ESCHERICHIA
COLI FROM BIOTIC AND ABIOTIC SOURCES

presented by

SETH TAYLOR WALK

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Department of Zoology and
Program in Ecology. Evolutionary
Biology and Behavior

 

 

 

Major Prgféssor’s Signature v V
g/ 5/ /2oo >
I /

Date

MSU is an afﬁnnative-action, equal-opportunity employer

_-—.-.--.—._.-.- —.-‘- -.--.-.-.-.-

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRC/DaleDue.indd-p.1

 

 

POPULATION GENETIC ANALYSES OF SALMONELLA ENTERICA SEROVAR
ENTERITIDIS AND ESCHERICHIA COLI FROM BIOTIC AND ABIOTIC
SOURCES
By

Seth Taylor Walk

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Zoology and Program in Ecology, Evolutionary Biology, and Behavior

2007

 

ABSTRACT
POPULATION GENETIC ANALYSES OF SALMONELLA ENT ERICA SEROVAR
ENTERITIDIS AND ESCHERICHIA COLI FROM BIOTIC AND ABIOTIC
SOURCES
By
Seth Taylor Walk
Humans and animals host a myriad of microbes including enteric bacteria of
the family Enterobacterz'aceae. While these organisms are related, they represent a
range of evolutionary strategies and interactions with the host gastrointestinal tract.
Clinically important enteric species, including Salmonella enterica and Escherichia
coli, have been thoroughly characterized by phenotypic, genetic, and epidemiologic
methods. Accordingly, these organisms serve as models for studying the genetic and
phenotypic diversity of this bacterial family. Although a few strains have been well
characterized, less information is available about the natural history of these
organisms and, speciﬁcally, how populations experience and adapt to selective
evolutionary pressures. The goal of this research is to present a novel perspective on
this topic through population genetic analyses. These methods, when applied to
longitudinal samples of strains, quantify the abundance and distribution of alleles,
genotypes, and phylogenetic groups in evolving populations. The results of each
individual research area add to the current understanding of microbial ecology and

evolution by incorporating information about populations from natural (biotic and

abiotic) sources.

DEDICATION
This work is dedicated to my parents and family for their unwavering commitment to
my intellectual pursuits, their support during times of ﬁnancial inadequacies, and,

above all, their love.

Also, this work is dedicated to Thomas Whittam, a leader, mentor, role model, and

friend. His dedication to my career and the careers of all students under his

advisement speaks loudly of his outstanding character and scientiﬁc professionalism.

iii

ACKNOWLEDGEMENTS
I wish to thank my mentor, Thomas S. Whittam, for his gracious advisement
and faith in my abilities. I thank my committee members, Elizabeth Alm, Vincent
Young, Richard Lenski, and Edward Walker for their guidance and suggestions.
Also, I thank Lisa Craft and Pat Resler for their secretarial support.
The members of the Microbial Evolution Laboratory have been instrumental E

to the completion of this research by providing valuable insight, experience, and

 

encouragement. I wish to thank Teresa Bergholz, David Lacher, Adam Nelson,
Lindsey Ouellette, Galeb Abu-Ali, Hans Steinsland, Priya Kailasan Vanaja, Amber
Cody Springman, James Riordan, Shannon Manning, Anthony Heidt, Julie
Cunningham, Jaclyn Middleton, Janice Mladonicky, Lukas Wick, Weihong Qi, Lisa

Calhoun, Shanda Birkland, and Christopher Moeller.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................ x
LIST OF FIGURES .......................................................................................... xi
Chapter 1 Overview .......................................................................................... 1
S. ENT ERICA SEROVAR ENTERITIDIS ................................................. 2
ANTIBIOTIC USE PROVIDES EXPERIMENTAL
OPPORTUNITY ................................................................................... 4
SECONDARY HABITAT E. COLI ARE UNDERSTUDIED .................. 4
A NOVEL EVOLUTIONARY PERSPECTIVE ....................................... 5
Chapter 2 The clonal structure of Salmonella enterica serovar
Enteritidis isolated from mice, chickens, and humans ................................ 6
INTRODUCTION ...................................................................................... 7
Reservoirs and transmission ................................................................. 7
Transmission within chickens ............................................................... 8
Transmission to humans ....................................................................... 9
Factors in the emergence of SE ............................................................ 9
EXPERIMENTAL PROCEDURES ........................................................... 12
SE isolates ............................................................................................. 12
Multilocus enzyme electrophoresis (MLEE) ........................................ 12
Dendrograms of genetic relatedness and eBURST analysis ................. 14
RESULTS ................................................................................................... 1 5
Genetic diversity and clonal analysis .................................................... 15
DISCUSSION ............................................................................................. 21
ACKNOWLEDGMENTS .......................................................................... 23

Chapter 3 The inﬂuence of antibiotic selection on the population
genetic composition of Escherichia coli from conventional and

organic dairy farms ........................................................................................... 24
INTRODUCTION ...................................................................................... 25
EXPERIMENTAL PROCEDURES ........................................................... 28

E. coli strain collection ......................................................................... 28
ECOR phylogrouping by multiplex PCR ............................................. 29
Resistance loci and class I integron PCR .............................................. 30
Statistical analyses ................................................................................ 30
RESULTS ................................................................................................... 32
Overall abundance of E. coli phylogroups ............................................ 32
Genetic composition of sensitive and resistant E. coli
populations ...................................................................................... 32
The inﬂuence of multi-drug resistance on genetic
composition ..................................................................................... 35
Tetracycline and ampicillin resistance determinants ............................ 39
Genetic composition and resistance determinants ................................ 4O

Genetic composition and class I integrons ............................................ 42

DISCUSSION ............................................................................................. 43
The rate of compositional change in conventional-resistant
populations ...................................................................................... 43
Evidence for clonal resistance dynamics .............................................. 44
Evidence for hitchhiking of resistance loci ........................................... 45
Age effects on genetic composition ...................................................... 46
Multi-drug resistant inﬂuences on genetic composition ....................... 47
Conclusion ............................................................................................ 48
ACKNOWLEDGMENTS .......................................................................... 49
Chapter 4 Genetic diversity and population structure of
Escherichia coli from freshwater beaches ........................................................ 50
INTRODUCTION ...................................................................................... 51
EXPERIMENTAL PROCEDURES ........................................................... 54
Beach site description and strain isolation ............................................ 54
Species delimitation .............................................................................. 54
Phenotypic proﬁling (biotyping) ........................................................... 56
Multilocus enzyme electrophoresis (MLEE) ........................................ 56
DNA isolation and multilocus sequence typing (MLST) ..................... 58
Estimation of diversity .......................................................................... 59
ECOR phylogrouping by multiplex PCR ............................................. 59
Phylogenetic analysis ............................................................................ 60
Linkage disequilibrium and recombination .......................................... 60
RESULTS ................................................................................................... 62
Biotype diversity ................................................................................... 62
Levels of protein polymorphism by MLEE analysis ............................ 62
DNA sequence polymorphism revealed by MLST analysis ................. 67
Abundance and distribution of phylogroups ......................................... 72
Recombination and linkage disequilibrium .......................................... 74
Phylogenetic analysis of genetic diversity ............................................ 76
Phylogroup transition events ................................................................. 80
DISCUSSION ............................................................................................. 84
Genetic diversity of environmental E. coli ........................................... 84
Lack of geographic differentiation among sites .................................... 85
Linkage disequilibrium and recombination in nature ........................... 86
Accuracy of the phylogrouping PCR technique ................................... 89
The ET-l clade ...................................................................................... 90
Conclusions ........................................................................................... 91
ACKNOWLEDGMENTS .......................................................................... 93
Chapter 5 A revised molecular phylogeny for Escherichia coli ....................... 94
INTRODUCTION ...................................................................................... 95
EXPERIMENTAL PROCEDURES ........................................................... 99
Strain collection and sequence source .................................................. 99
Conﬁrmation of E. coli ......................................................................... 103

vi

DNA isolation and multilocus sequence typing (MLST) ..................... 104
Descriptive statistics, Tajima’s test, and recombination

analysis ............................................................................................ 104
RESULTS ................................................................................................... 1 12
New primers for MLST of 24 housekeeping loci ................................. 112
MLST analysis ...................................................................................... 1 12
Recombination analysis ........................................................................ 116
Tajima’s test for group rates of evolution ............................................. 121
Phylogenetic construction and divergence time ................................... 121
DISCUSSION ............................................................................................. 124
Divergent bacterial lineages within E. coli ........................................... 124
The natural habitat of divergent E. coli strains ..................................... 125
Conclusions ........................................................................................... 127
ACKNOWLEDGMENTS .......................................................................... 128
References ......................................................................................................... 129

vii

LIST OF TABLES

Table 1. Sources of Salmonella enterica serovar Enteritidis by
years of isolation. .............................................................................................. 13

Table 2. Multilocus allele proﬁles deﬁning 26 ETs of 196 isolates
of Salmonella enterica serovar Enteritidis. ....................................................... 16

Table 3. The best-ﬁt models explaining the frequency of sensitive
and resistant E. coli in conventional and organic diary farm ............................ 34

Table 4. Log-linear modeling of signiﬁcant associations between
farm type (F), multi-drug resistance (D), and ECOR
phylogrouping (E). ............................................................................................ 37

Table 5. Log-linear modeling of farm type (F), resistance gene
(G), and ECOR phylogrouping (E). .................................................................. 41

Table 6. The number of bacterial isolates recovered at 6 sampling
sites along Lake Huron and the St. Clair River of Michigan. ........................... 63

Table 7. Phenotypic proﬁle, number, and overall percentage of the
four most common biotypes at each sampling site and nine

unknown biochemical biotypes ......................................................................... 64

Table 8. Electrophoretic variation of 18 housekeeping enzymes
analyzed by MLEE. .......................................................................................... 65

Table 9. Nucleotide variation within seven housekeeping genes
analyzed by MLST ............................................................................................ 69

Table 10. Recombination analysis based on sequence
comparisons. ..................................................................................................... 75

Table 11. Possible gene loss/acquisition events (A.) and

phylogrouping results (B.). ............................................................................... 81
Table 12. E. coli, E. albertii, and S. enterica strains information ....................... 100
Table 13. E. coli MLST primers for 24 housekeeping loci. ............................... 105
Table 14. Results of single locus analyses. ......................................................... 118

viii

LIST OF FIGURES
Images in this dissertation are presented in color.

Figure 1. Dendrogram of the genetic relationships of 26ETs of SE
inferred from electrophoretic allelic variation detected at 18
enzyme loci. ................................................................................................... 1 8

Figure 2. eBURST diagram of the main clone complex of SE ........................ 20

Figure 3. Histogram plots of ECOR phylogroups for sensitive and
resistant E. coli populations from conventional (A) and organic
(B) farms. ....................................................................................................... 33

Figure 4. Mosaic plot of the dependent associations between farm
type (F), multi—drug resistance (D), and ECOR phylogrouping
(E). ................................................................................................................. 38

Figure 5. Chaol diversity (36) estimates for the total number
multilocus genotypes (STs) based on MLST analysis. .................................. 71

Figure 6. Distribution of E. coli phylogroups across the 6
sampling sites. ................................................................................................ 73

Figure 7. Phylogenetic relationships of freshwater beach E. coli
based on MLST analysis and ECOR groupings. ........................................... 77

Figure 8. Divergent E. coli strains isolated ﬁ'om freshwater
beaches in Michigan. ..................................................................................... 97

Figure 9. Position of 24 housekeeping loci around the E. coli K-12
MG1655 chromosome. .................................................................................. 112

Figure 10. Neighbor-joining tree (panel A) and Neighbor-Net
(panel B) of E. coli, Environmental group 1, Environmental

group II, E. albertii, and S. enterica based on 24 housekeeping
loci .................................................................................................................. 113

Figure 11. Pairwise sequence divergence between E. coli and
environmental group 1, environmental group II, E. albertii, and S.
enterica. ......................................................................................................... 1 16

Figure 12. Neighbor-Net of E. coli, Environmental group I,

Envrionmental group II, E. albertii, and S. enterica based on 8
GARD fragments (fragments 1, 5, 6, 7, 13, 25, 37, and 39). ......................... 122

ix

CHAPTER 1

OVERVIEW

Enteric bacteria are members of the microbial milieu found in the
gastrointestinal (GI) tract of humans and animals. These organisms belong to the
bacterial family Enterobacteriaceae and represent >1000 different species (47). Most
organisms in this family have yet to be cultured and are known to exist based solely
on comparative genomic analyses (95, 166). The majority of what is known about the
biology of this family has been inferred from analyses using representative strains
that are easy to culture and amenable to laboratory growth and manipulation. In
addition, studies of enterics tend to focus on organisms that either cause disease or
have similar ecological requirements as pathogens so that discoveries have a direct
impact on public health (107).

Species within the Enterobacteriaceae are ecologically diverse. For example,
the enteric bacterium Escherichia coli is widely distributed among vertebrate hosts
(56), is one of the most commonly cultured enteric bacteria from mammals (57), and
often does not cause disease (174). Likewise, Salmonella enterica has a wide host
range and can be carried asyrnptomatically by hosts (147). However, certain strains
within each of these species are remarkably cell-type adapted, wherein they replicate
at the expense of the host (46). Such a wide diversity of host-bacterial interaction and
the rapidity with which they can be cultured are two useful characteristics that make
species, such as E. coli and S. enterica, excellent model organisms for studying the
natural history of enteric bacteria (147) and they are the subjects of this dissertation.

S. enterica serovar Enteritidis. The species designation, S. enterica,
contains >2,500 serotypes, or combinations of two antigens; the lipopolysaccharide

(0 antigen) and ﬂagellin proteins (H antigen) (29). S. enterica serovar Enteritidis

(SE) is one of these serotypes and, like most Salmonella, has the ability to invade
epithelial cells that line the gastrointestinal tract of warm-blooded vertebrates.
Human infection oﬂen results in massive ﬂuid secretion and diarrhea, fever,
bacteremia, and an estimated 16,000 hospitalization per year in the United States (86,
147). During the 1990’s, there was a human pandemic of Enteritidis and it soon
became the most commonly isolated Salmonella serovar in the United States (3).
Since then, the US. incidence of all types of sahnonellosis has decreased, while the
incidence of Enteritidis has increased (3). It remains a serious health problem in the
US. and other countries around the world.

Most information about the natural history of SE comes from epidemiologic
investigations of outbreaks and sporadic cases of disease. For example, human
infections in the US. are most often associated with shell eggs (142) and the
consumption of contaminated chicken products (86). Examination of poultry rearing
facilities has shown that wild animals, such as birds and mice, signiﬁcantly inﬂuence
the persistence of SE infection in chickens (39). There is also evidence that this
serotype can be transmitted vertically in chickens, as invading cells enter the
reproductive tract of the hen and into the developing egg (72). These observations
have led to a hypothesis that SE is maintained in nature by zoonotic infection, is
spread to chickens via the ingestion of contaminated fecal material, and is transmitted
to humans by chickens via undercooked eggs. The work presented in the second
chapter of this dissertation addresses this hypothesis through the description of a
population genetic based analysis of an assembled collection of Enteritidis strains

from these different hosts.

Antibiotic use provides experimental opportunity. Quantifying the
inﬂuence of evolutionary forces, like selection, in natural populations of bacteria
often requires the interpretation of many complex and interacting variables. One
approach to these types of analyses has been to evolve strains in the laboratory under
conditions where variables can be controlled (43). Another approach relies on
comparisons between populations that experience selection and those that do not.
The latter approach has the advantage of incorporating stochastic variation into the
experimental design. However, certain effects are difﬁcult to control and may result
in a variable effect size that is too subtle to observe under natural conditions.

Selection caused by the use of antibiotics is an anthropomorphic evolutionary
force, the impacts of which are not well understood. Speciﬁcally, therapeutic and
sub-therapeutic doses of drugs are common in clinical and agricultural settings, but
there are few data with which to assess these inﬂuences. The third chapter of this
dissertation takes advantage of a collection of E. coli from conventional farms that
commonly use antibiotics and organic farms, where antibiotic use is rare (136). This
type of study represents a natural experiment, which can be used to assess the
inﬂuence of selection on bacterial populations in nature.

Secondary habitat E. coli are understudied. Enteric bacteria, such as E.
coli, circulate between two habitats (137). The ﬁrst habitat, called the primary
habitat, is the GI tract of humans and animals. The secondary habitat is broadly
deﬁned as the environment (water, sediment, and soil). Much of the information
concerning the evolutionary ecology of E. coli has been collected from host-

associated (primary habitat) strains. Perhaps this was a historical oversight, but more

likely, it was the result of a bias (either funding or human interest or both) toward
strains of clinical and laboratory importance. Regardless, population dynamic
characterizations of E. coli in the secondary habitat are understudied, but important.
For example, cell counts of this species have long been used by the Environmental
Protection Agency as an indicator of fecal pollution (167). In addition, enteric
bacteria are oﬁen Spread via the fecal-oral route and it has been estimated that half of
all E. coli cells are currently in transit between host GI tracts, in the secondary habitat
(137). This leads to a number of questions: What, if anything, are they doing there?
Do certain microenvironments outside the host promote growth and reproduction?
How does selection operate on these populations? The results presented in the fourth
chapter of this dissertation build on current E. coli evolutionary biology by adding a
population genetic perspective with relevance to dynamics in a particular secondary
habitat — the freshwater beach environment.

A novel evolutionary perspective. As discussed brieﬂy in the previous
section, an interesting hypothesis is that certain aspects concerning the natural history
of E. coli have been biased by popular strains and strain collections. If this is true,
strains from under-represented sources will provide new insights into how this
species has evolved. The ﬁfth and ﬁnal chapter of this dissertation attempts to Show
how E. coli strains from a novel environment suggest a new evolutionary history for

this species.

CHAPTER 2
THE CLONAL STRUCTURE OF SALMONELLA ENTERICA SEROVAR

ENTERITIDIS ISOLATED FROM MICE, CHICKENS, AND HUMANS

INTRODUCTION

Salmonellae are common gastrointestinal pathogens of humans and animals.
In the last three decades, Salmonella enterica serotype Enteritidis (SE) has emerged
to become a major food-bome pathogen (142, 168). Investigations of both outbreak
and sporadic cases of infection identiﬁed eggs and chickens as vehicles for the
majority of cases (66, 158), which indicated that egg-laying chicken ﬂocks are a
major reservoir for SE. This zoonotic serotype led to a pandemic of the disease in
human populations in Europe and the United States (132). In 1987, 9 of 21 (43%),
countries reported SE as their most common serotype, and 8 of 9 (89%), were
European countries. Several epidemiological investigations have incriminated eggs
in human outbreaks of SE in the United States and Europe (66, 158). In the United
States, SE ranks second to Salmonella enterica serotype Typhimurium as the cause of
thousands of cases of salmonellosis each year (6, 105).

Reservoirs and transmission. As with many foodbome pathogens, SE is of
zoonotic origin and has various routes of transmission into humans. It is most readily
spread between animals via the fecal-oral route (i.e. through ingestion of
contaminated fecal material). Strains often circulates in wild animals and, in
particular, have often been recovered from rodent populations associated with
chicken farms (39, 67). As is also seen with infections caused by its relative,
Salmonella enterica serotype Typhimurium, SE is often highly pathogenic to
laboratory mice, causing systematic infection and mortality (165). However, the feral

domestic mouse (Mus musculus) is a frequent carrier that survives infection. In a

study of captured mice from hen houses on farms, approximately 20% of the
population had spleens positive for SE (61). It has been suggested that
physiologically diverse subpopulations of SE are propagated in mouse populations,
and some have an enhanced ability to infect hens (59).

Transmission within chickens. There are no easily observable disease
symptoms in poultry when SE colonizes the avian intestinal tract, which may in part
be due to the ability of the pathogen to actively mitigate signs of illness in the hen
(119). SE contaminates eggs either on the shell surface through contact with fecal
material or by contamination of the internal contents as a result of infection of the
reproductive tissue of laying hens, resulting in transovarian transmission (72, 108).
The pathogen has been cultured from the albumen (white) of the egg, the vitelline
(yolk) membrane, and the yolk (52, 72). Contamination of the vitelline membrane
and yolk follows adherence to and invasion of the granulosa cells of the preovulatory
follicles (161, 162), whereas the oviduct and fallopian tube are infected by bacteremic
spread (40) and may particularly involve colonization of tubular cells (40, 119).
While contamination of the preovulatory follicles may be facilitated by expression of
ﬁmbriae, the absence of ﬁmbriae appears to facilitate bacteremia and translocation of
the bacterium to the oviduct (41). Thus, access to the internal contents of the egg
may involve 2 fundamentally different anatomical routes, namely the albumen or the
yolk and its associated membranes. Both routes may contribute to the outbreak
potential of the organism. However, only oviduct transmission differentiates the
biology of SE from that of the classical pathology associated with S. enterica serovar

Pullorum, which is a serious poultry pathogen that generates substantial ovarian

pathology but does not cause illness in humans (150). The production of eggs with
Salmonella-infected contents not only serves as the primary vehicle for human
disease, but it can also perpetuate the bacteria through vertical inheritance via
infected chicks hatched from contaminated eggs (72, 161). This route of transmission
may contribute to additional outbreaks of SE that follow consumption of meat from
broiler chickens that were initially infected in hatcheries.

Transmission to Humans. Transmission of SE to humans occurs primarily
through contaminated food, most often via uncooked eggs, egg-containing foods, and
poultry. The Centers for Disease Control and Prevention (CDC) has reported that
80% of 371 outbreaks of SE in the United States between 1985 and 1999 were egg-
associated (120). SE-contaminated eggs are estimated to have accounted for
approximately 180 000 illnesses in the United States in 2000 (142). Non-egg-
containing foods have also been associated with SE infections. One notable example
is an outbreak of salmonellosis in North America in 2001 that was linked to
contaminated almonds (30). Subsequent investigation found that SE phage type (PT)
30 was detected in raw almonds collected from multiple sources, including
environmental swabs of orchards and their associated processing equipment, which
suggests that, under some environmental conditions, there is diffuse contamination
(76).

Factors in the emergence of SE. A variety of factors have been implicated
in the dramatic rise in the incidence of SE in the past 4 decades, including changes in
farm practices, eradication of competing Salmonella strains, and virulence evolution

of the pathogen. Farm practices clearly contribute to the incidence of Salmonella.

For example, molting strategies can inﬂuence bird susceptibility to infection (51, 70).
Rodents that inhabit farm environments are a principal source of infection for ﬂocks
(38). It has been suggested that the use of rodenticides containing certain strains of
SE for vermin control may have contributed to the initial spread of this organism
(116). However, the SE isolates from rodenticides used belong to PT6a, while the
majority of human SE epidemics were attributable to PT4 in England and PT8 and
13a in the United States (163).

Eradication of S. enterica serovars Pullorum and Gallinarum from chicken
ﬂocks during the 19605 has also been suggested as an important emergence factor. In
essence, some argue that an open niche resulted from the eradication, thereby
fostering the spread of SE (11). This hypothesis is based on the observation that all 3
avian serovars have the 09 surface antigen as the immunodominant epitope. In
addition, multilocus enzyme electrophoresis (MLEE) and sequencing of the ﬂagellin
gene (ﬂiC) indicate that the nonmotile Pullonun/Gallinarum complex is monophyletic
and shares a recent ancestor with SE (96). There is some evidence that S. enterica
serovar Typhi displaces SE in human populations, because India is one of the few
reporting countries that have a high incidence of Typhi and a low incidence of SE
(60). In addition, research has shown that both SE and Typhi are capable of
producing a type of capsule that is associated with production of high-molecular mass
1ipopolysaccharide(LPS; (60, 125)).

A third factor in SE emergence, the evolution of virulence evolution, garners
support from several types of observations. There is evidence for substantial

variation in virulence among different SE strains. For example, Solano and

10

colleagues discovered that SE strains with high virulence in a chicken model
produced ﬁlaments and aggregates in vitro and were, thus, phenotypically
distinguishable from low-virulence strains (155). Dominant PTs of SE isolated in the
United Kingdom (PT4) and in the United States (PT13a) Show substantial within-PT
variation that results in some strains having enhanced durability characteristics as
well as different abilities to contaminate eggs and to grow to high cell density (34, 60,
73). Strains of SE also vary in the production of a type of capsule associated with
production of high-molecular mass LPS. For example, isolates from eggs and avian
reproductive tissue are more likely to produce high-molecular mass LPS than isolates
from avian intestines or rodent samples, suggesting that this special LPS is critical for
infection of the reproductive tract (59).

