Final Report Summary - ANTIRESDEV (The effects of antibiotic administration on the emergence and persistence of antibiotic-resistant bacteria in humans and on the composition of the indigenous microbiotas at various body sites)
This project investigated the effects of antibiotic administration on the microbial communities which inhabit healthy volunteers and on the development of resistance to the antibiotics in members of these microbial communities.
Administration of minocycline had little effect on the cultivable microbiota of the skin, anterior nares or intestinal tract of the volunteers. In contrast, it induced a transient increase in the proportion of alpha-haemolytic streptococci in the oral cavity. There was also a transient increase in the proportion of minocycline-resistant bacteria in the anterior nares and oral cavity.
Administration of amoxicillin had little effect on the composition of the cultivable microbiota of the four body sites with the exception of the oral cavity where it induced a transient reduction in the proportion of Streptococcus salivarius. No increase in the proportion of amoxicillin-resistant bacteria was detected at any of the body sites.
Administration of ciprofloxacin had little effect on the cultivable microbiota of the skin, nose or oral cavity. However, in the intestinal microbiota, there was a transient decrease in the proportions of Escherichia coli and bifidobacteria. There was no increase in the proportion of ciprofloxacin-resistant bacteria in the nasal or skin microbiotas. However, the proportions of ciprofloxacin-resistant bacteria did increase in the intestinal tract and in the oral cavity.
Administration of clindamycin had little effect on the composition of the cultivable microbiota of the skin or nose. However, there were long-term changes in the intestinal and oral microbiotas. The proportions of clindamycin-resistant bacteria increased at all body sites except the skin.
None of the four antibiotics appeared to create selection pressure for the emergence of a number of pathogens of major clinical importance including Staphylococcus aureus, Pseudomonas aeruginosa and Acinetobacter spp.
Culture-independent analysis revealed that administration of minocycline, clindamycin and ciprofloxacin, but not amoxicillin, had a profound effect on the oral and intestinal microbiomes. However, after one month, the oral microbiome was similar to that which existed prior to antibiotic administration. One month after the administration of amoxicillin or minocycline, the intestinal microbiome was similar to that found prior to antibiotic administration. In contrast, a return to pre-administration values took 4 months in the case of ciprofloxacin and between 4 months and one year for clindamycin.
Three new DNA microarrays for the detection of antibiotic resistance genes in Acinetobacter baumannii, Gram-positive bacteria and Gram-negative bacteria were developed. Use of these micro-arrays revealed that many strains of Escherichia coli in the intestines are resistant to a number of antibiotics and that healthy people may harbour in the nose and on the skin staphylococci with an assortment of antibiotic resistance genes.
Those volunteers colonized with S. aureus were usually found to carry one single strain type and the administration of minocycline or amoxicillin had no measurable effect on the resistance or fitness of the S. aureus present. Many resistances and resistance combinations were found not to adversely affect the fitness of S. aureus.
A DNA micro-array for the detection of mobile genetic elements was developed and successfully evaluated. One particular mobile genetic element (having a Tn916-like core) appeared to be responsible for the spread of a large number of different antibiotic resistance genes. A novel genetic element on which antibiotic and antiseptic resistance genes are linked was found during the study. This shows that exposure to one type of antimicrobial agent could also promote the spread of resistance to a completely different type of antimicrobial agent.
Project Context and Objectives:
One of the main problems confronting medicine in the 21st century is the increasing prevalence of bacteria that are resistant to antibiotics. This, combined with a shortage of new antibiotics under development, means that our ability to treat infectious diseases is rapidly diminishing. Because the over-use and misuse of antibiotics in many countries is likely to continue, new antibiotic-resistant bacterial strains will continue to emerge. In order to develop new means of limiting resistance development and the further transmission of antibiotic-resistant strains, it is essential that a concerted effort is made to gain a greater understanding of how the administration of particular antibiotics affects the antimicrobial susceptibility of potentially-pathogenic and commensal members of the indigenous microbiota of humans. The genetic basis of such resistance needs to be clarified together with the persistence and mode of transmission of antibiotic-resistant strains and the biological cost to the organism arising from its conversion to a resistant phenotype. It is also important to assess the ecological impact of antibiotic administration on the indigenous microbiota.
The concept underlying the ANTIRESDEV project is that the administration of a particular antibiotic to humans results in the disruption of the indigenous microbiota and the emergence of antibiotic-resistant bacteria – the identities of which depend on the nature of the antibiotic. Different antibiotics, with their distinctive antimicrobial spectrum of activity and pharmacokinetics, induce the emergence of different resistant populations and the dynamics and persistence of such resistance is antibiotic-dependent. A thorough understanding of these processes can be achieved only by ascertaining the genetic basis underlying antibiotic resistance, the means by which the genes responsible are transmitted and an appreciation of the biological cost of resistance to the resistant bacterium.
