Long-term molecular epidemiology of extended-spectrum β-lactamase-producing Escherichia coli in a low-endemic

(1)

Long-term molecular epidemiology of extended-spectrum β-lactamase-producing Escherichia coli in a low-endemic setting

(2)

To my family

(3)

Örebro Studies in Medicine 207

A

NNA

F

AGERSTRÖM

Long-term molecular epidemiology of extended-spectrum β-lactamase-producing Escherichia coli in a low-endemic

setting

(4)

Title: Long-term molecular epidemiology of extended-spectrum β-lactamase-producing Escherichia coli in a low-endemic setting.

Publisher: Örebro University 2020 www.oru.se/publikationer-avhandlingar

Print: Örebro University, Repro 02/2020 ISSN1652-4063

ISBN978-91-7529-324-0

(5)

Abstract

Anna Fagerström (2020): Long-term molecular epidemiology of extended- spectrum β-lactamase-producing Escherichia coli in a low-endemic setting.

Örebro Studies in Medicine 207.

Escherichia coli is a commensal inhabitant in the gastro-intestinal tract of humans and animals but it is also the most common bacterial species causing urinary tract infection, which ranges in severity from distal cystitis to urosepsis and septic shock. During the past decades, the prevalence of antibiotic resistant E. coli has increased worldwide. Extended-spectrum β-lactamases (ESBL) causes resistance to β-lactam antibiotics, the most widely used class of antibiotics. The genes encoding ESBL, bla, are usually carried on conjuga- tive plasmids, which can be transferred between different bacterial lineages and different species. These plasmids frequently also carry resistance genes to additional antibiotic classes, and ESBL-producing E. coli are therefore of- ten multidrug-resistant. The aim of this thesis was to describe the long-term molecular epidemiology of ESBL-producing E. coli in Örebro County during the time when they first started to emerge. In addition, potential transmission to the environment was investigated by performing a comparative analysis on ESBL-producing E. coli isolated from patients and from the aquatic envi- ronment in Örebro city. In general, the E. coli population was genetically diverse, but the pandemic lineage ST131, first identified in 2004, appears to have been responsible for the dramatic increase of CTX-M-15-producing E.

coli observed during the late 2000s. CTX-M-15 was the most prevalent ESBL-type followed by CTX-M-14 and these genes were mainly found on plasmids belonging to the IncF or IncI1 families. Continuous horizontal transmission of IncI1 ST31 and ST37 plasmids between diverse E. coli line- ages have also contributed to the dissemination of blaCTX-M-15 in Örebro County. Extended spectrum β-lactamase-producing E. coli were found to be common in the aquatic environment in Örebro city and E. coli lineages ge- netically similar to those causing infections in humans were present in envi- ronmental waters indicating that transmission of ESBL-producing E. coli from humans to the aquatic environment likely has occurred.

Keywords: Escherichia coli, extended-spectrum β-lactamase, whole genome sequencing, plasmids, hybrid assembly, environment, IncI1, ST131

Anna Fagerström, School of Medical Sciences, Örebro University, SE-70182 Örebro, Sweden, anna.fagerstrom@regionorebrolan.se

(6)

(7)

Sammanfattning

Escherichia coli är en bakterieart som återfinns i normalfloran i tarmen hos människor och djur men är även den vanligaste orsaken till urinvägsinfekt- ion, vilket innefattar allt från enkel cystit till sepsis och septisk chock. Under de senaste årtiondena har en ökning av antibiotikaresistenta E. coli skett över hela världen. Extended-spectrum β-lactamase (ESBL) är enzymer som orsakar resistents mot betalaktam antibiotika, en av de antibiotikaklasser som används mest. Generna som kodar för dessa enzymer, bla, är oftast lokaliserade på plasmider som kan överföras mellan olika bakteriestammar och olika bakteriearter. Det är vanligt att samma plasmid bär på ett flertal olika resistensgener som orsakar resistens mot ytterligare antibiotikaklasser vilket innebär att ESBL-producerande E. coli ofta är multiresistenta. Syftet med den här avhandlingen var att beskriva epidemiologin avseende ESBL- producerande E. coli i Örebro län under den tidsperiod när de först började dyka upp. Dessutom samlades prover in från vattenmiljön i Örebro för att undersöka om spridning av ESBL-producerande E. coli från människor till miljön kan ha förekommit. Generellt var det en hög genetisk diversitet i E.

coli populationen som studerades. Dock kunde den pandemiska E. coli klo- nen ST131 identifieras från 2004 och framåt och förefaller ha varit en viktig orsak till den dramatiska ökningen av CTX-M-15 producerande E. coli un- der slutet av 2000-talet.

CTX-M-15 var den vanligast förekommande ESBL typen följt av CTX- M-14 och generna var framförallt lokaliserade på plasmider tillhörande grupperna IncF eller IncI1. Kontinuerlig horisontell överföring av IncI1 plasmider av typerna ST31 samt ST37 mellan diverse E. coli stammar tycks också ha bidragit till utbredningen av blaCTX-M-15 i Örebro län. ESBL-produ- cerande E. coli var vanligt förekommande i vattenmiljön i Örebro. Bakteri- estammar som var genetiskt lika de som isolerats från patienter återfanns även i vattnet vilket tyder på att spridning av ESBL-producerande E. coli från människor till vattenmiljön sannolikt har skett.

(8)

(9)

List of original papers

This thesis is based on the following original Papers and manuscripts, referred to in the text by their Roman numerals:

I. Önnberg A, Mölling P, Zimmermann J, Söderquist B. Molecular and phenotypic characterization of Escherichia coli and Klebsiella pneumoniae producing extended-spectrum β-lactamases with focus on CTX-M in a low-endemic area in Sweden. APMIS.

2011;119(4-5):287-295.

II. Önnberg A, Söderquist B, Persson K, Mölling P. Characterization of CTX-M-producing Escherichia coli by repetitive sequenced- based PCR and real-time PCR-based replicon typing of CTX-M- 15 plasmids. APMIS. 2014;122(11):1136-43.

III. Fagerström A, Aspelin O, Söderquist B, Sundqvist M, Mölling P.

Comparative analysis of blaCTX-M-15-IncI1 plasmids in clinical Escherichia coli isolated during a 5-year period in a low-endemic setting. (in manuscript)

IV. Fagerström A, Mölling P, Khan FA, Sundqvist M, Jass J, Söder- quist B. Comparative distribution of extended-spectrum beta-lac- tamase-producing Escherichia coli from urine infections and envi- ronmental waters. PloS one. 2019;14(11):e0224861

Papers I and II are reprinted with permission from John Wiley and Sons.

Paper IV is reprinted in accordance with the Creative Commons Attribution 4.0 (CC BY 4.0) International Public License.

