Skip to main content

Predicting antibiotic resistance

Final Report Summary - PAR (Predicting antibiotic resistance)

Executive summary:

Ever since effective antibiotic treatment was first introduced almost seventy years ago, it has been a remarkable success story and antibiotic treatment is without a doubt the single most important medical procedure or treatment ever invented as measured by the reduction in human morbidity and mortality. However, the intensive use and misuse of antibiotics have resulted in antibiotic resistance among almost all major human pathogens, and the resulting loss of therapeutic options might generate a post-antibiotic era where present and future medical advances are negated. As resistant bacteria dramatically reduce the possibilities of treating infections effectively and increase the risk of complications and fatal outcome for patients with severe infections, it represents a major public health concern and economic problem both within the European Union (EU) and globally.

The scientific rationale and uniting concept for this research program was to generate knowledge to describe and predict the rate and trajectory of resistance evolution and to use this knowledge to reduce or prevent it from occurring. To this end, we developed quantitative models that captured these complex dynamics, obtained relevant values for the important parameters and validated the models by testing them in suitable in vitro and in vivo models. To obtain molecularly based and quantitative descriptions of the emergence of antibiotic resistant pathogens, we integrated three levels of bioscience: biochemistry, cell biology and population genetics to better understand and predict the process from: biochemistry of the resistance mechanism -> impact on bacterial physiology -> effects on bacterial survival, persistence and transmission and bacterial ability to appear within individuals and transit. To obtain the most useful and relevant experimental knowledge possible, the above problems were addressed using a combination of in vitro studies, animal experiments and clinical studies in real-life settings. This integrative approach allowed us to obtain experimental data that is clinically relevant and addresses key important public health issues.

Except for a slight initial delay during recruitment of suitable Doctor of Philosophy (PhD) students and postdocs, the project has progressed as planned and the promised objectives, deliverables and milestones have largely been reached. The work has been performed by a group of researchers that has included 13 principal investigators and approximately 30 postdocs / researchers, PhD students and technical personnel that were partly or fully funded by PAR. Our group of researchers has had 4 major research meetings during the project period, allowing us to continuously plan and evaluate our progress, integrate the different disciplines and keep all involved people updated on the progress of the project. In addition, we have had several telephone conferences, email contacts and smaller research seminars.

The research constellation has during the project-period April 2010 to March 2013 published the following where PAR-funding has been acknowledged: 135 refereed publications (primary research publications and reviews) and 202 dissemination activities of the obtained results at international / national meetings (conference presentations and posters) and book chapters. In addition, our results have been presented and discussed in several other fora, including EU, EMEA, IMI and in contacts with industry. We have actively used our existing research networks to inform about our aims and results, to obtain feedback on our work and to seek new and relevant collaborations.

In the following text, we summarise the major findings of this research and their expected impact on health in Europe.

Project context and objectives:

One the most urgent problems regarding European and Global Public Health is the emergence and spread of antibiotic resistant bacteria. The PAR project has addressed this important issue and generated knowledge that can be used to stem both the emergence and transmission of resistant bacteria. The prevalence of antibiotic resistance among many human pathogens is increasing at alarmingly rapid rates all over the world and this increase in resistance is due both to the use / misuse of antibiotics and the spread of particular clones of resistant bacteria within and between countries. To optimise attempts to reverse the rise in resistance and to prevent the development of further resistance, intervention strategies need to be formulated and implemented at several levels. The urgency of the resistance problem makes the development of experimental and theoretical tools and methods to understand and predict (and by inference prevent) the development of antibiotic resistance a high priority.

The main objective of this project was to describe and predict the dynamics of antibiotic resistance development at the level of the drug target, the microbe and the host. Current knowledge about the mechanisms and evolutionary constraints that drive the emergence and survival of resistant strains is incomplete and this lack of knowledge means that we do not know how the various parts of the puzzle fit together, i.e. how do we connect antibiotic use patterns -> bacterial resistance mechanisms -> bacterial physiology and fitness -> bacterial survival within a host -> bacterial spread between hosts. This project has generated part of the knowledge needed to answer this question by developing novel conceptual and experimental approaches. We have also explored several approaches both with regard to new principles for rationally choosing drug targets and drugs with minimised risk of resistance development.

These questions have been studied in a number of bacterial species where antibiotic resistance generates problem with regard to treatment, including Salmonella typhimurium, Escherichia coli, Staphylococcus aureus, Pseudomonas aeruginosa, Mycobacterium tuberculosis. In addition, a number of different antibiotic classes are examined (e.g. fluoroquinolone (FQs), beta-lactams, aminoglycosides, macrolides, deformylase inhibitors, fusidic acid and rifampicin).

Project results:

The work performed was divided into 11 different work packages (WPs) and using a combination of genetics, molecular biology, biochemistry, clinical microbiology, epidemiology and clinical studies we address three major questions.

