1
De novo aerobic acquisition of
antimicrobial resistance in oxyR deficient
Escherichia coli bacteria
Bachelor’s Thesis
Tine Visser
11925434
01-07-2020
University of Amsterdam
Swammerdam Institute for Life Sciences – Science Park, Amsterdam
Prof. dr. Benno ter Kuile – Senior supervisor
PhD candidate Lisa Teichmann – Daily supervisor
Prof. dr. Brul – Second assessor
2
Content list
List of tables
3
List of figures
3
Abbreviations
4
Abstract
5
1. Introduction
6
1.1 Mechanisms contributing to AMR
6
1.2 Antibiotics: mechanisms of killing and inhibition
7
1.3 A central mechanism to the acquisition of AMR might be involved around reactive
oxygen species
9
1.4 The oxyR gene
10
1.5 Aims of this thesis
11
2. Material and Methods
12
2.1 Strains, media and antibiotics
12
2.2 Methods
13
3. Results
19
4. Discussion
27
5. Appendix
30
5.1 Materials
30
5.1.1 Composition and preparation of growth media
30
5.1.2 Composition and preparation of liquid solutions
31
5.1.3 Composition and preparation of antibiotic solutions
32
5.2 Protocols
33
5.2.1 Culturing and storage of mutant strains
33
5.2.2 Removal of kanamycin resistance
33
5.2.3 PCR and gel electrophoresis for confirmation of oxyR deletion
37
5.2.4 Growth curves
41
5.2.5 Photo spectrometer and plate reader measurements
44
5.2.6 Evolution experiment
45
5.2.7 Antibiotic dependent growth
46
5.3 Supplementary results
47
5.4 Supplementary background information
48
3
List of Tables
Table 1: E. coli strains and corresponding genotypes. 12
Table 2: Overview of antibiotics with additional class-specific information 13 Table 3: Oligonucleotide primers for amplifying the oxyR gene and its flanking DNA regions in ΔoxyR
mutant and wildtype E. coli strains. 15
Table 4: Parameters for PCR of oxyR DNA. 16
Table 5: Breakpoints used in this study to interpret susceptibility and resistance to amoxicillin (AMO),
enrofloxacin (ENR), kanamycin (KAN) and tetracycline (TET) 17
Table 6: Expected band lengths for PCR products. 19
Table 7: Minimal inhibitory concentrations (MICs) of amoxicillin, enrofloxacin, kanamycin and
tetracycline prior to evolution experiment. 20
Table 8: LB agar powder composition. 30
Table 9: LB liquid powder composition. 30
Table 10: Evans medium powder components. 31
Table 11: Evans medium liquid components. 31
Table 12: Compositions for calcium chloride, glucose, NaOH and HCl solutions. 31
Table 13: Trace elements solution composition. 32
Table 14: Antibiotic stock solutions (10ng/mL) preparation scheme. 32 Table 15: Minimal inhibitory concentration (MIC) of kanamycin (KAN) in ΔoxyR before and after
electroporation and FLP recombination with pcp20 plasmid DNA. 47
Table 16: OxyR regulated genes involved in the oxidative stress response. 48
List of figures
Figure 1: Steps in intracellular ROS formation leading up to intracellular (DNA) damage and
mutagenesis (Mendoza-Chamizo et al., 2018). 7
Figure 2: Overview of cellular targets for antibiotic classes relevant in this study. 8 Figure 3: Removal of kanamycin resistance (KANr) from ΔoxyR’s genome. 14 Figure 4: Schematic drawing of PCR primer sets for amplifying oxyR in ΔoxyR and MG1655 strains 15
Figure 5: Experimental set up of the evolution experiment. 17
Figure 6: Gel electrophoresis of ΔoxyR and MG1655 PCR products. 19 Figure 7: Growth curves ΔoxyR and MG1655 wild type (WT) strains. 20 Figure 8: acquisition of antimicrobial resistance to amoxicillin (AMO), enrofloxacin (ENR), kanamycin (KAN), and tetracycline (TET) in E. coli the ΔoxyR mutant strain in Evans medium under aerobic
circumstances. 21
Figure 9: Acquisition of antimicrobial resistance against amoxicillin (AMO) in ΔoxyR E. coli mutants in
aerobic and anaerobic conditions in Evans medium. 22
Figure 10: Acquisition of antimicrobial resistance against enrofloxacin (ENR) in ΔoxyR E. coli mutants
in aerobic and anaerobic conditions in Evans medium. 23
Figure 11: Acquisition of antimicrobial resistance against kanamycin (KAN) in ΔoxyR E. coli mutants in
aerobic and anaerobic conditions in Evans medium. 23
Figure 12: Acquisition of antimicrobial resistance against tetracycline (TET) in ΔoxyR E. coli mutants in
aerobic and anaerobic conditions in Evans medium. 24
Figure 13: LB agar plates with antibiotic resistant ΔoxyR E. coli cultures. 25 Figure 14: LB agar plates with amoxicillin (AMO) and kanamycin (KAN) resistant ΔoxyR E. coli
cultures. 26
4 Figure 16: Plate reader MIC measurements for AMO, ENR, KAN and TET prior to evolution
experiment. 48
Abbreviations
AB Antibiotic [AB]r [Antibiotic]resistance AMO Amoxicillin AMP AmpicillinAMR Antimicrobial resistance
bp Base pairs
CHL Chloramphenicol
E. coli Escherichia coli
ENR Enrofloxacin
KAN Kanamycin
LB Lysogeny broth
MIC Minimal inhibitory concentration
MQ Milli-Q (sterile water)
nt Nucleotides
OD Optical density
ROS Reactive oxygen species
ssDNA Single stranded DNA
TET Tetracycline
5
Abstract
Since the discovery of antibiotics, the de novo acquisition of antimicrobial resistance in bacteria has been an increasing threat to human health. A complex and wide variety of intracellular mechanisms ranging from phenotypical to genetic have been found to contribute to resistance against different antibiotic classes. Though mechanisms leading to antimicrobial resistance have been proven be partially class specific, a common contributing mechanism involved around DNA damaging reactive oxygen species has been proposed for the bactericidal β-lactams, fluoroquinolones, and
aminoglycosides.
One of the genes involved in the cellular response against reactive oxygen species is the oxyR regulon, from which downstream effects are contributing to the prevention of reactive oxygen species induced damage. In this study, the effect of oxyR genetic deletion in the de novo acquisition of antimicrobial resistance was investigated in Escherichia coli for three different bactericidal
antibiotic classes with different mechanisms of action, and one bacteriostatic antibiotic. Additionally, acquisition of antimicrobial resistance was compared in both aerobic and anaerobic circumstances to address the role of oxygen in the process. The overall outcome suggested that cellular mechanisms underlying the acquisition of antimicrobial resistance may vary among antibiotics of different classes in oxyR deficient mutants, though a common mechanism might also contribute. Furthermore, there could be hypothesized that presence of oxygen in absence of oxyR possibly interferes with the acquisition of resistance, though the exact effect and significance of oxyR on acquisition of antimicrobial resistance remain in need of further investigation.
6
1.
Introduction
It was September 1928 when Alexander Fleming discovered penicillin, later to be introduced to the medical world as the very first antibiotic substance deployed in the fight against microbial infections (Blair et al., 2015). Since then, a broad range of antibiotics was discovered, aiding in the reduction of deaths ascribed to bacterial infections as a consequence of chemotherapy and surgeries. Certainly, life span expectancy has drastically increased due to the influence of antibiotics in global health care (Blair et al., 2015; Ventola, 2015).