(12)

(13)

Abbreviations

AMC amoxicillin and clavulanate

AmpC ampicillin class C

API analytical profile index

BLASTN nucleotide basic local alignment search tool

bp base pair

BSI blood stream infection

cgMLST core genome multi-locus sequence typing

CI confidence interval

DDST double-disc synergy testing

DL DiversiLab

DNA deoxyribonucleic acid

EAEC enteroaggregative E. coli

EARS-Net European Antimicrobial Resistance Surveillance Network EHEC enterohaemorrhagic E. coli

EPEC enteropathogenic E. coli ESBL extended-spectrum β-lactamase ETEC enterotoxigenic E. coli

EUCAST European Committee on Antimicrobial Susceptibility Test- ing

ExPEC extraintestinal pathogenic E. coli

ID identity

Inc incompatibility

IncI1 incompatibility group I1 IncF incompatibility group F IncY incompatibility group Y

IntI Integron integrase

IS insertion sequence

MALDI-TOF MS matrix-assisted laser desorption/ionization time-of-flight mass spectrometry

MGE mobile genetic element

MLST multi-locus sequence typing

MST minimum spanning tree

NCBI National Center for Biotechnology Information

NJ neighbour-joining

PacBio Pacific Biosciences

PBRT PCR-based replicontyping

PCR polymerase chain reaction

PFGE pulsed-field gel electrophoresis

(14)

PhP Phene Plate

pMLST plasmid multi-locus sequence typing rep-PCR repetitive sequence-based PCR

SGS sequencing-by-synthesis

SMRT single-molecule real-time SNP single-nucleotide polymorphism

SRGA Swedish Reference Group for Antibiotics

ST sequence type

tnp transposase

UPEC uropathogenic E. coli

UPGMA unweighted pair-group method with arithmetic mean UTI urinary tract infection

WHO World Health Organization

WWTP wastewater treatment plant

(15)

Introduction

Escherichia coli

Escherichia coli is a species of rod-shaped Gram-negative bacteria that be- longs to the family Enterobacteriaceae. It is facultative anaerobic, non- sporulating, and is found as a commensal inhabitant in the gastrointes-tinal tract. However, it is also one of the most important human pathogens. Esch- erichia coli is a diverse species and is usually divided into three main clinical subsets, or pathotypes: commensal, diarrheagenic and extraintestinal path- ogenic E. coli (ExPEC). The commensal strains colonize the colon of their hosts and does not normally cause disease. Diarrheagenic strains cause different gastrointestinal syndromes and are further classified into sub-patho- types including enterohemorrhagic E. coli (EHEC), enterotoxigenic E. coli (ETEC), enteropathogenic E. coli (EPEC) and enteroaggregative E. coli (EAEC). The ExPEC strains often colonize the human gut but, unlike the commensal strains, they are also able to cause disease if they reach normally sterile extraintestinal body sites (1).

This thesis focuses primarily on ExPEC, which is a common cause of urinary tract infection (UTI) and blood stream infection (BSI) and can, in rare cases, cause neonatal meningitis. Since ExPEC most commonly causes UTI it is often also referred to as “uropathogenic E. coli (UPEC)” (2).

Escherichia coli has a highly dynamic genome with a constant flux of insertions and deletions. The bacterium can easily acquire new genes through horizontal gene transfer. It has been shown that recombination is the most important evolutionary mechanism in E. coli and clearly outweighs the impact of site-level mutation (3, 4). Despite the frequent occurrence of recombination events, the population structure is predominantly clonal, al- lowing the delineation of major phylogenetic groups (4, 5). There are four main phylogenetic groups termed A, B1, B2 and D. Commensal E. coli is typically derived from group A or B1, and the diarrheagenic pathotypes usually derive from phylogenetic groups A, B1 or D. Extra intestinal path- ogenic E. coli originates predominantly from group B2 and, to a lesser ex- tent, group D (2).

The average E. coli genome contains around 5,000 genes, and approxi- mately two-thirds of these are found in all E. coli genomes and constitutes the core genome (4, 6). The size of the core genome varies depending on how many genomes that are included in the analysis, and it has been shown to decrease with an increasing number of genomes (7, 8) however, only up

(16)

to a certain point then it becomes relatively constant (6) (Figure 1). The number of core gene families was 3,188 when 2085 E. coli genomes were analysed in January 2015 (6). By contrast, the pan genome, which consti- tutes the total number of genes found in all E. coli genomes, increased the more genomes were added. Land et al. reported in 2015 that the pan ge- nome consisted of around 90,000 gene families and was continuing to grow even after adding more than 2,000 genomes (6) (Figure 1).

Figure 1. Core genome and pan genome of 2,085 Escherichia coli genomes. Number of genomes are shown on the x-axis and number of gene families on the y-axis. Core gene families were defined as gene families with at least one member in at least 95%

of genomes. Reprinted from Land et al. (6).

Klebsiella pneumoniae

Klebsiella pneumoniae is a Gram-negative, rod-shaped, non-motile member of the family Enterobacteriaceae, and, just like E. coli, is a commensal in- habitant of the gastrointestinal tract in healthy humans and animals.

Klebsiella pneumoniae is mainly associated with hospital-acquired infec- tions. It is an opportunistic pathogen, primarily infecting immunocompro- mised individuals with severe underlying diseases. The most common infec- tion caused by K. pneumoniae is UTI, but the bacterium can also cause pneumonia, BSI and wound infections (9). Although K. pneumoniae was not the main focus of this thesis, it was included in Paper I as it was the second most common species associated with production of extended-spec-

(17)

Urinary tract infection

Urinary tract infection is one of the most common bacterial infections in humans and it is also the reason for a considerable number of antibiotic prescriptions (10). It has been estimated that 40% of women and 12% of men experience at least one symptomatic UTI episode during their lifetime, and up to 48% of the affected women experience recurrent UTIs. Further- more, UTIs constitutes about 40% of all hospital-acquired infections (11).

Urinary tract infection may include a wide spectrum of syndromes, ranging from simple distal cystitis to urosepsis and septic shock (10). They are often classified as complicated or uncomplicated, where complicated UTI involve functional or structural abnormalities in the urogenital system, which in- creases the risk of treatment failure or serious complications (12). Another common classification is based on the setting where the infection was acquired and includes community-acquired UTI, healthcare-associated UTI, and community-onset healthcare-associated UTI (10). Escherichia coli is the most common bacterial species causing UTI, accounting for approximately 80% of uncomplicated UTIs, 95% of community-acquired UTIs and 50%

of healthcare-associated UTIs (13).

Urine culture

Urine samples for culturing are primarily collected from patients with complicated UTI, patients with recurrent UTIs, and patients that has experi- enced treatment failure. The samples are commonly cultured on blood agar and a differential growth media, to identify and quantify the causative agent (14). In addition, susceptibility to clinically relevant antibiotics is tested, us- ing a standardized disk diffusion method (15). Enterobacteriaceae showing resistance to cefotaxime or ceftazidime are further tested for ESBL-production, using a double disk synergy test (DDST) (16).

Antibiotic resistance

The increasing prevalence of antibiotic resistant bacteria worldwide has been identified by the World Health Organization (WHO) as one of the most important threats to the global human health (17). In many parts of the world the levels of resistance in common bacteria has reached alarming levels. Effective antibiotics are, in addition to treating infectious diseases, a prerequisite for surgical procedures and immunosuppressive treatments,

(18)

such as chemotherapy. Since there is also a lack of new antibiotics in the pipeline (18) it is feared that we are heading towards a post-antibiotic era where people may die from common infections and minor injuries. Today, it is estimated that 700,000 people die every year as a result of infections caused by antimicrobial resistant microorganisms, and it has been further estimated that by 2050, as many as 10,000,000 people will die each year unless this situation is addressed (19).

Antibiotic resistance is an ancient natural phenomenon (20). Resistance genes have been found both in ancient environmental and human samples.