To begin to address how SE has evolved and adapted to its multiple hosts and
modes of transmission, we have assessed the clonal relatedness of isolates obtained
from various sources over the past 25 years. Clonal relatedness among isolates was
determined by multilocus analysis of conserved genes with housekeeping functions.
We found that most isolates of SE collected over the past few decades have the same
multilocus genotype, suggesting that the strains involved in the human epidemic mark
a single widespread clone. In addition, we found evidence that closely related, but
genetically distinct clonal variants may be shifting to new hosts in wild animal

populations.

11

EXPERIMENTAL PROCEDURES

SE isolates. A collection of SE isolates spanning the observed emergence of
this pathogen (1978-2004), including isolates from non-human sources (164) to
isolates that were recently collected from human and non-human sources are listed in
Table 1. All isolates were conﬁrmed biochemically and serologically at the National
Veterinary Service Laboratories (Ames, IA) and were stored at —80°C in tryptic soy
broth containing 15% glycerol. Three Enteritidis strains from the Salmonella
reference collection B (SARB; (22)) were obtained from E. Fidelma Boyd (National
University of Ireland - Cork) and served as genetic control strains.

Multilocus enzyme electrophoresis (MLEE). Enzyme extraction, gel
electrophoresis, and speciﬁc enzyme staining was carried out as described in Selander
et al. (144). Brieﬂy, lysates (whole cell enzymes) were extracted ﬁom overnight
nutrient broth cultures and frozen at -80°C. Lysate samples were individually loaded
and electrophoresed under non-denaturing conditions in a buffered starch gel matrix
at the appropriate concentration for the particular enzymes being stained. Gel slices
were incubated in enzyme speciﬁc staining solutions and ﬁxed for analysis.

The mobilities of 18 housekeeping enzymes were recoded for 196 isolates
spanning the years and sources of the collection (Table l). Mobilities were scored
relative to previously characterized Salmonella strains from the SARB collection.
The enzymes used in this study were: ADH (alcohol dehydrogenase), THD
(threonine dehydrogenase), SKD (shikimate dehydrogenase), G6P (glucose-6-
phosphate), MPI (mannose phosphate isomerase), GLUD (glutamate dehydrogenase),

MDH (malate dehydrogenase), NSP (nucleoside phOsphorylase), PEP (peptidase),

12

Table 1. Sources of Salmonella enterica serovar Enteritidis by years of isolation.

 

 

Source 1978-1987 1990-1992 1994-1999 2000-2004 Total
Environment 0 l 09 1 95 0 3 04
Mice O 10 1 0 1 1
Chickens 28 136 325 O 489
Eggs 0 42 63 0 105
Humans O 121 202 24 347
Others 2 2 4 9 1 7
Total 30 420 790 33 1273

 

13

GOT (glutamic oxalacetic transaminase), CAK (carbarnylate kinase), AK (adenosine
kinase), MPD (mannitol-l-phosphate dehydrogenase), PGD (6-phosphogluconate
dehydrogenase), PGI (phosphoglucose isomerase), IDH (isocitrate dehydrogenase),
ACO (aconitase), LDH (lactate dehydrogenase). Electrophoretic mobility variants
(electromorphs) were assigned scores in Side-by side comparisons relative the SARB
standards. Electromorphs were equated with alleles at the corresponding enzyme
locus based on the assumption that each mobility difference reﬂects at least one
amino acid replacement in the protein. Electrophoretic types (ETS) were assigned to
isolates with indistinguishable allele proﬁles across all enzyme loci studied. Allele
frequencies and ETs were used to estimate population genetic parameters as
described in Selander et al. (144).

To make quantitative estimates about the genetic relationships among isolates,
we equate electromorphs with alleles at an enzyme locus and electrophoretic types
(ETS) with multilocus genotypes. We assume that isolates with the same ET owe
their genotypic similarity to recent descent from a common ancestral cell; that is, they
are members of a naturally occurring bacterial clone.

Dendrograms of genetic relatedness and eBURST analysis. By comparing
the multilocus proﬁles of sampled populations, we assessed the genetic relationships
among ETs using a distance-based neighbor joining algorithm for dendrogram
construction and an eBURST analysis of allelic proﬁles (using the eBURST

algorithm) described in Feil et al. (49).

14

RESULTS

Genetic diversity and clonal analysis. Characterization of 196 SE strains by
MLEE of 18 housekeeping enzymes resolved an average of 2.7 alleles per locus and
distinguished 26 distinct electrophoretic types (ETS). Fourteen of the 18 loci were
polymorphic with the number of alleles per locus ranging ﬁom 1 to 6 for NSP (Table
4). The average single—locus genetic diversity was 0.051 (:1: 0.012) across 196 isolates
and 0.246 (:t 0.051) among the 26 ETS. The difference in these two measures of
diversity reﬂects the fact most SE strains belong to a single ET. Of the 26 ETS
identiﬁed, 19 (73%) were recovered only once. There were 7 ETs that were
represented by more than one isolate. The most common genotype, ET-3, accounted
for 151 (77%) of the 196 Enteritidis strains (Table 2). The most common ET also
included the SARB 16 reference strain. These MLEE results are consistent with the
previous ﬁndings of Boyd et al. (22), and indicate that most Enteritidis strains,
marked by ET-3, belong to a single widespread clone.

The dendrogram of the overall genetic relatedness among the ETS Shows that
ET-3 falls near the node of a cluster of 13 ETS at the top of the diagram (Figure 1,
cluster A). This cluster of ETS represents the most common genotypes of SE
circulating in various sources. The lower half of Figure 1 includes 13 divergent ETS
only one of which was isolated more than once. We suspect that these represent
recombinant genotypes that are rare in populations and that have evolved by
horizontal gene transfer.

To further assess genetic relatedness between the ETS of SE, we used the

allele proﬁles to perform an eBURST. The analysis focuses on discerning the

15

m m m m m o
m m m v m c
m m m m m We
m m m m m md
m m m m m md
m m m m m N
m m m m m md
m m m m m md
m m m m m md
m m m m m md
m m m m m md
m m. m m. M ms
m m m m m md
m v m o m md
m N m m m md
m N m m6 m md
m N m mNé m We

MMMMMMMMNVQMMMMMM

MMMMMMMMMMMMMMMWW

MMMMMMMMMMMVMNVQV

m N v m m ~ w 5
m m v m m _ v. c—
m N v N m H N. 2
m m v m m N N E
N v m m m m S 2
m m a m m g a N—
m m w m m ﬂ N~ :
m n v v m ﬂ : o_
m m v m m H VN a
m m v v m N S w
m m v m N .3 NN A.
m. m V m. m. N: m b
m m m m m _ oN m
m m v m m _ _N v
m m o m m N 0N m
m m v m m _ 2 N
m m w m m ~ ~ _

 

 

52 VH<U DAG mmm AGO QMm Ema. EOE mmZ

“22 Gem :3 03. me mamas am See

 

962 oE>Nco mo 86:3 23.888805

a

mo .02 no;

AXOD Gm 603.88 0.3 PM 208800 Dr: 69¢ .5me Hag

86—? 98 82m: 5 ﬂ Amémv Hm 5588 308 25. .N 2sz E Emewoaccop 2: mo 5—6.5 05 E @2585 can mhm Anna can
.900 02 40% 03908988 803 62 f 05 mo Sam .62 ”VERSE—om 3 “a 3382 9:85?» 3:395 33080583 3
tenant 8e moEoE .mmEEoEm 32:3 85:8 ezocogam mo mos—0mm 03 .«o mﬁm ON mic—mop 330E 22? mace—2:2 .N 033.

16

.5338 2835 038083 on .oH .223 :5: a moncmeoHc o

 

 

 

a
N m m m m H m m v v m v N m H m H N

m w m w v H m m v N m v N m H m H m N

N m m w m o m m v m m v N m H 3 VN

N m m m m H m m a We m v v m H N N

m m m m m H m m m m m v m m N w H NN

m m m e m H m m e m N v m m H m H N

m m m m m H m m w m N v m m H m ON

m m m m m mu... m m e m N v m m H N a H

m m m m m H m m w m N v m m H o w H
Hm: v20 DAG nHmHnH AGO @va HHQH. HHQSH AHmZ H H E QOAH HHQH OO< HHS monHOmH .H.mH .530
£52 0555 .3 66:3 mHHEoEoboonH .Ho .02 02H.

3

assuage N 29¢

17

 

 

1-SARB18 (1)
_| E 16-2T21 (1)
26-WSU 9103 (2)

21-KES13 (1)

‘— 20-C473 (1)

pdlstance 3'3““ (151)

*-—-* 22-WSU481 (15)

' 104120" (2)

24-WSU8780 (1)
114'6204’ (1)

A- 12-Hs1o+ (1)

17-H172 (3)

‘— 25—WSU9058 (2)

7-l061N342 (1)

4-ICB1H62 (1)

34%95251 (1)

640611-161 (1)

23-WSU2316 (1)

54061116531 (1)

9-I 061N522 (1)

LE— 13—SH 95 271 (2)
19—CECA95191 (1)

2-SARB17 (1)

141-0 95501 (1)

, [ 13-ce5521(1)
1scs5431r1)

 

 

 

   
  
  
   
  

 

 

 

 

 

 

 

 

 

 

 

Figure 1. Dendrogram of the genetic relationships of 26ETs of SE inferred from
electrophoretic allelic variation detected at 18 enzyme loci. The dendrogram was
constructed using a neighbor-joining algorithm and proportional distance (p-distance)
between ETS. The common ET-3 is highlighted in red. The number of isolates of

each ET is given in parentheses.

l8

relationships of multilocus genotypes in which both mutation and recombination can
be a source of genotypic divergence (48). Application of the eBURST analysis
identiﬁed ET-3 as the founding genotype based on the number of single locus
variants connected this electrophoretic type (Figure 2). The BURST diagram shows
that the 12 ETs of cluster A are closely related to ET-3 as single or double locus
variants. The eBURST analysis indicates that ET-3 represents the founding clonal

genotype of SE that has diversiﬁed with the epidemic spread of the clone.

19

. ET-ll

 

Mo 4
ET-12 ET“ 0
ET -21
_ 4
ET 16 pr().5 TDH4 ACO

ET-26 Ow:

CAK2

a TDH 2 ET-24
( ’
LDH 5
ET-ZO IDHZ
Nspz-5
POD 2 ”'2
o

ET—ZS
' ET-1 7

Figure 2. eBURST diagram of the main clone complex of SE. ET-3 is the founder
genotype (center) that is connected to related ETS that are single locus variants
(SLVs). The inferred allele substitutions, which create the SLVs, are given above the

lines. For example, ET-2, the second most commonly recovered clone, is a SLV that

differs by [DH2 ﬁ'om ET-3.

20

DISCUSSION

This genetic characterization demonstrates that most SE strains have an _
identical multilocus enzyme genotype (ET-3) regardless of PT, geographic origin, or
time of isolation that spanned over 2 decades (1978 to 2004). These data are
consistent with the hypothesis that these strains are members of a single widespread
clone (Table l). The ET—3 strains include 7 different PTs, with a predominance of
PT28 (42%), PT8 (28%), and PT13a (18%). PT4, the most observed type among
recent European outbreaks, was also found among ET-3 strains, but at a relatively
low frequency (~4%). While this genotype has been observed previously (22), it is
noteworthy that 23 of the 26 ETS found in this study are not represented among the
SE strains of the SARB collection.

An eBURST analysis identiﬁed ET—3 as the founding genotype for a group of
clonally related, but genetically distinct strains. These new ETs are deﬁned either by
novel alleles or new combinations of preexisting alleles. This analysis suggests that
ET—3 marks the ancestral clone that founded these genotypic clusters, but subsequent
nucleotide change (either single base pair mutations or recombination of horizontally
transferred alleles) has generated new variants. The identiﬁcation of a number of
more distantly related SE types suggests that these strains have radiated even further
through similar mutational events or through the spread of the SE-distinct LPS
antigen region into divergent backgrounds. In any case, these distant variants are
distinctly different from the ET-3 cluster, are rare in this collection, and do not appear

to be commonly associated with human disease.

21

SE strains of 3 major PTs (8, 13a, and 28) were distributed among 24
genotypes that were carried by a range of animal hosts, including chickens, cattle,
horses, and wild animals. These data suggest that SE strains have evolved the ability
to cross a number of host barriers, a research area that warrants further study. SE
strains from clinical cases of human gastroenteritis belong to the same bacterial clone
(ET-3) as strains from eggs, mice, and chickens. Our interpretation is that the ET-3
clone has evolved via successful adaptation to the avian host, which serve as a
reservoir for pathogenic strains. While this clone may not cause overt disease in
mammals, the ability of some strains to cause severe gastroenteritis in humans could
be due to the emergence of highly pathogenic variants. If this is the case, strains from
chickens and human disease may be differentiated by phenotypic virulence

characteristic in cell culture.

22

ACKNOWLEDGMENTS
This project is funded in part with federal funds from NIAID, NIH, and the
Department of Health and Human Services, under Contract No. N01-AI-30058. This
study was part of a published paper (Saeed, A.M., Walk, S.T., Arshad, M., Whittam,
TS. 2006. Clonal structure and variation in virulence of Salmonella Enteritidis
isolated from mice, chickens, and humans. Journal of AOAC International 89(2):

504-511.).

23

CHAPTER 3

THE INFLUENCE OF ANTIBIOTIC SELECTION ON THE POPULATION

GENETIC COMPOSITION OF ESCHERICHLA COLI FROM

CONVENTIONAL AND ORGANIC DAIRY FARMS

24

INTRODUCTION

Escherichia coli is an indicator species for a variety of anthropogenic
effects on microbial populations, such as the emergence and spread of antibiotic
resistance in agriculture (5, 13-15, 18, 19, 23, 42, 84, 127, 138, 139, 149, 152).
Although most strains are commensal bacteria and nonpathogenic to humans and
animals, there are well-recognized pathogenic strains that can cause a variety of
human and zoonotic diseases. In addition, some commensal populations are
known to carry a high level of antibiotic resistance (10, 117, 118). Such resistant
populations pose a public and veterinary health risk because of the potential
transfer of genetic resistance determinants to pathogens. In addition, certain
virulence factors may be mobilized on genetic elements and transferred to
normally commensal, but antibiotic resistant strains via horizontal exchange (128,
179, 182).

During antibiotic selection in the laboratory, resistance-conferring
mutations often have a measurable deleterious effect (i.e. a resistance cost) due to
a reduction in function of genes where resistance mutations arise. In order to
maintain a competitive advantage over other members of the population, it is
hypothesized that deleterious effects on ﬁtness are compensated by changes
elsewhere in the genome (93, 99, 100, 129, 134, 141). The occurrence of such
compensatory ﬁtness mutations makes it difﬁcult to determine whether the
abundance and distribution of resistant strains is a result of direct selection on the
original mutation that caused resistance, selection on compensatory changes, or

other ecological factors that limit population diversity (environmental selection,

25

bottlenecks, genetic drift, etc.). In addition, there is sound evidence that antibiotic
use increases the abundance of resistant phenotypes (136), but it is not clear if the
cessation of antibiotic use will decrease abundance after compensatory changes
have occurred (8, 94).

Antibiotic use in dairy cattle provides a useﬁrl opportunity to assess the
role of natural selection in bacterial populations for several reasons: the source of
the antibiotic selective pressure is known and the dosage is often recorded; the
common genetic determinants for certain resistant phenotypes have been
characterized and high throughput assays are available for their identiﬁcation;
hypotheses generated under laboratory conditions can be tested in vivo by
comparing bacteria from farms that regularly use antibiotics (conventional) and
bacteria from farms that rarely use antibiotics (organic) (136); and a number of
studies have previously characterized resistance dynamics on both farm types and
have identiﬁed variables that signiﬁcantly inﬂuence the abundance of resistant
phenotypes (5, 13, 15, 42, 127, 136, 138, 139).

The purpose of the present study was to assess the inﬂuence of antibiotic
selection on the genetic composition of E. coli populations from conventional and
organic dairy farms. First, we used a PCR based assay (33) to quantify the
abundance and distribution of 4 phylogenetic groups in populations cultured
during a longitudinal sampling of cattle from matched conventional and organic
dairy farms in Wisconsin (136). We then assessed the pattern of statistical
dependence among farm types (conventional vs. organic), cattle ages (cows vs.

calves), bacterial phenotypes (resistant vs. sensitive), and bacterial genetic

26

composition (ECOR groups A, B1, B2, and D) using hierarchical log-linear

modeling.

27

EXPERIMENTAL PROCEDURES

E. coli strain collection. A total of 678 E. coli strains (367 random
sensitive and 311 resistant strains) were assembled from a collection of 1,121
strains of a longitudinal sampling of 10 randomly selected cows and calves from a
matched set of 30 conventional and 30 organic dairy farms in Wisconsin (78).
Brieﬂy, a cluster of organic farms was selected and the geographically closest
conventional farm was selected for purposes of comparison so as to minimize the
effects of distance (cline effects). All organic farms were certiﬁed by a USDA-
accredited certiﬁcation agency as not having treated adult cows for at least 3 years
(mean, 8 years; range, 3-15 years) prior to this study. More information about
these farms is available at Sato et al. (136).

In the original study (136), fecal samples were taken from 5 lactating cows
and 5 calves (< 6 months of age) at each of two visits (once in March and once in
September) and conducted with aseptic technique. Laboratory isolation was
begun within 72 hours and a single E. coli colony was isolated from each fecal
sample so as to exclude any single farm or within-animal bias. All isolates were
conﬁrmed by standard biochemical assays. Minimum inhibitory concentrations
(MICs) of 17 antibiotics were determined for each strain as recommended by the
NCCLS (109) using a commercially available semiautomatic microbroth dilution
test (Sensititre, Trek Diagnostic Systems Inc, Cleveland, OH) and appropriate
quality control strains. These antibiotics included ampicillin, amoxicillin-
clavulanic acid, cephalothin, cefoxitin, ceftiofur, ceftriaxone, streptomycin,

kanamycin, gentamicin, apramycin, amikacin, tetracycline, sulfamethoxazole,

28

trimethoprim-sulfarnethoxazole, nalidixic acid, and ciproﬂoxacin). Resistant
phenotypes were conﬁrmed by the presence of overnight growth on LB broth
(Lennox; Becton, Dickinson, and Company, Sparks, MD) agar containing
antibiotic at the NCCLS cut-off concentrations. More details about the strain
collection and isolation procedures can be found at Sato et al. (136).

ECOR phylogrouping by multiplex PCR. Strains were grouped into 1
of 4 phylogenetic lineages (A, B1, B2, or D) based on methods adapted from
Clermont et al. (33). Genomic DNA was isolated from 2 mL of overnight culture
in LB broth (Lennox; Becton, Dickinson, and Company, Sparks, MD) using the
Puregene DNA isolation kit (Gentra Systems Inc., Minneapolis, MN.). DNA
preparations were quantiﬁed with a NanoDrop ND-lOOO UV-Vis
spectrophotometer (N anoDrop Technologies, Wilmington, DE.), diluted to a ﬁnal
concentration of 100 ng/uL, and stored at 4°C. Genomic DNA preparations were
tested using primers targeting a 650 bp region of the conserved housekeeping
gene, mdh (see www.shigatox.net/stec/mlst-new/index.htrnl for primer sequences
and reaction conditions), and AmpliTaq Gold ® DNA polymerase (Applied
Biosystems). This protocol has produced a positive amplicon in strains
representing the genotypic diversity of the species as well as E. coli’s most recent
common ancestor, Escherichia albertii. Genomic DNA was re-isolated if the
assay was negative. Strains that were negative for duplicate, independent
genomic isolations were considered to be members of species other than E. coli
and excluded from further analysis. Representative ECOR (E. coli reference

collection) strains were used as template controls for a duplex PCR targeting the

29

genes chuA and yjaA. We found that the following duplex conditions yielded
higher PCR speciﬁcity with AmpliTaq Gold ® than the published triplex:
denaturation at 94°C for 10 minutes; 35 cycles of 92°C for 1 minute, 59°C for 1
minute, 72°C for 30 seconds; and a ﬁnal elongation at 72°C for 5 minutes. A
separate PCR was run with primers targeting the TspE4.C2 anonymous DNA
locus using published conditions (3 3).

Resistance loci and class I integron PCR. Ampicillin resistant (ampR)
and tetracycline resistant (tetR) strains were screened for the presence of 6
previously identiﬁed resistance loci. A multiplex PCR was used to detect the
presence of blaTEM, blaSHv, and blaOXA-1 in ampR strains as reported by Colom et
al. (35). Fragments of the tetA, tetB, and tetC genes were targeted in tetR strains
using the primers and conditions published by Boerlin et al. (19). Three primer
sets were used to determine the presence of class 1 integrons in the resistant
strains. Primers sets targeting the class 1 integrase locus, intII, the conserved
region cassette region A and B, and the quartemary ammonium compound
resistance gene, qacEAI are given along with the reaction condition in Lindstedt
et al. (97). Integron presence was deﬁned as ampliﬁcation of all 3 loci.

Statistical analyses. Strains were categorized for analysis as follows: F =
farm type (conventional vs. organic), A = cattle age (calf vs. cow), D = resistant
phenotype (resistant vs. sensitive) or drug susceptibility level (high, medium, vs.
low), and E = ECOR group (A, B1, B2, and D), and recorded in the cells of
contingency tables. Hierarchical log-linear modeling with nested effects was used

to assess dependent associations using the CATMOD procedure and SAS

30

statistical software (SAS Institute, Cary, NC). Non-signiﬁcant, higher-order
interactions were removed until the most parsimonious model was found based on
the likelihood ratio chi-square statistic for testing goodness of ﬁt (G2). Non-
signiﬁcant G2 values indicated that the ﬁt model was not signiﬁcantly different
from the saturated model. Odds ratios were calculated based on parameter
estimates from the most parsimonious models. Higher-order (three-way)
interactions for multi-drug resistant phenotypes were visualized in mosaic plots,
which were obtained online at http://euclid.psych.yorku.ca/cgi/mosaics. The
original plots were redrawn and shaded with respect to the signiﬁcant (ct = 0.05)

associations from the SAS analysis.

31

RESULTS

Overall abundance of E. coli phylogroups. Strains belonging to all 4
ECOR phylogroups were identiﬁed (Figure 3) among the 678 E. coli strains from
calves and cows on dairy farms. The relative phylogroup composition of these
bacterial populations was used to compare different patterns of antibiotic use.

The populations analyzed here represent the natural variation between farm type
(conventional vs. organic), cattle age (calf vs. cow), and resistance phenotype
(resistant vs. sensitive). It is clear that phylogroup abundance was not evenly
distributed among the different types of dairy farms (Figure 3). The most
abundant phylogenetic groups were B1 (58.3%) and A (27.4%), whereas groups
D (11.5%) and B2 (2.8%) were rare. B2 strains were not sampled at each variable
level (no resistant B2 genotypes were found on organic farms), so these data were
combined with group D strain data (B2D) for statistical analyses.

Genetic composition of sensitive and resistant E. coli populations. Our
initial goal was to test for dependent associations among three nominal variables
(F = farm type, A = age of cattle, and E = ECOR phylogroup, E) by analyzing the
number of strains in these categories. The tests for associations in the sensitive
population (susceptible to 17 antimicrobials) by log-linear modeling of the 376
sensitive strains revealed no signiﬁcant interactions with ECOR phylogrouping
(Tables 3a and 3b). In other words, the distribution of phylogroups in sensitive E.
coli sampled from calves and cows on conventional and organic dairy farms was
similar and not signiﬁcantly different. A signiﬁcant negative association was

found between conventional farms and the number of sensitive calf strains (i.e.

32

A. Conventional farms

Sensitive

n=174
ao~

Percentage

 

 

 

A B1 82

B. Organic farms

Sensitive
n = 191

Percentage

 

 

 

32
E COR group

anal-1:1.

D

.1311

D

Resistant

n = 181
so

 

 

 

A B1 32 D

 

Resistant

80r- [1:130

70-

 

 

. IE.

A B1 B2 D
ECOR group

 

Figure 3. Histogram plots of ECOR phylogroups for sensitive and resistant E.

coli populations from conventional (A) and organic (B) farms (black bars = calf

strains, gray bars = cow strains).

33

Table 3. The best-ﬁt models explaining the frequency of sensitive and resistant E.

coli in conventional and organic diary farm. The analysis is based on testing

hierarchical log-linear models with nested effects in parentheses. Nominal

categorical variables are designated as A = animal age (calf or cow), E = ECOR

group (A, B1, B2, D), and F = farm type (conventional or organic). ,u designates

the overall main effect. A. Likelihood ratio chi-square statistic was used to test

for goodness of ﬁt (compared to the saturated model). B. Signiﬁcant interactions

for the sensitive and resistant populations.