However, soon after its introduction, resistant bacterial strains against antibiotics emerged (Blair et al., 2015). Evolution of antimicrobial resistance (AMR) in bacteria was found to occur naturally when bacteria adapt upon exposure to antimicrobial drugs. A consequence is decreased sensitivity of the pathogen to antimicrobial drugs, as well as increased persistence of possible infections (Hay et al., 2018). Since there is little progress in the development of new antibiotics due partially to lack of economic profit, and AMR is becoming more and more prevalent, pathogenic bacterial infections are forming an increasingly deadly threat to human health (Rodríguez-Rojas et al., 2013; Ventola, 2015). This problem is demonstrated to be accelerated by the overconsumption, inappropriate prescription, and the large-scaled inappropriate use of antibiotics in agriculture, as well as poor infection and hygiene control. All of these factors contribute to the build spread of resistance to human pathogens (CDC Global Health, 2017; Ventola, 2015).
As this thesis will explore the acquisition of AMR in bacteria on a more cellular level, a selection of mechanisms underlying this process will be explained in the following paragraphs.
1.1 Mechanisms contributing to AMR
Bacteria evade the killing mechanisms of antibiotics through a variety of strategies, one of which is stress-induced mutagenesis. Although the majority of mutations caused by simple errors in DNA replication have neutral effect, elevation of overall mutation rate increases the chances for improved generation fitness and better adaptation to antibiotics. In addition to stress-induced mutational adaptation to antibiotics, bacteria can phenotypically adapt cell permeability and efflux, as well as receive resistance genes through horizontal gene transfer (Slonczewski & Foster, 2013; Ventola, 2015). The latter is not important in this thesis, though mutagenic and physiological resistance mechanisms will be elaborated on in the following paragraphs.
Mutagenesis by stress-induced SOS response
Antibiotic environmental stress can increase DNA mutagenesis as generated by the SOS response (Galhardo et al., 2007). When a bacterial cell experiences severe DNA damage upon exposure to stressors like antibiotics, ssDNA (single stranded DNA) can accumulate within the cell, which triggers the intracellular SOS response. A first step in this response is the binding of RecA filaments to these ssDNA strands. In response to this, a coprotease enzyme is activated, which triggers autodigestion of the LexA repressor that normally inhibits transcription of SOS proteins. Among the proteins that are transcribed upon cleavage of LexA are error-prone DNA polymerases without proofreading capacity. A consecutive result is DNA replication with high mutation frequency, due to the sloppy activity of these error-prone DNA polymerases. The chance for point mutations in favour of augmented antibiotic resistance is then significantly increased (Slonczewski & Foster, 2013). An example of a stress-induced SOS mediated mutation in favour of AMR are point mutations in the gyrase genes, which are targets for quinolone antibiotics (Galhardo et al., 2007).
7
The controversial radical-based theory
A way in which antibiotics can increase the mutation rate in bacterial DNA has been proposed to be the formation and aggregation of reactive oxygen species (ROS), as depicted in Figure 1, by
overstimulation of the electron transport chain (Van Acker & Coenye, 2017). There are controversies regarding the radical-based theory, that proposes a correlation between the acquisition of AMR and ROS, as will be described in more detail in section 1.3. Nevertheless, studies supporting the theory have demonstrated that increased ROS levels in E. coli after treatment with sublethal levels of antibiotics correlate with increased mutagenesis (Dwyer et al., 2009; Händel et al., 2016; Kohanski, Dwyer, et al., 2010).
Figure 1: Steps in intracellular ROS formation leading up to intracellular (DNA) damage and mutagenesis. Image source: Mendoza-Chamizo et al. (2018). The first step in ROS formation occurs when a cell increases its metabolic rate, leading to NADH depletion . As a consequence, molecular oxygen (O2) is reduced into superoxide anions (O2-) by autooxidation of flavoproteins. O2- then releases iron ions (Fe2+) from iron proteins by reducing into the hydrogen peroxide (H2O2), or is directly converted into H2O2 by superoxide dismutases (SODs). Although O2- and H2O2 ROS molecules are lacking evidence of being directly associated with DNA damage, they (in)directly fuel the Fenton reaction. In this reaction, H2O2 is reduced into the highly reactive hydroxyl radical (OH), while Fe2+ is oxidized into Fe3+. In the final step of the process, the HO radicals cause DNA lesions by oxidizing DNA bases. These damaged bases are cut out, however sometimes not repaired. Subsequently, this leads to double stranded breaks and mutagenesis, which additionally calls for an intracellular SOS response, as well as reinforce an already induced SOS response (Dwyer et al., 2009; Mendoza-Chamizo et al., 2018; Rodríguez-Rojas et al., 2013).
Alterations in gene expression
In contrast to genetic adaptation, bacteria can also adjust phenotypical features to fight the damaging effect of antibiotics by differential gene expression. Bacteria can prevent access to the antibiotic targets by reducing membrane permeability, as well as increasing antibiotic efflux by upregulating drug efflux pumps. Another additional strategy is the direct modification or destruction of antibiotics by the bacterial cell. Examples are hydrolysis or transfer of a chemical group in order to inactivate the antibiotic (Blair et al., 2015).
1.2 Antibiotics: mechanisms of killing and inhibition
The initial cellular response to antibiotic stress is different among different antibiotics, for variations in mutational background and phenotypical adaptations have been found to be very diverse (Händel
8 et al., 2014). This can largely be explained by the wide variety of antibiotics, their different
mechanisms of action and cellular targets, as will be elaborated on in the following subsection.
Bactericidal and bacteriostatic
The first manner of categorization distinguishes two main groups of antibiotics. Bactericidal
antibiotics, independent of their cellular target, kill bacterial cells. Bacteriostatic antibiotics do not kill the bacteria, but simply inhibit growth (Kohanski et al., 2007; Pankey & Sabath, 2004).
In addition to whether the antibiotic induces cell death, drug target-interactions are another way to categorize antibiotics. The three main targets that can be characterized are DNA replication and repair, protein synthesis and cell-wall turnover (Kohanski et al., 2007; Kohanski, Dwyer, et al., 2010). Based partially on their cellular targets, antibiotics are divided into classes.
Relevant antibiotic classes
A schematic overview of antibiotic classes and their cellular targets is visualized in Figure 2 for antibiotic classes that were focussed on in this study. Furthermore, this subsection elaborates on mechanisms of action and resistance for each specific class.
Figure 2: Overview of cellular targets for antibiotic classes relevant in this study. Primary targets are subsequently; cell wall for β-lactams, the 30S ribosomal subunit for aminoglycosides and tetracyclines, and DNA gyrases for
(fluoro)quinolones.
Cell wall biosynthesis is the process targeted by the β-lactam class of antibiotics. The bacterial cell wall is built from peptidoglycan layers that are covalently cross-linked, and provides mechanical strength as well as protection against environmental stress. β-lactams inhibit the synthesis of the cell wall by inhibiting the transpeptidases that catalyse cross-linking of the peptidoglycan layers. β-lactams can hereby kill the bacterial cell in a lytic dependent, as well as in a lytic independent manner (Kohanski, Dwyer, et al., 2010). De novo resistance against β-lactams was demonstrated to be mainly driven by mutations in the ampC gene of beta-lactamase (Händel et al., 2014).
(Fluoro)quinolones are the class of antibiotics that target DNA replication. In E. coli bacteria, the primary target of fluoroquinolones is the topoisomerase enzyme DNA gyrase (Redgrave et al., 2014). Fluoroquinolone action results in inhibition of DNA synthesis and ultimately leads to cell death (Kohanski et al., 2007; Kohanski, Dwyer, et al., 2010). Fluoroquinolone resistance mechanisms involve mutagenesis in the topoisomerase targets. Bacteria become highly resistant when mutations are induced in the gyrase’s high affinity binding site of the gyrA gene, in combination with mutations in the parC gene or the gyrB gene (Händel et al., 2014; Redgrave et al., 2014). Another resistance
9 mechanism against fluoroquinolone stress is found in mutations that downregulate molecules providing membrane permeability. Furthermore, efflux systems are found to be overly expressed in fluoroquinolone resistant bacteria (Redgrave et al., 2014).