For instance, genes encoding resistance to tetracyclines, vancomycin and β- lactams were recovered from permafrost sediments in Canada that dated back 30,000 years (21). In another study, several pathogens possessing resistance genes against a number of antibiotic classes were identified in dental calculi from human skeletons dating back to c. 950-1200 CE (22). Antibi- otic resistance genes are believed to have evolved in antibiotic-producing microorganisms as a mechanism to protect themselves against the action of their own antimicrobial molecules (23, 24). The resistance genes have subsequently been enriched and extensively mobilized through the widespread use of antibiotics by humans (25).

Previously susceptible bacteria can acquire resistance to antibiotics through either chromosomal mutation or horizontal gene transfer (HGT) (26). Horizontal gene transfer is mediated by conjugative plasmids, which frequently carry multiple resistance genes; hence, a bacterium may become multi-resistant by simply acquiring one such plasmid.

Resistance in Escherichia coli

Escherichia coli is intrinsically susceptible to most clinically relevant antibi- otics. However, during the past few decades, the species has acquired an increasing number of resistance genes and today antibiotic resistance in E.

coli is considered one of the major global challenges in both humans and animals. An annual national surveillance programme showed continuously increasing levels of resistance to ampicillin, trimethoprim and fluoroquin- olones among E. coli causing UTI during the years 1996-2007 in Sweden (27) (Figure 2) but since 2009 the levels have remained relatively stable (28) (Figure 3). During the 2000s, resistance to third-generation cephalosporins became more and more common among E. coli, mainly owing to produc- tion of ESBLs. According to data from the European Antimicrobial Re-

(19)

to 8.3% in Sweden from 2002 to 2018, and in Italy during the same period, it increased with 25.8% (Figure 4). Resistance to carbapenems and poly- myxins (e.g. colistin) is also increasing worldwide (29, 30), which is particularly troublesome since these antibiotics are considered the last line of de- fence against many bacteria in case of serious infections.

(20)

Figure 2. Resistance levels to common antibiotics among Escherichia coli causing urinary tract infection in Sweden in 1996-2007. Reprinted from Swedres/Svarm 2007 (27).

Figure 3. Resistance levels to common antibiotics among Escherichia coli causing urinary tract infection in Sweden in 2009-2018. Reprinted from Swedres/Svarm 2018 (28).

(21)

Figure 4. Prevalence of invasive Escherichia coli isolates resistant to third-generation cephalosporins in Europe in (A) 2002 and (B) 2018, based on data from the Euro- pean Antimicrobial Resistance Surveillance Network (EARS-Net). Source: Surveil- lance Atlas of Infectious Diseases, European Centre for Disease Prevention and Control (ECDC) (https://atlas.ecdc.europa.eu/public/index.aspx).

(22)

Beta-lactam antibiotics

Beta-lactam antibiotics are the most widely used class of antibiotics (31).

They are bactericidal agents that act by inhibiting cell-wall formation in the bacterial cell as a result of covalent binding to penicillin-binding proteins in the bacteria (31). The core structure of the antibiotics in this class is centred around a structure known as a β-lactam ring (Figure 5).

Figure 5. Core chemical structure of penicillins. The β-lactam ring is marked in red.

Modified and reprinted from Lee et al. (32).

The β-lactams can be divided into four subclasses: penicillins, cephalosporins, carbapenems and monobactams (Figure 6). The first β-lactam antibiotic to be used clinically was benzylpenicillin, or penicillin G, which was approved in 1946 and was mainly used to treat streptococcal infections (31). The first cephalosporin was introduced in 1964 and since then the basic structure have been modified to improve pharmacokinetics and over- come bacterial resistance. The cephalosporins are grouped into “genera- tions” based on their general antimicrobial activity (33, 34) (Figure 6). The third-generation cephalosporins are broad-spectrum antibiotics that are useful in various clinical situations. Most of them penetrate cerebrospinal fluid and can therefore be used for treatment of bacterial meningitis (35).

(23)

Figure 6. The main subclasses of the β-lactam antibiotics and the different genera- tions of cephalosporins. Examples of antibiotics within each class are also listed.

Beta-lactamase

The most important factor contributing to resistance against β-lactam antibiotics in Gram-negative bacteria is the production of β-lactamases (36).

Beta-lactamases are bacterial enzymes that inactivate β-lactam antibiotics by hydrolysing the β-lactam ring, resulting in ineffective compounds (37).

The first β-lactamase, which had the ability to hydrolyse penicillin, was described in E. coli before penicillin had even been released for use in med- ical practice (37, 38). Since then, more than 2,000 different β-lactamases have been detected that differ from each other in their sequence homology, substrate profiles and inhibitor profiles (39).

Two classification systems exists for β-lactamases: Ambler’s molecular system, which divides the enzymes into the groups A, B, C and D based on amino acid sequence homology (40), and the Bush-Jacoby-Medeiros functional classification system that divides them into the groups 1, 2 and 3, based on substrate and inhibitor profiles (33, 41).

(24)

Extended-spectrum β-lactamases

Extended-spectrum β-lactamases are enzymes that are able to hydrolyse various types of β-lactam antibiotics, including the oxyimino-cephalosporins (also known as the third- and fourth-generation cephalosporins) and the monobactams, but not cephamycins or carbapenems (42). The ESBLs are inhibited by β-lactamase inhibitors such as clavulanic acid, tazobactam, sulbactam and avibactam (43, 44). The majority of ESBLs belong to class A in Ambler’s classification system (40) and group 2be in the Bush-Jacoby- Medeiros system (41) (Figure 7). The most common ESBLs are TEM, SHV and CTX-M. More than 200 different TEM variants and at least 193 SHV variants have been described (45) and they have evolved through point mu- tations from the ancestral β-lactamases TEM-1 and SHV-1 (37, 46). Ex- tended-spectrum β-lactamases of the TEM- and SHV-type are most com- monly found in E. coli and K. pneumoniae but they have also been found in other species within Enterobacteriaceae (37).

(25)

Proliferation of the ESBLs has been driven by selective pressure due to use and overuse of β-lactam antibiotics. The oxyimino-cephalosporins in the 1980s became widely used against infections caused by Gram-negative bacteria (37). In 1985 the first ESBL, SHV-2, was reported in a single strain of Klebsiella ozaenae isolated in Germany (46) and ever since then the number of ESBLs has been increasing rapidly. In the 1990s TEM and SHV were the dominant ESBL types but in the first decade of the 2000s enzymes of the CTX-M family started to increase and soon became the dominant ESBLs (48).

CTX-M

The first CTX-M ESBL was discovered in an E. coli strain in Munich, Ger- many in 1989 (49) and since then, more than 220 CTX-Ms have been described (43). The genes encoding these enzymes have been found in at least 26 bacterial species but are particularly prevalent in E. coli, K. pneumoniae and Proteus mirabilis (50). Studies on CTX-M phylogeny have shown that they can be differentiated into at least five main lineages or groups: CTX- M-1, CTX-M-2, CTX-M-8, CTX-M-9 and CTX-M-25 (Figure 8). Each group differs from the others by ≥10% amino acid residues and the enzymes within each group differ from each other by ≤5% amino acid residues (51- 53).

Figure 8. Tree showing the main CTX-M lineages. Reprinted from D’Andrea et al (53) with permission from Elsevier.

(26)

The genes encoding CTX-M, blaCTX-M, are believed to have originated from different species of Kluyvera, a member of the family Enterobacteriaceae that seldom causes human infections but is prevalent in the environment.