 

 

A. Population model Final model (32 (df, Pr > 112)
Sensitive ,u + A + E + F(A=calf) 8.85 (7, 0.26)
Resistant ,u + A + E + F(E=ECOR A) 5.32 (7, 0.62)

 

 

B. Population model Signiﬁcant interactions x2 (df, Pr > )8)
Sensitive F(A = calf) 4.27 (1, 0.04)
Resistant F(E = ECOR A) 21.1 (1, < 0.0001)

 

34

the F (A = calf) interaction in Table 1b). This result was expected because the
abundance of resistant strains was higher in calves on conventional farms than
calves on organic farms. Despite this discrepancy in abundance, however, these
data indicate that sensitive strains of the four phylogroups were circulating at
similar frequencies on both farm types in young and adult animals.

A similar analysis was applied to the 311 resistant strains and revealed a
signiﬁcant association between farm type and ECOR phylogrouping (Tables 1a
and 1b). Based on parameter estimates, the odds of recovering resistant E. coli of
the phylogroup A were signiﬁcantly greater on conventional farms than on
organic farms (df = 1, x2 = 21 .1, Pr > x2 < 0.0001). This overabundance of
phylogroup A strains was not seen in the sensitive population or the resistant
population ﬁ'om organic farms. In addition, there were no signiﬁcant farm-
phylogroup (i.e. FE) interactions when the sensitive populations from both farms
and the resistant population from organic farms were analyzed together (model
not shown). These data suggest that resistance determinants on conventional
farms were linked to the genetic backgrounds of phylogroup A and that these
strains increased in frequency as a result of antibiotic use. Interestingly, animal
age was not associated with the distribution of phylogroups in the resistant
p0pulation, suggesting that similar phylogroups circulate at similar frequencies in
young and adult dairy cattle.

The inﬂuence of multi—drug resistance on genetic composition. Strains
were categorized according to their level of drug susceptibility (D) as deﬁned by

the number of resistant antimicrobial phenotypes (low = 1 — 2, medium = 3 — 4,

35

and high = 5 or more). A log-linear model ﬁt to the data according to farm type
(F), multi-drug resistance level (D), and ECOR phylogrouping (E) revealed
signiﬁcant heterogeneity in the association between these variables (Tables 4a and
4b) including the presence of a signiﬁcant three-way (F DE) interaction. In other
words, the best-ﬁt model to these data included all three variables. The model
was simpliﬁed slightly by repararneterizing and nesting the variables (F DE
models I, H, and III in Table 4), which allowed non-signiﬁcant levels to be
removed.

To illustrate the complexity of the interactions affecting bacterial multi-
drug resistance, we summarized the components of the E. coli populations using
mosaic plots of three different parameterizations of the FDE log-linear model
(Figure 4). Odds ratios were estimated for signiﬁcant interactions with respect to
a ﬁxed or nested factor. For example, when the effect of multi-drug resistance
was nested, or ﬁxed, a signiﬁcant two-way interaction between farm type (F) and
phylogroup (E) was found and can be seen by comparing the size of the shaded
box to the size of the non-shaded boxes for a given level of drug susceptibility
(D). When the low multi-drug resistance level is considered, it is clear that the
shaded box representing group B] strains on organic farms is larger than the one
for conventional farms. The opposite is true for phylogroup A or B2D strains
(larger boxes for the conventional farm category); hence a signiﬁcant (positive)
interaction is represented by the shaded organic B1 box (df = 2, x2 = 6.3, Pr > x2
= 0.044). The odds of isolating phylogroup A strains with medium multi-drug

resistance were signiﬁcantly higher on conventional farms (df = 2, x2 = 8.3, Pr >

36

Table 4. Log-linear modeling of signiﬁcant associations between farm type (F),

multi-drug resistance (D), and ECOR phylogrouping (E). A. Three separate

parameterizations of the F DE model are given to show statistical dependence as a

function of nested effects. B. Signiﬁcant interactions (3-way) with E.

 

 

 

 

A. FDE model Final model G2 (df, Pr > X2)
I p+F+D+E+FE+DE+FD(E=ECORB1) 5.84 (4, 0.21)
11 ,u + F + D + E + FE + DE + DE(F = Organic) 1.79 (4, 0.78)
III p+F+D+E+FE+DE+FE(D=High)+ Saturated
FE(D = Medium) + FE(D = Low)
B. FDE model Signiﬁcant interactions with E x2(df, Pr > 12)
1 FD(E = ECOR B1) 9.55 (2, 0.008)
11 DE(F = Organic) 18.64 (4, < 0.001)
111 FE(D = High) 6.25 (2, 0.044)
111 FE(D = Medium) 8.29 (2, 0.016)
111 FE(D = Low) 10.34 (2, 0.006)

 

37

01
F7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ECOR group (E) A Bi 820
A B1 820 ' ‘

o .

9, 2 I ll [ " l

01 3

O

c ,

W

.0—0 ,

.9 E

VJ

e 2 jl: S

or g 8

3 2

3-

p ,

E 5 7

3 .U’ 7 Al to D

>: 2'- ' | || | c I «1.3.:
a l H l
i

Conv Org Conv Org ConvOrg
Conv Org Conv Org ConvOrg
Farm type (F)
C 131 d. B1 820

 

 

Low
m
Low
C_J

 

 

 

 

 

Medl um
Medi um
L_____3

 

 

 

 

 

 

Al grow 0

C3D [:1

Conv Org Org Org

th
High

 

Figure 4. Mosaic plot of the dependent associations between farm type (F), multi-
drug resistance (D), and ECOR phylogrouping (E). Shaded boxes mark
signiﬁcant odds ratio estimates (positive odds only). a. Overall mosaic plot for
FDE. b. FE interactions at ﬁxed levels of D. c. FD interactions at ﬁxed levels of

E. (1. DE interactions at ﬁxed levels of F.

38

x2 = 0.016), and the odds of isolating highly resistant, phylogroup B2D strains
were signiﬁcantly higher on organic farms (df = 2, x2 = 10.3, Pr > x2 = 0.006).
As mentioned above, there were no resistant phylogroup B2 strains isolated from
organic farms, so the shaded B2D box on organic farms represents group D
strains only. Phylogroup speciﬁc interactions were also found when model
effects were ﬁxed for E (df = 2, x2 = 9.6, Pr > x2 = 0.008) and F (df = 4, x2 =
18.64, Pr > x2 < 0.001). These data suggest that conventional farms are
associated with medium and highly resistant phylogroup A and B1 strains,
whereas, in contrast, organic farms with virtually no antibiotic use are associated
with low and highly resistant phylogroup B1 and D strains.

The association between cattle age (A), multi-drug resistance (D), and
ECOR phylogrouping (E) was also found to be heterogeneous. The high
abundance of resistant calf strains and limited overall resistance on organic farms
resulted in sampling zeros for 3 of the 9 possible categories in cows (no medium
resistant ECOR B2D strains, highly resistant phylogroup A strains, or highly
resistant phylogroup B1 strains were sampled). After correcting for sampling
zeros in the resistant cows categories, the three-way interaction term (ADE) was
not signiﬁcant in the model. These data suggest that cattle age inﬂuences the
abundance of multi-drug resistant strains, but does not inﬂuence the genetic
composition of this population.

Tetracycline and ampicillin resistance determinants. Of the 311
resistant strains analyzed, 129 (41.5%) were ampicillin resistant (ampR), 281

(90.4%) were tetracycline resistant (tetR), and 112 (36.0%) were resistant to both

39

drugs. Based on PCR screening for 3 common E. coli B-lactamase genes (blaTEM,
blagHv, and blaOXM), 119 (92.2%) ampR strains carried the blaTEM locus, while
the remaining 10 (7.8%) strains did not produce an amplicon for any of the
targeted loci. Similarly for 3 genes known to confer E. coli tetracycline resistance
(tetA, tetB, and tetC'), 268 (95.4%) tetR strains carried at least one of these loci,
while 13 (4.5%) did not. The tetB and tetA genes were the most abundant (64.8%
and 28.1%), while the tetC gene was rarely sampled (4.6%).

Genetic composition and resistance determinants. We created 4
datasets according to the 4 genetic determinants present in the resistant
population. Data from the sensitive population analyzed above were added to
each to create a two level factor (G) for log-linear modeling. The factor “G”
categorized strains that canied a resistance gene (blaTEM, tetA, tetB, tetC) or did
not (sensitive). Due to the low occurrence in the sample, data for tetC strains (11
= 7) were pooled with data for strains that were negative for all 3 loci (n = 13) and
called tetC/other. Data for strains that were negative for the 3 B-lactamase loci (n
= 10) were omitted. Log-linear models were ﬁt to each of the 4 datasets to test for
associations between F, G, and E (Tables 5a and 5b).

Interactions between the resistance loci and ECOR phylogroups were not
dependent on farm type (no FGE interactions). The genetic composition of the
sensitive population was not signiﬁcantly different from the tetA, tetB, or
tetC/other populations (models not shown). The only signiﬁcant GE association
was found in the blame: population (Table 5b), where odds of sampling the

blaTEM locus on conventional farms were signiﬁcantly associated with ECOR

4O

Table 5. Log-linear modeling of farm type (F), resistance gene (G), and ECOR

phylogrouping (E). A. Resistant populations were deﬁned by the determinant

they carried. Notice that GE was a signiﬁcant term in the blaTEM population only.

B. Signiﬁcant GE (two-way) interaction for the blaTEM population.

 

 

 

 

A. Resistance gene model Final model G2 (df, Pr > 02)
tetA p+F+G+E+FE+FG 7.32 (4, 0.12)
tetB p+F+G+E+FE+FG 4.10 (4, 0.39)
tetC/other p, + F + G + E 11.24 (7, 0.13)
bIaTEM 11+F+E+FE+FG+ 3.81 (2, 0.15)
GB
B. Resistance gene model Signiﬁcant interactions X2 (df, Pr > c2)
with E
blaTEM GE 20.84 (2, < 0.0001)

 

41

phylogroup A (df = l, x2 = 5.0, Pr > x2 = 0.025). These data suggest that the
genetic composition of resistant E. coli populations on dairy farms is dependent
on individual resistance determinants.

Genetic composition and class I integrons. All ampR and tetR strains
were screened for the presence of class I integrons based on the presence of 3 loci
(intII, qacEAl, and the conserved cassette region). Of the total 298 ampR and/or
tetR strains, 59 (19.8%) carried a class I integron. We created a 3 level factor
called “integron populations” that was comprised of resistant, integron positive
strains (int+); resistant, integron negative strains (int'); and sensitive, integron
negative strains (sensitive). Log-linear models were then used to test for
signiﬁcant associations between farm type (F), ECOR phylogrouping (E), and
integron populations (1).

There was no signiﬁcant phylogroup-integron (El) interaction with farm
type (no FEI interaction). However, the distribution of ECOR phylogroups was
dependent on integron presence in these populations (df = 4, x2 = 12.4, Pr > x2 =
0.015). The int' and sensitive populations were compositionally the same (Bl > A
> B2D), but phylogroups in the int+ population were evenly sampled. This
analysis suggests that the int+ population had signiﬁcantly more group A and B2D

strains than the other (int' and sensitive) populations.

42

DISCUSSION

In this study, we examined the dynamics of antibiotic selection on
conventional and organic dairy farms by comparing the genetic composition of
resistant and sensitive E. coli populations. The overall abundance of 4
phylogenetic groups was not similar, suggesting that phylogroup B1 strains
colonize at a higher abundance and, therefore, have a higher relative ﬁtness in
dairy cattle. Although phylogroup B1 strains are common in a variety of host
species (56), they were not the numerically dominant group in healthy swine (32)
or humans (112).

The rate of compositional change in conventional-resistant
populations. A key ﬁnding of this study is that there is an overabundance of
resistant phylogroup A strains on conventional diary farms compared to
phylogroup A strains on organic farms where antibiotic use has been limited.
Based on two observations, we are conﬁdent that this overabundance has been a
consequence of antibiotic use and not some other conventional management
practice. First, conventional-sensitive and organic-sensitive populations are
nearly identical in genetic composition and are not statistically different
suggesting that these E. coli populations experience similar selective pressures in
both agricultural environments. Second, the composition of the organic-resistant
population was not signiﬁcantly different from the sensitive populations. These
observations also suggest the possibility that the conventional-resistant population
will evolve to that of the sensitive, organic-resistant populations if antibiotic use

was stopped. Given that organic farms in this study were certiﬁed as having not

43

used antibiotics for at least 3 years (mean = 8 years, range = 3 — 15 years), we
estimate that, if antibiotic selective pressure to be removed, it will take at least
this long for the compositional transition.

Evidence for clonal resistance dynamics. These data say little about the
acquisition of resistance determinant by sensitive strains. However, we feel the
data can adequately describe dynamics after resistance is conferred. For example,
we expected to ﬁnd a signiﬁcant difference between the genetic composition of
resistant and sensitive populations if a resistant clone swept to high frequency
during drug use on conventional farms. We had the same expectation if clonal
interference was operating between resistant clones of the same phylogroup. The
signiﬁcant association between conventional farms, antibiotic resistance, and
phylogroup A strains supports this expectation. Further characterization is
needed, however, to differentiate between the spread of one or multiple closely
related clones.

Compositional similarity between sensitive populations and the organic-
resistant population suggests that there is an optimal genetic composition (OGC)
for the farms in this study. An overabundance of phylogroup A strains was
signiﬁcantly associated with the bIaTEM locus in the ampR population and the
presence of class I integrons in the overall resistant population. These data
suggest that blaTEM and class I integrons were linked to phylogroup A strains
during selection on conventional farms and resulted in a departure from OGC. If

this interpretation is correct, we predict that a more discriminate genetic

44

characterization of arnpR strains from conventional farms will reveal less genetic
diversity in group A strains compared to phylogroups B1 and B2D.

Evidence for hitchhiking of resistance loci. In contrast to the ampR
population, there was no evidence supporting an underlying clonal model for the
dynamics in the tetR population. Populations carrying tetR determinants (tetA,
tetB, tetC/other) were at OGC on both farm types. This observation is difﬁcult to
explain if antibiotic selection and clonal spread were occurring on a single farm
type. One explanation is that the organic farms received an occasional ﬂux of tetR
strains ﬁ'om conventional farms and the migration was sufﬁcient to maintain the
observed similarity. However, this explanation seems unlikely because the
occasional ﬂux would likely bring ampR strains from conventional farms as well,
which in turn would ameliorate the differences discussed above. If tetR loci were
linked to other compensatory, beneﬁcial mutations, then the composition of these
populations might appear similar regardless of antibiotic use.

Several lines of evidence support the hypothesis for the role of hitchhiking
or compensatory mutations in tetR antibiotic resistance spread. Bartoloni et al.
initially described a resistant E. coli population from humans living in a remote
Guarani Indian community in Bolivia (9). Individuals of the village had little
contact with outsiders, no veterinary or agricultural antibiotic use, relied on
rainwater for survival, and had limited available health care (every 3 months).
Yet, tetracycline resistance was found in 64% (69 of 108) of the individuals
tested. Pallecchi et al. recently characterized the underlying genetic determinants

and ECOR phylogroups for 113 resistant strains of the original collection (117).

45

The authors found that of the 103 tetR strains analyzed, 52 carried tetA and 51
carried tetB. These loci were distributed among all 4 E. coli phylogroups (same
procedure used in this study) and were found on all 5 conj ugative plasmids
identiﬁed in this study. The abundance and distribution of tetR strains in this
remote community supports the hypothesis that naturally occurring tetR
determinants circulate in hosts for reasons other than selection by drug use.

Support the hitchhiking hypothesis for tetR loci is also consistent with the
description of a “calf-adapted” E. coli population that was multiply resistant to
streptomycin, sulfadiazine, and tetracycline (SSuT) (84). Almost all strains (49 of
50) analyzed shared a ~140 kb plasmid, the same resistance loci (strA, su12, and
tetB), and were genetically diverse by pulsed-ﬁeld gel electrophoresis.
Khachatryan and colleagues showed that on average the SSuT population out-
competed sensitive strains in vitro and in neonatal calves (82). They also showed
that the resistance loci themselves do not inﬂuence this selective advantage (83).
Their main conclusion was that the combination of strA, sulZ, and tetB in the
original resistant population had hitchhiked with some other ﬁtness—conferring
locus.

Age effects on genetic composition. Sato et al. showed that the resistant
strains examined here were most prevalent in calves on conventional farms (136).
A similar positive association has been reported in other studies of pre-weaned
calves and adult cattle (13, 15, 84). However, cattle age had little inﬂuence on the
distribution of phylogenetic groups in either the sensitive or resistant populations

of this study. These data suggest that the abundance of resistant strains decrease

46

as cattle get older, but the genetic composition of this population remains stable.
Other analyses of human strains showed a signiﬁcant association between host
age and genetic composition, but the time reported for such change may be longer
than the average life span of dairy cattle (5 8).

Multi—drug resistant inﬂuences on genetic composition. We found a
rather complicated interaction between farm type, multi-drug resistance, and
ECOR phylogrouping (Figure 4). Signiﬁcant associations depended on the way
our log-linear model was pararneterized. However, all 3 possible
parameterizations resulted in a signiﬁcant association between low multi-drug
resistance, group Bl strains, and organic farms. These data suggest an inverse
relationship between multi-drug resistance and ﬁtness for group B1 strains on
organic farms. Since phylogroup Bl strains were the numerically dominant group
overall, this result should be encouraging for those seeking to reduce the amount
of multi-drug resistant strains in dairy cattle through limited antibiotic usage.

Two of the parameterizations showed an association with high multi-drug
resistance, group D strains, and organic farms. This result is important because a
number of human pathogens, including the strain most associated with human
enterohemorrhagic colitis, 0157:H7, belong to this group (according to the PCR
method used here). However, we are cautious to base generalizations on this
analysis because 1.) we did not design our sampling study to directly address this
question and 2.) the abundance of strains used for these comparisons were low.
For example, the F E(D = high) association (Table 2) between highly resistant

phylogroup D strains and organic farms becomes non-signiﬁcant if 2 fewer strains

47

were sampled on organic farms and 2 additional strains were sampled on
conventional farms. Similarly, we are cautious about the association between
high multi-drug resistant phylogroup B1 strains and medium multi-drug resistant
phylogroup A strains on conventional farms because the signiﬁcance of the
association depends on the model parameterization.

Conclusion. The genetic composition for the conventional-sensitive,
organic-sensitive, and organic—resistant E. coli populations was the same,
suggesting an optimal genetic composition (OGC) for the farms in this study. The
conventional-resistant population had an overabundance of ampR, group A strains,
that could be explained by linked loci (blaTEM and class I integrons) during a
selective sweep or clonal interference among closely related strains. Given the
amount of time since organic farms had abandoned conventional practices, the
rate of compositional change was estimated to be between 3 and 15 years (mean =
8 years). In contrast to the ampR population, the tetR populations analyzed here
showed no clonal dynamics and appeared to be at the OGC. These data add
support to the previously founded hypothesis that the abundance and distribution
of tetR determinants are weakly inﬂuenced by antibiotic use. We found that cattle
age had little inﬂuence on the genetic composition of the resistant or sensitive
populations. Finally, phylogroup Bl strains with low multi-drug resistance were
signiﬁcantly associated with organic farms, suggesting that these dairy farming
practices have a proportionately large, negative effect on the prevalence of multi-

drug resistant strains.

48

ACKNOWLEDGMENTS
This work was supported in part by MRU matching funds from the
College of Veterinary Medicine and Graduate College of Michigan State

University.

49

CHAPTER 4

GENETIC DIVERSITY AND POPULATION STRUCTURE OF

ESCHERICHIA COLI ISOLATED FROM FRESHWATER BEACHES

50

INTRODUCTION

Escherichia coli reaches high density in the gastrointestinal tract of
humans (~106 cells per gram of colonic content) and other warm-blooded animals
and are inevitably excreted to the external environment where hosts live (137).
These two habitats, called the primary (within host) and secondary (outside the
host), represent distinct ecosystems and differ in both the number and
heterogeneity of harmful stimuli. Secondary habitat stimuli such as UV radiation,
temperature, and predation have been shown to decrease the density of individual
strains to undetectable levels under controlled conditions (7, 81, 153). Based on
such evidence, it is often concluded that secondary habitats do not actively
support E. coli growth and therefore have little inﬂuence on the adaptive
evolution of the species (181). This conclusion underlies both the use of E. coli
as an indicator organism of fecal contamination and water pollution as well as its
use as a tool for bacterial source tracking (64, 167).

One implication of this hypothesis is that natural selection in the primary
habitat is a dominant inﬂuence on the genetic structure of populations sampled
from the secondary habitat (54). However, there are strikingly few data with
which to assess the contribution of selection and other genetic processes in the
secondary habitat on the variability and organization of natural E. coli
populations. Two observations suggest that adaptive evolution in the secondary
habitat can substantially inﬂuence population genetic structure of the E. coli
species as a whole; i.) First, the population size of E. coli in the secondary habitat

may be very large, as it is estimated that half of all living cells are presently

51

outside of a host (137). ii.) Second, data from multiple studies in both tropical
and temperate regions suggest that this organism can replicate and reach high
densities under favorable conditions outside of mammalian hosts (2, 4, 16, 156,
171) and in the absence of regular fecal input (92, 123).

To address questions about the fundamental evolutionary processes
operating in E. coli populations, two principal methods utilizing population
genetic approaches have been utilized. Multilocus enzyme electrophoresis
(MLEE) has been applied for more than 25 years to collections of E. coli isolates
obtained from a variety of human sources and other animals (53, 106, 115, 145,
157, 174). MLEE of protein polymorphisms has uncovered abundant allelic
variation and extensive multilocus linkage disequilibrium (176), that is,
populations or collections of pathogenic strains where certain genotypes are
numerically dominant and geographically widespread (37, 55, 101 , 179).
Recently, a worldwide analysis based on multilocus sequence typing (MLST)
revealed extensive allelic variation and homologous recombination, with
accelerated rates of evolution in pathogenic lineages of E. coli (182). However,
there have been relatively few studies of the genetic variability and population
structure of native E. coli isolated from environmental sources (55, 173).

In this paper, we characterize the phenotypic and genetic diversity of 190
E. coli strains isolated from six freshwater beaches along Lake Huron and the St.
Clair River in Michigan (2). We use standard biochemical assays to establish and
characterize individual phenotypic proﬁles (biotypes). Population genetic

analyses from MLEE of 18 housekeeping loci are compared to those obtained

52

using MLST of 7 conserved genes. We assess the abundance and distribution of
biotypes, multilocus genotypes, and the accuracy of a commonly used PCR-
based phylogrouping technique (33). We also analyze recombination rates and
linkage disequilibrium, which suggests that despite extensive recombination in
nature, natural selection is favoring certain E. coli genotypes, particularly those of

the B1 phylogroup, in the secondary habitat.

53

EXPERIMENTAL PROCEDURES

Beach site description and strain isolation. A detailed description of the
sampled beach sites and isolation procedures can be found elsewhere (i.e. air and
water temperatures, soil composition, and moisture content were monitored
throughout the study period) (2). Brieﬂy, six sites along Lake Huron and the St.
Clair River of Michigan were sampled from August 2001 to March 2003 with the
majority of samples taken during the summer months of 2002. For clarity, sites
were labeled here using the following scheme: Conger Lighthouse Beach, 1;
Holland Road Beach, 2; Lakeport State Park, 3; Lakeport State Campground, 4;
Marine City Beach, 5; Chrysler Park Beach, 6. Samples were taken from sand
cores (< 20cm) in the wave-wash zone and the overlying water column. The sites
were chosen to represent a variety of fecal input sources (agricultural, industrial,
and human septic runoff as well as wildlife inputs).

Single colonies of presumed E. coli strains were isolated and enumerated
from the water column of the beach sites based on a protocol published by the
US. Environmental Protection Agency (167). Strains associated with sand were
isolated in a similar manner from 9 x 20 cm sand cores that were sliced into 50m
sections (1 — 5 cm, 6 — 10 cm, 11 — 15 cm, 16 — 20 cm). Sediments were agitated
for ~1 minute to suspend cells and then processed as water samples.
Approximately 35 strains per site (a total of 205) were randomly selected and
characterized.