The 30S ribosomal subunits aiding in mRNA translation are targets of the aminoglycoside and tetracycline classes of antibiotics. When aminoglycosides bind the ribosome, faulty protein translation is promoted by incorporation of incorrect amino acids. Mistranslated proteins can subsequently bring damage to different cellular compartments. An example of this is the insertion of mistranslated proteins into the membrane, affecting permeability. In this manner, aminoglycosides may also contribute to β-lactam antibiotic activity (Kohanski, Dwyer, et al., 2010). Resistance mechanisms against aminoglycosides have been established as a combination of mutational and phenotypical adaptations (Garneau-Tsodikova & Labby, 2016). As opposed to the bactericidal effect of aminoglycosides, binding of tetracyclines to the ribosomal subunits merely arrests protein translation. Resistance to tetracyclines was mainly ascribed to phenotypical adaptations like efflux, enzymatic inactivation and protection of the ribosome target (Grossman, 2016).
1.3 A central mechanism to the acquisition of AMR might be involved around
reactive oxygen species
The acquisition of antimicrobial resistance has been demonstrated to be the result of a very complex and diverse combination of physiological and mutational adaptations that are induced upon
antibiotic stress. For bactericidal antibiotics, one contributor to AMR has been proposed to be the generation of reactive oxygen species (ROS), irrespective of initial mode of action as known for each separate antibiotic class (Dwyer, Belenky, et al., 2014).
Different antibiotic classes induce varying cellular responses
The acquisition of AMR in E. coli was found to be established by increasing antibiotic exposure stepwise. It was implied that sublethal levels of antibiotics can induce AMR, rather than high antibiotic concentrations that kill the cell or inhibit growth entirely (van der Horst et al., 2011).
In addition to this, prior research by Händel et al. (2014) pointed out that exposure to sublethal levels of different classes of antibiotics induce AMR in varying intracellular manners in Escherichia
coli (E. coli), as was previously described in section 1.2. The bacteriostatic antibiotic tetracycline,
which merely inhibits bacterial growth, was found to lead to AMR through reversible differential gene expression, though not by means of mutations.
In contrast, in E. coli exposed to bactericidal antibiotics, resistance was caused in more or lesser extend by consistent chromosomal mutations, and by adaptation of gene expression. The ratio between these two varied significantly among different classes of bactericidal antibiotics, indicating that mechanism of action influences intracellular route that leads to acquisition of AMR (Händel et al., 2014; Slonczewski & Foster, 2013). In line with these findings, additional research by Händel et al. (2016) further confirmed that cellular response is dependent on antibiotic class.
A common downstream mechanism independent of initial cellular response might lead to
AMR
In addition to class specific defence mechanisms it was suggested that irrespective of class, a central cellular mechanism, the downstream production of ROS, might be crucial for the development of de
10 In 2018, Hoeksema et al. further analysed the theory of a common mechanism underlying AMR acquisition. A central downstream mechanism of ROS production would suggest partially shared characteristics in the different cellular responses to β-lactams, fluoroquinolones and
aminoglycosides. These speculations were supported by the observation that resistance against one antibiotic class significantly increased the rate of AMR acquisition for another antibiotic class. It was further implied that resistance mechanisms induced by the first antibiotic class might have targeted the reduction of ROS, thereby protecting against production of ROS by stimulation with second antibiotic (Hoeksema et al., 2018). An additional study found that, under anaerobic conditions and thus in the absence of ROS, aminoglycosides, fluoroquinolones, and β-lactams all showed decreased killing efficacy (Dwyer, Collins, et al., 2014).
Although interesting, these findings are still a topic of debate. First of all, difficulties have been found in demonstrating ROS production, mainly because the use of hydroxyphenyl fluorescein (HPF) to measure ROS levels can yield biased results (Van Acker & Coenye, 2017). Besides this, studies have published contradictive data. Observations were made that quinolone, β-lactam and aminoglycoside killing were not significantly affected by the presence of oxygen in several experiments performed in aerobic and anaerobic circumstances (Liu & Imlay, 2013). Neither was there variation in ROS
production or ROS induced oxidative stress upon antibiotic exposure (Keren et al., 2013).
In order to address the role of ROS in the problem of increasing antimicrobial resistance, it is of great importance that the controversy is further investigated.
One of the genes involved in the response to elevated intracellular ROS levels is the oxyR gene, that was found to constitutively transcribe the OxyR protein. Additionally, OxyR functions as a
transcriptional regulator for a variety of genes involved in prevention of ROS induced DNA damage (Pomposiello & Demple, 2001), making oxyR an interesting target for experimentation with AMR. Therefore, oxyR was a main subject of focus in this thesis. Before the project itself will be described in more detail, the following section provides a background on the exact cellular function and role of
oxyR in the cell.
1.4 The oxyR gene
After transcription, the OxyR protein functions as a sensor for elevated ROS levels. When
endogenous H2O2 levels increase, the tetramer structured OxyR protein is oxidized by H2O2 into its active form through the formation of disulfide bonds in between its subunits. OxyR then stimulates transcription of downstream target genes by binding their promotor regions, resulting in a broad spectrum of active proteins (Pomposiello & Demple, 2001).
After a genome wide analysis, 79 genes were found to be downstream targets of OxyR in time of oxidative stress, some of which required additional co-activation by other transcriptional regulators (Seo et al., 2015). A selection of oxyR regulated genes important in the cellular response against oxidative stress are listed in section 5.4 (Table 16). Downstream effects of oxyR regulation are involved in the prevention of ROS inflicted cellular damage, as well as in reduction of intracellular ROS. First of all, Catalase and peroxidase enzymes are upregulated, resulting in the direct oxidation and removal of ROS (Mendoza-Chamizo et al., 2018; Vidossich et al., 2012). Furthermore, there are genes activated involved in decreasement of available iron for the Fenton reaction, stabilization of DNA, and thickening of the membrane (Mendoza-Chamizo et al., 2018). Lastly, oxidation of O2- and H2O2 into OH is prevented by expression of oxidation resistant isozymes. Since formation of the
11 damaging OH radical becomes less frequent through all of these these adaptations, further DNA damage is limited by oxyR’s transcriptional regulation (Mendoza-Chamizo et al., 2018).
1.5 Aims of this thesis
As the threat of antibiotic resistance is ever increasing and the demand for a solution remains, it is of great importance that the mechanisms behind the acquisition of AMR are better understood.
Consensus has been achieved in large extend for class-specific genetic and physiological adaptation that underlie AMR to quinolones, β-lactams, aminoglycosides and tetracyclines by extensive
experimentation over the years. Yet, one of the gaps in our knowledge about resistance mechanisms is the radical-based theory in relation to AMR acquisition. It is a topic of debate in need of further exploration. Given the central role of the oxyR gene in the prevention of oxidative stress, studying the acquisition of AMR in a oxyR deficient E. coli strain could give novel insights in the importance of ROS in this process.
The role of ROS as a secondary killing mechanism would suggest that upon deletion of oxyR in the E.
coli genome, elevated ROS levels on top of exposure to antibiotic stress would possibly induce
increased killing efficacy of bacteria. On the other hand, depending on the unknown extend of which ROS is involved in the acquisition of AMR, there may be an enhancing effect on the maximum concentration of antibiotics in which bacteria can grow when intracellular ROS is not effectively reduced upon oxyR deletion.
Therefore, documenting the acquisition of AMR in oxyR deficient E. coli by exposure to step-wise increasing antibiotic concentrations compared to a wild type strain could provide new insights of earlier findings (as in section 1.3) regarding the involvement of ROS in antibiotic resistance.
This thesis aimed to investigate the effect of genetic deletion of the oxyR gene in E. coli on the de
novo acquisition of antimicrobial resistance against amoxicillin, enrofloxacin, kanamycin, and
tetracycline. Hereby, there was focussed on the maximum antibiotic concentration that an oxyR deficient strain can adapt to, as well as the adaptation rate of the strain to reach these maximums. Subsequently, this study aimed to detect in what extend oxygen is involved in the acquisition of AMR to these antibiotics in oxyR deficient E. coli by comparison of experimental data of similar
experiments in aerobic and anaerobic conditions. Emphasis was again put on maximum adaptability and adaptation rate for all antibiotics.