(A) Formation and emergence of resistant bacteria (WPs 1 - 5).
(B) Survival and persistence of resistant strains (WPs 6 - 8).
(C) Transmission of resistant strains (WPs 9 - 11).

Below are summarised some of main results obtained from each of the 11 WPs. More detailed descriptions of results can be found in separate reports for each of the deliverables.

WP1: The intrinsic resistome

Objectives:
(1) To identify the chromosomally encoded elements which contribute to the observed phenotype of intrinsic resistance in relevant bacterial pathogens.
(2) To predict the ability of bacteria to develop mutation-driven resistance by loss of or over-production of gene products. These mutations define targets for the search of drugs that could be used in combination with antibiotics to reduce resistance.
(3) To study species-specific differences in antibiotic mechanisms of action, their physiological basis and role in resistance development. In particular, we examined why the identical resistance mechanism/mutation might have very different physiological effects in different species.

Deliverables:
- D1.1: Report on the determinants that contribute to intrinsic resistance of P. aeruginosa, E. coli, and A. baumannii.
- D1.2: Define new targets for the search of drugs that reduce the antibiotic susceptibility of P. aeruginosa and E. coli.
- D1.3: Report on drug specific and species-related differences in antibiotic mechanism of action.
- D1.4: Report on the physiological basis of drug- and species-related differences in antibiotic mode of action.

General summary:
This WP has been aimed at identifying and defining all genes in bacteria that have the potential to cause resistance when mutated, overproduced or removed. Several bacterial species (P. aeruginosa, A. baumanii and M. tuberculosis) have been examined. The project has overall been completed. Some major findings are:

(1) Identification of genes in P. aeruginosa that can confer resistance when altered or overproduced. For example, genes involved in cell wall, LPS and transport have been identified as well as regulator genes.
(2) Several resistance mutations have been introduced to several species of Mycobacteria to examine how the same resistance mutation in different species might confer very different physiological effects. By reconstructing bacterial sequence polymorphisms in drug binding pockets our results can explain differences observed in species-specific drug susceptibility patterns and species-specific resistance phenotypes conferred by mutational alterations.

WP2: Stable mutators

Objectives:
(1) Study stress resistance, nutritional competence, spontaneous mutation rates and capacity to generate mutations conferring resistance to different antibiotics of evolved E. coli laboratory strain lineages having different mutation rates.
(2) Characterise molecular mechanisms responsible for the small modulations of mutation rates in those strains.
(3) Verify presence of mutations resulting in modulation of mutation rates identified in E. coli laboratory strains in the genomes of E. coli natural isolates. Study stress resistance, nutritional competence, spontaneous mutation rates and capacity to generate mutations conferring resistance to different antibiotics of selected E. coli natural isolates.
(4) Use presence of these mutations as markers, to identify environmental niches / environmental conditions that play a role in the shaping of mutation rates.

Deliverables:
- D2.1: Report on the molecular mechanisms responsible for small modulations of mutation rates.
- D2.2: Report on the rate and trajectory of evolution to resistance as a function of mutation rate, initial resistance genotype and phenotype, selection pressure, and genetic bottleneck.
- D2.3: Report on the genetic alterations that accompany the evolution of resistance and fitness compensation.
- D2.4: Report on the epidemiological and phenotypic study of E. coli natural isolates concerning the studied phenotypes.

General summary:
The overall aim of this WP was to examine the importance of mutator bacteria for the rate of resistance development in laboratory and clinical settings with particular attention to the effect of small modulations of the mutation rates. Some major findings are:

(1) Demonstration that weak/strong genetic mutators can be enriched during selection for antibiotic resistance, especially under conditions where resistance develops via several mutations of small effect.
(2) Certain antibiotics (e.g. b-lactam antibiotics) can promote mutagenesis via RpoS-mediated MutS protein depletion. Thus, subinhibitory concentrations of b-lactam antibiotics induce the RpoS regulon, cause ROS production and induce PolIV-dependent mutagenesis in Escherichia coli.
(3) We have characterised the molecular mechanisms responsible for modulations of mutation rates and discovered that two MATE-family efflux pumps modulate the mutation rate, antibiotic resistance and oxidative stress in hypermutator strains of E. coli.

Highlights:
Demonstration that b-lactam antibiotics can promote mutagenesis via RpoS-mediated MutS protein depletion. Low subinhibitory concentrations of b-lactam antibiotics induce the RpoS regulon, cause production of reactive oxygen species and induce PolIV-dependent mutagenesis in Escherichia coli. As this mutagenesis can generate mutations conferring antibiotic resistance, it should be taken into consideration for the development of more efficient antimicrobial therapeutic strategies (Nature Communications 2013, Guiterres et al.).