Species delimitation. Tests of 21 standard biochemical reactions (see

AP120E in biotyping section below) revealed that some of the 205 isolates

54

originally selected for analysis had characteristics discordant with the traditional
(biochemical) E. coli species deﬁnition. Nine strains were positively identiﬁed as
species other than E. coli and were excluded from further analysis. To deﬁne
which of the remaining 196 isolates should be included as representative, we
compared the biochemical-based classiﬁcation with the genetic relatedness of
isolates. Clustering of isolates based on MLEE and MLST (dendrograms not
shown) identiﬁed two distinct groups of strains. Group 1 was the largest (190
isolates) and contained strains with excellent, acceptable, low, and unknown
AP120E proﬁles. Group 2 comprised 6 strains that clustered distinctly apart from
Group 1 and also had excellent, acceptable, and unknown proﬁles. To ensure
these strains were not typical, but divergent, E. coli, we attempted to verify their
genus and species based on 16S rRNA operon sequencing. Published universal
primers (8F and 1492R) targeting the 16S rRNA gene and appropriate PCR
conditions were used to amplify a 1484 bp product as described by Schmidt et al.
(90, 140). Sequencing was performed in both directions of the locus (2X
coverage). Genus and species identity was determined based sequence similarity
using the nucleotide BLAST algorithm (blastn) on the National Center for
Biotechnology Information (NCBI) website (www.ncbi.nhn.nih.gov/BLAST/).
Two isolates had 99% similarity to the distantly related, E. fergusonii. However,
four strains had 99% 16S rRNA sequence similarity to E. coli (3 to E. coli and 1
to Shigella boydii). It was unclear if these isolates were representatives of a more
distantly related Escherichia lineage or perhaps represented species “hybrids.” In

either case, we excluded all Group 2 isolates from further analyses.

55

Phenotypic profiling (biotyping). A phenotypic proﬁle was generated
for each strain using 20 reactions of the API20E bacterial identiﬁcation system
(BioMerieux, Inc.) and a ﬁlter paper oxidase test (Becton, Dickinson and
Company). Positive (1) or negative (0) results were recorded for the following
reactions: beta-galactosidase (ONPG), arginine dihydrolase (ADH), lysine
decarboxylase (LDC), ornithine decarboxylase (ODC), citrate utilization (CIT),
H28 production (H28), urease (URE), tryptophane deaminase (TDA), indole
production (IND), acetoin production (VP), gelatinase (GEL), and
fermentation/oxidation of glucose (GLU), mannitol (MAN), inositol (1N0),
soribitol (SOR), rharnnose (RHA), sucrose (SAC), melibiose (MEL), amygdalin
(AMY), and arabinose (ARA). Only the informative reactions (positive or
negative for at least two strains) were used to analyze phenotypic diversity.

Multilocus enzyme electrophoresis (MLEE). Enzyme extraction, gel
electrophoresis, and speciﬁc enzyme staining were carried out as described in
Selander et al. (144). Brieﬂy, lysates (whole cell enzymes) were extracted via
centriﬁigation from overnight nutrient broth cultures and frozen at -80°C. Of the
190 E. coli strains, 5 cultures failed to pellet when centrifuged and did not
produce sufﬁcient concentration of enzyme lysates for fluther MLEE analysis.
These cultures produced an excess exopolysaccharide or other capsular matrix
that inhibited sedimentation of cells under the centrifugation conditions. Lysate
for the remaining 185 samples were individually loaded and electrophoresed
under non-denaturing conditions in a buffered starch gel matrix at the appropriate

concentration for the particular enzymes being stained. Gel slices were incubated

56

in enzyme speciﬁc staining solutions and ﬁxed for analysis. The mobility of 18
housekeeping enzymes was recoded for all strains analyzed. Mobilities were
scored relative to previously published strains from our database. The enzymes
characterized in this study were: ADH (alcohol dehydrogenase), THD (threonine
dehydrogenase), SKD (shikimate dehydrogenase), G6P (glucose-6-phosphate),
MPI (mannose phosphate isomerase), GLUD (glutamate dehydrogenase), MDH
(malate dehydrogenase), NSP (nucleoside phosphorylase), PEP (peptidase), GOT
(glutamic oxalacetic transaminase), CAK (carbamylate kinase), AK (adenosine
kinase), MlP (mannitol-l-phosphate dehydrogenase), PGD (6-phosphogluconate
dehydrogenase), PGI (phosphoglucose isomerase), IDH (isocitrate
dehydrogenase), ACO (aconitase), G3P (g1yceraldehyde-phosphate
dehydrogenase).

Electrophoretic mobility variants (electromorphs) were assigned scores in
side-by side comparisons relative to the published strains. Electromorphs were
equated with alleles at the corresponding enzyme locus based on the assumption
that each electromorph represents at least one amino acid replacement in the
protein. The combination of 18 electromorphs (alleles) was used to generate a
multilocus genotype called an electrophoretic type (ET). The same ET was
assigned to strains with indistinguishable allele proﬁles. We assumed that strains
with the same ET owe their genotypic similarity to recent descent from a common

ancestral cell; that is, they are members of a naturally occurring bacterial clone.

57

Allele frequencies and the number of strains were used to estimate population
genetic parameters as described in Selander et al. (144). Computer programs
were used to calculate parameter estimates and can be found at
(www.foodsafemsu.edu/whittam/programs/index.htrnl). We assessed genetic
differences between beach sites by partitioning the total genetic diversity as
described previously (180). The coefﬁcient of genetic differentiation among
groups (GST) was calculated for each locus using the program ETdiv.

DNA isolation and multilocus sequence typing (MLST). Genomic
DNA was isolated from 2 m1 of overnight culture in LB broth, Lennox (Becton,
Dickinson, and Company, Sparks, MD) using a Puregene DNA isolation kit
(Gentra Systems Inc., Minneapolis, MN.). DNA preparations were quantiﬁed
with a NanoDrop ND-1000 UV-Vis spectrophotometer (N anoDrop Technologies,
Wilmington, DE.), diluted to a ﬁnal concentration of 100 ng/uL, and stored at
4°C.

Primers and sequencing methodology used for MLST were chosen and
carried out as part of a system described in detail elsewhere
(www.shigatox.net/cgi-bin/mlst7/index). Brieﬂy, the internal fragments of the
following 7 housekeeping genes were obtained for analyses: aspC, cle, fadD,
ich, lysP, mdh, and uidA. The program SeqMan II (DNASTAR, Inc., Madison,
WI) was used to edit and align the sequences. Sequences were concatenated and
uploaded into MEGA3 software (91) for descriptive statistics. For further
analysis, we assumed that each unique gene sequence represents an allele and

each unique concatenated sequence represents a multilocus genotype or sequence

58

type (ST). Descriptive statistic for MLST data including the number of variable
sites and codons, parsimoniously informative nucleotide sites, and estimates of d3
and did were done using MEGA3.1 (modiﬁed Nei-Gojobori model; Jukes-Cantor).
GST was calculated as described in the MLEE procedures above for each of the 7
sequenced loci.

Estimation of diversity. Nonparametric diversity estimates were
generated based on STs sampled per site and the Chaol algorithm (classic
formula) in the program Estimates (version 7.5.0, The University of Connecticut,
Storrs, CT) (36). We chose to use this algorithm to estimate genotypic diversity
because it is well suited for a small number of samples from diverse habitats (20)
and it has a closed-form solution for variance estimation (31). As more sites are
included in the analysis, the conﬁdence intervals around the diversity estimate
tend to decrease, allowing for direct comparisons between studies (71). The
estimate calculated here was based on the total number of observed STs, the
number of STs sampled once (singletons), and the number of STs sampled twice
(doubletons).

ECOR phylogrouping by multiplex PCR. All strains were grouped into
1 of 4 phylogenetic groups based on methods adapted from Clermont et al. (33).
To help prevent false negative results, the same genomic DNA used for MLST
analysis was used as template for a duplex PCR targeting the genes chuA and yjaA
(duplex conditions: denaturation at 94°C for 10 minutes; 35 cycles of 92°C for 1
minute, 59°C for 1 minute, 72°C for 30 seconds; and a ﬁnal elongation at 72°C

for 5 minutes). A separate PCR was run with primers targeting the TspE4.C2

59

anonymous DNA region using published conditions. An R X C test of
independence was done using the G-test to test for signiﬁcant association between
phylogroups and sampling sites (154). This test is a nonparametric method used
to determine if variables are signiﬁcantly associated, and the null hypothesis
considered was that the frequency of a phylogroup is independent of sampling
site.

Phylogenetic analysis. Two phylogenetic analyses were applied to the
sequence data for 130 sequence types (STs) determined by MLST. A neighbor-
joining tree was constructed using the Kimura 2-parameter model of nucleotide
substitution with the MEGA3 software (91) and the inferred phylogenies were
each tested with 1000 bootstrap replications. Phylogenetic network analysis was
conducted with the SplitsTree 4 (74) program using the neighbor-net algorithm
(26) and untransformed distances (p distance).

Linkage disequilibrium and recombination. We used allele proﬁles and
the program MultiLocus (version 1.2, Imperial College, available at
www.bio.ic.ac.uk/evolve/software/multilocus) to test for signiﬁcant linkage
disequilibrium. Two statistics used (1,, and rbard) are detailed by the authors of
MultiLocus (1). Brieﬂy, the index of association, IA, measures linkage
disequilibrium and is used to describe nonrandom associations of alleles at loci
(24, 65, 102). The rbard calculation is a standardized measure of I A that corrects
for the number of loci used in the analysis. Both statistics should equal 0 if there
is no linkage between alleles at different loci. After calculating each statistic the

program generates p-values based on comparisons between the observed and

60

expected (panmictic) datasets where alleles are randomly shufﬂed among strains
at loci. We used 1000 randomizations for all p-value calculations. The 0w
recombination test (25) as implemented by SplitsTree 4 was used to distinguish
recurrent mutation from recombination in generating genotypic diversity. The
ratio of recombination to mutation events was inferred by grouping sequence
types into BURST groups (49) and the counting method for groups with founding

sequence types (50).

61

RESULTS

Biotype diversity. Of the 21 biochemical reactions analyzed, 4 were
positive (ONPG, GLU, MAN, and ARA) for all strains, 6 were negative (HZS,
URE, TDA, GEL, INC, and OX) across the 190 E. coli strains, and 1 was positive
(CIT) in only one strain and was, therefore, not informative. The remaining 10
assays were variable with some strains either positive or negative for each test.
The results of the 10 variable tests were used to generate a strain- speciﬁc biotype
(Table 6). There were a total of 76 distinct biotypes among the 190 E. coli strains
and a similar number of biotypes were sampled across the six beach sites (average
= 12.6, range = 10 — 14). One proﬁle (Biotype 1) was recovered 86 times and
accounted for 45.3% of biotypic variation (Table 7). Biotype 1 was the most
frequently isolated biotype at all sites (average = 14.3, range = 11 — 19), and the
four most common proﬁles (Biotypes 1 — 4) accounted for 70% of total biotypic
variation.

Levels of protein polymorphism by MLEE analysis. To measure the
extent and organization of genetic diversity among environmental strains of E.
coli, we used the methods of MLEE to resolve protein polymorphisms at 18
enzyme encoding genes (144). Among the 185 strains analyzed (see MLEE
section in Experimental procedures), the number of electromorphs (alleles)
resolved per locus ranged from 3 (N SP and CAK) to 13 (PGD) with an average of
6.4 alleles per locus (Table 8). The average single-locus genetic diversity (h = the

probability that two randomly sampled strains have a different allele at a locus)

62

Table 6. The number of bacterial isolates recovered at 6 sampling sites along
Lake Huron and the St. Clair River of Michigan. The number of biotypes and

genotypes (ETs and STs) are given for all conﬁrmed E. coli isolates.

 

Sitea total Biotypes Est STs

l 32 13 26 30
2 31 13 24 24
3 32 13 26 26
4 34 10 26 28
5 33 13 28 3O
6 28 14 26 27
Total 190 76 143 130

 

aSite names are listed in Experimental procedures as published in Alrn et al.
(2003).
bFive isolates were not analyzed by MLEE (See MLEE section in Experimental

procedures).

63

Table 7. Phenotypic proﬁle, number, and overall percentage of the four most

common biotypes at each sampling site and nine unknown biochemical biotypes.

 

 

 

API call Site Overall

Biotype Proﬁlea 1% of call) A B C D E F percentage
1 0111011110 E. coli (99.6) 16 11 13 19 16 11 45.3

2 0101011010 E. coli (99.8) 4 2 5 4 1 4 10.5

3 0111011010 E. coli (99.9) 3 1 0 5 2 3 7.4

4 0101011110 E. coli (99.9) 3 1 2 1 4 2 6.8

5 0011111000 Unknown 0 1 0 0 0 0 0.5

6 0101001111 Unknown 1 0 0 0 0 0 0.5

7 0110111010 Unknown 0 0 1 0 0 0 0.5

8 0111011110 Unknown 0 0 0 0 l 0 0.5

9 0111101110 Unknown 0 0 0 1 0 0 0.5

10 0111111010 Unknown 0 0 0 1 0 0 0.5

11 0111110110 Unknown 0 l 0 0 0 0 0.5

12 0111011110 Unknown 0 0 0 0 0 2 1.0

13 1111011110 Unknown 0 0 2 0 0 0 1.0

 

aProﬁle is based on 10 informative (variable) biochemical reactions. Each digit

represents a positive (1) or negative (0) reaction for ADH, LDC, ODC, IND, VP,

SOR, RHA, SAC, MEL, and AMY, respectively.

64

Table 8. Electrophoretic variation of 18 housekeeping enzymes analyzed by

 

 

MLEE.
No. Single Frequency

Enzyme of locus of the most

locus allelesa heterozygositv (h)b common aLlelec G5_Td_
PGI 8 0.587 0.541 0.000
IDH 6 0.342 0.795 0.000
ACO 8 0.373 0.784 0.000
G3P 5 0.510 0.584 0.015
PGD 13 0.544 0.665 0.000
MlP 4 0.455 0.697 0.034
NSP 3 0.141 0.924 0.000
MDH 4 0.187 0.897 0.000
TDH 5 0.283 0.843 0.002
SkD 11 0.577 0.632 0.023
G6P 6 0.256 0.860 0.023
PEP 7 0.403 0.762 0.000
GLU 5 0.053 0.973 0.000
CAK 3 0.103 0.946 0.000
AK 4 0.200 0.892 0.014
GOT 5 0.161 0.914 0.000

65

Table 8 (continued).

 

 

No. Single Frequency
Enzyme of locus of the most
locus allelesa heterozygositv (h)b common allelec Gﬂd_
ADH 6 0.523 0.649 0.036
MP1 9 0.726 0.449 0.023

 

aAverage = 6.2
b . .

Average smgle locus heterozygosrty (H) = 0.357
CProduct of allele frequencies =

dAverage GST = 0.011

66

among the 185 strains was 0.357 (i 0.046; SE). The allelic variation was
organized into 143 distinct multilocus genotypes or electrophoretic types (ETS).

Across the six beach sites, the genetic diversity was relatively uniform
with similar numbers of ETs recovered at each site (average = 26 ETs per site,
range 24 to 28). There were 132 unique (sampled only once) and 11 common
(repeatedly sampled) ETS. The most common ET (ET-1) was isolated 30 times
(16% of all of the Group 1 isolates) from all 6 beach sites, over a 12 month
sampling period. This common genotype represents the modal ET, that is, the
combination of the most common allele at every locus (Table 3).

Is there evidence for genetic differentiation or subdivision of the E. coli
populations from different sampling sites? To assess this, we partitioned the total
genetic diversity per locus into within and between site components (110). On
average, the within-site component of diversity (HS = 0.356) accounted for nearly
all of the total genetic diversity (HT = 0.360). The average coefﬁcient of genetic
differentiation (GST) across the 18 loci was 0.011 and ranged from <0.0001 to
0.036 (Table 3), indicating that ~1% of the total diversity is accounted for by
differences between sites; thus, there is no evidence for local population
subdivision or the differentiation of allele frequencies across beach sites.

DNA sequence polymorphisms revealed by MLST analysis. In contrast
to MLEE, which assigns alleles indirectly via the electrophoretic mobility of their
gene products, alleles at multiple housekeeping loci can be deﬁned directly by
DNA sequencing via multilocus sequence typing (MLST). We sequenced the

internal ~500 bp on both strands of 7 housekeeping genes (aspC, cle,fadD,

67

ich, lysP, mdh, and uidA) and assembled consensus sequences for the genes in
every strain. Sequence comparisons among the 190 E. coli strains uncovered
substantial DNA polymorphism with an average of 61 .3 variable nucleotide sites
(range = 40 to 100) and 8.1 variable codons (range = 2 to 25) per locus. We
resolved a similar number of alleles at each of 7 housekeeping loci (average =
41.6) ranging from 30 alleles for lysP to 49 for uidA (Table 9).

An advantage of MLST data is that one can assess the nature of selection
on allelic variation by comparing rates of nonsynonymous (amino acid changing)
nucleotide substitutions to synonymous (non-amino acid changing) substitutions
(110). There was a wide range of variation observed in the percentage of
nonsynonymous nucleotide substitutions per nonsynonymous site (did x 100,
Table 9) with respect to the percentage of synonymous substitutions per
synonymous site (d5 x 100). This variation resulted in d3 to dye ratios that ranged
across loci by 2 orders of magnitude (16.2 for uidA to 1484.2 for cle). The low
percentage of nonsynonymous substitutions relative to synonymous substitutions,
especially for cle and lysP, suggests that natural selection at the molecular level
is strongly negative and acting to limit the amount of amino acid polymorphism at
housekeeping genes in this environment.

The number of distinct combinations of alleles at 7 MLST loci resolved
130 different multilocus genotypes or sequence types (STs); the number of STs
found for 7 MLST loci is comparable to, but slightly less than the number of ETs
resolved for 18 MLEE loci. Among the 130 STs, there are 103 singletons (STs

isolated only once) and 27 common STs (isolated more than once). The

68

GHQ 89: 0020108 no Bang/Hm

 

 

we 28.8 :3 508 RS. no? 6i 3 a: c as. $2 mega E.

E 898 mg 22.8 em: a. 8.3 a 2.3 E wwm we;

mm 298 2 S 258 $3 a. 2.8 m 24$ 8 new «es

com 238 adv 64.8 82 cm 2.: N 3% 3. Re one

2: 298 86 82V :e 34 :d m 215% new 363

82 20.8 35 am: one 3 2.8 2 2.8V 2: Se Qeex

em: 2o.ovv So and 8.x 2 2.: N . an: no Sm case

on 228 36 ﬁne e2 mm 2i w as 8 am one

laminae mm H 52¢ 25 a eeoé 63% c5 388 @365 ammo; 263
8:26 2: a we do .62 gene; ozone»

 

 

. {51:2 43 cowbﬁa macaw manage—8:0: so>Om E55 2052.2; 0282032 .m 033.

69

distribution of the 12 most common STs (sampled 3 or more times) was similar to
the distribution of ET-1 from MLEE analysis. For example, one of these common
STs was independently isolated 7 times, from 5 of 6 sites, in the water column, at
all 5 depths in the sand cores, and at separate times over the 35 months of the
collecting times. These observations suggest that recovery of isolates of the same
ST is a result of the repeated isolation of a widespread genotype that is in high
frequency in the beach habitat and is not an artifact of the sampling protocol.

For comparison to MLEE, we used the allelic variation determined by
MLST to assess the extent of genetic differentiation of the bacterial populations
across the sampling localities. The calculated GST for the MLST alleles ranged
from 0.003 to 0.023 with an average of 0.012, a value almost identical to that
calculated by the MLEE analysis. Based on the similarity of GST values
calculated by both the MLEE and MLST analyses, it is likely that the E. coli
samples from the different beach sites are genetically indistinguishable and thus
represent random samples from a single, diverse and well mixed bacterial
population.

If the population genetic diversity is relatively uniform across sites, as
suggested by GST, then estimates of allelic and genotypic diversity based on the
observed dataset should be very similar to estimates obtained when alleles or STs
are randomly assigned to sites. We tested this hypothesis by estimating the total
number of STs across beach sites based on the observed (open circles) dataset and
a randomized (black circles) dataset (Figure 5). There was no signiﬁcant

difference between estimates made using the Chao 1 algorithm (36), although the

70

1600

 

 

1400 1
1200 —
11’ 1000 ~
in ,.
k -—
o .11—
3 800 - ,_ -
Q ‘1—
§
2 600 . 6

 

 

 

400 a i
200 a i

Number of sampling sites

 

 

 

Figure 5. Chaol diversity (36) estimates for the total number multilocus
genotypes (STs) based on MLST analysis. Average estimates are marked by
circles and 95% conﬁdence intervals are designated by brackets. Estimates based
on the observed dataset (open circles) are not signiﬁcantly different from

estimates based on a random site assignment (black circles).

71

95% conﬁdence intervals were large. This ﬁnding agrees with the small GST
values, and indicates that there is virtually no difference in the diversity of STs
across sampling sites.

Abundance and distribution of phylogroups. Phylogenetic studies have
shown that commensal and pathogenic E. coli strains can be subdivided into 4
major phylogroups referred to as the ECOR A, B1, B2, and D groups (68, 145).
These groups appear to differ in important aspects of their population biology.
For example, many commensal strains belong to groups A and B1 whereas E. coli
associated with human extraintestinal infections belong to groups B2 and D (58,
111-113). There are, however, exceptions as certain human populations have
relatively high counts of naturally occurring B2 and D strains in their fecal
microbial ﬂora (45).

To determine the distribution of ECOR phylogroups among the E. coli
from the beach environment, we classiﬁed all 190 strains into 1 of 4 phylogenetic
groups based on PCR methods (33). Group B1 strains were isolated more often
than any other group across 5 of 6 sites and 56% (106/190) of all strains were
classiﬁed into this group (Figure 6). Group B2 strains were infrequently
recovered across all 6 sites and only 12 strains (6%) were classiﬁed into this
group. Group A and Group D strains were isolated at intermediate frequencies
(23% and 15%, respectively). Overall, the distribution of groups was similar
across the sites (Figure 6), and the frequency of phylogroup isolation was
statistically independent of beach sites (G-test of independence, x2 005,15 = 25, Gad};

= 5.9). At site 6, there were more Group A strains than B1, but the difference was

72

25 Site 1 (n=32) 25 SI. 2 (n=31) 25 Site 3 (IP32)

g 20 20 20
g
.2 15 15 15
k
E 10 10- 10
3
2 5 5 5
o o o
25 Site 4 (n=34) 25 Site 5 (n=33) 25 SI. 6 (rt-28)
E 20 20 20
E
"’ 15 15 15
‘9
3 10 10 10
E
Z 5 5 s
. o I .
A B1 82 D A at 32 D A B1 B2 0
ECOR phylogroup ECOR phylogroup ECOR phylogroup

Figure 6. Distribution of E. coli phylogroups across the 6 sampling sites. The
frequency of each phylogroup (A, B1 , B2, and D) was independent of site (i.e.
there was no difference in the overall distribution of the phylo groups from site to

site).

73

 

m1. w.
.2

not signiﬁcant. These data are consistent with the previous inferences drawn
from the GST values and Chao 1 estimates that the genetic composition of the E.
coli samples is virtually the same and across the various beach sites.

Recombination and linkage disequilibrium. To test statistically for
recombination, we used the (I)... test, which has been shown to discriminate
between recurrent mutation and recombination in a variety of circumstances (25).
Application to each of the 7 MLST loci found that 4 out of the 7 loci show
evidence of signiﬁcant recombination in generating allelic variation (Table 10).
We also tested for the role of recombination in generating multilocus genotypes
by concatenating the sequences of the 130 STs. There were 279 informative sites
and the ow test found statistically signiﬁcant evidence of recombination (p <0.001)
overall including both the within-gene recombination, which generates new
alleles, and assortative recombination, which shufﬂes existing alleles into new
genotypic combinations.

What is the relative contribution of recombination and mutation to
genotypic variation in the natural environment? To address this question, we used
the ratio of number of recombination events, R, to the number of mutation events,
M, based on organizing STs into distinct BURST groups (49, 50). There were 83
STs that could be classiﬁed into BURST groups and for which an ancestral or
founder genotype could be inferred. Comparison of the allele proﬁles to that of
the founder genotype of each group identiﬁed a total of 30 genetic events.
Twenty-one of the 25 recombination events (R) are accounted for by changes at

the 3 loci, all of which had signiﬁcant p values in the 111w test (Table 10). Overall

74

Table 10. Recombination analysis based on sequence comparisons. There is
signiﬁcant evidence for recombination (p values for 111w recombination test) was
detected for 4 loci among all isolate (n=190) and across sequence types (STs,

n=l 30). However, there was no signiﬁcant recombination detected for strains
assigned to BURST groups (n=83). Also given are the number of recombination
(R) and mutation (M) events inferred within BURST groups in which an ancestral

sequence type was predicted.