Close homologues of acquired blaCTX-M genes have been detected in the chro- mosome of strains of Kluyvera ascorbata, K. cryocrescens and K. georgiana.

It is likely that each CTX-M group has been derived from one or more dif- ferent Kluyvera strains (53).

The CTX-M variant CTX-M-15, which belongs to group CTX-M-1, is the most common variant found in E. coli today. It was first discovered in India in 2001 (54), and since then it has been detected worldwide and is the predominant ESBL in most countries (55).

Carbapenemases

The clinical use of carbapenem antibiotics has increased over the past decade because of the increasing prevalence of ESBL-producing bacteria. This has in turn resulted in increasing numbers of bacterial isolates carrying β- lactamases capable of hydrolysing carbapanems, called carbapenemases (56). These enzymes are biochemically diverse and are found in class A, B and D in the Ambler classification system (Ambler 1980). One of the most prevalent groups of carbapenemases globally are the KPC enzymes (57), which belongs to class A, and are found within group 2f in the functional classification system (47) (Figure 7). In Sweden, OXA-48 was the most prev- alent carbapenemase in Enterobacteriaceae in 2018 (28).

Dissemination of extended-spectrum β-lactamases

The global dissemination of ESBL-producing E. coli is considered to have been caused by the spread of a few successful “high-risk” E. coli clones in combination with horizontal transmission of ESBL genes between diverse lineages (43).

Clonal spread

A clone can be defined as a group of bacterial isolates that, when character- ized by biochemical or molecular methods, share similar traits, which indi- cates that they belong to the same lineage or cluster and have a common ancestor (43). In this thesis, the terms “clone” and “lineage” have been used interchangeably to describe strains believed to have a common origin.

(27)

Bacterial typing

There is a variety of different methods available to investigate potential clonal relationship between bacterial isolates. Some are based on bacterial phenotype, such as the commercially available PhenePlate (PhP) system.

This system quantitively measures the reaction products formed by the bacteria’s metabolism for a number of different substrates, rendering bio-chemical fingerprints, which can then be compared. The method is based on the assumption that isolates belonging to the same clone have identical genotypes and therefore also share identical metabolic properties, while isolates with different genotypes will also show differences in one or more of the measured metabolic processes (http://www.phplate.se/?page_id=43).

During the past decades the scientific basis for identification and subtyp- ing of bacteria has shifted more and more to genetic methods. Pulsed-field gel-electrophoresis (PFGE) is a genotypic method that has long been considered the gold standard for bacterial strain typing. This method uses re- striction enzymes to cut the bacterial DNA at specific sites and the resulting DNA fragments are subsequently separated using electrophoresis where electric fields are applied at different angles. DNA polymorphism at the re- striction enzyme recognition sites will result in different banding-patterns (58).

Another genetic method used for bacterial typing is repetitive sequence- based PCR (rep-PCR). All bacteria have naturally occurring repetitive sequences dispersed in multiple copies throughout their genome, which can be used for DNA fingerprinting. In rep-PCR, primers complementary to the repetitive sequences are used to amplify the DNA fragments in between the repetitive sequences. The DNA fragments are separated by electrophoresis, creating a fingerprint for each bacterial isolate (58).

A method commonly used for investigating long-term epidemiology, which also allows comparisons on a global scale, is multi-locus sequence typing (MLST). This is a sequence-based method where fragments (c. 450- 500 bp) of a number of different house-keeping genes are sequenced (58).

For E. coli there are two MLST schemes in use. The Achtman (or Warwick) scheme (http://enterobase.warwick.ac.uk/species/ecoli/allele-st-search), which includes seven loci and the Pasteur scheme (https://bigsdb.pasteur.fr/ecoli/ecoli.html), where eight loci are included. Each unique sequence for each locus included in the scheme is assigned an arbitrary and unique allele number. When combining the allele numbers for each locus, an allelic profile is obtained, which is assigned a sequence type (ST) designation (59).

(28)

Today, most bacterial typing methods have been replaced by whole genome sequencing (WGS), which allows comparisons based on the entire bacterial genomes.

Whole genome sequencing is performed on massively parallel sequencing, or next generation sequencing, platforms. Several commercially available systems are available which utilize different sequencing techniques. The Il- lumina platforms uses a “sequencing by synthesis” (SGS) technique where DNA fragments are attached to the surface of a flow-cell and subsequently amplified by bridge-amplification. The fragments are sequenced using reversible fluorophore-labelled nucleotides, which are optically read from the flow-cell. The generated reads have high accuracy but are relatively short, up to 300 bp, which can be problematic for genomes with large repeats (60).

Single-molecule real-time (SMART) sequencing, developed by Pacific Bio- Sciences (PacBio) offers much longer read lengths and faster sequencing runs than the SGS methods, but has the drawbacks of lower yield, higher error rate, and higher cost per base (61).

Most traditional sequence-based methods, including MLST, can be per- formed in silico on WGS data, and, in addition, bioinformatics tools have been developed making it possible to determine E. coli serotype and fimH- type from the WGS data (62). Serotyping is a method for classification of E. coli that has been in use since the 1940s (63), which is based on deter- mining the combination of the immunogenic structures, lipopolysaccharide (O antigen) and the flagellar (H) antigen (62). Typing of the E. coli fimH gene has become a commonly used method for sub-typing of specific E. coli lineages (64, 65). FimH is an adhesin located at the tip of the type 1 fim- briae, involved in the binding of the bacterial target cells (64).

High-risk clones

Several international “high-risk” clones have been identified, which are defined as clones with a global distribution showing enhanced ability to colonize, spread, and persist in a variety of niches (66, 67). It is believed that these high-risk clones may possess biological factors which provide them with increased “fitness”, giving them a competitive advantage over other bacterial lineages that reside in the same niche (67, 68). High-risk clones have contributed to the spread of different plasmids and resistance genes among Gram-negative bacteria and play an important role in the global spread of antibiotic resistance (67, 69). Several international multi-resistant

(29)

high-risk clones have been described among Enterobacteriaceae, for exam- ple E. coli ST131, ST38 and ST155, and K. pneumoniae ST14, ST37 and ST258 (66, 67).

Escherichia coli sequence type (ST)131

Escherichia coli ST131 is the most important lineage involved in the global dissemination of ESBLs. It usually displays multi-resistance and is associated with CTX-M-15. This lineage was first identified in 2008 in three different continents (70, 71), and today it is found worldwide (72-74) (Figure 9). Retrospective studies of international ST131 isolates showed that isolates from the end of the 20th century were susceptible to antibiotics. Be- tween 2000 and 2005, fluoroquinolone-resistant ST131 isolates first emerged, and subsequently, fluoroquinolone resistant ST131 isolates carry- ing blaCTX-M-15 appeared (75).

Foreign travel, particularly to the Indian subcontinent, has been impli- cated as a potential major player in the initial global spread of ST131 car- rying blaCTX-M-15 (76, 77).

Figure 9. Global dissemination of Escherichia coli ST131 in 2013. The red stars indicate extended-spectrum β-lactamase (ESBL)-producing isolates. The blue stars indicate isolates that are fluoroquinolone-resistant but not ESBL-producing. Re- printed from Nicolas-Chanoine et al. (74).

(30)

Phylogenetic analyses of ST131 have shown that it belongs to phylogenetic group B2 and consists of different clades, or lineages (78, 79) (Figure 10).