WP3: Inducible mutators

Objectives:
(1) To identify the antibiotic classes that can influence the inducible mutation rates.
(2) to identify the main determinants responsible for the antibiotic-based inducible mutagenesis;
(3) to determine if ribosomal inhibitors and ribosomal mutations (ram mutations) can by virtue of induction of mistranslation cause an increase in mutation rates;
(4) to determine at what extent the inducible mutagenesis of human commensal and pathogen isolates can be affected by low antibiotic concentrations;
(5) to determine whether patients being treated for tuberculosis with FQs are at greater risk of hypermutability than patients treated with conventional regimens.

Deliverables:
- D3.1: List of the currently used antibiotics with capacity to increase mutation rate in E. coli/S. typhimurium and report on which genes are involved in antibiotic-induced mutagenesis.
- D3.2: Report on the influence of ribosomal accuracy on mutation frequency and on the mechanisms involved in mistranslation-induced mutagenesis.
- D3.3: Recommendations on antibiotic regimes to avoid induction of mutagenesis and resistance development.
- D3.4: Report on the risk of FQ based treatments for tuberculosis increasing the rate of drug resistant strains and establishment of a consensus protocol to identify the presence of hypermutable bacteria in sputum of patients undergoing treatment with antibiotics.

General summary:
The general aim of this WP was to experimentally study mechanisms of antibiotic mediated mutagenesis induction and to determine the importance of antibiotic-inducible mutators for the rate of resistance development in laboratory, and clinical settings. Questions addressed include the importance of radical-scavenging ensymes, stress- and mistranslation-inducing antibiotics in the generation of antibiotic resistance. Some major findings are:

(1) We have shown that many antibiotics can weakly induce mutation rates in a RecA-dependent way. 11 out 12 antibiotics conferred a 2- to 17-fold induction of mutation rates. Similar to the studies reported in WP2, these results imply that subinhibitory levels of drugs ought to be avoided since apart from directly allowing selection of resistant mutants these drug concentrations may also stimulate formation of the mutants.
(2) A very interesting clinical study (REMoxTB clinical trial) has examined if FQs used for treatment of M. tuberculosis might speed up resistance development to other drugs. Results obtained suggest that FQs are not responsible for an increase in mutation rates.

Highlights:
FQ antibiotics have been shown in vitro to have an effect on bacterial resistance generation at sub-MIC concentrations. The REMoxTB clinical trial has allowed a test of the idea that FQ can increase mutation rates in bacteria of treated patients. 50 strains from patients with recurrence have been obtained and subjected to whole genome sequencing. As this is a regulatory trial, we will not be able to break the blind until the end of this year, but preliminary results suggest that FQ are not responsible for hypermutability in practice. This is the first study that addressed this question in any bacterial species and is of particular importance in the context of tuberculosis.

WP4: Recombination and gene amplification

Objectives:
(1) to determine the significance of gene amplification mechanisms and recombination rates for resistance in the laboratory and in clinical settings;
(2) to identify if currently used antibiotics are able to influence recombination rates;
(3) to identify the main genetic determinants responsible for inducible recombination;
(4) to test whether inhibition of these determinants may abolish or reduce the antibiotic-based induction of recombination;
(5) to determine if stable hyper-recombination is present in clinical isolates and serves to facilitate antibiotic resistance development;
(6) to determine how the combined effect of gene amplification, mutation and recombination may drive the evolution of antibiotic resistance and how they are modified by antibiotic treatments.

Deliverables:
- D4.1: A list of the currently used antibiotics with capacity to increase recombination and/or gene amplification rates in E. coli, S. typhimurium and P. aeruginosa.
D4.2: Report on the evolution of genes encoding ESBLs based on gene amplification-mutation-recombination.
General summary:
The general objective of this WP was to determine the significance of gene amplification mechanisms and recombination rates on the rate of resistance development in laboratory and clinical settings. This WP has progressed very well and all objectives and deliverables have been achieved. Some major findings are:

(1) demonstration that gene amplification is common in bacterial populations and a significant driver of antibiotic resistance development both in vitro and in clinical settings;
(2) demonstration that the order of appearance of mutations can influence the trajectory of resistance evolution;
(3) determination of the costs of gene amplification and stability/segregation rates of amplified arrays;
(4) demonstration that only FQs (among a number of antibiotic classes tested) have the effect of increasing intrachromosomal recombination of homologous deoxyribonucleic acid (DNA) sequences in E. coli.;
(5) demonstration of how antibiotic resistant small colony variants (SCV) of S. enterica escape the fitness costs associated with resistance by gene amplification of the mutated resistance gene followed by mutation. Apart from its clinical implications with regard to SCV and difficult-to-treat recurring infections, these results demonstrate how rapidly natural selection can occur even when it acts on very small fitness differences;
(6) construction of a number of genetic gadgets (e.g. transposons) and strains that are available for the scientific community and that can be used to study gene amplifications and their biological roles.