 

 

All STs BURST
isolates only groups R M
aspC 0.166 0.142 0.084 3 0

prX <0.001 <0.001 0.118 4 1
fadD <0.001 <0.001 0.121 11 1
1'ch <0.001 <0.001 0.132 6 0
lysP 0.116 0.116 1.000 0 1

uidA <0.001 <0.001 0.258 1 2

 

75

 

m 3. w.
.rA

the R:M ratio was 5:1 which indicates that the rate of recombination is ~5 times
the rate of mutation in generating new STs within the BURST groups.

We also analyzed sequences for nonrandom associations between alleles
at the MLST loci. Two test statistics, IA and rbard, were used to compare the
observed statistical associations to those expected from a population at linkage
equilibrium (i.e. undergoing extensive recombination). Both statistics show
signiﬁcant departures from the expected value (test statistic of zero) when all
strains were analyzed together. Further analyses with the strains subdivided based
on STs, beach sites, or dendrogram clustering (clusters with bootstrap support >
90) found that all statistics for that subdivided data were signiﬁcant (data not
shown) indicating the presence of strong and substantial multilocus linkage
disequilibrium. Thus, there is signiﬁcant linkage disequilibrium in this natural
population despite the fact the recombination is generating new genotypes at a
rate 5 times the point mutation rate. Therefore, the rate of recombination is not
sufﬁcient to randomize alleles in genotypes and remove linkage disequilibrium
regardless of potential beneﬁts of ongoing recombination in the natural habitat.

Phylogenetic analysis of genetic diversity. To compare the PCR-based
classiﬁcation of phylogroups to the level of sequence divergence as measured by
MLST, we constructed a neighbor-joining (NJ) dendrogram showing the genetic
relatedness among the 130 E. coli STs (Figure 7A). There were 15 highly
supported bifurcating nodes identiﬁed that ranged in bootstrap conﬁdence values
from 90% to 100% (boxes in Figure 7A). As would be expected from MLST

analysis where a small number of loci are analyzed, almost all highly supported

76

Figure 7. Phylogenetic relationships of freshwater beach E. coli based on MLST
analysis and ECOR groupings. A. Phylogenetic tree based on the neighbor-
joining algorithm and genetic distances for 130 sequence types (STs). There are
15 nodes with high bootstrap values (2 90%) from 1000 replicates that are marked
with boxes (black box is the only interior node with high support). Circles on tips
of branches represent STs and colors represent results ﬁ'om phylogrouping by
multiplex PCR. B. Phylogenetic network based on the neighbor-net algorithm.
Parallelograms denote branches with conﬂicting phylogenetic signals as a result

of recombination or recurrent mutation.

77

Figure 7.

 

 

 

 

 

nodes contain a small number of STs (2 to 10) and are located at the tips of the
clusters. However, one highly supported node (bootstrap value = 91%) comprised
27 STs representing 44 strains and is located at a deeper interior branch (black
box in Figure 7A). This clade contains 23 of 30 (77%) strains identiﬁed as the
ET-l genotype by MLEE analysis and we will refer to it as the ET-l clade.
Strains of the ET-l clade are closely related and genetically distinct from other
strains in the tree, but show little uniqueness with respect to their biotypic
proﬁles. For example, over half of the ET-l clade strains (55%) had identical
biotypes. However, all 8 biotypes represented in this clade were also represented
in other clades.

The results of the phylo grouping PCR were plotted on the NJ tree to
determine the correspondence between ST clusters and the ECOR phylogroups
(Figure 7A). There is some evidence for clustering of STs of the same
phylogroup. Group B2 strains clustered with high bootstrap support (92%),
whereas group A strains clustered together, but were not well supported (50%).
Groups D and B1 were found in multiple clusters and some were well supported.
It is noteworthy that the ET-l clade is a subset of the B1 group and appears to be
numerically over-represented in the environmental E. coli populations.

How has recombination inﬂuenced the genetic relationships among
sequence types? To address this, we examined a phylogenetic network (Figure
7B), based on Splitstree analysis (74). This analysis does not force the sequence
data into a bifurcating tree and allows for numerous parallel paths indicative of

the presence of phylogenetic incompatibilities in the divergence of STs. Such

79

incompatibilities could arise ﬁ'om recombination or recurrent mutation in the
MLST loci. Interesting, despite the abundant recombination, the four main ECOR
phylogroups are separated and relatively intact (Figure 7B). It also appears that
the effect of recombination varies among phylogroups with genotypes of the B1
group showing the most extensive amount of recombination.

Phylogroup transition events. We compared the position of strains in
the phylogenetic tree with the ECOR phylogrouping to detect potential gene loss
and acquisition events among the three discriminating loci (chuA, yjaA, and
TspE4.C2). A phylogroup transition matrix was made based on the possible
transitions between the phylogroups A, B1, B2, and D (Table 11A). As can be
seen in Figure 7, some STs cluster between these 4 major groups and, perhaps,
represent previously discovered hybrid lineages that have a mixture of
phylogenetic signals (182). If so, we would not expect the phylo grouping method
used here to correctly classify these strains. However, we can answer the
question, which transition events are more likely to occur in the freshwater beach
environment? Of the 10 possible kinds of transitions (gene losses/acquisitions)
between the groups, only ﬁve were seen. A total of 14, 3, and 2 false positive
PCR results were discovered for certain strains of group A, B1, and B2,
respectively, based on their position in the tree (Table 11B). There were
statistically more transitions explained by gene loss as strains classiﬁed as
phylogroup A by PCR clustered with strains of B1 (11 of 43 strains) or D (3 of 43

strains) by MLST. There were 2 transitions identiﬁed among 12 group B2 strains

80

,
J.
.

 

 

Table 11. Possible gene loss/acquisition events (A.) and phylogrouping results

(B.). False positive phylogrouping results were highest for gene loss events.

A. Gene loss/acquisition events

Possible Events

1 loss of chuA

2 gain of chuA

3 loss of yjaA

4 gain of yjaA

5 loss of TspE4.C2
6 gain of TspE4.C2

 

B. Phylo grouping results

 

Strains of Clustered w/ # of gene

 

phylogroup other strains of losses acquisitions total (%)a

A A 0 0 29/43 (67)
A B1 1 0 11/43 (26)
A D 1 0 3/43 (7)

B1 B1 0 0 103/106 (97)
B1 D 1 1 2/ 106 (2)

B1 A 0 1 1/ 106 (1)

B2 B2 0 0 10/ 12 (83)

81

Q ..,_

 

Table 11 (continued).

 

 

Strains of Clustered w/ # of gene

phylogroup other strains of losses acquisitions total (%)a
B2 D 0 1 2/ 12 (17)

D D 0 0 29/29 (100)

 

a . . . .
For the rows where stralns of a glven phylogroup clustered wrth other strains of

the same phylogroup, the values represent true positive results. For the rows
where strains of a given phylo group did not cluster with other strains of the

phylogroup, the values represent false positive results.

82

(17%) and 3 transitions among 106 strains of group B1 (3%). All group D strains

identiﬁed by PCR clustered together by NJ.

83

m

 

 

DISCUSSION

Genetic diversity of environmental E. coli. Although the presence of E.
coli in natural waters has long been used as an indicator of fecal pollution, there is
a growing body of data suggesting that there exists a specialized subset of E. coli
that can reproduce and persist in secondary environments outside animal hosts in
both tropical (16, 131, 156) and temperate climates (27, 55, 77, 103, 171). The
purpose of the current study was to characterize the genetic variation and
population structure of E. coli recovered from the water and sand in temperate
freshwater beaches. We found that the average allelic diversity (h = 0.357 i
0.046) as measured by MLEE among the beach isolates is within the range (0.34-
0.54) reported for natural populations of E. coli isolated from humans and a
variety of other sources (145, 174). The allelic diversity in the beach habitat is
~l .5 times greater than that found among E. coli isolated from the secondary
habitat represented by septic tanks in two households based on comparable
methods (55).

Although several studies have revealed extensive genotypic diversity in E.
coli populations from environmental sources (27, 103), it is difﬁcult to directly
compare measures of diversity based on different molecular methods. DNA
ﬁngerprinting techniques, such as pulsed-ﬁeld gel electrophoresis (PFGE) of
digested genomic DNA (104) and amplicon proﬁles produced by repetitive
element-PCR (rep-PCR) (80), have been widely applied for microbial source
tracking (160). These techniques simultaneously detect many classes of genomic

change including point mutations, insertions, duplications, and deletions of DNA

84

 

M? w
r.»

(89). As a consequence, DNA ﬁngerprints can change rapidly through movement
of horizontally transmitted genetic elements (17).

In contrast to DNA ﬁngerprinting methods, sequence based techniques
such as MLST, provide precise information on the nucleotide differences
underlying allelic and genotypic variation (98). MLST analysis of 7
housekeeping genes revealed an average 40 alleles per locus among the
environmental strains, a level of allelic diversity less than that seen previously for
the uidA gene among hundreds of E. coli strains recovered from a wide variety of
environmental sources and wild animal species (126). The lower average allelic
diversity observed here may be a result of sampling or the effect of negative
selection in the beach environment, especially as observed in two of the MLST
genes (cle and lysP). Overall, the genotypic diversity was extensive: a total of
130 distinct sequence types (STs) were resolved with 103 STs recovered only
once in the sampling. Despite this diversity, several of the genotypes were
repeatedly recovered at multiple sites and sampling times, suggesting that natural
selection is favoring certain genotypes.

Lack of geographic differentiation among sites. The freshwater beach
habitat of Lake Huron and the St. Clair River in Michigan is an open system with
many possible inputs of fecal contamination and a variety of habitats and
microhabitats in which E. coli could survive, reproduce, and locally adapt. We
therefore anticipated that the E. coli samples recovered at different sites would be
genetically distinct, reﬂecting both the variety of indigenous inputs as well as the

local environmental pressures. In contrast to this view, however, several lines of

85

 

'm
/
J.

,i ’_.'> ‘

 

evidence indicate a lack of geographic subdivision of the population. First, allele
frequencies are relatively uniform across sites as reﬂected in small between-site
component of total genetic diversity (GST < 0.02 for MLEE and MLST). Second,
similar mixtures of biotypes and genotypes (ETs, and STs) were recovered from
all sites. Third, there was no signiﬁcant difference in diversity estimates, based
on the Chaol estimators, from the observed and expected datasets randomized
across sites. Finally, the frequency of the 4 major phylogroups was relatively
uniform across samples and statistically independent of site. Based on these
ﬁndings, the population genetic analyses support the hypothesis that the bacterial
isolates from the different beach sites represent samples ﬁ'om a single, well mixed
E. coli population.

Linkage disequilibrium and recombination in nature. One of the main
observations reported here is the presence of extensive linkage disequilibrium in
the natural population of E. coli inhabiting the beach environment. Linkage
disequilibrium, the statistical situation in which two or more alleles are found
together in haplotypes more frequently than expected (48), can arise and be
maintained by a variety of population genetic processes. For example, a simple
situation occurs in asexually reproducing organisms where genotypes evolve as
distinct genetic lineages or clones within the population (174). Under these
circumstances, clones that increase in frequency in the population by clonal
expansions or selective sweeps (63, 85), can predominant and drive speciﬁc
multilocus genotypes to high frequencies, and account for high levels of strong

and persistent linkage disequilibrium. Clonal genotypes can also be over-

86

 

represented in bacterial species associated with human disease, particularly in
samples of isolates from clinical cases, which may artiﬁcially or temporally
generate patterns of linkage disequilibrium. For example, among strains of
Listeria monocytogenes examined by MLEE, one ET marked a clone that was
common in epidemics in the 19803 (122). Widespread clonal genotypes were also
found in bacteria such as the nontyphoidal serovars of Salmonella (12, 143) and
Shiga toxin-producing E. coli 0157 (177, 178) for which many of the strains were
obtained ﬁom clinical sources at different times and places. The linkage
disequilibrium often seen in these studies could be a reﬂection of the biased
sampling of special genotypes in clinical collections or a consequence of
epidemic spread of pathogenic clones (102).

Biased sampling or epidemic spread alone, however, cannot account for
the linkage disequilibrium and population genetic structure seen in the
environmental E. coli. The study here is based on the characterization of E. coli
isolated from a randomly sampled set of beach locations (2) with the frequency of
recovery reﬂecting the actual abundance of genotypes in the natural habitat. This
population-based sampling contrasts the situation often encountered in studies
based on historical collections of clinical strains or strain collections assembled
from various researchers. In such cases, linkage disequilibrium may be
artiﬁcially created by biased sampling or may not be accurately estimated. Here,
however, the linkage disequilibrium found by both the MLEE and MLST analyses
reﬂects allele frequencies in the population and identiﬁed several multilocus

genotypes (11 ETs and 17 STs) that were numerous and widespread among the

87

sampling sites. There are signiﬁcant associations of alleles in environmental
genotypes, as seen for example in the IA and rbard values. The linkage
disequilibrium exists despite the clear evidence for recombination at the sequence
level generating new alleles at loci and assorting existing genes into new
multilocus combinations. These data provide compelling evidence for complex
linkage disequilibrium persisting at multiple levels from individual phylogenetic
clusters to the entire population.

Our working hypothesis is that natural selection favors certain genotypes
in the environment, particularly the ECOR B1 group, which generates linkage
disequilibrium despite frequent recombination. The clearest example of such an
adaptive clone is ET-l, which was found at a frequency of 30 out of 185 strains,
and was recovered at multiple sites in the environment over the course of the
study. This clone falls into the B1 phylogroup and appears to be diversifying at
the sequence level by recombination based on the MLST data. Thus, it seems
likely that recombination in nature is occurring between similar lineages more
frequently than between distantly related ones. This idea is consistent with the
ﬁndings of Wirth et al. who show evidence for extensive recombination between
strains of the 4 ECOR groups, particularly between members of groups A and B1
(1 82).

Based on eBURST analysis, we estimate that recombination generates
new alleles 5 times faster than mutation (R:M ratio of 5 tol), which is less than
the theoretical threshold of between 10 and 20 recombination to mutation events

required to keep a natural population at linkage equilibrium (102). Assuming that

88

$1

 

'!

recombination exchanges fragments that are approximately the size of the loci we
sequenced (~500 bp), we can also express an R:M ratio in terms of generating
new nucleotide sites. This is accomplished by ﬁrst expressing recombination and
mutation in terms of changes per site. For recombination, we count up the
number of pair wise nucleotide differences in recombinant alleles between
members of clonal groups and divide by the length of the sequence. Likewise for
mutation, it is a single base pair change divided by the length of the locus. The
per site R:M ratio is then the ratio of these two numbers. If each recombinant
allele differed from the alleles of other clonal group members by a single
nucleotide, then the R:M ratio would approach unity. On the other hand, if
recombinant alleles differed by more than one nucleotide, then the inﬂuence of
recombination would be greater than mutation and the ratio would increase. For
the clonal groups found in the freshwater beach habitat, recombination is
generating novel nucleotide sites 8 times faster than mutation. While it appears
that recombination is effective at creating genetic diversity on a per site basis, this
value is less than the estimated ~50:l ratio reported for within the ECOR A group
and similar to the 5:1 estimate for all ECOR groups (62).

Accuracy of the phylogrouping PCR technique. Clermont et al.
originally described a decision tree based on a triplex PCR that groups strains into
one of the four phylogenetic lineages of the E. coli reference (ECOR) collection
(33). This technique has been applied extensively to characterize diverse
collections of strains including commensal and pathogenic E. coli from humans

(133) and a variety of animal hosts (32, 56, 151). We have shown here that the

89

Q.

 

 

 

false positive rate across the four groups is ~10%, which is one misclassiﬁcation
per 10 strains analyzed. It is noteworthy that the distribution of false positives is
not even across the groups. For example, all group D strains clustered together
and thus appear to be correctly classiﬁed, while the false positive rate for group A
strains was 33%. These ﬁndings suggest that the diagnostic absence of a PCR
amplicon, presumably resulting from gene deletion or mutation within the PCR
primer sites, leads to the misclassiﬁcation of group A strains. It is clear that a
more discriminatory PCR assay is needed to distinguish strains of group A.

The ET-l clade. There are few studies of secondary habitat E. coli
populations with which to compare the frequency of the B1 phylogroup and,
indirectly, the importance of the ET-l clade. However, numerous studies have
used ECOR phylogrouping to assess the genetic composition of strains from
different hosts and host populations (44, 45, 56, 58, 111-113, 135). Strains of
group B1 were infrequently recovered from various human populations (45, 58,
113), and were the most frequently sampled from certain types of wild animals
(56). For example, Gordon et al. reported that B1 strains were most frequently
isolated from ectotherms (ﬁsh, frogs, and reptiles) living in or near water,
suggesting that the occurrence of B1 strains in these hosts was associated with the
abundance of B1 strains in the secondary habitat (56). Bacteria of the Bl
phylogroup were also identiﬁed in the characterization of natural E. coli bloom
events in an Australian freshwater lake where fecal contamination was unlikely to

account for abundance (123). Finally, 70% of strains from river and surface water

around Munich, Germany were classiﬁed as either phylogroup B1 or A, based on

 

.JJM. . d1 . 1
gr.

the presence of the chuA gene and the Clermont et al. PCR primers (69). These
phylogrouping studies are in agreement with our hypothesis that the ET-l clade
represents a closely related group of secondary habitat adapted E. coli. If our
interpretation of the ET-l clade is correct, then approximately 23% of strains are
adapted to the freshwater beach environment. In addition, the utility of these
strains for use in water quality assessment or microbial source tracking will be
suspect.

Conclusions. Colonization and persistence of certain E. coli genotypes in
humans and other animals have been well documented (28, 79). When excreted
outside the host, the average strain, however, probably does not persist and dies
off at a rapid rate (172, 181). This results in a net-negative growth rate under
some secondary habitat conditions (137). Consequently, adaptation to primary
habitat stimuli is considered to dominate evolutionary processes in the E. coli
species as a whole (137). However, certain groups of E. coli have been shown to
colonize and persist autochthonously (without fecal input) in secondary habitats
(121, 123, 131), and recent data suggest that certain genotypes are adapted to
these conditions (4, 55, 103, 173). lshii and colleagues (77) have proposed the
term “naturalized” strains to refer the persistent E. coli genotypes that comprise
the autochthonous members of the microbial community in the environment. The
ﬁndings reported here add support to the emerging model that naturalized E. coli
strains are a signiﬁcant component of the environmental coliform microbiota. In

addition, the persistent genotypes are concentrated in a distinct clade, the B1

91

phylogroup, whose members may possess special traits that allow them to survive

and reproduce in the environment under temperate conditions.

92

Cl.

 

 

ACKNOWLEDGMENTS
We would like to acknowledge those who participated in this study ﬁ'om
Central Michigan University: Janice Burke, Erica Francis, Erin Hagan, AJ
Matthews, Anne Spain, and Erin Zacharias. This work was supported in part by
MRU matching funds from the College of Veterinary Medicine and Graduate

College at Michigan State University.

93

Q

 

 

CHAPTER 5

A REVISED MOLECULAR PHYLOGENY FOR ESCHERICHIA

C 0L1

94

INTRODUCTION

The species Escherichia coli belongs to the bacterial family,
Enterobacteriaceae, and is known to circulate within and between humans and
animals (21). The host gastrointestinal tract is referred to as the primary habitat
for E. coli. However, it is estimated that half of all strains are presently outside of
hosts in the secondary habitat, or the environment (137). The consensus view is
that E. coli cannot survive in the environment (21, 181), although recent evidence
suggests that autochthonous strains do exist (2, 4, 16, 54, 92, 123, 156, 171).
There are few data with which to assess the natural history of E. coli in the
secondary habitat because most studies are concerned with commensal strains
from the primary habitat (28, 45, 53, 55, 56, 58, 145, 146) or strains that cause
disease (128, 148, 175, 179, 180).

In a recent population genetic characterization of strains from a secondary
habitat (169), we discovered 6 strains that were highly divergent in nucleotide
sequence using multilocus sequence typing (Figure 8). These strains were
originally isolated from freshwater beaches along Lake Huron and the St. Clair
River in Michigan and were phenotypically identiﬁed as E. coli using standard
biochemical tests (2, 169). It appears that such strains represent novel genetic
diversity that may be used to make evolutionary inferences about the species as a
whole. For example, it has been hypothesized that a similarly divergent lineage
of E. coli represents the breadth species-level diversity that existed some 10 -—- 15
million years ago (182). An alternate or additional hypothesis is that multiple

lineages of

95

m

3.. _.,_

 

-4)

Figure 8. Divergent E. coli strains isolated from freshwater beaches in Michigan.
Neighbor-joining tree is based on concatenated sequences from 7 housekeeping
loci (aspC, cle,fadD, ich, lysP, mdh, and uidA). The white circle contains
represents the typical genetic diversity of E. coli and the grey circle represents

that of the divergent strains.

96

 

Figure 8.

190 of 196 E. coli strains
from freshwater beaches

   
 
  
  
 
 

6 of 196 E. coli strains
from freshwater beaches.
Strain 1D:

TW09231

TW09254

TW09266

TW09276

TW09308

TW14182

 

S. enterica

97

/
J.

 

 

do. J.

biochemically similar, but genetically distinct bacteria have evolved, so that
multiple species currently exist within what is now considered to be E. coli.

The purpose of this study was to assemble a collection of genetically
divergent E. coli and present a novel phylogenetic context for these strains. We
developed a novel multilocus sequence typing (MLST) protocol based on 24
housekeeping loci that can be applied to divergent and typical strains of E. coli as
well as their closest known relative, Escherichia albertii. Using Salmonella
enterica as an outgroup, we constructed dendro grams based on individual gene
fragments, assessed inter- and intra-genic recombination, and tested for
differences in the evolutionary rate between clusters of strains. Finally, we
present a new evolutionary relationship for two distinct bacterial lineages within
E. coli, and discuss the evidence that these strains are prevalent in the secondary

habitat.

98

EXPERIMENTAL PROCEDURES

Strain collection and sequence source. All E. coli and E. albertii strains
included in this study (Table 12) are part of a strain repository maintained at
Michigan State University. More information on each isolate can be found online
at www.shigatox.net. The E. coli strains were assembled from 5 sources; 1). Six
strains (TW09231, TW09254, TW09266, TW09276, TW09308, and TW14182)
were isolated between 2001 and 2003 in Michigan from freshwater beaches along
Lake Huron and the St. Clair River as part of a previously described study (2); 2).
One strain (TW11588) was isolated from soil in the Puerto Rican rainforest by Dr.
Gary A. Toranzos (Department of Biology, University of Puerto Rico — Rio
Piedras). This strain was sampled as part of a collection of strains that are
unassociated with human activity. It was isolated on February 26, 2003 at a depth
of 6-10 cm; 3). Five strains (TW14263, TW14264, TW14265, TW14266, and
TW14267) were isolated by Jeffrey L. Ram (Department of Physiology, Wayne
State University) in 2005 as part of a source tracking experiment in the Great
Lakes region. Four strains were isolated from the environment (water), and one
was isolated from a raccoon; 4). Two previously published (182) strains
(TW14351 and TW14352) were obtained from Mark Achtrnan (Department of
Molecular Biology, The Max Planck Institute) and were part of a previously
published study (182). One strain was isolated from a parrot and one strain was
isolated from a dog; 5). Four strains (TW11930, TW11966, TW12018, and

TW14421) were included from an ongoing cohort study of enterotoxigenic E. coli

99

(ETEC) infection in infants from Africa and were supplied by Hans Steinsland

(Microbial

100

 

beam 23 62628265 22 6822 666260 968622 54:5

been 22:. 6am .2 2 <3 22 2863 $2325

25:. 22:. :am .2 2. <3 22 2263 686;:

been may 52 .2 2 <3 22 .2863 @2225

been any :52 .2 2 <3 22 26263 68322:.

.263... m3 862 .2 .2 <3 22 688622 82:39

.263... may 52 .3 .2 <3 22 26263 365:.