The approach we have adopted in the ANTIRESDEV project is to carry out longitudinal intervention studies using antibiotics with different modes of action, antimicrobial spectra and pharmacokinetic properties and to identify the antibiotic-resistant strains that arise from their use. The studies will involve healthy volunteers and the dynamics (emergence and persistence) of the antibiotic-resistant strains isolated from them and the ecological impact of antibiotic administration on the indigenous microbiota will be subjected to statistical modelling. The genetic basis of resistance in these isolates will be identified using state-of-the-art microarrays and the genetic elements mediating the spread of some of these will be further investigated in order to help characterise their likely mode of transmission. The biological cost to the organisms of developing resistance to one or more antibiotics will be determined as this impinges on the dynamics and transmission of antibiotic resistance throughout the population. In parallel with these culture-based studies, a culture-independent approach (454 pyrosequencing) will also be used to determine the composition of the oral and faecal microbiotas (including not-yet-cultivated members), the full complement of antibiotic resistance determinants (the “resistome”) in these communities and the effect of antibiotic administration on these.
The scientific and technological objectives of the project are to use a multidisciplinary approach involving clinical, pharmacological, genetic, bacteriological, molecular biological and epidemiological expertise to study the impact of different antibiotics in selecting resistance among pathogenic and commensal members of the indigenous microbiota of humans. This overall objective will be achieved by the following sub-objectives:
Objective #1: to identify and quantify those cultivable antibiotic-resistant bacteria that emerge during the administration of four different antibiotics to healthy volunteers
Objective #2: to investigate, using 454 pyrosequencing, the full complement of resistance determinants (the resistome) in the cultivable and not-yet-cultivable microbiota and the effect on this of antibiotic administration
Objective #3: to ascertain the dynamics of resistance development and the persistence of antibiotic-resistant strains
Objective #4: to compare the pattern of antibiotic resistance development induced by different classes of antibiotics).
Objective #5: to investigate the ecological impact of antibiotic administration on the cultivable indigenous microbiota
Objective #6: to investigate, using 454 pyrosequencing, the ecological impact of antibiotic administration on the cultivable and not-yet-cultivable indigenous microbiota
Objective #7: to characterise, using state-of-the-art microarrays, the antibiotic resistance determinants in the isolates obtained from the clinical studies
Objective #8: to determine the mobility of the resistance genes detected in the isolates obtained from the clinical studies
Objective #9: to ascertain the biological cost of antibiotic resistance in a number of clinically-important organisms isolated during the clinical studies
Objective #10: To disseminate the project findings to the clinical and scientific communities and to the general public and to ensure access to the ANTIRESDEV databases to enable future studies by other groups
Objective #11: To use the ANTIRESDEV findings to inform health care decision makers of some of the factors influencing the emergence and persistence of antibiotic-resistant bacteria following the administration of particular antibiotics and thereby provide opportunities for them to implement appropriate policies concerning antibiotic use
This study of the effects of antibiotic administration on the microbial communities inhabiting humans and on the development of antibiotic resistance in these communities was conducted using healthy volunteers from the general public in South West London, UK. It was designed as an open labelled parallel group clinical trial. A total of 44 healthy subjects were recruited including 31 women and 13 men aged between 18 and 40 years. They showed normal findings in their medical history and on physical examination. Written informed consent was obtained from all subjects prior to the study. The study was approved by the South West London Research Ethics Committee of St George’s, University of London and by the Medicines and Healthcare Products Regulatory Agency, UK.
15 of the subjects were allocated to a “minocycline group”, 15 to an “amoxicillin group” and 14 to a “placebo group”. The dosage used was that recommended by the British National Formulary for amoxicillin i.e. 250 mg (Amoxil capsules, GlaxoSmithKline, UK) three times daily for seven days and for minocycline 100 mg (Aknemin capsules, Almirall S.A.) twice daily for five days. Two doses of placebo daily for five days were given to the control group. Unfortunately, 9 subjects withdrew due to adverse effects and personal reasons including 3 from the amoxicillin group, 5 from the minocycline group and 2 from the placebo group which left 12 subjects in the amoxicillin and placebo groups and 10 in the minocycline group.
Saliva, skin, faecal and nasal samples were taken from these volunteers immediately before and after treatment, and 1, 2, 4 and 12 months after treatment. The cultivable bacteria present in these samples were identified and also the susceptibility of these bacteria to a range of antibiotics was determined. The samples of saliva and faeces were also sent to other partners so that they could use a different method of analysis, based on the DNA present, to identify the bacteria (cultivable and not-yet-cultivable) in the samples.
One week of minocycline administration to the volunteers appeared to have an impact on the microbial communities in saliva. The proportion of α-haemolytic streptococci increased from 60% to 73% while the proportion of Streptococcus salivarius decreased from 40% to 27%. This could be due to the susceptibility of S. salivarius to minocycline. However, larger reductions in the proportion of this species took place spontaneously in the absence of antibiotics in the placebo group. Consequently, the reduction in the proportion of S. salivarius could be coincidence. Also, minocycline treatment appeared to disturb the composition of the microbial communities in the intestinal tract. One week after treatment, the proportion of Escherichia coli decreased from 85% to 45% and there was a corresponding increase in the proportion of Enterococcus species. Minocycline treatment had no effect on the composition of bacterial communities in the anterior nares or on the skin.