Highlights:
(1) Demonstration that the order of appearance of mutations can influence the trajectory of resistance evolution: lon mutants can evolve high-level resistance faster than a lon+ strain by allowing selection of frequent spontaneous acrAB duplications that generate high-level resistance. This is because amplification of acrAB co-duplicates the nearby lon gene and as increased gene dosage of wild type lon is deleterious, only bacteria with an inactivated lon gene can develop high-level resistance by acrAB duplication. As a result, when the acrAB duplication occurs first it represents a dead-end evolutionary pathway towards further increased antibiotic resistance via arcAB amplification.
(2) We have demonstrated how antibiotic resistant small colony variants (SCV) of S. enterica escape the fitness costs associated with resistance by gene amplification of the mutated resistance gene followed by mutation. Apart from its clinical implications with regard to SCV and difficult-to-treat recurring infections, these results demonstrate how rapidly natural selection can occur even when it acts on very small fitness differences (Mol Microbiol 2011, Pränting et al., see Commentary).

WP5: Horizontal gene transfer

Objectives:
(1) to ascertain the factors and limitations of plasmid transfer among clinical strains in vitro and in vivo;
(2) to initiate a high through put plasmid characterisation platform to determine the full set of plasmid sequences from clinical bacteria;
(3) to determine the mechanism of Tn402 and ISCR elements with respect to moving antibiotic resistance genes and DNA hot spots;
(4) to characterise the role of fused class 1 gene cassettes and the role of genetic factors in integron expression and gene movement;
(5) to determine rates of HGT via transformation for S. pneumoniae in culture and biofilm;
(6) to determine rates of HGT between intracellular bacteria during growth in vitro and in cell lines.

Deliverables:
D5.1: Initiate and establish a large scale high-through put plasmid characterisation platform using clinically derived and environmental bacteria.
- D5.2: Examine the risk factors and species/strains limitations on plasmid transfer.
- D5.3: Examine the roles of individual genes in the Tn402 transposition and whether Tn402 is species specific. Determine whether Tn402 and Tn21 respond to DNA hotspots on bacterial plasmids and chromosomes and to examine the role of GC content.
- D5.4: Determine the mechanism of ISCR mobilisation and whether rolling circle intermediates can transfer independently of plasmids.
- D5.5: Examine the role of sub-MIC concentrations of antibiotics on transformation, conjugation and rolling circle transposition.
- D5.6: Determine the rate of horizontal exchange of the pbp2X gene in planktonic cultures between the same species of S. pneumoniae and between different species of streptococci; that can colonise the oro- and nazo-phayrnx, e.g. S. oralis (donor) and S. pneumoniae (recipient), and in biofilm cultures exposed to variants of CSP.
- D5.7: Determine the rate of HGT between intracellular bacteria during growth in vitro and in cell cultures.
- D5.8: Determine the rates of plasmid transfer for E. coli and Klebsiella in mice.

General summary:

This WP was aimed at determining rates and efficiencies of conjugation and transformation in vitro and in mouse models with clinical strains and naturally occurring plasmids and to examine whether sub-lethal concentrations of antibiotics influence HGT rates. Some major findings are:

(1) In a series of papers we have demonstrated the global dissemination and analysed the genetic properties of the very problematic NDM-1 gene that confers resistance to all beta-lactam antibiotics. It has been shown that the NDM-1 resistance mechanism is community-acquired and closely linked to poor sanitation and polluted drinking water in East Asia and often transmitted by Medical tourism and also by tourism in East Asia. These studies have had very important political impact and have led the Indian government to draw up an antibiotic prescription and control policy for the first time in its history.
(2) We have identified several of the factors that underlay the recent success of the NDM-1 gene. Firstly the NDM-1 gene now has a strong promoter that function in a wide range of different species including Eukaryotes. Secondly, numerous different insertion sequences are found upstream of the NDM-1, which in turn gives access to many new genetic positions in numerous plasmids via homologous recombination events. Finally, the NDM-1 is often present in a plasmid that appears to increase fitness of the host bacterium.
(3) Demonstration that sepiolite, a dietary coadjuvant in animal feed, promotes the direct horisontal transfer of antibiotic resistance plasmids between bacterial species.

Highlights:
Our work has shown that NDM-1 has been constructed by a recent rare genetic fusion event that is likely mediated by a new ISCR element ISCR27. The recent success of NDM-1 in transmitting globally appears to be a result of its high expression, high recombination ability and association with certain plasmids. This work has resulted in several highly cited publications.

WP6: Effects of resistance on bacterial physiology, fitness and virulence

Objectives:
(1) Determine using a set of defined resistant mutants how the resistance mechanisms affect fitness (growth and survival within and outside hosts) of several pathogenic bacterial species under laboratory conditions and in experimental animals. By experimentally identifying the combinations of antibiotics and bacteria where the fitness costs of resistance are the highest we can decide which antibiotics should be used in order to maximise reversibility of resistance as well as to predict reversibility in clinical situations.
(2) To determine the physiological reasons for why fitness is reduced in antibiotic resistant mutant bacteria.