.2636 may 66626.62on 22 68622 68260 265622 2 SN :5

been one 6623on .m 66622 666260 6226622 82 5:.

been 25 65662on .2 682m 666260 e668: 8% :5

263a 23 62.28-886.23 .< .0 822 6:652 :8 £2 :5

A63.... we; 82 .3 2 <3 22 an? 8895

been ea. 62 .3 .2 <3 22 826.3 62825

been we? 82 .3 .2 <3 22 2863 8895.

been one 82 .3 .2 <3 22 2863 .6289:

been 23 82 .3 .2 <3 22 2863 28825

682 26 so 523 §§eo< 22 866 oz Head BAN

88 26 so 563 626622 .2 866 oz mom 623222

88 26 so 2626? <2 <3 .22 6686: 58.32 33

one 26 no 86566 <22 62262262 e656: 836862 3.22

82 26 so Baez <22 Boa .6522 e683 2268 om<m

88 26 so 6862 <2 <3 22 ease: 2829.: 03m

88 26 so 263 <22 682 6686: .562 868% a.
32 26 so 266622 <22 366 oz 565: 236$: .6202 .266 .2
858.3% 8.50m 58% 283 283mm 9 Embm momoomw

m.

22:29:82: 382m 3.20:6 .m. USN 2.533 .m. 220.6 .M .N_ 2an

101

.Mm<m.:8 8m 8% 088% 22:28 8:20:25 «:26

 

 

 

88 26 26 2666:6662 <2 62% oz 66:5: ~52 83.3523.
88 26 26 man <2 6266 oz 668622 N5 :35 66.26266 .m

38 26 26 653: 6662265 .< .22 amoﬁaﬁm 66532 32

38 26 26 6:62: ~82 on? 366 oz 668:: 26-0

38 26 s 6:52 66:62 22 2 2866238 668622 $62
88 26 26 6652 66662 22 2 266662263 5&6: 6626 2.666226 .62
8:20.232 08:8 Eabm 283 Eta—mm e Ewbm $6on

n

2666622682 E 6365

102

Evolution Laboratory, Michigan State University). The E. albertii strains (9194,
19982, C-425, and K-l) all came from Bangladesh and were part of a previously
published study (75). Type strains for E. coli and S. enterica (K-12 M61655,
Shigellaﬂexneri 2457T, EDL933, 042, e2348/69, CFTO73, S. enterica serovar
Typhi TY2, and S. enterica serovar Typhimurium LT2) were analyzed in silico by
downloading sequences from the comparative genomics database, coliBASE
(colibase.bham.ac.uk).

Confirmation of E. coli. The strains in this study satisﬁed 4 biochemical
criteria. 1). A phenotypic proﬁle was generated for each strain using 20 reactions of
the APIZOE bacterial identiﬁcation system (BioMerieux, Inc.). Positive (1) or
negative (0) results were recorded for the following reactions: beta-galactosidase
(ONPG), arginine dihydrolase (ADH), lysine decarboxylase (LDC), ornithine
decarboxylase (ODC), citrate utilization (CIT), HZS production (HZS), urease (URE),
tryptophane deaminase (TDA), indole production (IND), acetoin production (VP),
gelatinase (GEL), and fermentation/oxidation of glucose (GLU), mannitol (MAN),
inositol (INO), soribitol (SOR), rhamnose (RHA), sucrose (SAC), melibiose (MEL),
amygdalin (AMY), and arabinose (ARA). 2). Strains were screened for cytochrome c
oxidase activity using a ﬁlter paper test. Single colonies from overnight growth on
LB agar plates were streaked onto ﬁlter paper containing the reagent N,N,N ', N '-
Tetramethyl-p-phenylenediamine (Oxidase test kit, Becton-Dickinson and Company).
3). Strains were tested for ,B-glucuronidase activity by streaking single colonies onto
Luria-Bertani (LB) agar plates containing the flourogenic compound 4-

methylumbelliferyl-ﬂ-D-glucuronide (MUG, Sigma Aldrich). 4). Strains were tested

103

for glutamate decarboxylase (GAD) activity using the protocol described by Rice et
al. (130). All strains were positively identiﬁed as E. coli by APIZOE, negative for
oxidase activity, positive for MUG activity, and positive for GAD activity.

DNA isolation and multilocus sequence typing (MLST). Genomic
DNA was isolated from 2 ml of overnight culture in LB broth, Lennox (Becton,
Dickinson, and Company, Sparks, MD) using a Puregene DNA isolation kit
(Gentra Systems Inc., Minneapolis, MN .). DNA preparations were quantiﬁed
with a NanoDrop ND-lOOO UV-Vis spectrophotometer (N anoDrop Technologies,
Wilmington, DE.), diluted to a ﬁnal concentration of 100 ng/uL, and stored at
4°C.

Sequencing methodology used for MLST was carried out as part of a
system described in detail elsewhere (www.shigatox.net/cgi-bin/mlst7/index).
Brieﬂy, we analyzed the internal fragments of 24 housekeeping genes (see Table
13). We selected 20 genes from two publicly available protocols
(www.shigatox.net (124) and www.mlst.net), 1 gene (metG) from a previous E.
coli MLST study (170), and 3 genes (kdsA, torC, and yde) from a recent analysis
of E. coli MLST gene candidates (87). The program SeqMan H (DNASTAR,
Inc., Madison, WI) was used to edit and align the sequences.

Descriptive statistics, Tajima’s test, and recombination analysis.
Sequences were concatenated and uploaded into MEGA3 software (91).
Descriptive statistics for MLST data including the number of variable sites,
parsimoniously informative nucleotide sites, and estimates of ds and dN were done

using MEGA3.1 (modiﬁed Nei-Gojobori model; J ukes—Cantor). Tests for

104

 

$2262 6a <<ou<ou<0<ooooo<<o<0 I m
.2326”... mm NS SN <<o<5<5885035 I 3 m5 .36
$2262 28 ueo<<30<<50058<<< I x
52262 mm 6% S oewo<<05<ou35050 I ”2 a: 9&6
363... 625 22% 5055505900580 I a
363m .025 mm an 2%- 55005355000030 I “2 a; .2666
$223 23 3335353335 I 2
$2262 mm m? 66 0053303002053 I ,2 5 E 2.96
56523 6% 55580055022503 I m
c6822 62 6mm 2 00035300050085: I m as 266
08:80.28 0:538:22 2026qu @302 we 2.28m «A. mummoocosuom ant ommm 3004
SEE 3:23:23 we 03m 68m :oEmoa .2255

.385 2.22.233 .m 93 .228 M
Eow5>€ 98 :2me @2352“ 8 com: 803 ob: “23m: EoEtg 2C. .62 3303030: vm Mom ﬁcEﬂQ Ema—2 28 MN .3 033.

105

 

Hmdzom Rm OU<HO<UUUOH<<H<UGHUH I m
.5425 mm wwv om OOH<UH<HH<<<O<<OOUUU I m vow MEM
:aEEo< o _ w m09h<0<<<<<000<000<90 I m
€822 3 8w 2 moﬁuuuw<ooooewo<u<ue I m 3.: 95%
Fwd/Mom M32 H<UHHUHHUUH<<O©<UOUO I m
quzom mm 3% won HHH<U<UH<HOHUOUUOHUO I m 33 Sag
33m 25. $3 <uoeo<<ooo<§<<oeohg I a
33m 35 mm 80 m2 Samueiuouﬁoeuoo I m 8: was.
Hmdzom mww U<<<UOHOUHOUUOUHUH<< I m
quzom mm Km 3 m HHO<<<UUOO<H©UUHOUHU I m 5mm V90
008288 833388 85on 9.0.32 mo 28m armsmﬂmoocozﬂoom Era oﬁm 300A
58th mam—meg .8 oﬁm Bob 5568 SEE

$35303 2 2an

106

 

50:00 30 00<<0<0<<<005000500 I m
5.0200 mm 20 cm 00<<<5<<0500000<50 I m 2.: «3
33m 3:. o; 5<<0<<<0005<00000<00 I m
303... $5 mm 8m m 005<00<50050<<<<< I 0 am 33
53200 82 <0<<<0500050005000< I E
51—200 mm $0 5. 505<0050<<00<0000050 I E
33... 35 N9. 532000032005 I 2
33m $5 mm 3 £3- 0<0<000500500<<00 I E $2 $3..
3% $5 25 50505505000505 I m
35. 25 mm 30 0% 00550000000030 I m 23 38
00002008 83000082 82005 @302 mo tﬁm «ﬁnnmv woos—Baum 3.53% 983
025.5 wczmoﬁa mo oﬁm Sow 02:80 00:55

€020.53 2 035

107

 

$5292 :2 <0<000<000<<5000<20 I m
5852 E 5 0 w 3.5. <020<0<<<020500000 I 5 3S 2&5
53200 Ema 50500520002200 I m
53:00 wm 0% $2 <0<50<<05000<250000 I 5 $2 55::
53200 80 200<2500000<500<00 I a
53.25 mm $0 NN 0050050005<2<500<00 I m a: 0 SE
595. 35 w? 0500502000525520 I m
8222 mm mm mm 050505<05000<0020<0 I 5 Sam 035
53200 o5 020500005000<050<5 I m
53:00 mm owe o: 0<500022000<05020050 I 5 ma 5%
08:80.08 chambebﬂ 8:080 9952 mo 93m armsﬂ $055503 38 oﬁm 2604
525.5 mammoqé 50 3% 805.5 .8380 525.5

$830800 2 295

108

 

>wa £5 28 0205050050<2<5005000< I a
59% 0:5 mm 58 85 <02000000<000<05<0 I 5 :2 $0.:
59% £5 82 <500<0002<5000<<5000<05 I m
53:00 mm £5 RN 52500050502000255 I 5 mm: $0.5
5?; £55 82 <0<50050<0005000000 I m
52% 35 mm 30 m5». <0<<<02<00000002<05 I 5 m5: 0§
53225 Now <500<00<055<<0000<0 I x
53200 mm ms EN <<50<0<00202000000 I 5 m3 55%
$6292 a; “<0000<000<502<<005005 I m
5822 mm :5 a: a00205000255500052000 I 5 83 EB
000.50.50.52 0530501953 50300.5! £9.52 .50 5.85m MARJWV moonozdnum ﬁne 0N5 mzooq
BEE mam—0053‘ mo oﬁm 82.5 22:80 5085.5

.635380 2 235

109

doom .3 8 588363 «a 055.“ on SS 8:80.58 .0832. RF 5038258
Ava: geckoﬁwﬁmég 058 8528.??? F. 893320 Fwd—2 00:00 03256 323:0 9 .8005 quzom 05 55E€<o
40,32 05 mo 58% 05 3 2538 5585* 5853: 8865 EB? o>cmmoz .303 05 USA 562309 GOES 055385

05 E :5 .230me: 0882 8:». ﬁ 303 50.5 BEE 03350.5 BB .905 E 8359. 0882 mm @050 8.5 5255 08350.5 235

douﬁaoto 850on 05 E on 9 :3QO >05 :5 60:80.58 05: E Hogan >05 mm 80: 08m: Ba 3253 80:80

$350800 2 2&5

110

unequal rates of molecular evolution were performed using the Tajima’s test in
Mega3.1. This test has been used elsewhere (43) to display differences in the rate
of evolution compared to the null hypothesis that neutral mutations accrue
through time at a uniform rate (molecular clock).

The approach we used to assess the effects of recombination was called
Genetic Algorithm for Recombination Detection (GARD) (88). This analysis
screens for evidence of phylogenetic incongruence and identiﬁes the number and
location of breakpoints in aligned sequences. There are basically 3 parts to the
analysis: 1). Upload a ﬁle containing aligned sequences, 2). Select an appropriate
model of nucleotide evolution, and 3). Run the analysis. GARD is publicly
available and the analysis can be executed on clusters of computers through the

website www.datamonkey.org.

111

RESULTS

New primers for MLST of 24 housekeeping loci. The strains included
in this analysis were assembled to represent the genetic diversity between E. coli
and E. albertii (Table 12). To maximize efﬁciency and cost, we developed an
MLST protocol that utilizes one PCR primer pair per locus for PCR ampliﬁcation
and sequencing. Published primers for E. coli MLST analysis were initially used
for PCR ampliﬁcation and resulted in adequate sequence for 16 of 21 (76%) loci
(Table 13). New primer pairs were designed and tested for the 5 loci that failed to
work under published conditions. In addition, 3 other primer pairs were designed
for candidate E. coli MLST loci (87). These new loci were selected to maximize
sequence coverage around the E. coli chromosome (Figure 9). Primers were
optimized so that annealing occurred at the same temperature (58°C), which
meant that multiple loci could be ampliﬁed in the same PCR themocycling
reaction. As expected, the uidA locus was absent in E. albertii and S. enterica
strains, but was present in all E. coli lineages.

MLST analysis. Sequencing of 24 housekeeping loci resulted in 12,191
bp of nucleotide sequence, encoding for approximately 3,861 codons. There were
2,703 variable sites, of which 2,453 were parsimoniously informative (occurred in
more than one strain). Pairwise strain comparisons of the concatenated sequence
resulted in clusters of ﬁve well-supported groups in a neighbor-joining (NJ)
dendrogram (Figure 10A). As expected, all strains of deﬁned taxa (S. enterica, E.
albertii, and E. coli) clustered together. However, two additional groups of

strains were found. We labeled them as environmental group I and environmental

112

 
 

prX (9.8)

 
 

purA (94.9)

   

arcA (100.0)

  
   
 
  
 

adk (10.7)

cyaA (85.9)
gyrB (83.5)

 
 

aspC (21.2)

 
    

mt/D (81.3)
torC (22.8)

 
 
 
 

Position of 24 houskeeping loci 1'ch (25,7)

around the genome sequence of
K12 MG1655 deA (27.3)

 
  

   
 

      
 

aroE (73.8)
mdh (72.8)

  
 

dnaG (69.1)
fumC (36.2)

   
  

uidA (36.4)

 

   

mutS (61.5) fadD (40.6)

$7pr (59- 1) metG (47.3)

     

   
 

rpoS (61.7)
recA (60.8)

  

lysP (48.3)

Figure 9. Position of 24 housekeeping loci around the E. coli K-12

M61655 chromosome.

113

Figure 10. Neighbor-joining tree (panel A) and Neighbor-Net (panel B) of E.
coli, Environmental group I, Environmental group II, E. albertii, and S. enterica

based on 24 housekeeping loci (see Table 13).

114

Figure 10.

S. enterica

A

0.01

  
    

Environmental group II

    

S. enterica

   

Environmental
group I

i—v—«i

o.o1 Envuronmental group II

E. albertii

Environmental
grow I

115

group 11 because they contained the 6 divergent freshwater beach strains from
Michigan (Figure 8).

The environmental groups also contained 4 additional strains from
Michigan water and one strain from pristine (non-human associated) soil in a
remote area of a Puerto Rican jungle. In addition, two previously reported (182)
E. coli strains (one from a dog and one from a parrot) and one strain from a
raccoon in Michigan clustered with environmental group 11 strains. Pairwise
group comparisons revealed that both of the enviromnental groups had a similar
level of sequence divergence from typical E. coli. The extent of this divergence
approached that of E. albertii for silent, or synonymous, substitutions (Figure
11A) and was identical to that of E. albertii for amino acid altering, or
nonsynonymous, substitutions (Figure 11B). These data suggest that the
environmental groups may be distinct taxa within the named E. coli species.

Recombination analysis. To visualize the overall contribution of
recombination between these groups, we constructed a phylogenetic network
(N eighbor-Net) (Figure 10B) using SplitsTree analysis (74). This analysis does
not force sequence data into a bifurcating tree and allows for numerous parallel
paths, which indicate the presence of phylogenetic incompatibilities. Such
incompatibilities could arise from recombination or recurrent mutation in the loci.
It is interesting that despite the inﬂuence of recombination, the ﬁve groups
deﬁned in the NJ tree remain distinct and relatively intact (Figure 10B). It also

appears that the effect of recombination varies among groups as more alternate

116

A. Synonymous substitutions

 

 

 

 

 

 

% s 30 ~15—
.‘ O
E 9 25
\ LU 1
3 E
t E
‘5 a. 20 -
(I)
0 a:
5‘3, g 15
E g
g .35 10
a) ‘o ‘
EL
5 .
o .
Env Env E. albertii S. enterica

Group I Group II

B. Nonsynonymous substitutions

 

Percentage of nucleotide
differences from E. coli

 

 

 

 

 

0.4 - . . Twin-..
ni .
0.0 i , ll

Env Env E. albertii S. enterica
Group I Group II

Figure 11. Pairwise sequence divergence between E. coli and environmental
group 1, environmental group II, E. albertii, and S. enterica. Environmental
groups appear to be as divergent from E. coli as E. albertii at synonymous (panel

A) and nonsynonymous (panel B) sites.

117

paths can be seen within environmental group I and between this group and E.
coli.

To further assess the inﬂuence of recombination, we characterized the
number of complete loci and fragments of loci that were either shared or clustered
with divergent groups. Genetic Algorithm for Recombination Detection (88), or
GARD, was used to infer recombination breakpoints within the 24 loci (Table
14). No breakpoint was found for 9 loci (arcA, aroE, aspC, dnaG,fumC, kdsA,
mdh, purA, and torC) and at least one was inferred for the remaining 15, resulting
in 39 non-recombinant fragments. For each fragment, we constructed a neighbor-
joining dendro gram to assess the extent of recombination between groups.
Evidence for recombination was found if a non-monophyletic relationship was
found for the groups (strains of each group did not cluster together on a single
branch).

For the 9 loci where no breakpoint was found, 4 (44%) produced
monophyletic relationships for all groups. Of the remaining 30 fragments, only 7
(23%) produced monophyletic relationships among the groups, suggesting that
recombination of gene fragments (intra-genic) occurred more often than the
transfer of entire loci. Three groups were monophyletic at all or nearly all
ﬁagments. The S. enterica group was always monophyletic (39 out of 39
ﬁ'agments); strains of E. albertii clustered together with 36 of the 39 (92%)
ﬁagments; and environmental group II was monophyletic for 35 of the 39 (90%)
ﬁ'agments. In contrast, strains of environmental group I and E. coli were rarely

monophyletic (18 of 39 and 15 of 39, respectively). These data suggest that

118

Table 14. Results of single locus analyses. Recombination (intra-genic)

breakpoints were identiﬁed using GARD analysis. Inter-genic recombination was

deﬁned by loci with no breakpoints that do not exhibit a monophyletic

 

relationship.

GARD Position of Signiﬁcant Monophyletic
fragment Locus breakpoint Taﬂma’s test relationshioa
1 arcA None No Yes
2 aroE None No . No
3 aspC None No No
4 dnaG None No No
5 fumC None No Yes
6 kdsA None No Yes
7 mdh None No Yes
8 purA None Yes No
9 torC None No No
10 adk 1-394 No No
11 adk 395-536 No No
12 cle 1-240 No No
13 cle 241-567 No Yes
14 cyaA 1-294 No No
15 cyaA 295-498 No No
16 fadD 1-186 No No
17 fadD 187-492 No No
1 8 grpE 1-277 No No
19 grpE 278-417 No No
20 gyrB 1-270 Yes No
21 gyrB 271-460 No No
22 ich 1-472 No No
23 1'ch 473-826 No No
24 lysP 1-223 Yes Yes
25 lysP 224-477 No Yes
26 metG 1-193 No No
27 metG 194-406 No No
28 metG 407-588 No No
29 mtID 1-363 Yes Yes
30 mtlD 364-540 Yes Yes
31 mutS 1-240 Yes No
32 mutS 241-393 No No
33 mutS 394-507 No No
34 recA 1 -1 8 1 No No

119

Table 14 (continued).

 

GARD Position of Signiﬁcant Monophyletic
ﬁagment Locus breakpoint Taiima’s test relationship3
35 recA 182-510 Yes No
36 rpoS 1-360 No No
37 rpoS 361-5 85 No Yes
38 )2de 1-145 No No
39 yidB 146-430 No Yes

 

aMonophyletic relationship indicates that strains of all ﬁve groups labeled in.

120

environmental group I is as recombinationally isolated from E. coli as it is from E.
albertii, while environmental group II and E. coli are sharing loci much more
frequently.

Tajima’s test for group rates of evolution. Here, we wanted to test the
hypothesis that groups were not evolving at a similar rate (that of the molecular
clock). We used Tajima’s test (x2 test, df = 1, p < 0.050) to examine the
nucleotide sequence from a representative strain of each group. The analysis was
done on the third codon positions only, to minimize effects of selection. For each
fragment deﬁned by the GARD analysis, we conducted pairwise tests between E.
coli strain K-12 and one strain from E. albertii (strain 9194), environmental group
I (strain TW09231), and environmental group II (TW09308) using S. enterica
(strain Typhimurium LT2) as an outgroup. Of the 39 fragments, 7 (18%) showed
unequal rates of evolution (Table 14). The rate of the environmental group II
lineage was unequal to E. coli at all 7 fragments. It had more unique differences
than K-12 (accelerated rate of evolution) at 3 fragments and fewer unique
differences than K—12 (slower rate of evolution) at the other 4 fragments.
According to 2 fragments, environmental group I evolved slower than E. coli, and
only one fragment suggested that E. albertii experienced accelerated evolution
compared to E. coli. These data suggest a fairly uniform rate of evolution for
these groups.

Phylogenetic construction and divergence time. In order to construct
the most robust phylogeny possible, we used only the loci that satisﬁed two

criteria: 1) they show no evidence of between- group recombination

121

(monophyletic relationships) and 2) they must satisfy the molecular clock
hypothesis of uniform neutral substitution (non-signiﬁcant Tajima’s test). Of the
total 39 GARD ﬁ'agments, 8 satisﬁed both criteria (fragments 1, 5, 6, 7, 13, 25,
37, and 39) and were used to generate another phylogenetic network (Figure 12).
Adherence to these 2 criteria removed most of the phylogenetic incompatibility
between the groups. This analysis clearly shows that E. coli is more related to the
environmental groups than to E. albertii. In addition, the data suggest that E. coli,
environmental group I, and environmental group II diverged from one another in
a short period of time. This divergence occurred at such a rate that it is not clear
which of these groups is ancestral. This observation is in contrast to the
relationship inferred by the NJ tree (Figure 10A), where it appears that

environmental group II is ancestral to environmental group I and E. coli.

122

S. enterica

E. albertii

 

Environmental group I

Environmental group n
Figure 12. Neighbor-Net of E. coli, Environmental group I, Environmental group
II, E. albertii, and S. enterica based on 8 GARD fragments (fragments 1, 5, 6, 7,
13, 25, 37, and 39). These ﬁagments were selected based on criteria presented in
Table 14 (a monophyletic relationship for the groups and non-signiﬁcant Tajima’s

Test).

123

DISCUSSION

Divergent bacterial lineages within E. coli. All strains analyzed in this
study were conﬁrmed by standard biochemical tests. However, some strains are
remarkably divergent in their nucleotide sequence when compared to typical E.
coli strains that have been completely sequenced. The analysis presented here
shows that divergent E. coli strains represent two distinct evolutionary lineages.
Based on synonymous and nonsynonymous nucleotide substitutions in conserved
housekeeping genes, we have shown that one of these lineages, environmental
group II, is as divergent from E. coli as E. coli is from its recent ancestor E.
albertii. In addition, this group appears to be recombinationally isolated from
closely related lineages (E. albertii and E. coli).

The phylogenetic relationship between environmental group I and E. coli
is not as clear. Based on nucleotide substitutions, environmental group I is as
divergent from E. coli as environmental group II and E. albertii. However, there
does not appear to be a similar limitation on recombination between E. coli and
environmental group I as there is between E. coli and other groups. This suggests
that the environmental group I lineage is diverging away from E. coli, but has not
yet accrued enough genetic variation to be recombinationally isolated.

An interesting hypothesis generated by our analysis is that the
environmental groups are actually unique species. Two “gold standards” for
describing a bacterial species are similarity based on DNA-DNA hybridization
and a molecular characterization of 16S rRNA gene (159). While it is not the

purpose of this study to propose a formal new species designation, we feel that it

124

is noteworthy that we have sequenced 1,217 bp of the 16S rRNA gene in all
strains. Based on preliminary analyses, the 16S rRNA gene for E. coli,
environmental group 1, environmental group II, and E. albertii are no more than
98.7% similar. We recommend that lineages with this degree of similarity in the
16S rRNA gene undergo further analysis by DNA-DNA hybridization (159).
Regardless, these data further support our interpretation that environmental
groups I and H are as divergent from E. coli as E. coli is from its close relative E.
albertii.