Clade A and B most likely diverged from a single E. coli progenitor at some point before the year 2000 (79), while recent studies have shown that clade C have evolved from clade B, and that this might have occurred in North America already during the late 1980s (80-82). Isolates belonging to clade A are usually associated with serotype O16:H5 and variant 41 of the type I fimbrial adhesin gene, fimH. Isolates belonging to clade B and C are sero- type O25b:H4 and are associated with fimH22 and fimH30, respectively (Figure 9). Clade C, which is also called H30, is strongly connected to fluoroquinolone resistance and has further diverged into two sub-lineages, C1 (H30R) and C2 (H30Rx). Clonal expansion of sub-lineage C2/H30Rx is considered to be the most important reason for the increasing prevalence of E. coli carrying blaCTX-M-15 globally (67).

Figure 10. Schematic illustration of the lineages identified within E. coli ST131.

Several PCR assays have been developed for rapid detection of ST131. An example is the PCR described by Clermont et al. which is based on ST131 specific SNPs in the pabB gene from the Pasteur MLST scheme (83).

(31)

Horizontal gene transfer

Horizontal gene transfer is the process whereby DNA from one bacterial cell is physically transferred to another bacterial cell and then incorporated into the recipient’s genome (25). This is in contrast to vertical gene transfer where DNA is transferred from a mother to a daughter cell during cell divi- sion. Many of the antibiotic resistance genes found in E. coli today were acquired through horizontal gene transfer (84). Horizontal gene transfer is mediated by mobile genetic elements (MGEs). These are DNA segments encoding enzymes and other proteins that mediate the movement of DNA within genomes (intracellular mobility) or between bacterial cells (intercellular mobility) (85). There are three main mechanisms, by which intercellular movement of DNA occurs between bacteria, known as transformation, transduction, and conjugation (Figure 11).

Figure 11. The main mechanisms involved in horizontal gene transfer: transfor- mation, transduction and conjugation. Reprinted from Bello-López et al. (86).

(32)

Transformation

When a bacterial cell dies, it deteriorates and its DNA is degraded and released into the surrounding environment. This naked DNA can then be picked up by another bacterial cell and be incorporated into its genome, a process called “transformation” (23) (Figure 11). Bacterial species that are able to acquire genetic material through natural transformation are referred to as “naturally competent”. Escherichia coli is generally not considered to be naturally competent but recent studies have shown that it can express modest natural competence under feasible environmental conditions (87).

Transduction

Transduction is a mechanism whereby genetic exchange is mediated by bacterial viruses, called “bacteriophages”. Bacteriophages can accidentally package segments of host DNA in their capsid and then inject this DNA into a new host where it can recombine with the chromosome (85) (Figure 11).

Conjugation

Conjugation involves cell-to-cell contact and is a process whereby DNA, usually in the form of a plasmid, is moved from one bacterial cell to another via a hair-like transfer appendage called a pilus, which forms a bridge between the two bacterial cells (23) (Figure 11).

Mobile genetic elements

Plasmids

Plasmids are the most important MGEs responsible for intercellular movement of DNA. They act as vehicles for other MGE, and the genes associated with these elements, such as antibiotic resistance genes (88). Plasmids are extra-chromosomal circular DNA molecules that replicate independently of the host genome (43). The genes that are located on plasmids are not essential for the hosts survival but they often encode traits that gives the host a selective advantage, such as virulence determinants and antibiotic resistance genes (23). The plasmid genome consists of a backbone of “selfish” modules encoding functions that are necessary for replication and copy number control, stability and conjugation. In addition to these core genes modules, they carry accessory regions that contain the resistance genes or virulence genes

(33)

row or broad host range plasmids. Narrow host range plasmids are re- stricted to a limited number of similar bacterial species, while plasmids with broad host range can be transferred between widely different species (89).

Plasmids with the same replication mechanism cannot stably co-exist within the same bacterial cell, a phenomenon called “incompatibility (Inc)”, which forms the basis for plasmid classification (90, 91). Incompatible plasmids are likely related and are classified in the same Inc goup (88). Cur- rently, 28 Inc groups have been described in Enterobacteriaceae, of which incompatibility group F (IncF), which is mainly found in E. coli, is the most frequently described from both human and animal sources (92). The sudden increase of E. coli carrying blaCTX-M-15 in the mid-2000s is believed to be mainly due to the acquisition of specific IncF plasmids harbouring this gene by the high-risk clone ST131 (67).

Plasmid typing

Since the spread of antibiotic resistance is driven, in part, by dissemination of resistance plasmids it is important to include plasmid analysis when investigating suspected outbreaks of antibiotic resistant bacteria. Replicon typing is the most commonly used method for plasmid classification, target- ing genes encoding functions involved in plasmid replication, which are also the same features that determine which Inc-group a plasmid belongs to. A PCR-based replicon typing (PBRT) scheme was developed by Carattoli et al., identifying the replicons of the major plasmid families occurring in En- terobacteriaceae (93), which has been widely used. In addition, plasmid MLST schemes have been developed for the most frequently identified Inc groups (94). Replicon typing and pMLST can be used to rule-out plasmid transmission between bacterial isolates but, unfortunately, none of them provide sufficient resolution to confirm that plasmid transmission has occurred (95). The most accurate method to characterize a plasmid is to determine the complete DNA sequence by whole genome sequencing (96).

Before plasmid typing can be performed, the plasmid of interest must be isolated. This can be achieved by transferring the plasmid into a plasmid- free E. coli strain, via electroporation or conjugation. Another method is to use S1 nuclease, an enzyme that cuts single stranded DNA, to linearize the plasmids, and, subsequently, separate the linearized plasmids by PFGE (97).

Since most E. coli carry several plasmids, it can be challenging and time- consuming to isolate the plasmid of interest, and therefore, today, much effort is put into developing bioinformatics tools for extracting plasmid sequences directly from WGS data.

(34)

Transposable elements

In addition to plasmids, several other MGEs are involved in the mobilization of antibiotic resistance genes. Transposable elements are DNA segments that can move almost randomly from one location to another on a plasmid or a chromosome, or between the plasmid and the chromosome, or even between two different plasmids (88). The simplest transposable elements are the insertion sequence (IS) elements. They usually contain a single open reading frame and a transposase (tnp) gene, encoding the enzyme re- sponsible for transposition, i.e. the movement of the DNA segment from one location to another. When two copies of the same, or related, IS elements capture a resistance gene in between, they form a composite trans- poson, which can move as a single unit (88). An IS element that has fre- quently been described associated with blaCTX-M is ISEcp1 (55).

Another type of transposable elements are the unit transposons, which have traditionally been thought of as elements that are larger than the IS elements, and are bounded by inverted repeats instead of two IS (88).

Integrons

Integrons are another type of MGE that use site-specific recombination to move resistance genes between defined sites (88). They are capable of capturing individual genes that are part of mobile gene cassettes. These are free circular DNA molecules, which are not expressed by themselves since they lack a promoter (23, 25). The gene cassettes have a recombination site called attC. Integrons have three essential features, responsible for capturing and subsequently expressing the captured genes. First, they all have the gene intI, which encodes integron integrase (intI1). Secondly, they have an integron- associated recombination site, attI, and, finally, they all have an integron- associated promoter, Pc. The gene cassettes are integrated by site-specific recombination between attI and attC, which is mediated by IntI1. This is a reversible process, and the gene cassettes can be excised as free circular DNA elements again (98). Several different classes of integrons have been defined based on the sequence of IntI (88). It is mainly the class 1 integrons that are associated with the acquisition and mobilization of antibiotic resistance genes in Gram-negative bacteria (25).