Deliverables:
D6.1: Report on the relationship between specific combinations of resistance mutations and bacterial fitness in vitro.
- D6.2: Report on the relationship between specific combinations of resistance mutations and bacterial fitness in vivo.
- D6.3: Establishment of a consensus protocol for rapid, relevant and precise measurements of fitness costs associated with resistance mutations.
- D6.4: Report on the correlation between fitness of resistant strains and their patterns of altered gene expression.

General summary:

This WP has very broad and basic aims and we have experimentally determined how different types of antibiotic resistances affect fitness (within and outside hosts) of several pathogenic bacterial species and determined the physiological reasons for fitness-reductions in antibiotic resistant bacteria. The WP has progressed very well and all objectives and deliverables have been reached. We have studied the fitness costs of several types of resistances including efflux pumps, target alteration and modifying ensymes to several antibiotic classes (beta-lactams, diarylquinolone, tetracyclines, FQ, macrolides, aminoglycosides and antimicrobial peptides) in bacterial species such as E. coli, S. enterica, S. aureus and P. aeruginosa. A general conclusion that has emerged is that practically all types of resistances are associated with a reduced fitness that can be observed as a slower growth rate in vitro and/or in animal models. This WP has also been very productive with regard to publications and several papers covering different aspects of fitness costs have been published. Some major findings are:

(1) Demonstration for several different classes of antibiotics (e.g. aminoglycosides, FQs, beta-lactams) and bacterial species (E. coli, S. typhimurium, S. aureus M. tuberculosis and P. aeurginosa) that various resistance mechanisms are generally associated with a reduced fitness that can be observed as a slower growth rate in vitro and/or in animal models.
(2) Demonstration that resistance towards mecillinam, tigecycline and different antimicrobial peptides confer costs due to alterations in cellular physiology and cell wall structure.
(3) Demonstration that the constant over-expression activity of efflux pumps increases the cellular energy burden and rate of respiration. This cost might be compensated by the activation of an alternative respiratory chain.

Highlights:
Demonstration that resistance to several antimicrobial peptides (LL-37, CNY100HL and wheat germ histone) can, in contrast to common belief, develop rapidly with small fitness costs and with cross-resistance to several AMPs. Resistant mutants can have a competitive advantage over the wild type strain at AMP concentrations similar to those found near human epithelial cells. These results suggest that resistant mutants could both be selected de novo and maintained by exposure to our own natural repertoire of defence molecules.

WP7: Compensation to reduce fitness costs

Objectives:
(1) to determine the extent and type (i.e. reversion, intragenic or extragenic suppression) of genetic compensation for resistant strains that evolve in the absence / presence of antibiotic in laboratory medium, cell cultures, and experimental animals.
(2) to determine the mutation spectra of compensatory mutations for bacteria that evolve in mice and laboratory media.
(3) to examine how bacterial physiology is altered in response to compensatory mutations;
(4) We will examine the validity of one of the previously proposed in-vivo examples of compensatory evolution.

Deliverables:
D7.1: Report on the genetics of in vitro and in vivo fitness compensation for different types of resistance mutations.
- D7.2: Report on the degree of fitness compensation in resistant strains, in vitro and in vivo.
- D7.3: Establishment of a consensus protocol for rapid, relevant and precise measurements of the compensation of fitness costs associated with resistance mutations.

General summary:

This WP is directly related to WP6 in that mutants with fitness costs examined in WP6 will be tested here for their ability to reduce the costs by compensatory evolution. The aims have been to determine if and at what rates the fitness costs of resistance can be reduced by mutation or physiological mechanisms (e.g. induction of chaperones) in vitro and in vivo. We have shown that compensatory evolution will always (!) occur in response to costly resistance mechanisms and that such evolution quite often can restore fitness to essentially wild type level without loss of resistance. Thus, resistance to, for example, aminoglycosides, FQs, tetracyclines, beta-lactams and deformylase inhibitors can be reduced by second-site mutations. The genetic analysis show that there exist a number of different mechanisms by which compensation can occur, indicating that compensatory evolution will be rather rapid process. This WP has been productive and several papers have been published. Some major findings are:

(1) demonstration that compensatory evolution will (without exception) occur in response to any costly resistance mechanism;
(2) demonstration that there exist a number of different mutations and mechanisms by which compensation can occur, explaining why compensatory evolution can be such a rapid process;
(3) demonstration that secondary mutations in RNA polymerase genes of ripampicin-resistant and ribosomal ribonucleic acid (rRNA) of aminoglycoside resistant M. tuberculosis are fitness-compensatory mutations;
(4) demonstration in P. aeruginosa that the cost of over-expressing efflux pumps might be compensated by the activation of an alternative respiratory chain.