12
2 Materials and Methods
In order to investigate the influence of the oxyR gene on the de novo acquisition of AMR to various antibiotic classes, a mutant ΔoxyR E. coli strain was exposed to step-wise increasing sublethal concentrations of three bactericidal antibiotics of β-lactam, fluoroquinolone and aminoglycoside classes, and one bacteriostatic tetracycline antibiotic in an evolution experiment in aerobic conditions. Additional data of a similarly performed evolution experiment in anaerobic conditions was obtained by Yke Sijpestijn.
Sublethal antibiotic concentrations were derived from the minimal inhibitory concentrations, defined as the lowest concentration of antibiotic that prohibited bacterial growth.
Additionally, prior to the evolution experiment FLP recombination was used to remove kanamycin resistance (KANr) from the ΔoxyR genome, growth curves were conducted to compare ΔoxyR’s growth rate to MG1655’s, and PCR and gel electrophoresis were performed to confirm absence of the oxyR gene in the ΔoxyR strain. Furthermore, ΔoxyR cultures resistant to amoxicillin and
kanamycin were grown with and without varying concentrations of corresponding antibiotics in order to test for suspected dependency for growth.
2.1 Strains, media and antibiotics
Escherichia coli strains
In this study, the Escherichia coli (E. coli) JW3933 ΔoxyR strain was used to study the acquisition of antimicrobial resistance against amoxicillin, enrofloxacin, kanamycin and tetracycline. This strain was derived from a E. coli K-12 BW25113 parent strain and obtained from E. coli Keio Knockout Collection (E. coli Keio Knockouts, 2020). oxyR was deleted and replaced with a kanamycin resistance cassette by means of FLP recognition target (FRT) recombination (Baba et al., 2006). An overview of additional strains can be accessed in Table 1.
Table 1: E. coli strains and corresponding genotypes.
Name Strain Chromosomal properties
ΔoxyR JW3933 F- Δ(araD-araB)567, ΔlacZ4787(::rrnB-3), λ-, rph-1, Δ(rhaD-rhaB)568, ΔoxyR749::kan, hsdR514
K-12 wild type (WT)
MG1655 F- λ- rph-1
pCP20 DH5α fhuA2 ΔlacU169 phoA glnV44 Φ80' ΔlacZM15 gyrA96 recA1 relA1 endA1 thi-1 hsdR17
Growth media and chemicals
Sterilization of chemicals, agar and liquid media was performed by autoclaving for 20 min at 121 °C, unless stated otherwise. Powder bases were prepared according to section 5.1 in which components for all media, liquid solutions and antibiotic solutions are listed.
E. coli ΔoxyR and WT strains were grown on LB agar. Agar plates were incubated at either 37°C or
30°C, depending on the method. Optionally, sterilized antibiotics were added to LB agar in required concentrations after autoclaving.
Evans medium with 55mM glucose was used as a limited nutrient source medium, and was applied to ensure control and possibility of adaptation of the nutrient supply in the medium if necessary.
13 Trace elements and calcium chloride dihydrate were added before adjusting pH to 6.9 with 4M NaOH and 1M HCl. Sterilized glucose solution, autoclaved for 10 min at 110°C and prepared according to section 5.1.2, was added to Evans medium after sterilization. LB liquid medium was used as a rich nutrient source medium.
Antibiotics
An overview of antibiotics that were utilized in the experiments can be consulted inTable 2, along with abbreviations that will be used throughout the materials and methods sections, corresponding classes and their cellular targets. Stock solutions for all antibiotics were prepared according to section 5.1.3. For sterilization of antibiotic solutions, sterile syringes and 0.2 μL filters were used. Amoxicillin stock was stored at 4°C for a maximum of three days. All other antibiotics were stored at -20°C and, when thawed, kept in the 4°C fridge for a maximum of three days.
Table 2: Overview of antibiotics with additional class-specific information
Antibiotic Abbreviation Bactericidal/Bacteriostatic Class Cellular target
Amoxicillin AMO Bactericidal Β-lactam Cell wall
Enrofloxacin ENR Bactericidal Fluoroquinolone DNA
synthesis
Kanamycin KAN Bactericidal Amino-glycoside Protein
synthesis
Tetracycline TET Bacteriostatic Tetracycline Protein
synthesis
Chloramphenicol CHL Bactericidal Phenicols Protein
synthesis
Ampicillin AMP Bactericidal Β-lactam Cell wall
2.2 Methods
Protocols for all described methods in the following paragraphs are listed and can be consulted in section 5.2.
Culturing and storage of strains
For culturing, a single colony was selected from an LB agar plate with an inoculation loop to obtain a pure culture. This colony was inoculated in liquid LB medium or Evans medium with 55mM glucose and grown overnight in a 200 rpm 37°C shaker. When stored, liquid cultures were frozen in a 1:1 ratio with sterilized 60% glycerol at -80°C.
Removal of kanamycin resistance
KAN resistance was removed from the ΔoxyR mutants’ chromosome in order to prevent interference of a resistance background with the de novo acquisition of other antibiotics, as was previously observed (Hoeksema et al., 2019).
14
Figure 3: Removal of kanamycin resistance (KANr) from ΔoxyR’s genome. pCP20 plasmid DNA encoded for
chloramphenicol resistance (CHLr) and ampicillin resistance (AMPr), and was isolated from the DH5α strain with miniprep. Before transformation, electroporation made the ΔoxyR strain competent for taking up pCP20 DNA. Resistance genes were transcribed upon incubation of the transformed strain at 30°C. KANr was cut out by means of FLP recombination upon activation of the heat sensitive origin of replication of FLP recombinase in the pCP20 plasmid at 44°C, after which resistance genes for CHL, AMP, and KAN were degraded and/or discarded of at 30°C.
For deletion of kanamycin resistance from the chromosome of ΔoxyR mutants (Figure 3), bacteria were electroporated and transformed with pCP20 Flp (flippase) recombinase carrying plasmid DNA according to protocols in section 5.2.2. Yield of isolated plasmid DNA was measured with Nanodrop. Prior to transformation, pCP20 plasmid DNA was isolated from the DH5α strain with the GeneJET Plasmid Miniprep Kit K0502. Plasmid DNA contained resistance genes for CHL and AMP. Transcription of plasmid genes was induced at 30°C, after which growth of the ΔoxyR strain could be observed on LB agar containing CHL, AMP, and KAN. Excision of chromosomal KANr with FLP recombination was induced at 44°C by activating the heat sensitive origin of replication in the pCP20 plasmid. The mechanism of Flp recombination works by recognition and cutting of FRT (Flp recombinase recognition target) sites flanking the target gene by FLP recombinases, after which the latter mediates homologous recombination at the FRT sites (Park et al., 2011). Afterwards, the excised KANr gene was gradually lost in the ΔoxyR culture by degradation and cell division, and pCP20 DNA with CHLr and AMPr was discarded of by the cells at 30°C.
Removal of kanamycin resistance was also performed for the ΔrelA mutant strain, that was further investigated by Niels Boek.
Determination of growth rate
Growth curves were generated to determine and compare generation time, defined as the time it takes before a bacterial culture has doubled, between ΔoxyR and wild type strains. For both LB and Evans media, strains were cultured as described in 1.2.1, after which liquid cultures were brought to
15 an OD of 0.05 in corresponding medium. Optical density was measured at 595nm in a Multiscan FC plate reader over a total time period of 23 hours (138 reads, 10 min/read) as in section 5.2.4.
Amplification of oxyR from E. coli Mg1655 and Jw3933 ΔoxyR genomic DNA with PCR
Full-length DNA of both the ΔoxyR mutant and Mg1655 wild type (WT) strains was inoculated overnight in a 37°C shaking incubator in Evans medium (55mM glucose). Optical density was measured, after which both strains were resuspended in fresh Evans medium to a similar OD600 of 2.0.