Epidemiology

Ever since they first emerged, the prevalence of ESBL-producing Enterobac-

(35)

ESBL type CTX-M-15 became widespread and was often found in E. coli causing community-acquired infections (71).

The highest prevalence of ESBL-producing bacteria is observed in India where, already in 2007, up to 80% of E. coli and 70% of K. pneumoniae were ESBL-producers (99). A high prevalence have been observed also in China where up to 70% of E. coli and 60% of K. pneumoniae are producing ESBL (100). A slightly lower prevalence has been reported in North America where in U.S. hospitals, 18.3% of E. coli urine isolates in 2014 were ESBL- producers, and the majority produced CTX-M-15 (101). In Canada, the prevalence of ESBL-producing E. coli and K. pneumoniae was less than 10% in 2007-2011, but it appeared to be increasing (102). Figure 3B shows the prevalence of invasive E. coli resistant to third generation cephalospor- ins in Europe in 2018, which is mainly due to ESBL-production. The prevalence varies significantly between different regions. In the Nordic countries it is still relatively low (just below 10%), while some countries in the south- ern and eastern part of Europe have a prevalence above 25%.

Faecal carriage of ESBL-producing E. coli in healthy humans in the com- munity is common. Before 2008, the community-carriage rate was less than 10% in most parts of the world, but since then it has increased everywhere.

In South East Asia and the Eastern Mediterranean (WHO areas) the carriage rate is over 60% (103). In Sweden, 4.7-6.6% of the population may be car- riers of ESBL-producing E. coli (104, 105).

Extended-spectrum β-lactamase-producing Escherichia coli in non-human sources

The term “One Health” was coined in 2004 (106) but is essentially a re- conceptualization of historic ideas, where the foundation may be dated all the way back to Hippocrates (460-367 BCE) and Aristotle (384-322 BCE) (107). The concept of One Health recognizes that the health of humans, animals and the environment are tightly connected and stipulates that the responsibility for this health needs to be shared between multiple disciplines and sectors (106), which is particularly evident in the case of antibiotic resistance. ESBL-producing bacteria have been extensively reported both in animal and environmental sources (108-115), and transmission of ESBL- producing E. coli have been shown between humans and animals (116), and from animals to the environment (117, 118). Large amounts of ESBL-producing bacteria are released into the environment from wastewater treatment plants (WWTP) (119, 120), which most likely originate from human sources. Presence of E. coli ST131 with bla have been shown in both

(36)

WWTP effluents and other environmental waters in several countries (113, 121-123). A recent population-based modelling study from the Netherlands estimated that approximately two-thirds of ESBL-producing E. coli carriers in the community were attributable to human-to-human transmission, while non-human sources, such as food, animals and environment, ac- counted for the remaining one-third (124). This clearly shows that a One Health approach is necessary to tackle the growing global challenge of an- tibiotic resistance. In the Nordic countries, ESBL-producing E. coli have been reported in wild birds (125), in environmental water (113, 126), in food and in farm animals (127), but it is still not clear whether transmission of these bacteria and/or the ESBL-genes has occurred between these different niches.

(37)

Aims

The general aim of this thesis was to investigate the long-term molecular epidemiology of ESBL-producing E. coli in Region Örebro County, which is an area with low prevalence of antibiotic resistance.

The specific aims of the studies were:

I. To investigate the epidemiology of ESBL-producing E. coli and K.

pneumoniae in 1999-2008 using a phenotypic method for bio- chemical fingerprinting and by sequencing the bla genes.

II. To further characterize the CTX-M-producing E. coli isolated in 1999-2008 using a genotypic method, to estimate the prevalence of the high-risk clone ST131, and, in addition, to characterize the plasmids carrying blaCTX-M-15, by replicon typing.

III. To investigate if the high proportion of IncI1 plasmids carrying blaCTX-M-15 was associated with dissemination of a single specific- IncI1 plasmid, by extracting the complete plasmid sequences from whole genome sequence data.

IV. To perform a comparative analysis of ESBL-producing E. coli iso- lated from urine samples and from the aquatic environment in Örebro city in 2013, using whole genome sequencing.

(38)

Materials and methods

Setting

The bacteria studied in this thesis were isolated from patients in Örebro County and from the aquatic environment in the city of Örebro. The entire region has a population of approximately 304,500 inhabitants, approximately 150,000 of whom lives in Örebro city (www.scb.se/en accessed on 20 December 2019). The municipal WWTP in Örebro city processes 45,000 m³ of wastewater per day (www.orebro.se), from households, a university hospital, local industries, and a small airport. After treatment, effluent water from the WWTP is discharged into Svartån River, which flows through the urban area and terminates in Lake Hjälmaren. The majority of the ag- ricultural activities are outside of the urban area, and there are no pharma- ceutical industries in Örebro city. There are 16 primary care centres located in the city and one large tertiary hospital (Örebro University Hospital). The hospital wastewater is transported to the municipal WWTP without prior treatment.

Clinical isolates

All clinical isolates were collected at the Department of Laboratory Medi- cine, Clinical Microbiology, at Örebro University Hospital, which serves the entire county.

The bacteria in Papers I-III were isolated in 1999-2008, mainly from urine and blood. They were identiﬁed by routine diagnostic methods. Anti- biotic susceptibility testing, including screening with cefotaxime and ceftazidime, was performed by disc diffusion according to the recommen- dations by the Swedish Reference Group for Antibiotics (SRGA) (http://www.srga.org). The species identity was veriﬁed by API 20 E (bio- Mérieux, Marcy l’Etoile, France).

In Paper I both ESBL-producing E. coli (n = 171) and K. pneumoniae (n

= 29) were included. They were isolated from 183 hospitalized patients and outpatients from the whole county.

The E. coli isolates found to be positive for blaCTX-M in Paper I (n = 152) were subsequently included in Paper II. They were obtained from 142 patients.

In Paper III, 17 of the blaCTX-M positive E. coli isolates that had been found to carry bla on IncI1 plasmids were included.

(39)

The clinical isolates in Paper IV (n = 45) were collected in February-Sep- tember 2013 from routine urine samples from patients infected for the first time with ESBL-producing E. coli. The patients were either hospitalized or outpatients that had visited one of 16 primary care centres located in Öre- bro city. The bacteria were identified and isolated with routine diagnostic methods and species identification was performed with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) using Microflex LT and Biotyper 3.1 (Bruker Daltonik, Bremen, Germany).

Antibiotic susceptibility testing, including screening with cefotaxime and ceftazidime, was performed by disk diffusion according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST methodology) (http://www.eucast.org).

Environmental isolates

Water samples in Paper IV were collected from the WWTP, from Svartån River downstream of the wastewater effluent discharge, from Lake Hjälma- ren, and from several ponds in Oset, a migratory bird sanctuary close to, but not directly connected to, the lake (Figure 12). Sampling was carried out on two occasions: in June and in October 2013. Samples from the WWTP (from incoming wastewater and from effluent water after treatment) were collected at three separate times on each sampling occasion. For environmental water, three independent surface-water samples were collected on each occasion.