Highlights:
(1) A long unresolved question is how significant compensatory evolution is in clinical isolates of bacteria. In a landmark study (Mol Microbiol 2010, Shcherbakov et al.), it was shown that compensatory evolution occurs in drug-resistant clinical isolates of M. tuberculosis and that compensatory evolution contributes to the spread of drug-resistant tuberculosis disease.
(2) During study of the compensatory mechanism of resistance to deformylase inhibitors, it was shown that mutations in translation initiation factor 2 could compensate for the lack of a formyl group on Met-tRNA. Detailed study of these IF2 mutants has provided a general model for the action of GTP-binding proteins, and how GTP and other ligands may modulate their activities (EMBO J 2011, Pavlov et al.).

WP8: Target choice to reduce resistance development

Objectives:
(1) Examine the 'multiple targets concept', suggesting that drugs that act at multiple targets have a reduced risk of resistance development.
(2) Examine the 'high pleiotrophic cost concept'. Thus, targets will be examined with respect to their potential of conferring a high pleiotrophic biological cost when mutated towards resistance. That is, potential resistance mechanisms should severely reduce pathogen fitness by interfering with bacterial physiology at many different levels. It is expected that defects associated with resistance to an antibiotic activity that disturbs more than one cellular process are very difficult to reverse by additional mutations, effectively trapping a resistant mutant in a low fitness state.

Deliverables:
- D8.1: Establishing consensus protocols and documents for the pharmaceutical industry and researchers that allow assessment of the likelihood of resistance development in clinical settings.
- D8.2: Report on the feasibility of reducing antibiotic resistance development by choosing targets and drugs that act pleiotrophically and confer high costs.

General summary:
This WP was assumed to examine various antibiotic derivatives obtained from Pharmaceutical companies. Unfortunately they have been unable to supply these compounds because of discontinued research on these compounds. Thus, we have instead examined commercially available antibiotics (e.g. fuszidic acid, tetracyclines, antimicrobial peptides) for their resistance potential and relation to both the 'multiple target' and 'high pleiotrophic' cost concepts. We have shown that the basic concepts are valid and that resistance development generally is slower for drugs that have multiple targets and/or high pleiotropic fitness costs. Some major findings are:
(1) Demonstration for commercially available antibiotics (e.g. fusidic acid, tigecycline and mecillinam) that resistance development / compensatory evolution generally is slower for drugs with multiple targets and / or high pleiotrophic fitness costs.

WP9: Resistance in clinical settings

Objectives:
(1) to determine the relationship between resistance (rate of appearance and fitness costs) and transmission of M. tuberculosis in clinical settings;
(2) to validate the clinical importance of compensation by examining its occurrence in human patients;
(3) to determine how the mechanism, rate and trajectory of compensatory evolution is influenced by genetic background and environment (e.g. transmission between hosts);
(4) to determine the extent to which trimetoprim resistance in E. coli can be reversed by reduced antibiotic use. We will analyse a unique clinical intervention study where the use of an antibiotic has been significantly curtailed in a community in an attempt to reduce the frequency of resistant bacteria.

Deliverables:
- D9.1: Report on fitness cost of drug resistance from information gathered on the frequency of different mutations in clinical strains and in laboratory generated drug resistant M. tuberculosis mutants.
- D9.2: A report providing a comprehensive assessment of the impact of mutations of M. tuberculosis and its relationship to the genetic background. Information on the effect of both genetic background and specific resistance mutation on transmission.
- D9.3: A report indicating compensatory mutations identified by whole genome sequencing and associated with amelioration of fitness deficit following transmission of drug resistant M. tuberculosis (rifampicin, isoniasid) between patients.
- D9.4: Report on the feasibility of reducing antibiotic resistance by reducing antibiotic use and on why intervention studies might not be successful and the explanations for absence of reversibility.

General summary:
This WP is aimed at determining the relationship between resistance and transmission in three clinical studies and to link basic biology with resistance epidemiology. We have characterised the molecular mechanisms involved in clinical drug resistance and determined resistance-associated fitness cost both in real life and under experimental settings. Furthermore, to determine if resistance can be reversed by reduced antibiotic use we have performed and completed the analysis of a unique clinical intervention study where the use of trimethoprim was significantly curtailed in a community in an attempt to reduce the frequency of resistant bacteria. Some major findings are:

(1) Demonstration that the combination of a low cost of trimethoprim resistance and co-selection with other resistance markers caused a very slow reversibility in community intervention study where trimethoprim use was significantly curtailed in an attempt to reduce the frequency of resistant E. coli bacteria.
(2) Demonstration in clinical isolates of M. tuberculosis that epistatic interactions between different drug resistance mutations and the strain genetic background modulate phenotypic levels of drug resistance.

Highlights:
In a unique clinical prospective intervention study we showed that a severe reduction (85 %) in the use of trimethoprim for treatment of uncomplicated UTIs did not result in a reduction of trimethoprim resistant E. coli in spite of 2-years of intervention. The reasons for this disappointing result were likely a combination of the low biological cost of trimethoprim (providing a weak driving force for reversibility) and co-selection for other resistance markers that were present in the trimethoprim resistant clones (JAC 2010, Sundqvist et al).