The natural habitat of divergent E. coli strains. The current
understanding of the evolutionary ecology of E. coli appears to be biased by
analyzing strains from hosts. While the ﬁrst divergent lineage of E. coli to be
described came from host samples (182), these strains (2 strains of the same
genotype) were rare in the overall collection of 462 strains. While it is enticing to
describe this observation as a frequency (2/462 or 0.4%), it is worth noting that
the collection was not a random sample of strains ﬁom hosts. For example,
included in this collection were 72 strains from the E. coli reference collection
(ECOR) that represent much of the known diversity of the species (114). These
strains were selected ﬁ'om 2,600 strains that, among other criteria, represent the
breadth of genetic diversity across host species and geographic distribution. In
addition to the ECOR strains, the collection also included 15 strains representing
the breadth of genetic diversity found in a collection of 1,844 strains from humans
and their septic tanks (55). Based then on these numbers, a more accurate

estimate of how frequent this divergent E. coli lineage is in humans and animals is

125

on the order of 2 in 4,819 or 0.04%. However, this value is still inﬂated with
respect to other population genetic characterizations of E. coli isolated from
human and animal hosts (28, 53, 55, 56, 58, 180).

The most diverse E. coli collection to be analyzed by similar population
genetic methods was perhaps a worldwide sampling of 202 strains ﬁ'om mammals
and birds (157). ECOR strains were included in the analysis and based on their
position in a NJ dendrogram it appears that this collection still does not account
for the divergent lineages discussed here, although some strains do cluster outside
the diversity of ECOR. Regardless, we observed 2 divergent lineages among 6
strains of a randomized collection (196 strains) ﬁ'om freshwater beaches in
Michigan (169). Although it has not been addressed directly, both the literature
and our recent observations (169) suggest that- divergent E. coli strains are rare in
humans and animals, but abundant in freshwater beaches in Michigan (~0.04% vs.
~3%).

Further evidence that divergent E. coli strains are rare in human hosts
comes ﬁ'om an ongoing analysis of a collection of 715 enterotoxigenic E. coli
(ETEC) strains. This collection comes from a cohort sampling of infants from
Guinea Bissau in Africa (Table 12). Based on MLST analysis, 3 strains in this
collection (TW11930, TW12018, and TW14421) are indistinguishable at 7
housekeeping loci from 3 strains of environmental group 1(TW09231, TW09266,
and TWO9254). A fourth strain (TW11966) was also identiﬁed as environmental
group I based on clustering in a NJ tree. These 4 pathogens carry certain plasmid

encoded virulence factors that are common among ETEC. However,

126

environmental group I is not strictly an ETEC group because the strains analyzed
in this study do not carry the common virulence factors. If this collection is
representative of the normal E. coli ﬂora in humans, which is questionable, then
the frequency of divergent strains is 4 in 715 or 0.6%, an estimate that is still an
order of magnitude less than that observed in freshwater beaches. Our
interpretation of these data is that divergent E. coli lineages have an adaptive
advantage in certain environments outside the host where they are found more
frequently.

Conclusions. Divergent bacterial lineages exist within the biochemically-
deﬁned species, E. coli, and they may represent novel species. Similarly
divergent strains are rarely sampled ﬁ'om human and animal hosts. This
difference in abundance may be a result of adaptive evolution because they appear
to be much more prevalent in the secondary habitat. Finally, the divergent
lineages presented here represent excellent opportunities for comparative genomic
studies, as they can be used to test the rates of evolutionary processes, such as the

limits of gene ﬂow, between closely related, but distinct bacterial taxa.

127

ACKNOWLEDGMENTS
We would like to acknowledge Konstantinos Konstantinidis for his
supplying data on potential MLST loci for this study. This work was supported in
part by MRU matching funds from the College of Veterinary Medicine and

Graduate College at Michigan State University.

128

10.

11.

REFERENCES

Agapow, P., and A. Burt. 2000. MultiLocus, 1.2 ed. Dept. of Biology,
Imperial College, Silwood Park, Ascot, Berks.

Alm, E., J. Burke, and A. Spain. 2003. Fecal indicator bacteria are abundant
in wet sand at freshwater beaches. Water Res 37 :3978-82.

Altekruse, S. F., N. Bauer, A. Chanlongbutra, R. DeSaguu, A. Naugle, W.
Schlosser, R. Umholtz, and P. White. 2006. Salmonella enteritidis in broiler
chickens, United States, 2000-2005. Emerg Infect Dis 12:1848-52.

Anderson, K. L., J. E. Whitlock, and V. J. Harwood. 2005. Persistence and
differential survival of fecal indicator bacteria in subtropical waters and
sediments. Appl Environ Microbiol 71 :3041-8.

Anderson, M. A., J. E. Whitlock, and V. J. Harwood. 2006. Diversity and
distribution of Escherichia coli genotypes and antibiotic resistance phenotypes
in feces of humans, cattle, and horses. Appl Environ Microbiol 72:6914-22.

Angulo, F. J., and D. L. Swerdlow. 1998. Salmonella enteritidis infections in
the United States. J Am Vet Med Assoc 213:1729-31.

Arana, 1., A. Irizar, C. Seco, A. Muela, A. Fernandez-Astorga, and I.
Barcina. 2003. gfp-Tagged cells as a useful tool to study the survival of
Escherichia coli in the presence of the river microbial community. Microb
Ecol 45:29-38.

Barbosa, T. M., and S. B. Levy. 2000. The impact of antibiotic use on
resistance development and persistence. Drug Resist Updat 3:303-311.

Bartoloni, A., F. Bartalesi, A. Mantella, E. Dell'Amico, M. Roselli, M.
Strohmeyer, H. G. Barahona, V. P. Barron, F. Paradisi, and G. M.
Rossolini. 2004. High prevalence of acquired antimicrobial resistance
unrelated to heavy antimicrobial consumption. J Infect Dis 189:1291-4.

Bartoloni, A., L. Pallecchi, M. Benedetti, C. Fernandez, Y. Vallejos, E.
Guzman, A. L. Villagran, A. Mantella, C. Lucchetti, F. Bartalesi, M.
Strohmeyer, A. Bechini, H. Gamboa, H. Rodriguez, T. Falkenberg, G.
Kronvall, E. Gotuzzo, F. Paradisi, and G. M. Rossolini. 2006. Multidrug—
resistant commensal Escherichia coli in children, Peru and Bolivia. Emerg
Infect Dis 12:907-13.

Baumler, A. J., B. M. Hargis, and R. M. Tsolis. 2000. Tracing the origins of
Salmonella outbreaks. Science 287:50-2.

129

 

12.

13.

14.

15.

16.

l7.

l8.

19.

20.

21.

Beltran, P., J. M. Musser, R. Helmuth, J. J. F. 111, W. M. Frerichs, I. K.
Wachsmuth, K. Ferris, A. C. McWhorter, J. G. Wells, A. Cravioto, and
R. K. Selander. 1988. Toward a population genetic analysis of Salmonella:
Genetic diversity and relationships among strains of serotypes S. cholerasuis,
S. derby, S. dublin, S. enteritidis, S. heidelberg, S. infantis, S. neWport, and S.
typhimurium. Proceedings of the National Academy of Sciences USA
85:7753-7757.

Berge, A. C., E. R. Atwill, and W. M. Sischo. 2005. Animal and farm
inﬂuences on the dynamics of antibiotic resistance in faecal Escherichia coli
in young dairy calves. Prev Vet Med 69:25-38.

Berge, A. C., P. Lindeque, D. A. Moore, and W. M. Sischo. 2005. A
clinical trial evaluating prophylactic and therapeutic antibiotic use on health
and performance of preweaned calves. J Dairy Sci 88:2166-77.

Berge, A. C., D. A. Moore, and W. M. Sischo. 2006. Field trial evaluating
the inﬂuence of prophylactic and therapeutic antimicrobial administration on

antimicrobial resistance of fecal Escherichia coli in dairy calves. Appl
Environ Microbiol 72:3872-8.

Bermudez, M., and T. C. Hazen. 1988. Phenotypic and genotypic
comparison of Escherichia coli from pristine tropical waters. Appl Environ
Microbiol 54:979-83.

Bielaszewska, M., R. Prager, W. Zhang, A. W. Friedrich, A. Mellmann,
H. Tschape, and H. Karch. 2006. Chromosomal dynamism in progeny of
outbreak-related sorbitol-ferrnenting enterohemorrhagic Escherichia coli
0157:NM. Appl Environ Microbiol 72: 1900-9.

Bjorkman, J., I. Nagaev, O. G. Berg, D. Hughes, and D. I. Andersson.
2000. Effects of environment on compensatory mutations to ameliorate costs
of antibiotic resistance. Science 287 :1479-82.

Boerlin, P., R. Travis, C. L. Gyles, R. Reid-Smith, N. Janecko, H. Lim, V.
Nicholson, S. A. McEwen, R. Friendship, and M. Archambault. 2005.
Antimicrobial resistance and virulence genes of Escherichia coli isolates from
swine in Ontario. Appl Environ Microbiol 71 :6753-61.

Bohannan, B. J ., and J. Hughes. 2003. New approaches to analyzing
microbial biodiversity data. Curr Opin Microbiol 6:282-7.

Bopp, C. A., F. W. Brenner, J. G. Wells, and N. A. Strockbine. 1999.
Escherichia, Shigella, and Salmonella, p. 459-474. In P. R. Murray, Baron,

130

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

E.J., Jorgensen, J .H., Pfaller, M.A., Yolken, R.H. (ed.), Manual of Clinical
Microbiology, 7th ed. American Society for Microbiology, Washington, DC.

Boyd, E. F., F. S. Wang, P. Beltran, S. A. Plock, K. Nelson, and R. K.
Selander. 1993. Salmonella reference collection B (SARB): strains of 37
serovars of subspecies I. J Gen Microbiol 139 Pt 6:1125-32.

Brinas, L., M. A. Moreno, T. Teshager, M. Zarazaga, Y. Saenz, C.
Porrero, L. Dominguez, and C. Torres. 2003. Beta-lactamase
characterization in Escherichia coli isolates with diminished susceptibility or
resistance to extended-spectrum cephalosporins recovered from sick animals
in Spain. Microb Drug Resist 9:201-9.

Brown, A. H. D., M. W. Feldman, and E. Nevo. 1980. Multilocus structure
of natural populations of Hordeum spontaneum. Genetics 96:523-536.

Bruen, T. C., H. Philippe, and D. Bryant. 2006. A simple and robust
statistical test for detecting the presence of recombination. Genetics 172:2665-
8 1 .

Bryant, D., and V. Moulton. 2004. Neighbor-net: an agglomerative method
for the construction of phylogenetic networks. Mol Biol Evol 21 :255-65.

Byappanahalli, M. N., R. L. Whitman, D. A. Shively, M. J. Sadowsky, and
S. Ishii. 2006. Population structure, persistence, and seasonality of

autochthonous Escherichia coli in temperate, coastal forest soil from a Great
Lakes watershed. Environ Microbiol 8:504-13.

Caugant, D. A., B. R. Levin, and R. K. Selander. 1981. Genetic diversity
and temporal variation in the E. coli population of a human host. Genetics
98:467-90.

CDC. 2003. Salmonella surveillance summary, 2002. US Department of
Health and Human Survices.

Chan, E. S., J. Aramini, B. Ciebin, D. Middleton, R. Ahmed, M. Howes, I.
Brophy, I. Mentis, F. Jamieson, F. Rodgers, M. Nazarowec-White, S. C.
Pichette, J. Farrar, M. Gutierrez, W. J. Weis, L. Lior, A. Ellis, and S.
Isaacs. 2002. Natural or raw ahnonds and an outbreak of a rare phage type of
Salmonella enteritidis infection. Can Commun Dis Rep 28:97-9.

Chao, A. 1987. Estimating the population size for capture-recapture data with
unequal catchability. Biometrics 43:783-91.

Chapman, T. A., X. Y. Wu, 1. Barchia, K. A. Bettelheim, S. Driesen, D.
Trott, M. Wilson, and J. J. Chin. 2006. Comparison of virulence gene

131

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

proﬁles of Escherichia coli strains isolated from healthy and diarrheic swine.
Appl Environ Microbiol 72:4782-95.

Clermont, 0., S. Bouacorsi, and E. Bingen. 2000. Rapid and simple
determination of the Escherichia coli phylogenetic group. Appl Environ
Microbiol 66:4555-8.

Cogan, T. A., and T. J. Humphrey. 2003. The rise and fall of Salmonella
Enteritidis in the UK. J Appl Microbiol 94 Suppl:114S-119S.

Colom, K., J. Perez, R. Alonso, A. Fernandez-Aranguiz, E. Larino, and R.
Cisterna. 2003. Simple and reliable multiplex PCR assay for detection of
blaTEM, b1a(SHV) and blaOXA-l genes in Enterobacteriaceae. FEMS
Microbiol Lett 223: 147-5 1 .

Colwell, R. 2005. Estimates: Statistical estimation on species richness and
shared species from samples, 7.5 ed.

Czeczulin, J. R., T. S. Whittam, I. R. Henderson, F. Navarro-Garcia, and
J. P. Nataro. 1999. Phylogenetic analysis of enteroaggregative and diffusely
adherent Escherichia coli. Infect Irnmun 67 :2692-9.

Davies, R., and M. Breslin. 2001. Environmental contamination and
detection of Salmonella enterica serovar enteritidis in laying ﬂocks. Vet Rec
149:699-704.

Davies, R. H., and C. Wray. 1995. Mice as carriers of Salmonella enteritidis
on persistently infected poultry units. Vet Rec 137 :337-41.

De Buck, J., F. Pasmans, F. Van Immerseel, F. Haesebrouck, and R.
Ducatelle. 2004. Tubular glands of the isthmus are the predominant
colonization site of Salmonella enteritidis in the upper oviduct of laying hens.
Poult Sci 83:352-8.

De Buck, J., F. Van Immerseel, F. Haesebrouck, and R. Ducatelle. 2004.
Effect of type 1 ﬁmbriae of Salmonella enterica serotype Enteritidis on

bacteraemia and reproductive tract infection in laying hens. Avian Pathol
33 :3 14-20.

Donaldson, S. C., B. A. Straley, N. V. Hegde, A. A. Sawant, C. DebRoy,
and B. M. J ayarao. 2006. Molecular epidemiology of ceﬁiofur-resistant

Escherichia coli isolates from dairy calves. Appl Environ Microbiol 72:3940-
8.

132

 

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

Elena, S. F., and R. E. Lenski. 2003. Evolution experiments with

microorganisms: the dynamics and genetic bases of adaptation. Nat Rev Genet
4:457-69.

Escobar—Paramo, P., O. Clermont, A. B. Blanc-Potard, H. Bui, C. Le
Bouguenec, and E. Denamur. 2004. A speciﬁc genetic background is

required for acquisition and expression of virulence factors in Escherichia
coli. Mol Biol Evol 21:1085-94.

Escobar-Paramo, P., K. Grenet, A. Le Menac'h, L. Rode, E. Salgado, C.
Amorin, S. Gouriou, B. Picard, M. C. Rahimy, A. Andremont, E.

Denamur, and R. Ruimy. 2004. Large-scale population structure of human
commensal Escherichia coli isolates. Appl Environ Microbiol 70:5698-700.

Falkow, S. 1996. The evolution of pathogenicity in Escherichia coli, Shigella,
and Salmonella, p. 2723-2729. In F. C. Neidhardt, Ingraham, J.L., Lin, E. C.
C., Low, K. B., Magasanik, B., Reznikoff, W. S., Riley, M., Schaechter, M.,
and Umarger, H.E. (ed.), Escherichia coli and Salmonella: Cellular and
Molecular Biology, 2nd ed, vol. 2. ASM Press, Washington, DC.

Farmer, J. J. I. 2003. Enterobacteriaceae: Introduction and Identiﬁcation, p.
636-653. In P. R. Murray, Baron, E.J., Jorgensen, J .H., Pfaller, M.A., Yolken,
R.H. (ed.), Manual of Clinical Microbiology, 8th ed, vol. 1. ASM Press,
Washington, DC.

Feil, E. J. 2004. Small change: keeping pace with rnicroevolution. Nat Rev
Microbiol 2:483-95.

Feil, E. J., B. C. Li, D. M. Aanensen, W. P. Hanage, and B. G. Spratt.
2004. eBURST: inferring patterns of evolutionary descent among clusters of
related bacterial genotypes from multilocus sequence typing data. J Bacteriol

186: 1518-30.

Feil, E. J ., and B. G. Spratt. 2001. Recombination and the population
structures of bacterial pathogens. Annu Rev Microbiol 55:561-90.

Garber, L., M. Smeltzer, P. Fedorka-Cray, S. Ladely, and K. Ferris.
2003. Salmonella enterica serotype enteritidis in table egg layer house
environments and in mice in US. layer houses and associated risk factors.
Avian Dis 47:134-42.

Gast, R. K., and P. S. Holt. 2001. Multiplication in egg yolk and survival in

egg albumen of Salmonella enterica serotype Enteritidis strains of phage types
4, 8, 13a, and 14b. J Food Prot 64:865-8.

133

1‘-

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

Gordon, D. M. 1997. The genetic structure of Escherichia coli populations in
feral house mice. Microbiology 143 ( Pt 6):2039-46.

Gordon, D. M. 2001. Geographical structure and host speciﬁcity in bacteria
and the implications for tracing the source of coliform contamination.
Microbiology 147 :1079-85.

Gordon, D. M., S. Bauer, and J. R. Johnson. 2002. The genetic structure of
Escherichia coli populations in primary and secondary habitats. Microbiology
148:1513-22.

Gordon, D. M., and A. Cowling. 2003. The distribution and genetic structure
of Escherichia coli in Australian vertebrates: host and geographic effects.
Microbiology 149:3575-86.

Gordon, D. M., and F. FitzGibbon. 1999. The distribution of enteric bacteria
ﬁom Australian mammals: host and geographical effects. Microbiology 145 (
Pt 10):2663-71.

Gordon, D. M., S. E. Stern, and P. J. Collignon. 2005. Inﬂuence of the age
and sex of human hosts on the distribution of Escherichia coli ECOR groups
and virulence traits. Microbiology 151:15-23.

Guard-Bouldin, J., R. K. Gast, T. J. Humphrey, D. J. Henzler, C.
Morales, and K. Coles. 2004. Subpopulation characteristics of egg-
contarninating Salmonella enterica serovar Enteritidis as deﬁned by the
lipopolysaccharide O chain. Appl Environ Microbiol 70:2756-63.

Guard-Petter, J. 2001. The chicken, the egg and Salmonella enteritidis.
Environ Microbiol 3:421-30.

Guard-Potter, J., D. J. Henzler, M. M. Rahman, and R. W. Carlson. 1997.
On-farm monitoring of mouse-invasive Salmonella enterica serovar enteritidis
and a model for its association with the production of contaminated eggs.
Appl Environ Microbiol 63:1588-93.

Guttman, D. S., and D. E. Dykhuizen. 1994. Clonal divergence in
Escherichia coli as a result of recombination, not mutation. Science 266:1380-
3.

Guttmau, D. S., and D. E. Dykhuizen. 1994. Detecting selective sweeps in
naturally occurring Escherichia coli. Genetics 138:993-1003.

Hamilton, M. J., T. Yan, and M. J. Sadowsky. 2006. Development of

goose- and duck-speciﬁc DNA markers to determine sources of Escherichia
coli in waterways. Appl Environ Microbiol 72:4012-9.

134

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

Haubold, B., M. Travisano, P. B. Rainey, and R. R. Hudson. 1998.
Detecting linkage disequilibrium in bacterial populations. Genetics 150: 1 341-
8.

Hedberg, C. W., M. J. David, K. E. White, K. L. MacDonald, and M. T.
Osterholm. 1993. Role of egg consumption in sporadic Salmonella enteritidis

and Salmonella typhimurium infections in Minnesota. J Infect Dis 167:107-
1 l.

Henzler, D. J., and H. M. Opitz. 1992. The role of mice in the epizootiology
of Salmonella enteritidis infection on chicken layer farms. Avian Dis 36:625-
3 1 .

Herzer, P. J ., S. Inouye, M. Inouye, and T. S. Whittam. 1990. Phylogenetic
distribution of branched RNA-linked multicopy single-stranded DNA among
natural isolates of Escherichia coli. J Bacteriol 172:6175-81.

Hoffmann, H., M. W. Hornef, S. Schubert, and A. Roggenkamp. 2001.
Distribution of the outer membrane haem receptor protein ChuA in

environmental and human isolates of Escherichia coli. Int J Med Microbiol
291 :227-30.

Holt, P. S. 1995. Horizontal transmission of Salmonella enteritidis in molted
and unmolted laying chickens. Avian Dis 39:239-49.

Hughes, J. B., J. J. Hellmann, T. H. Ricketts, and B. J. Bohannan. 2001.
Counting the uncountable: statistical approaches to estimating microbial
diversity. Appl Environ Microbiol 67 :4399-406.

Humphrey, T. J. 1994. Contamination of egg shell and contents with
Salmonella enteritidis: a review. Int J Food Microbiol 21 :31-40.

Humphrey, T. J., E. Slater, K. McAlpine, R. J. Rowbury, and R. J.
Gilbert. 1995. Salmonella enteritidis phage type 4 isolates more tolerant of

heat, acid, or hydrogen peroxide also survive longer on surfaces. Appl
Environ Microbiol 61 :3 161-4.

Husou, D. H., and D. Bryant. 2006. Application of phylogenetic networks in
evolutionary studies. Mol Biol Evol 23:254—67.

Hyma, K. E., D. W. Lacher, A. M. Nelson, A. C. Bumbaugh, J. M. Janda,
N. A. Strockbine, V. B. Young, and T. S. Whittam. 2005. Evolutionary
genetics of a new pathogenic Escherichia species: Escherichia albertii and
related Shigella boydii strains. J Bacteriol 187 :619-28.

135

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

Isaacs, S., J. Aramini, B. Ciebin, J. A. Farrar, R. Ahmed, D. Middleton,
A. U. Chandran, L. J. Harris, M. Howes, E. Chan, A. S. Pichette, K.
Campbell, A. Gupta, L. Y. Lior, M. Pearce, C. Clark, F. Rodgers, F.
Jamieson, l. Brophy, and A. Ellis. 2005. An international outbreak of
salmonellosis associated with raw almonds contaminated with a rare phage
type of Salmonella enteritidis. J Food Prot 68:191-8.

Ishii, S., W. B. Ksoll, R. E. Hicks, and M. J. Sadowsky. 2006. Presence and
growth of naturalized Escherichia coli in temperate soils ﬁ'om Lake Superior
watersheds. Appl Environ Microbiol 72:612-21.

Jacobs, J. L., T. L. Carroll, and G. W. Sundin. 2005. The role of
pigmentation, ultraviolet radiation tolerance, and leaf colonization strategies
in the epiphytic survival of phyllosphere bacteria. Microb Ecol 49:104-13.

Jenkins, M. B., P. G. Harte], T. J. Olexa, and J. A. Stuedemann. 2003.
Putative temporal variability of Escherichia coli ribotypes from yearling
steers. J Environ Qual 32:305-9.

Johnson, L. K., M. B. Brown, E. A. Carruthers, J. A. Ferguson, P. E.
Dombek, and M. J. Sadowsky. 2004. Sample size, library composition, and
genotypic diversity among natural populations of Escherichia coli from

different animals inﬂuence accuracy of determining sources of fecal pollution.
Appl Environ Microbiol 70:4478-85.

Jones, T., C. O. Gill, and L. M. McMullen. 2004. The behaviour of log
phase Escherichia coli at temperatures that ﬂuctuate about the minimum for
growth. Lett Appl Microbiol 39:296-300.

Khachatryan, A. R., T. E. Besser, D. D. Hancock, and D. R. Call. 2006.
Use of a nonmedicated dietary supplement correlates with increased
prevalence of streptomycin-sulfa-tetracycline-resistant Escherichia coli on a
dairy farm. Appl Environ Microbiol 72:45 83-8.

Khachatryan, A. R., D. D. Hancock, T. E. Besser, and D. R. Call. 2006.
Antimicrobial drug resistance genes do not convey a secondary ﬁtness
advantage to calf-adapted Escherichia coli. Appl Environ Microbiol 72:443-8.

Khachatryan, A. R., D. D. Hancock, T. E. Besser, and D. R. Call. 2004.
Role of calf-adapted Escherichia coli in maintenance of antimicrobial drug
resistance in dairy calves. Appl Environ Microbiol 70:752-7.

Kilmartin, D., D. Morris, C. O'Hare, G. Corbett-Feeney, and M.
Cormican. 2005. Clonal expansion may account for high levels of quinolone
resistance in Salmonella enterica serovar enteritidis. Appl Environ Microbiol
71 :25 87-91.