The administration of minocycline resulted in a statistically-significant increase in the proportion of minocycline-resistant bacteria isolated from the anterior nares. For example, the proportion of minocycline-resistant Propionibacterium acnes increased immediately after treatment, then gradually decreased at one, two and 4 months post-treatment although the background level of minocycline-resistant P. acnes in the treatment group was much higher than in the placebo group. After one year, there were almost no minocycline-resistant P. acnes remaining. Also the proportion of facial skin minocycline-resistant P. acnes in the minocycline group increased one week after treatment, but it returned to the pre-treatment level during the subsequent sampling points. These data demonstrate that minocycline treatment is likely to induce an increase in the proportion of minocycline-resistant P. acnes. The clinical implications of these findings are that minocycline treatment, even over a short course of one week, can induce increased resistance which lasts for months. In addition, minocycline treatment appeared to impact on the proportion of minocycline-resistant bacteria in the oral cavity, especially α-haemolytic streptococci. After treatment with minocycline, the proportion of minocycline-resistant α-haemolytic streptococci increased about 10%, then deceased at the other sampling points. However, no such increase in the proportion of minocycline-resistant bacteria was found in the intestinal tract or on the skin.
Interestingly, administration of amoxicillin to the healthy volunteers also appeared to have little effect on the composition of the microbial communities inhabiting the four body sites with the exception of the oral cavity where it induced a reduction in the proportion of S. salivarius. Immediately after amoxicillin treatment, the proportion of α-haemolytic streptococci increased from 45% to 94% while that of S. salivarius decreased from 54% to 6%. S. salivarius gradually regained its pre-treatment level by two months post-treatment. Also one week of amoxicillin treatment did not appear to induce resistance to this antibiotic as demonstrated by analysis of antibiotic resistance profiles from the four body sites over a one year period. Very low numbers of amoxicillin-resistant bacteria were isolated from the nose, skin and saliva. Surprisingly, large numbers of amoxicillin-resistant bacteria were isolated from the faecal samples in both the treatment and placebo groups. In the treatment group, at pre-treatment, the average proportion of amoxicillin-resistant bacteria was 15%. After one week of amoxicillin treatment, the proportion increased to 30%, remained at this level at one month post-treatment and returned to the pre-treatment levels at 2, 4 and 12 months. In the placebo group, the average proportion of amoxicillin-resistant bacteria was high at pre-treatment and at 4 months post-treatment. However, there was no significant difference between the amoxicillin and placebo groups.
A total of 441 amoxicillin-resistant bacteria (178 aerobes and 263 anaerobes) were isolated from the amoxicillin group, 550 minocycline-resistant bacteria (300 aerobes and 250 anaerobes) from the minocycline group and more than 1000 amoxicillin- or minocycline-resistant bacteria from the placebo group. The Minimum Inhibitory Concentrations (MICs) of amoxicillin or minocycline for each group of bacteria isolated from the antibiotic-treated and the placebo groups were determined. The MICs of the amoxicillin-resistant and minocycline-resistant bacteria isolated were similar in the placebo and the test groups.
The concentrations of amoxicillin or minocycline in the faeces and saliva were measured to determine the compliance of the volunteers. Minocycline was found in both the saliva and faeces of the minocycline group and reached a peak following 7 days of administration. However, no amoxicillin was detected in either the saliva or faeces of the amoxicillin group.
Statistical analysis of the data revealed that the only statistically-significant effect resulting from the administration of minocycline was an increase in the proportional abundance of minocycline-resistant bacteria in the anterior nares immediately following minocycline administration. In some cases, there appeared to be a significant population level change in the placebo group one year after minocycline administration, however, unintentional inequalities in sampling effort cannot be ruled out. There were no clear effects resulting from amoxicillin administration with respect to amoxicillin- or minocycline-resistant bacteria at any of the body sites. In the oral cavity, the absolute and proportional abundance of S. salivarius were significantly reduced following amoxicillin administration. Analyses using the Bray-Curtis dissimilarity index in the oral cavity indicated that amoxicillin administration perturbed the oral microbiota primarily through changes in abundance.
Due to delays in the start of WP3 (as explained in the 1st periodic report) it was decided that the clinical trials involving the administration of ciprofloxacin (WP3) and clindamycin (WP4) would be conducted simultaneously. This decision resulted in the completion of both work packages during Period 2.
In work package WP3, 10 volunteers (5 males and 5 females) who received ciprofloxacin (500 mg bid) for 10 days and 10 volunteers (5 males and 5 females) in the control group were given a placebo. Within the ciprofloxacin group, one volunteer (female) was excluded 4 months after the dosing due to treatment with another antibiotic because of an infection. The remaining volunteers (9 in the ciprofloxacin group and 10 in the placebo group) completed the trial. The results presented below are based on 9 volunteers (5 males and 4 females) in the ciprofloxacin group and 10 volunteers (5 males and 5 females) in the placebo group.
In work package WP4, 10 volunteers (5 males and 5 females) received clindamycin (150 mg qid) for 10 days. Within this group, one volunteer (female) was excluded directly after the treatment due to personal reasons. The 9 volunteers who received clindamycin completed the trial. The results presented below are based on 9 volunteers (5 males and 4 females) in the clindamycin group and 10 volunteers (5 males and 5 females) in the placebo group (the same volunteers as in WP3).