WP10: Transmission of resistant mutants

Objectives:
(1) to develop novel animal experimental models for study of transmission;
(2) to use several animal models to examine in controlled experimental situations the relative rates of transmission of susceptible and resistant bacteria;
(3) to determine the role of key breakpoints in the transmission cycle, such as desiccation and innate defences tolerance, for the transmission of susceptible and resistant bacteria.

Deliverables:
D10.1: Establishment of novel animal models for rapid and precise measurements of the transmission rates of bacteria.
D10.2: Report on the effect of different resistance mutations on transmission rates as measured in animal models and on survival on surfaces.

General summary:
WP10 was aimed at experimentally determining how different types of antibiotic resistances affect survival and transmission rates and to develop several animal experimental models for study of transmission. We have set up several different transmission studies, including studies of ESBL E. coli between ducks, human-human contact transmission of E. faecium, P. aeruginosa transmission between C.elegans and surface-human transmission of S. aureus. The experiments have progressed well and the objectives and deliverables have been reached. Some major findings are:

(1) Established a transmission model to study fecal-oral transmission of ESBL E. coli between ducks and demonstrated that FQ resistant bacteria are outcompeted by susceptible bacteria in the absence of drug but that very low concentrations of ciprofloxacin in the ducks' drinking water are sufficient to allow the resistant strain to outcompete the susceptible variant.
(2) Demonstration that individuals differ largely in their ability to transmit resistant E. faecium clones by finger-finger transmission.
(3) Demonstration that mutants over-expressing MDR efflux pumps increase transmission rates between C. elegans nematodes.

Highlights:
We have set up a new model system for studies of transmission of E. coli between dabbling ducks. Using this model we have shown that transmission occurs very rapidly between ducks (hours) kept in the same pond and that different E. coli strains differ drastically in their ability to establish intestinal colonisation and transmit. Furthermore, we have shown by competition experiments between susceptible and FQ resistant E. coli that the resistant bacteria are outcompeted by susceptible bacteria in the absence of drug but that very low concentrations of ciprofloxacin in the ducks' drinking water are sufficient to allow the resistant strain to outcompete the susceptible variant (manuscripts in preparation). These findings further support our notion (see highlight WP11) that extremely low antibiotic concentrations in the environment can maintain the resistant bacteria.

WP11: Integration and mathematical modelling

Objectives:
(1) to develop models for how different microbe population parameters (population sise, mutation rate, fitness, strength of selection, genetic drift etc. affect the emergence of resistant bacteria in a population;
(2) to test these models under well-defined conditions using parameter values obtained from WPs 1 - 9.

Deliverables:
D11.1: Generation of models that can be used for predicting the impact of mutators (stable and inducible) on resistance development.
- D11.2: Generation of models that can be used for predicting the impact of fitness costs of mutations and mobile genetic elements and compensatory evolution on resistance development.
- D11.3: Generation of models that can be used for predicting the impact of transmission rates on resistance development.
- D11.4: Generation of a generic integrated model that incorporates the rate of formation of resistant mutants, fitness costs, compensatory evolution and transmission rates of a pathogen in a host population to allow prediction of resistance development.

General summary:
This WP has aimed to develop a quantitative setting and rigorous modelling to understand and interpret the results obtained from some of the other WPs. One main aim has been to study how different microbe population parameters (e.g. population size, mutation rate, fitness, strength of selection etc.) affect the emergence and maintenance of resistant bacteria in a population and to integrate this with the obtained experimental results. This WP has been very successful and provided us with explanatory models for much of the experimental results, in particular when it comes to understanding and interpreting the impact of fitness costs on the success of bacterial clones. Thus, in many of the published papers modelling has provided us with suitable tools to interpret the results and direct us towards new experiments. The objectives and deliverables have been fulfilled. Some major findings are:

(1) We have generated mathematical models that describe how different microbe population parameters (e.g. population size, mutation rate, fitness, strength of selection) affect the emergence of resistant bacteria in a population.
(2) These models have been applied to explain experimental results obtained in other WPs (in particular WPs 5-7) and guide new experiments.

Highlights:
In a widely discussed publication, we have shown for several clinically used antibiotics that extremely low antibiotic concentrations (1/200 of the MIC of a susceptible bacterium) can select for resistant bacteria (PLoS Pathogens 2011, Gullberg et al.). These antibiotic concentrations are similar to the concentrations found in natural environments, suggesting that antibiotic release into the environment might be a significant contributor to the emergence and maintenance of resistance and emphasise the importance of introducing measures to reduce antibiotic pollution. This publication has generated extensive interest from the public health sector, physicians, environmental agencies, agriculture organisations and media.