136

 

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

Kimura, A. C., V. Reddy, R. Marcus, P. R. Cieslak, J. C. Mohle-Boetani,
H. D. Kassenborg, S. D. Segler, F. P. Hardnett, T. Barrett, and D. L.
Swerdlow. 2004. Chicken consumption is a newly identiﬁed risk factor for
sporadic Salmonella enterica serotype Enteritidis infections in the United
States: a case-control study in F oodNet sites. Clin Infect Dis 38 Suppl
3:S244-52.

Konstantinidis, K. T., A. Ramette, and J. M. Tiedje. 2006. Toward a more
robust assessment of intraspecies diversity, using fewer genetic markers. Appl
Environ Microbiol 72:7286-93.

Kosakovsky Pond, S. L., D. Posada, M. B. Gravenor, C. H. Woelk, and S.
D. Frost. 2006. GARD: a genetic algorithm for recombination detection.
Bioinformatics 22:3096-8.

Kudva, I. T., P. S. Evans, N. T. Perna, T. J. Barrett, F. M. Ausubel, F. R.
Blattner, and S. B. Calderwood. 2002. Strains of Escherichia coli 0157:H7
differ primarily by insertions or deletions, not single-nucleotide
polymorphisms. Journal of Bacteriology 184:1873-1879.

Kuehl, C. J., H. D. Wood, T. L. Marsh, T. M. Schmidt, and V. B. Young.
2005. Colonization of the cecal mucosa by Helicobacter hepaticus impacts the
diversity of the indigenous microbiota. Infect Immun 73:6952-61.

Kumar, S., K. Tamura, and M. Nei. 2004. MEGA3: Integrated software for
molecular evolutionary genetics analysis and sequence alignment. Brief
Bioinfonn 5:150-63.

Lasalde, C., R. Rodriguez, and G. A. Toranzos. 2005. Statistical analyses:
possible reasons for unreliability of source tracking efforts. Appl Environ
Microbiol 71 :4690-5.

Levin, B. R., and C. T. Bergstrom. 2000. Bacteria are different:
observations, interpretations, speculations, and opinions about the
mechanisms of adaptive evolution in prokaryotes. Proc Natl Acad Sci U S A
97 :698 1-5.

Levy, S. B. 2002. The 2000 Garrod lecture. Factors impacting on the problem
of antibiotic resistance. J Antimicrob Chemother 49:25-30.

Ley, R. E., F. Backhed, P. Turnbaugh, C. A. Lozupone, R. D. Knight, and

J. 1. Gordon. 2005. Obesity alters gut microbial ecology. Proc Natl Acad Sci
U S A 102:11070-5.

137

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

106.

Li, J., N. H. Smith, K. Nelson, P. B. Crichton, D. C. Old, T. S. Whittam,
and R. K. Selander. 1993. Evolutionary origin and radiation of the avian-
adapted non-motile salmonellae. J Med Microbiol 38:129-39.

Lindstedt, B. A., E. Heir, I. Nygard, and G. Kapperud. 2003.
Characterization of class I integrons in clinical strains of Salmonella enterica

subsp. enterica serovars Typhimurium and Enteritidis from Norwegian
hospitals. J Med Microbiol 52: 141-9.

Maiden, M. C., J. A. Bygraves, E. Feil, G. Morelli, J. E. Russell, R. Urwin,
Q. Zhang, J. Zhou, K. Zurth, D. A. Caugant, I. M. Feavers, M. Achtman,
and B. G. Spratt. 1998. Multilocus sequence typing: a portable approach to
the identiﬁcation of clones within populations of pathogenic microorganisms.

Proc Natl Acad Sci U S A 95:3140-5.

Maisnier-Patin, S., and D. I. Andersson. 2004. Adaptation to the deleterious
effects of antimicrobial drug resistance mutations by compensatory evolution.
Res Microbiol 155:360-9.

Maisnier-Patin, S., O. G. Berg, L. Liljas, and D. I. Andersson. 2002.
Compensatory adaptation to the deleterious effect of antibiotic resistance in
Salmonella typhimurium. Mol Microbiol 46:355-66.

Maslow, J. N., T. S. Whittam, C. F. Gilks, R. A. Wilson, M. E. Mulligan,
K. S. Adams, and R. D. Arbeit. 1995. Clonal relationships among
bloodstream isolates of Escherichia coli. Infect Irnmun 63:2409-17.

Maynard Smith, J., N. H. Smith, M. O'Rourke, and B. G. Spratt. 1993.
How clonal are bacteria? Proc Natl Acad Sci U S A 90:4384-8.

McLellau, S. L. 2004. Genetic diversity of Escherichia coli isolated from
urban rivers and beach water. Appl Environ Microbiol 70:4658-65.

McLellau, S. L., A. D. Daniels, and A. K. Salmore. 2001. Clonal
populations of therrnotolerant Enterobacteriaceae in recreational water and

their potential interference with fecal Escherichia coli counts. Appl Environ
Microbiol 67 :4934-8.

Mead, P. S., L. Slutsker, V. Dietz, L. F. McCaig, J. S. Bresee, C. Shapiro,
P. M. Grifﬁn, and R. V. Tauxe. 1999. Food-related illness and death in the
United States. Emerg Infect Dis 5:607-25.

Milkman, R. 1973. Electrophoretic variation in Escherichia coli from natural
sources. Science 182:1024-6.

138

 

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

Miller, W. A., M. A. Miller, I. A. Gardner, E. R. Atwill, B. A. Byrne, S.

J ang, M. Harris, J. Ames, D. Jessup, D. Paradies, K. Worcester, A. Melli,
and P. A. Conrad. 2006. Salmonella spp., Vibrio spp., Clostridium
peifringens, and Plesiomonas shigelloides in marine and freshwater
invertebrates ﬁ'om coastal California ecosystems. Microb Ecol 52:198-206.

Nakamura, M., N. Nagamine, M. Norimatsu, S. Suzuki, K. Ohishi, M.
Kijima, Y. Tamura, and S. Sato. 1993. The ability of Salmonella enteritidis

isolated from chicks imported from England to cause transovarian infection. J
Vet Med Sci 55:135-6.

NCCLS. 2002. Performance standard for antimicrobial disk and dilution
susceptibility tests for bacteria isolated from animals. Approved standard,
2nd ed, Wayne, PA.

Nei, M., and S. Kumar. 2000. Molecular evolution and phylogenetics.
Oxford University Press, New York, NY.

Nowrouzian, F., B. Hesselmar, R. Saalman, I. L. Strannegard, N. Aberg,
A. E. Wold, and I. Adlerberth. 2003. Escherichia coli in infants' intestinal
microﬂora: colonization rate, strain turnover, and virulence gene carriage.
Pediatr Res 54:8-14.

 

,‘ﬂ': -. .-

Nowrouzian, F. L., I. Adlerberth, and A. E. Wold. 2006. Enhanced
persistence in the colonic microbiota of Escherichia coli strains belonging to

phylogenetic group B2: role of virulence factors and adherence to colonic
cells. Microbes Infect 8:834-40.

Nowrouzian, F. L., A. E. Wold, and I. Adlerberth. 2005. Escherichia coli
strains belonging to phylogenetic group B2 have superior capacity to persist in
the intestinal microﬂora of infants. J Infect Dis 191:1078-83.

Ochman, H., and R. K. Selander. 1984. Standard reference strains of
Escherichia coli from natural populations. J Bacteriol 157 :690-3.

Ochman, H., T. S. Whittam, D. A. Caugant, and R. K. Selander. 1983.
Enzyme polymorphism and genetic population strucutre in Escherichia coli
and Shigella. Journal of General Microbiology 129:2715-2726.

Painter, J. A., K. Molbak, J. Sonne-Hansen, T. Barrett, J. G. Wells, and
R. V. Tauxe. 2004. Salmonella-based rodenticides and public health. Emerg
Infect Dis 10:985-7.

Pallecchi, L., C. Lucchetti, A. Bartoloni, F. Bartalesi, A. Mantella, H.

Gamboa, A. Carattoli, F. Paradisi, and G. M. Rossolini. 2007. Population
structure and resistance genes in antibiotic-resistant bacteria ﬁ'om a remote

139

118.

119.

120.

121.

122.

123.

124.

125.

126.

community with minimal antibiotic exposure. Antimicrob Agents Chemother.
51 :1 179-84.

Pallecchi, L., M. Malossi, A. Mantella, E. Gotuzzo, C. Trigoso, A.
Bartoloni, F. Paradisi, G. Kronvall, and G. M. Rossolini. 2004. Detection
of CTX-M-type beta-lactamase genes in fecal Escherichia coli isolates from
healthy children in Bolivia and Peru. Antimicrob Agents Chemother 48:4556-
61.

Parker, C. T., B. Harmon, and J. Guard-Petter. 2002. Mitigation of avian
reproductive tract function by Salmonella enteritidis producing high-
molecular-mass lipopolysaccharide. Environ Microbiol 4:538-45.

Patrick, M. E., P. M. Adcock, T. M. Gomez, S. F. Altekruse, B. H.
Holland, R. V. Tauxe, and D. L. Swerdlow. 2004. Salmonella enteritidis
infections, United States, 1985-1999. Emerg Infect Dis 10:1-7.

Perez-Rosas, N., and T. C. Hazen. 1989. In situ survival of Vibrio cholerae
and Escherichia coli in a tropical rain forest watershed. Appl Environ
Microbiol 55:495-9.

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 U S’A 86:3818-22.

Power, M. L., J. Littlefield-Wyer, D. M. Gordon, D. A. Veal, and M. B.
Slade. 2005. Phenotypic and genotypic characterization of encapsulated
Escherichia coli isolated from blooms in two Australian lakes. Environ
Microbiol 7 :63 1-40.

Qi, W., D. W. Lacher, A. C. Bumbaugh, K. E. Hyma, L. M. Ouellette, T.
M. Large, C. L. Tan, and T. S. Whittam. 2004. Presented at the
Proceedings of the 2004 IEEE Computational Systems Bioinforrnatics
Conference CSB2004, Los Alarnitos, CA.

Rahman, M. M., J. Guard-Petter, and R. W. Carlson. 1997. A virulent
isolate of Salmonella enteritidis produces a Salmonella typhi-like
lipopolysaccharide. J Bacteriol 179:2126-31.

Ram, J. L., R. P. Ritchie, J. Fang, F. S. Gonzales, and J. P. Selegean.

2004. Sequence-based source tracking of Escherichia coli based on genetic
diversity of beta-glucuronidase. J Environ Qual 33:1024-32.

140

127.

128.

129.

130.

131.

132.

133.

134.

135.

136.

137.

138.

Raymond, M. J., R. D. Wohrle, and D. R. Call. 2006. Assessment and
promotion of judicious antibiotic use on dairy farms in Washington State. J
Dairy Sci 89:3228-40.

Reid, S. D., C. J. Herbelin, A. C. Bumbaugh, R. K. Selander, and T. S.
Whittam. 2000. Parallel evolution of virulence in pathogenic Escherichia
coli. Nature 406:64-7.

Reynolds, M. G. 2000. Compensatory evolution in rifarnpin-resistant
Escherichia coli. Genetics 156: 1471-81.

Rice, E. W., C. H. Johnson, M. E. Dunnigan, and D. J. Reasoner. 1993.
Rapid glutamate decarboxylase assay for detection of Escherichia coli. Appl
Environ Microbiol 59:4347-9.

Rivera, S. C., T. C. Hazen, and G. A. Toranzos. 1988. Isolation of fecal
coliforms ﬁ'om pristine sites in a tropical rain forest. Appl Environ Microbiol
54:5 13-7.

Rodrigue, D. C., R. V. Tauxe, and B. Rowe. 1990. International increase in
Salmonella enteritidis: a new pandemic? Epidemiol Infect 105:21-7. E

 

Sabate, M., E. Moreno, T. Perez, A. Andreu, and G. Prats. 2006.
Pathogenicity island markers in commensal and uropathogenic Escherichia
coli isolates. Clin Microbiol Infect 12:880—6.

Sander, P., B. Springer, T. Prammananan, A. Sturmfels, M. Kappler, M.
Pletschette, and E. C. Bottger. 2002. Fitness cost of chromosomal drug
resistance-conferring mutations. Antimicrob Agents Chemother 46: 1204-1 1.

Sannes, M. R., M. A. Kuskowski, K. Owens, A. Gajewski, and J. R.
Johnson. 2004. Virulence factor proﬁles and phylogenetic background of

Escherichia coli isolates from veterans with bacteremia and uninfected control
subjects. J Infect Dis 190:2121-8.

Sato, K., P. C. Bartlett, and M. A. Saeed. 2005. Antimicrobial susceptibility
of Escherichia coli isolates from dairy farms using organic versus
conventional production methods. J Am Vet Med Assoc 226:5 89-94.

Savageau, M. A. 1983. Escherichia coli habitats, cell types, and molecular
mechanisms of gene control. The American Naturalist 122:732-744.

Sawant, A. A., N. V. Hegde, B. A. Straley, S. C. Donaldson, B. C. Love, S.

J. Knabel, and B. M. Jayarao. 2007. Antimicrobial-resistant enteric bacteria
from dairy cattle. Appl Environ Microbiol 73: 156-63.

141

139.

140.

141.

142.

143.

144.

145.

146.

147.

148.

Sawant, A. A., L. M. Sordillo, and B. M. Jayarao. 2005. A survey on
antibiotic usage in dairy herds in Pennsylvania. J Dairy Sci 88:2991-9.

Schmidt, T. M., and D. A. Relman. 1994. Phylogenetic identiﬁcation of
uncultured pathogens using ribosomal RNA sequences. Methods Enzymol
235:205-22.

Schrag, S. J., V. Perrot, and B. R. Levin. 1997. Adaptation to the ﬁtness
costs of antibiotic resistance in Escherichia coli. Proc Biol Sci 264:1287-91.

Schroeder, C. M., A. L. Naugle, W. D. Schlosser, A. T. Hogue, F. J.
Angulo, J. S. Rose, E. D. Ebel, W. T. Disney, K. G. Holt, and D. P.

Goldman. 2005. Estimate of illnesses from Salmonella enteritidis in eggs,
United States, 2000. Emerg Infect Dis 11:113-5.

Selander, R. K., P. Beltran, N. H. Smith, R. Helmuth, F. A. Rubin, D. J.
Kopecko, K. Ferris, B. D. Tall, A. Cravioto, and J. M. Musser. 1990.
Evolutionary genetic relationships of clones of Salmonella serovars that cause
human typhoid and other enteric fevers. Infection and Immunity 58:2262-
2275.

Selander, R. K., D. A. Caugant, H. Ochman, J. M. Musser, M. N.
Gilmour, and T. S. Whittam. 1986. Methods of multilocus enzyme

electrophoresis for bacterial population genetics and systematics. Appl
Environ Microbiol 51:873-84.

Selander, R. K., Caugant, D.A., and Whittam, T.S. 1987. Genetic structure
and variation in natural populations of Escherichia coli., p. 1625-1648. In F.
C. Neidhardt, Ingraham, J .L., Magasanik, B., Schaechter, M., and Umarger,
H.E. (ed.), Escherichia coli and Salmonella typhimurium: Cellular and
Molecular Biology. American Society for Microbiology, Washington, DC.

Selander, R. K., and B. R. Levin. 1980. Genetic diversity and structure in
Escherichia coli populations. Science 210:545-7.

Selander, R. K., Li, J., Nelson, K. 1996. Evolutionary genetics of Salmonella
enterica, p. 2691-2707. In F. C. Neidhardt, Ingraham, J .L., Lin, B. C. C., Low,
K. B., Magasanik, B., Reznikoff, W. S., Riley, M., Schaechter, M., and
Umarger, H.E. (ed.), Escherichia coli and Salmonella: Cellular and Molecular
Biology, 2nd ed, vol. 2. ASM Press, Washington, DC.

Selander, R. K., J. M. Musser, D. A. Caugant, M. N. Gilmour, and T. S.

Whittam. 1987. Population genetics of pathogenic bacteria. Microb Pathog
3:1-7.

142

149.

150.

151.

152.

153.

154.

155.

156.

157.

158.

Sherley, M., D. M. Gordon, and P. J. Collignon. 2004. Evolution of multi-
resistance plasmids in Australian clinical isolates of Escherichia coli.
Microbiology 150: 1 539-46.

Shivaprasad, H. L. 2000. Fowl typhoid and pullorum disease. Rev Sci Tech
19:405-24.

Simpson, K. W., B. Dogan, M. Rishniw, R. E. Goldstein, S. Klaessig, P. L.
McDonough, A. J. German, R. M. Yates, D. G. Russell, 8. E. Johnson, D.
E. Berg, J. Harel, G. Bruant, S. P. McDonough, and Y. H. Schukken.
2006. Adherent and invasive Escherichia coli is associated with
granulomatous colitis in boxer dogs. Infect Irnmun 74:4778-92.

Singh, R., C. M. Schroeder, J. Meng, D. G. White, P. F. McDermott, D. D.
Wagner, H. Yang, S. Simjee, C. Debroy, R. D. Walker, and S. Zhao. 2005.
Identiﬁcation of antimicrobial resistance and class 1 integrons in Shiga toxin-

producing Escherichia coli recovered from humans and food animals. J
Antimicrob Chemother 56:216-9.

Sinton, L. W., C. H. Hall, P. A. Lynch, and R. J. Davies-Colley. 2002.
Sunlight inactivation of fecal indicator bacteria and bacteriophages from
waste stabilization pond efﬂuent in fresh and saline waters. Appl Environ '-
Microbiol 68:1 122-31.

 

Sokal, R. R., and F. J. Rohlf. 1995. Biometry, 3rd ed. W. H. Freeman and
Company, New York.

Solano, C., B. Sesma, M. Alvarez, T. J. Humphrey, C. J. Thorns, and C.
Gamazo. 1998. Discrimination of strains of Salmonella enteritidis with
differing levels of virulence by an in vitro glass adherence test. J Clin
Microbiol 36:674-8.

Solo-Gabriele, H. M., M. A. Wolfert, T. R. Desmarais, and C. J. Palmer.
2000. Sources of Escherichia coli in a coastal subtropical environment. Appl
Environ Microbiol 66:230-7.

Souza, V., M. Rocha, A. Valera, and L. E. Eguiarte. 1999. Genetic
structure of natural populations of Escherichia coli in wild hosts on different
continents. Appl Environ Microbi0165:3373-85.

St Louis, M. E., D. L. Morse, M. E. Potter, T. M. DeMelfi, J. J. Guzewich,
R. V. Tauxe, and P. A. Blake. 1988. The emergence of grade A eggs as a
major source of Salmonella enteritidis infections. New implications for the
control of salmonellosis. J ama 259:2103-7.

143

159.

160.

161.

162.

163.

164.

165.

166.

167.

168.

169.

170.

Stackebrandt, E., and J. Ebers. 2006. Taxonomic properties revisited: ‘
tarnished gold standards. Microbiology Today:6-9.

Stoeckel, D. M., M. V. Mathes, K. E. Hyer, C. Hagedom, H. Kator, J.
Lukasik, T. L. O'Brien, T. W. Fenger, M. Samadpour, K. M. Strickler,
and B. A. Wiggins. 2004. Comparison of seven protocols to identify fecal

contamination sources using Escherichia coli. Environ Sci T echnol 38:6109-
1 7.

Thiagarajan, D., A. M. Saeed, and E. K. Asem. 1994. Mechanism of
transovarian transmission of Salmonella enteritidis in laying hens. Poult Sci
73:89-98.

Thiagarajan, D., M. Saeed, J. Turek, and E. Asem. 1996. In vitro
attachment and invasion of chicken ovarian granulosa cells by Salmonella
enteritidis phage type 8. Infect Irnmun 64:5015-21.

Threlfall, E. J ., A. M. Ridley, L. R. Ward, and B. Rowe. 1996. Assessment
of health risk from Salmonella-based rodenticides. Lancet 348:616-7.

Truchanowicz, J., E. Burek, and Z. Gorzelak. 1970. [Clinical observations
on the course of Salmonella enteritidis infections in children]. Przegl
Epidemiol 24: 101-6.

Tsolis, R. M., L. G. Adams, T. A. Ficht, and A. J. Baunrler. 1999.
Contribution of Salmonella typhimurium virulence factors to diarrhea] disease
in calves. Infect Irnmun 67 :4879-85.

Turnbaugh, P. J., R. E. Ley, M. A. Mahowald, V. Magrini, E. R. Mardis,
and J. 11. Gordon. 2006. An obesity-associated gut microbiome with
increased capacity for energy harvest. Nature 444: 1027—31.

USEPA. 2000. Improved enumeration methods for the recreational water
quality indicators: Enterococci and Escherichia coli. USEPA.

Velge, P., A. Cloeckaert, and P. Barrow. 2005. Emergence of Salmonella
epidemics: the problems related to Salmonella enterica serotype Enteritidis
and multiple antibiotic resistance in other major serotypes. Vet Res 36:267-
88.

Walk, S. T., E. W. Alm, L. M. Calhoun, J. M. Mladonicky, and T. S.
Whittam. 2007. Genetic diverstiy and population structure of Escherichia
coli isolated from freshwater beaches. Environ Microbiol In Press.

Weissman, S. J., S. Chattopadhyay, P. Aprikian, M. Obata-Yasuoka, Y.
Yarova-Yarovaya, A. Stapleton, W. Ba-Thein, D. Dykhuizen, J. R.

144

171.

172.

173.

174.

175.

176.

177.

178.

179.

180.

Johnson, and E. V. Sokurenko. 2006. Clonal analysis reveals high rate of
structural mutations in ﬁmbrial adhesins of extraintestinal pathogenic
Escherichia coli. Mol Microbiol 59:975-88.

Whitman, R. L., and M. B. Nevers. 2003. Foreshore sand as a source of

Escherichia coli in nearshore water of a Lake Michigan beach. Appl Environ
Microbiol 69:5555-62.

Whitman, R. L., M. B. Nevers, G. C. Korinek, and M. N. Byappanahalli.
2004. Solar and temporal effects on Escherichia coli concentration at a Lake
Michigan swimming beach. Appl Environ Microbiol 70:4276-85.

Whittam, T. S. 1989. Clonal dynamics of Escherichia coli in its natural
habitat. Antonie Van Leeuwenhoek 55:23-32.

Whittam, T. S. 1996. Genetic variation and evolutionary processes in natural
populations of Escherichia coli, p. 2708-2720. In F. C. Neidhardt, Ingraham,
J .L., Lin, B. C. C., Low, K. B., Magasanik, B., Reznikoff, W. S., Riley, M.,
Schaechter, M., and Umarger, H.E. (ed.), Escherichia coli and Salmonella:
Cellular and Molecular Biology, 2nd ed, vol. 2. ASM Press, Washingtion,
D.C.

Whittam, T. S., and A. C. Bumbaugh. 2002. Inferences from whole-genome
sequences of bacterial pathogens. Curr Opin Genet Dev 12:719-25.

Whittam, T. S., H. Ochman, and R. K. Selander. 1983. Multilocus genetic
structure in natural populations of Escherichia coli. Proc Natl Acad Sci U S A
80:17 5 1-5.

Whittam, T. S., I. K. Wachsmuth, and R. A. Wilson. 1988. Genetic
evidence of clonal descent of Escherichia coli 0157:H7 associated with

hemorrhagic colitis and hemolytic uremic syndrome. Journal of Infectious
Diseases 157:1124—1133.

Whittam, T. S., and R. A. Wilson. 1988. Genetic relationships among
pathogenic Escherichia coli of serogroup 0157. Infection and humanity
56:2467-2473.

Whittam, T. S., M. L. Wolfe, I. K. Wachsmuth, F. Orskov, I. Orskov, and
R. A. Wilson. 1993. Clonal relationships among Escherichia coli strains that
cause hemorrhagic colitis and infantile diarrhea. Infect Irnmun 61 : 1619-29.

Whittam, T. S., M. L. Wolfe, and R. A. Wilson. 1989. Genetic relationships

among Escherichia coli isolates causing urinary tract infections in humans and
animals. Epidemiol Infect 102:37-46.

145

181. Winfield, M. D., and E. A. Groisman. 2003. Role of nonhost environments
in the lifestyles of Salmonella and Escherichia coli. Appl Environ Microbiol
69:3687-94.

182. Wirth, T., D. Falush, R. Lan, F. Colles, P. Mensa, L. H. Wieler, H. Karch,
P. R. Reeves, M. C. Maiden, H. Ochman, and M. Achtman. 2006. Sex and

virulence in Escherichia coli: an evolutionary perspective. Mol Microbiol
60: 1 136-5 1 .

146

 

i"(lijigiilgijgElli