Subsequently, the oxyR gene fragment and its flanking regions were amplified with 10μM primers (Table 3, as designed as in section 5.2.3) in a polymerase chain reaction (PCR). Primer sets used for amplification of two distinct fragments including oxyR are schematically visualized and described in Figure 4.
Figure 4: Schematic drawing of PCR primer sets for amplifying oxyR in ΔoxyR and MG1655 strains. Primer set 1 was designed to amplify oxyR to confirm absence of the gene in ΔoxyR’s genome. Primer set 2 was designed to amplify oxyR and its flanking regions to confirm deletion of the gene from its specific position in ΔoxyR’s genome.
Table 3: Oligonucleotide primers for amplifying the oxyR gene and its flanking DNA regions in ΔoxyR mutant and wildtype E. coli strains. Primer set 1 was used to confirm absence of the gene in the ΔoxyR mutants. Primer set 2 was used to confirm deletion of oxyR from its specific genomic position in the mutants.
Forward Reverse Range (no. of
nucleotides) Primer set 1
(oxyR) 5’-TGATCTTGAGTACCTGGTG-3’ 5’-CAGATAAACAACCCCATCGC-3’ 760 Primer set 2
(oxyR + flanking regions)
5’-AGGCTCGGTTAGGGTAAG-3’ 5’-AGTTTAAACATCTGGCACG-3’ 1430
In 25 μL aliquots, PCR samples were prepared with 2x MyTaq Mix (1x), Forward primer (0.4μM), Reverse primer (0.4μM), sterile MQ and 1.5μL liquid culture of ΔoxyR, WT (positive control), or MQ (negative control). All reagents were kept on ice at all times. After spinning PCR tubes down, samples were loaded in a Thermocycler PCR that was ran with parameters as listed in Table 4.
16
Table 4: Parameters for PCR of oxyR DNA. PCR ran with ΔoxyR and MG1655 WT strains and primer sets as listed in Table 3.
Step °C Time (min:s) Cycles
Denaturation 94 10:00 1 Denaturation 94 00:40 - Annealing 45 00:40 - Extension 72 01:00 30 Extension 72 05:00 1 4 ∞ ∞
Visualisation of PCR products with gel electrophoresis
For visualising DNA fragments as obtained from PCR, a 1.5% agarose gel was prepared with 1x TAE buffer, to which Midori Green dye was added. Samples were loaded after resuspension in 1μL Thermo Scientific 6x Orange Dye Loading Dye. As a ladder, the Thermo Scientific O’GenerRuler 1kb DNA Ladder (0.1μg/μL) was used.
Gel electrophoresis was performed twice with the same PCR samples. Best visibility of bands was observed when gel ran at 80V and 400A for 2 hours, as opposite to a running time of 1 hour.
A detailed protocol for PCR and gel electrophoresis can be consulted insection 5.2.3.
Optical density measurements
Optical Density (OD) of liquid E. coli cultures was determined in plastic cuvettes at 600nm in a photo spectrometer (Lightwave II Isogen Life Science), according to protocol in section 5.2.4. For calibration and dilution of cultures, growth medium equivalent to the medium applied for culturing was used.
MIC measurements
Minimal Inhibitory Concentration (MIC) of all antibiotics was measured to determine the minimal concentration of antibiotic at which no bacterial growth was observed above an OD600 of 0.2. This cut off value was chosen based on the assumption that at least two consecutive cycles of bacterial replication are sufficient to be considered growth in its most minimal form. MICs were measured to define antibiotic starting concentrations that could later be used in the evolution experiment.
MIC measurements were performed in a Multiskan FC plate reader, and executed according to protocol in section 5.2.5. Antibiotic starting concentrations were established in the first column of a 96-wells plate, and duplicate diluted in 2-fold up until the tenth column in Evans or LB medium. Starting concentrations were based on expected MICs as documented in a large collection of previously encountered MIC values (Eucast2, 2020:
https://mic.eucast.org/Eucast2/SearchController/search.jsp?action=performSearch&BeginIndex=0& Micdif=mic&NumberIndex=50&Antib=-1&Specium=162).
Subsequently, columns eleven and twelve functioned as positive and negative controls to test for growth without the influence of antibiotics and contamination of the medium. An OD600 of 0.05 of inoculated bacterial culture was added to each column but the twelfth. MIC values for all antibiotics were determined in μg/mL during 138 readings, 10 minutes each. Sensitivity or resistance to every antibiotic was ascribed to the strain based on breakpoint guidelines determined by CLSI, and are listed in Table 5 (CDC, 2019).
17
Table 5: Breakpoints used in this study to interpret susceptibility and resistance to amoxicillin (AMO), enrofloxacin (ENR), kanamycin (KAN) and tetracycline (TET)(CDC, 2019). *Due to lack of available breakpoints for AMO and ENR specifically, ampicillin and ciprofloxacin breakpoints were considered the most reliable alternatives for interpretation of MIC values as obtained from AMO and ENR cultures.
CLSI class Antibiotic MIC breakpoint (μg/mL) Susceptible Intermediate Resistant
Penicillins Ampicillin* ≤8 16 ≥32
Quinolones Ciprofloxacin* ≤0.25 0.5 ≥1
Aminoglycosides Kanamycin ≤16 32 ≥64
Tetracyclines Tetracycline ≤4 8 ≥16
Evolution experiment
To follow the development of antimicrobial resistance against bactericidal antibiotics of different classes, an evolution experiment was performed with the ΔoxyR mutant strain after removal of chromosomal kanamycin resistance, according to Figure 5. The experiment was similarly performed by Yke Sijpestijn under anaerobic circumstances to investigate the role of oxygen on acquisition of AMR in ΔoxyR mutants. To monitor acquisition of AMR, MIC measurements were performed twice a week. Interpretation of susceptibility and resistance was based on breakpoints in Table 5. An MIC measurement for the positive control was done every other week.
Figure 5: Experimental set up of the evolution experiment. ΔoxyR mutants were cultured in Evans medium (55mM glucose), followed by duplicate incubation of the culture in Evans medium with antibiotics (AB). AB starting concentrations were determined as ¼ of the initial MIC concentration of amoxicillin, enrofloxacin, kanamycin or tetracycline. Bacterial culture was added to each tube with an OD600 of 0.1. After overnight inoculation at 37°C (or 30°C for two-night incubations)
in a 200rpm shaker, OD600 was measured daily and if sufficient growth was observed (OD600 >60% of OD600 whereby AB
concentration was last increased), antibiotic concentration was increased in 2-fold. The duplicate culture with the highest OD600 was continued each day. A positive control of ΔoxyR culture in medium without antibiotics was carried along
18
Antibiotic-dependency test with AMO and KAN resistant ΔoxyR cultures
During an advanced stage of the evolution experiment where mutants had acquired the ability to grow at antibiotic concentrations that would be regarded as resistant, observations were made that AMO and KAN resistant mutants had trouble growing without, and in low AMO/KAN concentrations. To investigate whether ΔoxyR mutants had become dependent on the antibiotics that they had become resistant to during the evolution experiment for growth, a protocol (section 5.2.7) was set up with LB agar plates. 10-fold and 100-fold dilutions of AMOr and KANr mutants were plated on LB agar with the antibiotic concentration of their last measured MIC (2048 μg/mL), the antibiotic concentration in which best growth was observed in a previous plate reading (64 μg/mL), and on LB agar without antibiotics. Additionally, gradient plates were made for both antibiotics, with a 0-2048 μg/mL AMO/KAN gradient. Plates were incubated at 37°C. Higher dilutions are recommended for repetition of this experiment, after finding no possibility to determine CFU/mL (>300 colonies/plate).
19
3 Results
Confirmation of oxyR gene deletion
To confirm deletion of the oxyR gene from ΔoxyR mutants’ genomic DNA, an oxyR gene fragment and fragments of flanking region with or without oxyR were amplified with two distinct primer sets in PCR with ΔoxyR and MG1655 WT strains. Expected product band lengths are listed in Table 6.