Saliva, faecal, skin and nose samples were collected from all the volunteers before the administration of the antibiotic or placebo, on day 11 (i.e. immediately after the dosing), 1, 2, 4 and 12 months post-dosing. All samples were processed according to Standard Operating Procedures (SOPs). Quantitative and qualitative cultures on selective and non-selective media were performed on all the samples taken at the different time points. In addition, all samples were cultured on antibiotic-containing agar plates for isolation of resistant microorganisms. The Minimum Inhibitory Concentrations (MICs) of a range of antibiotics (ciprofloxacin, clindamycin, minocycline, amoxicillin and erythromycin) were determined for all the resistant isolates. The concentration of each antibiotic was determined in the saliva and faecal samples at each time-point.
In the oral cavity, no differences between the ciprofloxacin group and the placebo group regarding Staphylococcus aureus, Streptococcus salivarius, Candida albicans, fusobacteria, leptotrichia, prevotella, veillonella or lactobacilli were observed. However, there was a decrease of Alpha-Streptococci directly after treatment (day 11), which normalised at 2 months post-treatment. There were no differences between the placebo group and ciprofloxacin group in the proportion of resistant Alpha-Streptococci, Streptococcus salivarius, prevotella, leptotrichia and fusobacteria in the oral cavity. However, a significantly higher proportion of ciprofloxacin-resistant veillonella was observed in the ciprofloxacin group as compared to the placebo. This effect lasted throughout the whole study period, i.e. up to 12 months post-dosing. There was no difference between the ciprofloxacin and placebo groups concerning the MICs for ciprofloxacin, clindamycin and amoxicillin. Some Alpha-Streptococci, Streptococcus salivarius, prevotella, leptotrichia and fusobacteria had high MICs for minocycline and erythromycin. In the clindamycin-treated group no differences were observed between the treated group and the placebo regarding Alpha-Streptococci, Staphylococcus aureus, Candida albicans, prevotella, veillonella or lactobacilli. However, there was a decrease of Streptococcus salivarius directly after treatment (day 11) and 1-month post-treatment. The total numbers of fusobacteria decreased on day 11. In addition, there was a decrease of leptotrichia at day 11, 1 and 2 months post-treatment. There were no differences between the placebo and clindamycin group in the relative proportions of clindamycin-resistant fusobacteria and leptotrichia. There was an increase in the relative proportion of clindamycin-resistant Alpha-Streptococci, Streptococcus salivarius, prevotella, veillonella and lactobacilli in saliva in the clindamycin treated group at various time points.
In the faecal samples, no differences between the ciprofloxacin group and the placebo group were detected with regard to Enterobacteriaceae, Candida or bacteroides. However, there was a decrease of Escherichia coli directly post-dosing (day 11) that normalised at Month 2. The numbers of enterococci were stable until Month 1 and increased at Month 2, the normal level was reached at 12 months post-treatment. A decrease in bifidobacteria and lactobacilli was observed directly post-dosing (day 11). No difference in the relative proportion of ciprofloxacin-resistant enterococci, bacteroides or lactobacilli was seen. There was a significant increase in ciprofloxacin-resistant Escherichia coli at day 11. The relative proportion of ciprofloxacin-resistant Bifidobacteria showed an increase on day 11, 1 and 2 months post-treatment. No differences between the clindamycin-treated group and the placebo group regarding Escherichia coli, Enterobacteriaceae, enterococci or candida were seen. However, there was a decrease in bacteroides (day 11 and 1 month post-treatment), bifidobacteria (in all samples up to 4 Months post-treatment) and lactobacilli (day 11 post-treatment). No differences in the relative proportion of clindamycin-resistant Escherichia coli, bifidobacteria or lactobacilli were seen. There was an increase in the relative proportion of clindamycin-resistant enterococci at day 11, 1 and 2 Months post-dosing. In addition, there was a significant increase in clindamycin-resistant bacteroides at 1, 2, 4 and 12 Months post-treatment. No Clostridium difficile and very few pseudomonas and acinetobacter were recovered from the faecal samples.
At the skin sites, administration of ciprofloxacin or clindamycin to volunteers did not result in any significant changes of coagulase-negative staphylococci, Propionibacterium acnes or other bacteria at any time point or skin sample location. There was no difference between the placebo group and the ciprofloxacin group in the relative proportion of ciprofloxacin-resistant coagulase-negative staphylococci or Propionibacterium acnes at any time point and skin sample location. Administration of clindamycin did not result in any significant changes in coagulase-negative staphylococci, Propionibacterium acnes or other bacteria at any time point and skin sample location. Only minor changes in the relative proportion of clindamycin-resistant coagulase-negative staphylococci were present in the neck skin samples from the clindamycin-treated group at 1 month post-dosing.
In the anterior nares, administration of ciprofloxacin or clindamycin resulted in a decrease in coagulase-negative staphylococci directly post-dosing (Day 11) but this normalised at Month 1. There was a minor decrease in Propionibacterium acnes directly post-dosing of ciprofloxacin (Day 11) but this normalised at Month 1. No difference between the placebo and ciprofloxacin groups was observed for other bacteria in the nasal microbiota. There was no difference between the placebo group and ciprofloxacin group in the relative proportion of resistant coagulase-negative staphylococci. There was a minor increase in ciprofloxacin-resistant Propionibacterium acnes at Month 2 and at Month 12 in the ciprofloxacin group as compared to the placebo group. In the clindamycin group, there was a minor increase in clindamycin-resistant coagulase-negative staphylococci at 1, 2 and 12 months post-treatment in the clindamycin group as compared to the placebo group.