Potential impact:

The results produced in the project have generated great interest from the public health sector, physicians, environmental agencies, agricultural organisations and media. Thus, we have as representatives of the PAR project been invited to several seminars and discussions organised by these various organisations where we have presented our findings and their implications. The obtained results also have general life science implications with regard to our understanding of molecular evolution, adaptation, genetics and physiology in bacteria as well as for virulence and transmission in various types of host organisms. Most importantly, our results have several medically relevant applications.

Thus, our final results have:

(1) Generated parameter values for most of the bacterial factors that are important for resistance development (i.e. rate of emergence, persistence and survival, and transmission). These parameter values are of value for researchers, industry and the public health sector since they will allow more qualified predictions about how rapidly resistance will emerge and spread in a population.
(2) Generated a deeper understanding of how different levels in the complex pathways leading from drug sensitivity to drug resistance and high fitness are integrated. These results are of great value when it comes to understanding which drug targets to choose as the most optimal with regard to slow rates of resistance development.
(3) Provided the experimental knowledge required to model and perform risk assessments for the development and spread of resistance to a given antibiotic. Based on our mathematical models and determinations of emergence, survival and transmission for individual antibiotics and bacteria, we can make improved risk assessments.
(4) Provided the knowledge base required to develop novel diagnostic test systems for bacteria and compounds with a high risk of resistance development; e.g. high mutation rate, low fitness cost, efficient compensation etc. We have developed methodological tools and protocols to measure the parameters that are of importance in increasing the risk of resistance evolution.
(5) Generated model systems and measurements of how rapidly susceptible and resistant bacteria will transmit between infected hosts. These results are of importance when modelling transmission of resistance and evaluating effects of interventions to reduce spread of antibiotic resistance.
(6) Generated knowledge and strategies to reduce the rate of resistance of development by exploiting novel drug targets and drugs. A key concept to slow resistance development is that resistance is associated with a high pleiotropic cost. Our results in this area identify which types of targets are the best with regard to slow resistance evolution.
(7) Created predictive tools for industry and regulatory agencies for pre-clinical and clinical development of novel antibiotics. The combination of various experimental models and model development have generated a set of protocols and approaches that can be applied to any drug development programme.

These results have been presented in 135 publications and 202 dissemination activities (i.e. seminars and talks at conferences), generating a very strong research impact in the field of antibacterial agents and resistance development. In particular, our very high output in terms of publications has made most researchers in the field aware of the importance and experimental possibilities of making qualified predictions and assessments of how rapidly resistance will evolve depending on drug target, bacterium and selection pressure. Furthermore, with regard to these results, academic researchers as well as the pharmaceutical industry have shown great interest in utilising these tools and methods to assess risks of resistance development for new drugs in development (see point 7 above). As an illustration of this, five of the partners in the PAR project are now participants of a stage-2 proposal within the 8th call of the Innovative Medicines Initiative, ND4BB TOPIC 3: 'Discovery and development of new drugs combatting gram - negative infections'. A central part of the microbiology program within this large drug discovery program deals exactly with the subject matter of PAR, namely making predictions about risks of resistance. Thus, we will utilise much of the PAR-generated concepts and methodology within this new IMI project and apply them in the context of the pharmaceutical industry and drug discovery.

Two other specific findings generated within PAR are also especially noteworthy. First, we have shown in a widely noticed and discussed publication that extremely low concentrations of several clinically used antibiotics (1/200 of the MIC of a susceptible bacterium) can select for resistant bacteria (PLoS Pathogens 2011, Gullberg et al.). As these antibiotic concentrations are similar to the concentrations found in many natural environments due to anthropogenic activities, our findings suggest that antibiotic release into the environment might be a significant contributor to the emergence and maintenance of resistance. This work emphasise the importance of introducing measures to reduce antibiotic pollution and various agencies (environmental, water sector) have seriously begun evaluating the potential benefits of introducing various purification methods at sewage water plants to reduce the levels of pharmaceutical residues, such as for example antibiotics.

A second very important result is the demonstration (which have been published in a series of papers, e.g. Lancet Infectious Diseases 2011, Walsh et al; Antimicrobial Agents and Chemotherapy 2012, Toleman et al) of the global dissemination and analysis of the genetic properties of the very problematic NDM-1 gene that confers resistance to all beta-lactam antibiotics. This work has shown that NDM-1 has been constructed by a recent rare genetic fusion event that is likely mediated by a new ISCR element ISCR27. The recent success of NDM-1 in transmitting globally appears to be a result of its high expression, high recombination ability and association with certain plasmids. We have also shown that the NDM-1 resistance mechanism is community-acquired and closely linked to poor sanitation and polluted drinking water in East Asia. It is often transmitted by Medical tourism and by tourism in East Asia. These studies have had very important political impact and have led the Indian government to draw up an antibiotic prescription and control policy for the first time in its history (see Nature 2012, 489:192).

List of websites: 1