Extended-spectrum β-lactamase-producing E. coli were isolated from water samples by filtering each sample through polyethylene sulfonate membrane filters (Sartorius Stedim Biotech, Goettingen, Germany), which were subsequently placed on chromID ESBL agar plates (bioMérieux). After incubation for 18 hours at 37°C presumptive E. coli colonies were sub-cul- tured onto Chromocult Coliform Agar (Merck, Darmstadt, Germany) to confirm the species and obtain pure isolates. Isolates identified as E. coli were subsequently verified with MALDI-TOF MS (Bruker Daltonik).

(40)

All colonies from each sample identified as E. coli were included for fur- ther analysis (one to twelve per sampling site) at the sampling in June, while for the October sampling, all isolates identified as E. coli were included, except for samples with more than ten isolates, from which ten isolates were randomly selected for further analysis. In total, 82 isolates were included from water samples.

Figure 12. Schematic map of the study area in Paper IV. The red crosses mark the water sampling locations. Modified and reprinted from Fagerström et al. (128).

Extended-spectrum β-lactamase confirmation

Production of ESBL was cofirmed with double-disc synergy testing (DDST) on Iso Sensitest Agar (Oxoid Ltd, Basingstoke, UK) (Papers I-III) or Mueller-Hinton agar 3.8 w/v (BD Diagnostic Systems, Sparks, MD, USA) (Paper IV) using cefotaxime (5 µg), ceftazidime (10 µg), cefepime (30 µg) and amoxicillin and clavulanate (AMC) (30 µg) (Oxoid). Two agar plates were inoculated for each isolate and the cephalosporin discs were placed at a distance of 20 mm and 25 mm from the AMC disc on the two plates, respectively. A dilated inhibition zone between AMC and one or more of the cephalosporin discs was considered a positive result. A cefoxitin disc (10 µg) was included on one of the plates to identify AmpC producing isolates.

(41)

Biochemical fingerprinting

In Paper I, biochemical fingerprinting was performed on E. coli and K.

pneumoniae using the PhenePlate (PhP) system (PhPlate Stockholm AB, Stockholm, Sweden) according to the manufacturer’s instructions. The resulting fingerprints were compared pairwise for all isolates and the similarities were calculated as correlation coeﬃcients that were clustered using the unweighted pair group method with arithmetic mean (UPGMA) to produce a dendrogram. Isolates with identical or very similar biochemical fingerprints (correlation coeﬃcient ≥0.975) were assigned to the same PhP type (Möllby 1993). The PhP software (PhPlate Stockholm AB) was used for all data processing including optical readings, calculations, clustering and con- struction of dendrograms.

DNA extraction

In Paper I, DNA from all clinical isolates was extracted by boiling bacteria, suspended in 10 mM Tris-HCl, for 15 min. After centrifugation for 30 seconds at 12,000 × g the supernatant, containing DNA, was transferred to new tubes. The DNA from a strain used as a PCR control, was extracted using QIAamp DNA mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions.

In Papers II and IV, the MoBio UltraClean Microbial DNA Isolation Kit (MO BIO Laboratories Inc., Carlsbad, CA, USA) was used to extract DNA.

In Paper III, the Wizard Genomic DNA Purification Kit (Promega, Mad- ison, WI, USA) or the Qiagen Genomic-tip 500/G kit (Qiagen, Hilden, Ger- many) followed by the MoBio PowerClean Pro DNA Clean-Up Kit (MO BIO Laboratories) was used.

Real-time PCR

Real-time PCR was performed in Papers I and II on a Rotor-Gene Q (Qi- agen). All PCR amplifications were performed in a volume of 20 µl containing 1× Rotor-Gene SYBR Green PCR Master Mix (Qiagen), 2 µl tem- plate DNA, and primers added to a final concentration of 0.5 µM each.

After amplification, melting curve analysis was performed for specific am- plicon detection.

In Paper I, real-time PCR was used in a first step to screen for blaCTX-M, using primers CTX-M-F and CTX-M-R (Table 1). In blaCTX-M positive isolates, primer pairs CTX-M-1 grp, CTX-M-2 grp and CTX-M-9 grp were used to detect subgroups CTX-M-1, 2 and 9, respectively, and amplify the

(42)

entire gene for downstream sequencing. In addition, real-time PCR using primer pairs SHV F and R, and TEM F and R was performed on all isolates to detect and amplify the entire TEM gene and the major part of the SHV gene. The cycling conditions are listed in Table 2.

In Paper II, the pandemic E. coli clone ST131 was detected by real-time PCR using two ST131-specific pabB PCR assays (129) (Tables 1 and 2).

Repetitive sequence-based PCR

Repetitive sequence-based PCR (rep-PCR) was performed using the Diver- siLab (DL) Escherichia kit (bioMerieux) on a GeneAmp 9700 thermal cy- cler (Applied Biosystems, Warrington, UK) according to the manufacturer’s instructions. The resulting DNA fragments were separated by electrophoresis in microfluidics DNA LabChips on an Agilent 2100 Bioanalyzer (Agilent Technologies Inc., Paolo Alto, CA, USA) and the rep-PCR patterns were analysed with the DL software version 3.4 as recommended by the manufacturer. Pearson’s correlation coefficient was used to calculate pairwise similarities. Isolates showing >97% similarity with no peak diﬀerences in their rep-PCR patterns were considered indistinguishable and were assigned to the same DL type. DL types showing >95% similarity were considered similar.

Plasmid replicon typing

In paper II, plasmid replicon typing was performed on E. coli isolates that carried blaCTX-M-15 (n = 82). First whole genome DNA from the bacteria was prepared in agarose plugs and treated with S1 nuclease (Promega) to linearize plasmid DNA. The DNA fragments were separated by pulsed-field gel electrophoresis (PFGE) on a GenePath apparatus (Biorad Laboratories, Hercules, CA, USA). Plasmid DNA bands were cut from the gel and puriﬁed using the QIAquick Gel Extraction Kit (Qiagen). The plasmids were screened for the presence of blaCTX-M using real-time PCR (Tables 1 and 2).

PCR-based replicon typing (PBRT) was performed on plasmids that were positive for blaCTX-M by real-time PCR using primers described by Carattoli et al. (93) and cycling conditions listed in Table 2.

(43)

Table 1. Primers used in Papers I and II.

Primer Sequence (5’ to 3’) Product size (bp)

Reference

CTX-M-F ATGTGCAGYACCAGTAARGT- KATGGC

(130)

CTX-M-R ATCACKCGGRTCGCCGGRAT 300 (130),

modified

CTX-M-1 grp F CCAGAATAAGGAATCCCATG (131)

CTX-M-1 grp R GCCGTCTAAGGCGATAAAC 948 (131)

CTX-M-2 grp F ATGATGACTCAGAGCATTCG (132)

CTX-M-2 grp R TGGGTTACGATTTTCGCC 876 (132)

CTX-M-9 grp F ATGGTGACAAAGAGAGTGCA (132)

CTX-M-9 grp R CCCTTCGGCGATGATTCTC 876 (132)

SHV F ATGCGTTATATTCGCCTGT (133)

SHV R CGTTGCCAGTGCTCGATC 859 (133)

SHV R2 GGCGAGTAGTCCACCAGAT (134)

TEM F ATGAGTATTCAACATTTYCGT (135)

TEM R TTACCAATGCTTAATCAGTGA 861 (135),

modified

ST131TF GGTGCTCCAGCAGGTG (129)

ST131TR TGGGCGAATGTCTGC 40 (129)

ST131AF GGCAATCCAATATGACC (129)

ST131AR ACCTGGCGAAATTTTTCG 49 (129)

bp = base pairs; grp = group; K = G or T; R = A or G; Y = C or T.