Ciprofloxacin or clindamycin were detected in faeces only on day 11 (directly after the treatment) in those volunteers who were administered the respective antibiotic. No measurable concentrations of ciprofloxacin or clindamycin in faeces were detected in any volunteer from the placebo group. No measurable ciprofloxacin or clindamycin were detected in any of the saliva samples from any groups at any time point.
There was no obvious correlation between the MICs of isolates from the saliva, faeces, anterior nares or skin from placebo group and baseline samples from any of the treated groups compared to the MICs of isolates from the samples collected from either the ciprofloxacin- or clindamycin-treated volunteers.
Statistical analysis of the data revealed that, in the case of ciprofloxacin administration, there were ambiguous effects on the absolute and proportional abundance of clindamycin-resistant bacteria in the anterior nares and intestine. It seems likely that with a greater sample size these effects could have been reported with more certainty. Clindamycin administration negatively impacted the abundance of P. acnes on the skin, the abundance of Leptotrichia spp. in the oral cavity and the abundance of Bifidobacterium spp. in the intestine. The Bray-Curtis index suggested that clindamycin administration caused significant disturbances in the oral cavity and intestine communities, which later recovered many months after the end of administration. It cannot be said with confidence that any of the effects observed at any of the body sites were attributable to the administration of ciprofloxacin.
Next generation sequencing techniques that enable analysis of the composition and function of the oral and gut ecosystem through the genetic properties of the bacteria were used to gain insights into the effects of antibiotic treatment on 1) the composition of the saliva microbiome, 2) the composition the faecal microbiome, 3) on the full repertoire of antibiotic resistance genes present in the oral ecosystem (salivary resistome) and 4) on the full repertoire of antibiotic resistance genes present in the gut ecosystem (faecal resistome). Importantly, the approach does not require any cultivation under laboratory conditions and thus complements work performed by other partners that used cultivation-dependent techniques. As it is known that only a small fraction (10-30%) of the oral and faecal bacteria can be cultivated under laboratory conditions, cultivation-independent methods offer an unbiased and comprehensive insight into the oral and faecal microbial ecosystems.
Saliva and faecal samples from healthy individuals, collected during the four clinical studies before, and at different timepoints after, the antibiotic or placebo administration were obtained. DNA was extracted from these samples and used to selectively amplify a small part of the ribosomal DNA by a polymerase chain reaction (PCR). As all bacterial lineages are believed to stem from the same ancestor, their components, such as the ribosome, share common properties. Importantly, as these bacterial lineages have evolved, unique changes have occurred in the structure of the shared components. To phrase it more exactly, ribosomal genes are found within each bacterial lineage but the unique DNA basepair order is indicative of the originating bacterium. Therefore, by looking at the DNA basepair order, one can predict what bacterium it originated from, and the frequency with which it is found is related to the proportional abundance in the community. Importantly, next generation sequencing techniques have allowed us to establish the DNA basepair structure of ~10.000 unique ribosomal DNA fragments for each sample, thereby revealing what species are present in the saliva and faecal samples and in what relative abundance.
Saliva samples were assessed by a 16S rRNA gene amplicon sequencing approach. The microbiomes from the placebo groups showed stable species richness and diversity throughout the year of the study. The samples obtained from the same individual showed high phylogenetic relatedness. The baseline and week 1 samples from the same individual in the placebo groups were highly similar. Some fluctuations in microbial composition were observed between more distant timepoints. Samples from individuals that were exposed to antibiotics (all but Amoxicillin) showed reduced species richness directly after the administration of antibiotics (week 1) and a fast recovery of microbiome diversity (day 30) thereafter. In the Amoxicillin group there was no effect on species richness or microbiome diversity after the antibiotic administration period. However, this antibiotic was associated with strong individual responses, resulting in increased phylogenetic distances between the baseline and week 1 samples compared to the other groups (Placebo HP and Minocycline). Administration of Minocycline, Ciprofloxacin or Clindamycin resulted in distinct qualitative changes in the microbiome at week 1 (clusters of the week 1 samples in unweighted UniFrac plots), while on day 30 no difference from the baseline was observed. Certain taxa and phylotypes (OTUs) showed significant shifts in relative abundance after the administration of antibiotics.
Faecal samples were assessed in the same way as described above and the microbiomes from the placebo groups showed stable species richness and diversity throughout the year of the study. The samples obtained from the same individual showed high phylogenetic relatedness. The baseline and week 1 samples from the same individual in the placebo groups were highly similar. Some fluctuations in microbial composition were observed between more distant timepoints. Faecal samples from the Minocycline group showed a significant decrease in species richness up to day 30. A shift in composition was observed immediately after exposure to Minocycline and was recovered by day 30. Exposure to Amoxicillin had no significant effect on the diversity of the faecal microbiome. Shifts in microbial proportions (quantitative changes) were observed immediately after exposure to the antibiotic and showed a recovery by day 30. Exposure to Ciprofloxacin had a strong and prolonged effect on the diversity of the faecal microbiome: even one year following antibiotic exposure the species richness and the diversity index were significantly lower than at the baseline. Pronounced qualitative and quantitative compositional shifts were observed by day 30 and a recovery in composition – by day 120. Exposure to Clindamycin had a strong effect on the diversity of the faecal microbiome that was still significant on day 120. A qualitative microbial shift in composition was observed until day 120, with the most pronounced shift present on day 30. Only one year following exposure to clindamycin did the microbiomes show a recovery in diversity and composition.