Table 6: Expected band lengths for PCR products. Lane numbers correspond with lane contents as described in Figure 6.
oxyR oxyR + flanking regions
Lane no. 1 2 3 4 5 6
Band length (bp)
- 760 - 550 1450 -
Analysis of PCR products with gel electrophoresis (Figure 6) showed no visible band for oxyR in ΔoxyR and a ~760bp band for MG1655. This indicated absence of the gene in the mutant´s, and presence in the WT´s chromosomal DNA. Deletion of oxyR from its specific position in the chromosome was implied by a band showing at ~550bp for ΔoxyR. An additional band at ~1450bp for MG1655 indicated presence
of the gene in the WT strain. No bands were visible for the negative control, indicating no
contamination. Bands in the agarose gel were considered sufficient for assuming absence of the gene in the ΔoxyR, and presence in the WT strain.
Figure 6: Gel electrophoresis of ΔoxyR and MG1655 PCR products. Lanes 1-3 contain PCR products of primer set 1 (oxyR gene), and lanes 4-6 contain the products of primer set 2 (oxyR gene + flanking regions). Strains are ΔoxyR (lanes 1 and 4), MG1655 (lanes 2 and 5) and a negative control (lanes 3 and 6). L=ladder with band lengths as indicated right of the ladder. Expected lengths (bp) are, in the order of the numbered lanes; 0, 760, 0, 550, 1450, 0.
20
Generation of growth curves and determination of growth rate
Growth curves were generated in a plate reader to compare growth rate between E. coli ΔoxyR and MG1655 (WT). This was done in order to interpret comparability of MIC values between these strains as obtained from plate reader measurements. Difference in generation time, as determined from the exponential phase of growth curves (Figure 7) was considered to be negligible: on average, in a 24 hour incubation, there would be an expected variance of 2 doublings between ΔoxyR and WT cultures.
ΔoxyR and MG1655 strains were susceptible to all antibiotics prior to evolution
experiment
After successful removal of KAN resistance (Appendix ), MIC values were determined for AMO, ENR, KAN and TET in order to define appropriate starting concentrations for the evolution experiment as described in Figure 5. MIC values for ΔoxyR and MG1655 WT strains are listed in Table 7, and were similar for all antibiotics. Both strains were considered susceptible to all antibiotics based on CLSI breakpoints (Table 5). Graphs with supporting visual data of MIC measurements can be consulted in section 5.3.
Table 7: Minimal inhibitory concentrations (MICs) of amoxicillin, enrofloxacin, kanamycin and tetracycline prior to evolution experiment. MIC values for the MG1655 strain were obtained by Lisa Teichmann.
Antibiotic Strain MIC (μg/mL)
Amoxicillin ΔoxyR 4 MG1655 4 Enrofloxacin ΔoxyR 0.125 MG1655 0.125 Kanamycin ΔoxyR 16 MG1655 16 Tetracycline ΔoxyR 1 MG1655 1 0 60120 180240300360420480540600660720780 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Growth curves exponential phase
oxyR and MG1655 strainsTime (min) O D5 9 5 n m oxyR aerobic MG1655
Figure 7: Growth curves ΔoxyR and MG1655 wild type (WT) strains. An initial log phase was followed by an exponential phase, from which generation time was determined for both strains: 98 min (WT) and 113 min (ΔoxyR). Growth curves for both strains were generated from the mean values of triplicate measurements of three separate plate readings (138 reads, 10 min each) in Evans medium with 55mM glucose.
21
De novo acquisition of AMR in ΔoxyR mutants by exposure to sublethal antibiotic
concentrations
To induce antibiotic stress, ΔoxyR was cultured and inoculated with sublethal antibiotic concentrations as determined as ¼ of priorly determined MIC values (Table 7). Starting
concentrations were 1 μg/mL (AMO), 0.03125 μg/mL (ENR), 4 μg/mL (KAN) and 0.25 μg/mL (TET) in limited nutrient source Evans medium (55mM glucose). These sublethal concentrations were chosen based on the assumption that ¼ MIC would induce enough cellular stress to stimulate resistance mechanisms for the acquisition of AMR, and also not kill cells/inhibit growth entirely. Antibiotic concentrations were increased accordingly as described in Figure 5. Number of generations, as determined from OD at the moment of inoculation and OD as measured the subsequent day, varied between 3-6 per overnight incubation, and between 7-9 for two-night incubations for all antibiotics. Average increase in generations was considered comparable between different antibiotics.
Figure 8: acquisition of antimicrobial resistance to amoxicillin (AMO), enrofloxacin (ENR), kanamycin (KAN), and tetracycline (TET) in E. coli the ΔoxyR mutant strain in Evans medium under aerobic circumstances. A: minimal inhibitory concentration of antibiotic at which no growth was observed (OD<0.2). B: fold change in antibiotic concentration in which growth of ΔoxyR was observed (OD>0.2). Antibiotic concentrations in which cells were inoculated was increased in two-fold if growth was comparable to growth in the previous concentration.
Acquisition of AMR was followed for 31 days by biweekly MIC measurements, and by daily documentation of fold change in antibiotic concentration in which growth (OD>0.2) was observed after overnight inoculation.
MIC was observed to initially increase at roughly the same rate for all bactericidal antibiotics (8A), as was also seen for increase in fold change of concentration (8B). This was observed until ENR’s MIC started to increase more exponentially from 16 μg/mL onwards. A similar maximum MIC of 2048 μg/mL was reached at roughly 150 generations for AMO, ENR and KAN, while the maximum MIC measured for TET was 32 μg/mL (Figure 8A). ΔoxyR acquired resistance to all antibiotics, for all maximum MIC values surpassed resistance breakpoints (Table 5). Altogether, from the graphs in Figure 8 there can be deduced that irrespective of class, E. coli can grow in remarkably higher bactericidal than TET bacteriostatic concentrations in the absence of oxyR. Among bactericidal antibiotics, graph patterns (i.e. linearity/exponentiality) differ in some extend.
On a side note, a 5-step difference in fold change of antibiotic concentration at which growth
occurred was ultimately observed for ENR, in comparison to AMO and KAN (Figure 8B). Given E. coli’s low ENR susceptibility breakpoint (Table 5), this difference is suggested to be insignificant on grounds
22 of initial inoculation with a lower starting concentration of ENR in comparison to the other
antibiotics, while growth occurs in a similar maximum concentration.
Comparison of acquisition of AMR in ΔoxyR in aerobic and anaerobic conditions
For the purpose of investigating whether oxygen has an impact on the acquisition of AMR in oxyR deficient E. coli mutants, datasets of the evolution experiment under aerobic circumstances were compared to datasets of a similarly performed evolution experiment under anaerobic circumstances. All raw data for anaerobic experiments with ΔoxyR were obtained from Yke Sijpestijn, and further analysed for use in this thesis. Yke generated distinct datasets of two ΔoxyR cultures for each antibiotic. To correct for the slower growth rate of anaerobically grown ΔoxyR (~3
generations/incubation, opposed to ~5 generations/incubation when grown aerobically), MIC values and fold change were plotted against the number of generations in addition to time in days.
Results focus on maximum MIC, and fold change in concentration of antibiotic that cultures were able to grow in. Additionally, results focus on rate of acquisition of AMR as expressed in both time and number of generations, as relative to MIC values or fold change.
Figure 9: Acquisition of antimicrobial resistance against amoxicillin (AMO) in ΔoxyR E. coli mutants in aerobic and anaerobic conditions in Evans medium. A: fold change in AMO concentration in days. B: fold change in AMO concentration in generations. C: minimal inhibitory concentration of AMO in which no growth was observed (OD<0.2).