(44)

Table 2. PCR-programmes used in Papers I and II.

Assay PCR-programme

CTX-M detection, CTX-M group 1

95°C 5 min, followed by 40 cycles of 95°C 5 s, 60°C 40 s

CTX-M group 2, CTX-M- group 9

95°C 5 min, followed by 40 cycles of 55°C 10 s, 60°C 40 s

TEM, SHV 95°C 5 min, followed by 35 cycles of 55°C 10 s, 60°C 40 s

ST131 detection 95°C 5 min, followed by 30 cycles of 95°C 5 s, 58°C 10 s

Real-time PBRT, replicon FII

95°C 5 min, followed by 35 cycles of 95°C 5 s, 52°C 10 s, 60°C 30 s

Real-time PBRT, replicons HI1, HI2, I1, X, L/M, N, FIA, FIB, W, Y, P, FIC, A/C, T, FIIs, K, B/O

95°C 5 min, followed by 35 cycles of 95°C 5 s and 60°C 30 s.

Cycle sequencing 96°C 10 min, followed by 25 cycles of 55°C 5 s, 60°C 4 min

PBRT = PCR-based replicon typing

Sanger sequencing

In Paper I, nucleotide sequencing was performed on the TEM, SHV and CTX-M genes. The PCR products from the real-time PCR assays described above were first purified using Multiscreen PCRµ96 Filter Plates (Millipore, Bedford, MA, USA). After vacuum filtration the products were recovered by adding 50 µl of 10 mM Tris-HCl and agitating the plates for 10 minutes.

The purified PCR products were cycle-sequenced using the same primers as in the PCR, except for SHV where one additional primer, SHV R2, was

(45)

used (Table 1). Sequencing was performed in 96-well plates and each reaction (10 µl) contained 2 µl of BigDye Terminator version 3.1 Cycle Sequenc- ing Kit (Applied Biosystems, Warrington, England, UK), 1.6 pmol primer and 1 µl of purified PCR product. Cycle sequencing was carried out on the GeneAmp PCR System 2700 (Applied Biosystems). The products were pre- cipitated by using 3 M sodium acetate, pH 4.6 and 95% ethanol with a final ethanol concentration of 65%, washed in 70% ethanol, and resuspended in 10 µl formamide (Applied Biosystems) before separation on ABI PRISM 3130XL Genetic Analyzer (Applied Biosystems).

ChromasPro version 1.5 was used to create consensus sequences that were subsequently aligned and translated into amino acid sequences using BioEdit sequence alignment editor version 7.0.9.0. The default nucleotide

basic local alignment search tool (BLASTN)

(http://www.ncbi.nlm.nih.gov/blast/Blast.cgi) was used to compare the sequences against previously sequenced ESBL genes in the National Center for Biotechnology Information (NCBI) database.

Pyrosequencing

Isolates that were untypable by the pabB SNP assays in Paper II (n = 6) were verified by pyrosequencing on a PSQ96 MA (Qiagen) following the manufacturer’s instructions. Biotin-labelled primers ST131TF and ST131AF were used (Table 1).

Whole genome sequencing

Illumina sequencing

In Papers III and IV, whole genome sequencing was performed using Illu- mina technology. The Nextera XT DNA library preparation kit and Nex- tera XT index kit (Illumina, San Diego, CA, USA) were used to create DNA libraries for sequencing, according to the manufacturer’s instructions. Se- quencing was performed on a MiSeq sequencer (Illumina) using the MiSeq reagent kit v3, 600-cycle (Illumina). Quality trimming of reads was performed using the FastQ toolkit version 2.0.0 on the Illumina BaseSpace server (BaseSpace Labs) with a Q threshold set to 20. De novo assembly of paired-end reads was performed using SPAdes Genome Assembler version 3.6.0 with default settings.

(46)

PacBio sequencing

DNA from the E. coli isolates in Paper III was also sent to the Norwegian Sequencing Centre (University of Oslo, Oslo, Norway) for PacBio sequencing. Libraries were prepared using the Pacific Biosciences 20 kb library preparation protocol (Pacific Biosciences, Menlo Park, CA, USA). Size selection of the final libraries was performed using BluePippin (Sage Science, Beverly, MA, USA). The libraries were sequenced on a Pacific Biosciences RS II in- strument using P6-C4 chemistry with 240-360 minutes movie time. For each library, one or two single-molecule real-time (SMRT) cells were used for sequencing.

Whole genome data analysis

The Ridom SeqSphere+ version 3.4.0 software (Ridom GmbH, Münster, Germany) was used for MLST and core genome MLST (cgMLST) in Papers III and IV.

In Paper III, the E. coli cgMLST scheme available in SeqSphere, contain- ing 2,513 targets and using the same loci and reference gene sequences as the EnteroBase Escherichia/Shigella cgMLST version 1 scheme (https://enterobase.warwick.ac.uk/species/index/ecoli), was used.

In Paper IV, the cgMLST Target Definer version 1.4 in SeqSphere was used to create a local ad hoc cgMLST scheme. This scheme was created by extracting open reading frames from the reference genome, CFT073 (NC_004431.1), and seven additional query genomes: NC_020163.1, NZ_CP009644.1, NC_009800.1, NC_022648.1, NZ_CP016007.1, NZ_CP015834.1 and NZ_CP009578.1, which resulted in 2,758 targets.

However, only the strict core genome was used in the subsequent analyses, i.e. only targets identified in all isolates were included: 1,957 when all isolates were included in the analysis, and 2,682 when only ST131 isolates were analysed.

The cgMLST results were visualized by constructing neighbour-joining (NJ) trees and minimum spanning trees (MST) based on the distance matrix that was built from pairwise comparisons of the cgMLST allele profiles of all isolates included in the analysis.

A number of web-based bioinformatics tools available at the Center for Genomic Epidemiology (http://www.genomicepidemiology.org/) were used in Paper IV. For ESBL genotyping and screening for other acquired antibiotic resistance genes, ResFinder version 2.1 was used with selected ID

Long-term molecular epidemiology of extended-spectrum β-lactamase-producing Escherichia coli in a low-endemic

Örebro Studies in Medicine 207

A

F

Long-term molecular epidemiology of extended-spectrum β-lactamase-producing Escherichia coli in a low-endemic

setting

Abstract

Sammanfattning

Table of Contents

List of original papers

Abbreviations

Introduction

Escherichia coli

Klebsiella pneumoniae

Urinary tract infection

Antibiotic resistance

Beta-lactam antibiotics

Beta-lactamase

Dissemination of extended-spectrum β-lactamases

Clonal spread

Horizontal gene transfer

Mobile genetic elements

Epidemiology

Extended-spectrum β-lactamase-producing Escherichia coli in non-human sources

Aims

Materials and methods

Setting

Clinical isolates

Environmental isolates

Extended-spectrum β-lactamase confirmation

Biochemical fingerprinting

DNA extraction

Real-time PCR

Repetitive sequence-based PCR

Plasmid replicon typing

Sanger sequencing

Pyrosequencing

Whole genome sequencing

Whole genome data analysis