The saliva and faecal samples were also assessed for changes in their antibiotic resistance gene repertoire. DNA extracted from saliva and faecal samples taken before and after antibiotic treatment of four individuals was analysed by deep metagenomic shotgun sequencing. Subsequently, sequence data were analysed for the presence of known antibiotic resistance genes.
Deep metagenomic sequencing of microbial DNA from saliva generated good baseline characteristics of antibiotic resistance gene repertoires for the oral microbiome of four individuals participating in the study. A great diversity of genes conferring antibiotic resistance was found in the oral microbial community, albeit generally at low abundance. The most abundant antibiotic resistance protein classes found in saliva samples in the four individuals analysed included bacitracin efflux pump BcrA, Macrolide efflux pumps CarA, OleB, SrmB, TlrC, VgaA, and VgaB, and class A penicillin binding protein pbp1a. Importantly, abundant resistance proteins were not ubiquitously present among all four individuals, but were found abundantly in two individuals: subject 10 (HP) and subject 15 (KI) and found at lower abundance in the other two individuals. Interestingly, individuals in whom antibiotic genes were found in higher abundance displayed a wider array of antibiotic resistance genes at higher abundance. Whether or not this is indicative of the presence of bacteria carrying multiple resistance markers is not known.
For three of the four antibiotics tested, a significant increase was observed in distinct antibiotic resistance proteins. In most cases, genes were present at low levels at the baseline of the intervention. In a number of cases, antibiotic resistance genes were not detected at baseline but were detected after the antibiotic intervention. The most likely explanation is that the genes were present at levels below the detection limit. On the whole, selective enrichment under the conditions tested appeared to be modest and impacting low abundant genes. Importantly, selective enrichment of antibiotic resistance genes does not directly imply a possible functional relationship as an increase may also be the consequence of increased abundance of bacteria by competitive advantage not linked to these antibiotic genes while carrying these resistance genes in their genome. Care should be taken not to draw firm conclusions from individual cases described in detail through resistome analysis. The methodology applied shows that it may be a valuable tool to examine the effects of selective pressure by antibiotics on resistance parameters in the oral community as a whole. For future work it would be of importance to include gene hotspot mutation analysis providing a more comprehensive insight.
Deep metagenomic sequencing of microbial DNA from faeces generated good
baseline characteristics of antibiotic resistance gene repertoires for four individuals participating in the study.
The most abundant and ubiquitous antibiotic resistance genes found in faecal samples encoded bacitracin efflux pump BcrA, Macrolide efflux pumps OleB, SrmB, TlrC, VgaA, VgaB and MacB, ribosomal protection proteins tetPB and tetQ and vancomycin resistance operon VanRA – VanRE. The most apparent individual differences included a 10-fold higher abundance of Macrolide-Lincosamide-Streptogramin B efflux pump Aac6I / Aac6Ic / Aac6Ie / Aac6If / Aac6Ig in one individual, and a five- to ten-fold higher abundance of TetQ in another individual. Overall, resistance abundance appeared to be somewhat lower in samples from the Karolinska intervention as compared to the Helperby intervention. For each of the four antibiotics tested, significant increases were observed of distinct antibiotic resistance genes. In most cases, genes were present at low levels at the baseline of the intervention. For a number of cases, antibiotic resistance genes were not detected at baseline but were detected after the antibiotic intervention. The most likely explanation is that the genes were present at levels below the detection limit. Care should be taken not to draw firm conclusions based on individual cases described in detail through resistome analysis. The methodology applied shows that it may be a valuable tool to examine the effects of selective pressure by antibiotics on resistance parameters in the faecal community as a whole. For future work it would be of importance to include gene hotspot mutation analysis providing a more comprehensive insight.
A total of 2863 antibiotic-resistant Gram-positive bacteria were subjected to micro-array analysis to determine which antibiotic resistance genes were present. 1479 of these were from Helperby Therapeutics in the UK while 1384 were from the Karolinska Institutet in Sweden. Of these, 2215 were aerobic and 648 were anaerobic isolates. The isolates were identified to the species level by Maldi-TOF. The main genera of identified Gram-positive bacteria were Streptococcus (1224), Staphylococcus (674), Propionibacterium (486), Enterococcus (206) and Lactobacillus (111). Various species were identified with Streptococcus pneumoniae/mitis group (550), Propionibacterium acnes (457), Staphylococcus epidermidis (454), Enterococcus faecalis (86) and Lactobacillus salivarius (48) being the largest groups of species per genus. Genomic DNA of all Gram-positive isolates was extracted and tested for the presence of antibiotic resistance genes by micro-array analysis. For this, a new DNA amplification and labelling system was developed which consisted of an amplification by phi-29 polymerase followed by a classic PCR amplification step using biotinylated primers. The system was tested using a new generation of microarray which contains 116 antibiotic resistance genes abundant in Gram-positive bacteria. The labelling system and microarray were validated with 73 reference strains from our collection, which cover 95% of the antibiotic resistance genes represented on the microarray. Of note, fluoroquinolone resistance in Gram-positive bacteria is associated with point mutations in the topoisomerase and gyrase genes. Such point mutations could not be detected by the microarray.