23
Figure 10: Acquisition of antimicrobial resistance against enrofloxacin (ENR) in ΔoxyR E. coli mutants in aerobic and anaerobic conditions in Evans medium. A: fold change in ENR concentration in days. B: fold change in ENR concentration in generations. C: minimal inhibitory concentration of ENR in which no growth was observed (OD<0.2).
Figure 11: Acquisition of antimicrobial resistance against kanamycin (KAN) in ΔoxyR E. coli mutants in aerobic and anaerobic conditions in Evans medium. A: fold change in KAN concentration in days. B: fold change in KAN concentration in generations. C: minimal inhibitory concentration of KAN in which no growth was observed (OD<0.2).
24
Figure 12: Acquisition of antimicrobial resistance against tetracycline (TET) in ΔoxyR E. coli mutants in aerobic and anaerobic conditions in Evans medium. A: fold change in TET concentration in days. B: fold change in TET concentration in generations. C: minimal inhibitory concentration of TET in which no growth was observed (OD<0.2).
Comparing maximum fold change and MIC values
For AMO, ENR, and TET, a negligible difference was seen in fold change of antibiotic concentration in which ΔoxyR mutants were able to grow between aerobic and anaerobic circumstances, as was also the case for maximum MIC (Figures 9, 10, and 12). Maximum adaptation to TET was again observed to be considerably lower than in bactericidal antibiotics.
In contrast, a 3-step difference in maximum fold change was observed between (an)aerobic conditions for KAN(Figure 11AB). In addition, maximum MIC values for KAN in anaerobic conditions (512 μg/mL, KAN anaerobic I & II) were 3 steps lower than for aerobic conditions (2048 μg/mL, KAN aerobic) (Figure 11C). However, no apparent long-term stagnation of increase in fold change or MIC can be deduced from Figure 11 for KAN. A concluding deduction from Figures 9-12 is that absence of oxygen might limit the maximum concentration of KAN that E. coli can grow in in absence of oxyR, although this is not certain until the evolution experiment would be continued for KAN under anaerobic circumstances. Additionally, presence or absence of oxygen does not seem to interfere with maximum values for AMO, ENR and TET.
Comparing the rate of adaptation to increasing antibiotic concentrations
The number of generations and time in days relative to the increase in MIC and fold change were compared between (an)aerobic conditions for each separate antibiotic, in order to investigate the role of oxygen in the rate of AMR acquisition in ΔoxyR.
Based on Figures 9, 10, and 12, presence of oxygen appears to correlate with a faster adaptation to increasing AMO, ENR, and TET concentrations in terms of number of generations.
However, a small bias should be recognized for Figures 9B-12B. When antibiotic concentration was increased multiple days in a row for both conditions, as is the case mainly in the beginning stage of the experiment, an observation of less generations would be merely due to the lower generation time of anaerobically grown ΔoxyR.
25 With this in mind, significance of the observed difference between (an)aerobic conditions was
determined with a Wilcoxon matched-pairs signed rank test. Adaptation rate in terms of generations was found to be significantly different between conditions (p<0.05) for AMO (Figure 9B, p=0.0137), ENR (Figure 10B, p=0.0008), and TET (Figure 12B, p=0.0156). No significant difference between conditions was found for KAN (Figure 11B, p=0.2813). However, a slower adaptation is seen for KAN in time in days in anaerobic conditions.
Data analysis
To analyse significance of ΔoxyR’s adaptation rate to increasing concentrations of all antibiotics, the number of generations corresponding to every step when mutants were able to grow in an antibiotic concentration for the first time was determined for both conditions. A mean concentration value was taken for the datasets in anaerobic conditions. Generations at each step were compared with a Wilcoxon matched-pairs signed rank test in GraphPad, and data point were compared only when a shared antibiotic concentration was available in both conditions. P-values lower than 0.05 were considered significant.
Antibiotic-dependent growth of antibiotic resistant strains (post evolution experiment)
After observations of limited growth in positive controls of AMO resistant (AMOr) and KAN resistant (KANr) ΔoxyR cultures during MIC measurements in comparison to better growth at (high) antibiotic concentrations, the cultures adapted to maximum concentrations (2048 μg/mL) were plated out on LB agar (Figure 13). Growth on LB agar plates was limited for AMOr and KANr cultures. Accordingly, these were diluted and plated out on antibiotic plates as described previously in the Methods section (Figure 14). Insufficient diluting made quantification by CFU/mL impossible (Colony count > 300), though comparison by sight showed no colony formation at AMO/KAN concentrations of 2048 μg/mL. Furthermore, growth of AMOr and KANr cultures was observed on agar with 64 μg/mL AMO/KAN, and agar without antibiotics. From this second set of results in Figure 14, no dependency on higher concentrations of antibiotics can be associated with the resistant strains.
Figure 13: LB agar plates with antibiotic resistant ΔoxyR E. coli cultures. A: Kanamycin resistant culture, B: Enrofloxacin resistant culture, C: Amoxicillin resistant culture, D: Tetracycline resistant culture. All resistant cultures were streaked on the plates after the final incubation of the evolution experiment.
26
Figure 14: LB agar plates with amoxicillin (AMO) and kanamycin (KAN) resistant ΔoxyR E. coli cultures. On plates A (LB agar +2048 μg/mL AMO), B (LB agar + 64 μg/mL AMO) and C (LB agar), AMO resistant culture was plated in a 100x dilution. On plates A(LB agar +2048 μg/mL KAN), B (LB agar + 64 μg/mL KAN) and C (LB agar), KAN resistant culture was plated in a 10x dilution. Resistant cultures were selected from a frozen sample of the last day of the evolution experiment.
27
4. Discussion
Intent was to establish in what extend oxyR deficiency in E. coli affected the de novo acquisition of AMR against β-lactam, fluoroquinolone, and aminoglycoside bactericidal antibiotics, and a
tetracycline bacteriostatic antibiotic. Subsequently, the role of oxygen on acquisition of AMR was investigated in absence of oxyR to address controversy regarding the radical-based theory. Altogether, this study suggests the influence of class specific resistance mechanisms, as well as possible involvement in unknown extend of oxygen in the acquisition of AMR to AMO, ENR, and KAN bactericidal antibiotics in oxyR deficient E. coli mutants. Findings leading up to this conclusion are discussed in the following section.
As a first insight on the acquisition of AMR in ΔoxyR E. coli, it appears that the mutant strain can adapt to a substantially higher concentration of bactericidal antibiotics irrespective of class, than to the bacteriostatic antibiotic TET (Figure 8). This indicates that a possibly (partly) shared mechanism may underlie acquisition of AMR to bactericidal antibiotics as opposed to bacteriostatic antibiotics. This mechanism is presumably mutation based, as the ability to grow in a higher antibiotic
concentration is usually associated with genetic adaptation (Händel et al., 2014). Accordingly, adaptation to TET is known to be limited mainly to adaptations at gene expression level (Grossman, 2016). The maximum concentration of TET that ΔoxyR adapted to was the same as in WT E. coli (van der Horst et al., 2011)
In addition, graph patterns for ENR, AMO and KAN vary, and are suggested to be a reflection of a diversity of class specific resistance mechanisms involved in de novo acquisition of AMR (Blair et al., 2015; Händel et al., 2014). An example is the described sudden increase in the ENR ΔoxyR culture that was not observed for AMO and KAN cultures. The occurrence of a S83L point mutation in gyrA, or mutations in parC or gyrB are suggested to underlie this observation for ENR as these are
mutations known to largely contribute to ENR resistance (Händel et al., 2014). Certainly, these speculations are in need of support and therefore, conducting genetic mutation profiles for ΔoxyR mutants at multiple points during the evolution experiment is suggested for all antibiotics with whole genome sequencing.