With regard to the amoxicillin, minocycline, ciprofloxacin and clindamycin treatment groups, a total of 73 different antibiotic resistance genes were detected in the samples from the UK and Sweden. The most frequently-detected antibiotic resistance genes were those conferring resistance to beta-lactam antibiotics [blaZ], macrolide, lincosamides and streptogramin B (MLSB) antibiotics [erm(B)], macrolides [msr, mph(C), mef], tetracyclines [tet(M), tet(K)], and pleuromutilin, lincosamide and streptogramin A antibiotics [vga(A)]. Many different aminoglycoside resistance genes were also frequent among the Gram-positive isolates. Of note, a substantial number of Staphylococcus (S. epidermidis, S. haemolyticus, S. hominis) isolates from the Helperby placebo and treated groups contained the methicillin-resistance gene mecA which confers resistance to all beta-lactam antibiotics.
In addition, a novel SCCmec element was found in methicillin-resistant S. hominis isolated from the placebo group using next generation sequencing. SCCmec typing also indicated that new types of SCCmec elements were also present in methicillin-resistant S. epidermis from the Helperby placebo group. The whole genome sequence of S. hominis also resulted in the identification of the topoisomerase genes grlA, grlB and gyrA, gyrB for the first time in this species.
While the distribution of species was comparable between samples obtained from the UK and Sweden, the relative abundance of antibiotic resistance genes differed in the isolates from both countries with a higher number of resistance genes in the UK isolates. The distribution of resistance genes was determined in detail for all Staphylococcus and Streptococcus species, since a significant number of bacteria could be isolated at each visit. These analyses could not be made for the other species due to both the low number of isolates and the absence of isolates at certain visits.
In both treatment groups (amoxicillin and minocycline) as well as in the placebo group from the UK, the abundance of antibiotic resistance genes was remarkably high, with a maximum of 100% strains carrying at least one resistance gene in the genus Staphylococcus. In volunteers who received minocycline (Helperby), the number of tet(M)-containing staphylococci increased compared to placebo and other treatment groups. In volunteers who received clindamycin (Karolinska), the abundance of antibiotic resistance genes increased from visit 2 (before treatment) to visit 3 (immediately after treatment) by 22%, the majority of these were tetracycline resistance genes, within staphylococci. In volunteers who received ciprofloxacin (Karolinska), an increase of 40% in the abundance of antibiotic resistance genes was observed between visit 2 and 3 in Staphylococcus species. The majority of these were macrolide resistance genes. This indicates co-selection of resistance genes by the use of clindamycin and ciprofloxacin. In volunteers who received amoxicillin (Helperby), no significant increase of any antibiotic resistance gene was observed. No further correlation between the abundance of antibiotic resistance genes and antibiotic treatment was observed in staphylococci, this is likely due to the high number of resistance genes already present in isolates before treatment. Within the genus Streptococcus, the abundance of strains carrying at least one resistance gene was comparable in isolates from the UK and Sweden. A slight increase in the abundance of strains carrying at least one resistance gene could be observed in both the Helperby and Karolinska treatment groups at visit 4 (one month post administration) compared to the placebo group. An increase of macrolide, lincosamide, streptogramin B (MLS) resistance genes, predominantly erm(B), and tetracycline resistance genes was also observed in the genus Streptococcus in the clindamycin treatment group at visit 4 from the Karolinska samples. Furthermore, an increase of macrolide resistance genes at visit 4 was also observed in streptococci isolates from volunteers who received ciprofloxacin, again indicating co-selection of resistance genes by the use of ciprofloxacin.
Strains with multiple antibiotic resistance genes were found within different species. A total of 295 Gram-positive isolates (10.1%) [179 from Helperby (12.1%) and 116 from Karolinska (8.4%)] contained three or more antibiotic resistance genes. The Gram-positive isolates carrying multiple antibiotic resistance genes were identified as Enterococcus, Staphylococcus and Streptococcus. Very diverse resistance profiles were observed among the different species, indicating that not a single multiple-resistant strain emerged due to the antibiotic treatment. In general, genes conferring resistance to tetracycline [tet(M), tet(K), tet(L)], to beta-lactam antibiotics [blaZ], macrolide, lincosamides and streptogramin B (MLSB) antibiotics [erm(B), erm(C)], macrolides [msr, mph(C)], were the most frequent genes found in the multiple-resistant isolates. However, The combination of tet(M)-erm(B)-mef was the most frequent in the Streptococcus pneumoniae/mitis group. The frequency of strains with multiple resistance genes was assessed for each visit, each treatment and placebo groups from both Helperby and Karolinska. No obvious correlation between treatment and abundance of multiple-resistant isolates was observed, except for a slight increase in the relative abundance of strains with multiple resistance genes in the clindamycin group when compared to the placebo group immediately after the antibiotic treatment. This is likely due to the presence of the tet(M)-erm(B)-mef genotype in the Streptococcus pneumoniae/mitis group, as erm(B) is a known macrolide – lincosamide resistance gene. Strains harbouring these genes are likely to have been selected by the use of clindamycin.