In the light of involvement of oxygen in the acquisition of AMR, a second important set of insights contributing to the final conclusion in this study was provided by comparison between aerobic and anaerobic data (Figures 9-12). Oxygen did not appear to interfere with the maximum achievable AMO, ENR and TET concentrations in which ΔoxyR could grow. This maximum was again considerably lower for TET than for bactericidal antibiotics in both aerobic and anaerobic conditions. Contrary to this, absence of oxygen possibly limited the maximum achievable KAN concentration wherein ΔoxyR could grow, or at least had presumable influence on adaptation rate in days. A critical note must be given that the evolution experiment should be continued under anaerobic circumstances to ensure that the maximum concentration of KAN was indeed reached, for no stagnation in maximum value was observed.
Furthermore, absence of oxygen was suggested to possibly lower the adaptation rate to increasing AMO, ENR and TET concentrations. The difference in adaptation rate in terms of generations is possibly not as significant as apparent at first sight, as was explained in the results section.
Accordingly, a suggestion could be to equalize number of generations per incubation by shortening of incubation time in aerobic conditions to check if rate of acquiring AMR still differs significantly between (an)aerobic conditions in ΔoxyR. Nevertheless, it should be recognized that the observed
28 difference in adaptation rate is possibly big enough to speculate about absence of oxygen in ΔoxyR mutants as a contributing factor besides experimental set up.
Together, findings led to the interpretation that absence of the oxyR gene in presence of oxygen may be contributing to the acquisition of AMR for AMO, ENR, and KAN in some extend. This would leave room for speculation about a common mechanism contributing to AMR involved around oxygen, as was previously proposed (Händel et al., 2016b; Hoeksema et al., 2018; Kohanski, Dwyer, et al., 2010). The extend of this involvement could vary depending on mechanism of action, given the distinct results for aerobic and anaerobic KAN ΔoxyR cultures in comparison to AMO an ENR cultures.
Furthermore, a methodological obstruction found in this study was the simple inability to dissolve antibiotic powder into more concentrated stock solutions prevented growth of the ΔoxyR strain in higher antibiotic concentrations than 2048 μg/mL, rendering this a maximum measurable MIC value by technical limitation. In addition, acquisition of resistance to all antibiotic was only documented once in aerobic conditions, and twice in anaerobic conditions, thus in demand of repetition. Lastly, it is difficult to make a solid assumption on the effect of oxyR deletion and its function in AMR
acquisition, given the current lack of comparable data for a WT E. coli strain. Still, comparison of
ΔoxyR acquisition of AMR in this study to acquisition of AMR in WT E. coli as in earlier studies does
suggest possibility of a lower adaptation rate of E. coli to AMO, ENR and TET in absence of oxyR (Händel et al., 2016; Hoeksema et al., 2018).
Results of this study are contributing to a better understanding of the complex cellular processes underlying AMR, and give a beginning insight on the importance of the oxyR gene in this process. More specifically, if in the future significance of downstream oxyR involvement in ROS inflicted damage can be further specified in relation to AMR, there could be a more certainty about the need for focus on the radical-based theory for the development of novel antibiotics. Enhancing ROS production by targeting the oxyR gene with new antibiotics might lead to increased killing efficacy, though caution must be taken, for low ROS concentrations might actually induce mutations driving resistance (Van Acker & Coenye, 2017). However, this reasoning might currently be a few steps too far ahead and further investigation is highly recommended.
The need for this must be stressed, given the large financial and health related burdens AMR currently puts on the world due to lack of effective antibiotics (CDC Global Health, 2017; Ventola, 2015). Therefore, future research in ΔoxyR deficient bacteria should point out if focus should be shifted to oxidative damage in the stress response to antibiotics.
Accordingly, a suggestion to further study the involvement of oxidative damage as a common contributing mechanism to AMR would be to test for cross resistance of ΔoxyR resistant cultures to other bactericidal antibiotics, like Hoeksema et al. (2018) did for WT E. coli strains. If MIC values are increased for other antibiotic classes than the culture is resistant for, despite never being exposed to these other antibiotics, this would suggest that a common mechanism partially contributes to AMR irrespective of class. Furthermore, this observation in combination with absence of oxyR might point towards involvement of ROS.
To further enforce the effect of oxyR deletion on interference with intracellular ROS, possibility of another transcriptional regulator of oxidative stress partially taking over its role should be prevented. One of such transcriptional regulators in the soxS gene, responding to superoxide (Händel et al., 2016b). ΔsoxS mutants were investigated before in an evolution experiment by Händel et al. (2016), and the gene was found to affect acquisition of AMR at least in some extend. Therefore, a suggestion is that a mutant with a double deletion for soxS and oxyR is constructed and investigated in an evolution experiment.
29 An additional suggestion is measuring of ROS levels at various steps during the evolution experiment in (an)aerobic ΔoxyR and WT strains, in order to find correlations between oxyR function, differences in ROS levels and acquisition of AMR, although caution should be taken regarding difficulty and bias in current methods for measuring ROS levels (Van Acker & Coenye, 2017).
As a concluding remark, the results of this study are in demand of further investigation of the effect of the oxyR gene in relation to antimicrobial resistance. There could be hypothesized that presence of oxygen in absence of oxyR has possible association with the acquisition of resistance, though the exact effect and significance of oxyR, oxidative damage, and ROS on AMR remain in need of further investigation.
30
5. Appendices
5.1 Materials
Sterilization of liquid solutions, agar and liquid media was performed by autoclaving for 20 min at 121 °C, unless stated otherwise. Powder bases for media and liquid solutions were diluted to required total volumes in Milli-Q ® (MQ) ultra-pure water. For suspending powder bases into MQ, a magnetic stirrer was used. Antibiotic powder bases were dissolved in different solutions and/or MQ as listed.
5.1.1 Composition and preparation of growth media
LB agar
Table 8: LB agar powder composition.
Compound Concentration (g/L)
NaCl 10
Yeast extract 5
Bacto-tryptone 10
Agar 20
After preparation with powder base composition as listed in Table 8 and sterilization, LB agar was kept at 55 °C until further use. For pouring agar plates, LB agar was taken out of the stove and cooled down until comfortable to the touch. Subsequently, LB agar was used either directly, or antibiotics were mixed in in required concentrations if necessary. 20 mL of LB agar was poured per plate in a laminar airflow cabinet, and plates were afterwards stored at 4°C.
LB liquid medium
Table 9: LB liquid powder composition.
Compound Concentration (g/L)
NaCl 10
Yeast extract 5
Bacto-tryptone 10
Liquid LB medium was prepared with powder base composition as listed in Table 9, sterilized and stored at room temperature until further use.
31
Evans medium (55mM glucose)
Table 10: Evans medium powder components.
Compound Concentration (g/L) NaH2P4*2H2O 15.60 KCl 0.75 MgCl2*6H2O 0.25 NH4Cl 53,49 Na2SO4 142.04 Titriplex 0.39
Table 11: Evans medium liquid components.
Compound Volume for 1L (mL)
CaCl2*2H2O 1
Trace elements 5
MQ was added to the Evans medium powder base (Table 10) to approximately three quarters of the total required volume before adding liquid solutions as listed in Table 11. pH was set to 6.9 with 4M NaOH and 1M HCl, as prepared according to section 5.1.2. Before measuring, the sensor of the pH meter was rinsed with MQ until pH 7.0, and NaOH and HCl were added while mixing. The Evans medium was then brought to the desired total volume with MQ and sterilized, where after 55mM sterile glucose was added prepared as described in section 5.1.2. Evans medium with 55mM glucose was stored at room temperature until further use.
5.1.2 Composition and preparation of liquid solutions
Calcium chloride, glucose, NaOH and HCl solutions
Table 12: Compositions for calcium chloride, glucose, NaOH and HCl solutions.
Solution Compound Molarity (M) Molar weight
(g/mol)
Calcium chloride CaCl2*H2O 0.02 147.01
Glucose C6O12H6 1 180.16
NaOH NaOH 4 40.00
1 40.00
HCl HCl 1 36.46
0.1 36.46
All solutions were prepared with powder compounds as listed in Table 12, before being sterilized and stored at room temperature. Sterilization of glucose solution was performed by autoclaving for 10 minutes at 110 °C.