All the data were further analyzed by comparison of genotypes with phenotypes. While most of the genotypes from all tested isolates matched the phenotype, one Staphylococcus saprophyticus had no corresponding genotype. This indicates the presence of a novel mechanism in this species. The strain has been completely sequenced using IonTorrent technology.
Taken together, species identification by Maldi-TOF revealed a normal distribution and abundance of Gram-positive bacteria within the sampled groups. The analysis of the antibiotic resistance genes by microarray revealed only very subtle effects of the antibiotic treatment on Staphylococcus and Streptococcus. However, administration of minocycline clearly selected for a tet(M)-containing Staphylococcus microbiota and clindamycin selected for an erm(B)-carrying Streptococcus microbiota. Interestingly, effects such as the selection for MLS- and tetracycline resistance genes in streptococci isolated from volunteers treated with ciprofloxacin indicated the presence of co-resistance.
A Gram-negative antimicrobial resistance (AMR) gene microarray was expanded so that 159 probes/primers representing 124 AMR genes present in a broad range of Gram-negative bacteria were detectable. During the course of the study, probes/primers representing more than 90 AMR genes were validated; control strains were not available for the remaining genes. The updated microarray (AMR06) was used to detect antibiotic resistance genes in 3694 Gram-negative isolates recovered mainly from faeces and saliva from the four clinical studies.
E. coli was the most abundant Gram-negative bacterium isolated from faeces. To compare the numbers of resistance genes harboured in E. coli and non-E. coli isolates, data from WP1, 2 and the corresponding placebo group were used. E. coli generally harboured much greater number of resistances than non-E. coli isolates. In fact, the maximum number of resistance genes present in the E. coli isolates ranged from 11 to 15. E. coli harbours mobile genetic elements (MGE) such as plasmids which can encode large numbers of transferable resistance genes. Multi-resistant E. coli were found to be currently more prevalent in the individuals studied than other Gram-negative bacteria. The array results also indicated that certain resistance genes were much more prevalent than others, with blaTEM being the most prevalent across all groups, irrespective of treatment. Anaerobic bacteria were unlikely to harbour resistance genes common in aerobes, although sul2, a gene moderately prevalent in aerobes, was also detected in a small number of anaerobic bacteriaThe percentage of isolates harbouring resistances to 2 or more classes of antibiotic (i.e. multiple resistance) were similar (approximately 35%) for Minocycline (WP1), Amoxicillin (WP2) treatments and the respective placebo group. Only in the Amoxicillin treatment group did the percentage of multiple resistance isolates increase immediately after treatment. This may have been due to increases seen in blaTEM positive E. coli in this group. To determine gene linkages associated with the presence of multiple resistances, all E. coli isolates from selected participants were studied. Pulse field gel electrophoresis (PFGE) was performed and showed that isolates with identical PFGE profile generally harboured the same resistance gene pattern, even though they may have been collected from several different visits that were months apart. Often a core set of resistance genes was present, indicating loss of some resistances due to the mobile nature of the transferable elements harbouring these genes.
A further updated array (AMR07) was used for analysis of isolates received from WP3 and WP4. In volunteers receiving either Ciprofloxacin or Clindamycin, there was an increase in the percentage of resistant isolates immediately following treatment, which was not detected in the placebo group. However, only one isolate was collected from the Ciprofloxacin study at this time point. In general, there was a dramatic decrease in the percentage of susceptible isolates in the two treatment groups following antibiotic administration which was not seen for the placebo group; indicating a possible effect of antibiotic treatment on the resident microbiota. As for WP1 and WP2, isolates harbouring identical resistance genes were found at different sampling time points from both treatment groups and the placebo. Several genes were found to cluster together, indicating possible gene-linkages in these isolates. In one instance, the β-lactam gene blaACC which is known to be located in the chromosome of Hafnia alvei, and detected in an H. alvei isolate was detected in an E. coli isolate from the same participant and visit, indicating possible in vivo horizontal gene transfer.
In all treatment groups, anaerobic bacteria had much fewer detectable resistance genes, with tetQ being the most prevalent. In fact, there was a significant increase in the proportion of tetQ positive Bacteroides detected immediately after Minocycline administration, which was not apparent in the other groups. The sul2 gene, which is prevalent in aerobic Gram-negative bacteria, was present in anaerobes from all study groups.
Four isolates (ARD120; ARD289; ARD1053; ARD1258) were chosen for further analysis. Each isolate represented a cluster showing a similar antimicrobial resistance profile and identical PFGE pattern. The isolates were screened for the presence of plasmids by plasmid profiling and typed by incompatability (Inc) group. Many of the isolates possessed IncF replicon plasmids. To determine the location of resistance genes in these isolates, a directed plasmid curing approach was undertaken. In addition, an isolate naturally cured of its plasmid was also included. The four isolates chosen belonged to the most abundant multilocus sequence types (ST10, ST95 and ST648), which are predominantly found in humans. Each isolate harboured between 2 and 4 plasmids which ranged in size from approximately 150kb to 63kb. All the isolates carried an IncF replicon plasmid as well as plasmids harbouring other replicons such as IncI1, N, R, B/O and U plasmids. Maintenance of plasmids within their host was encoded by different addiction systems (pemK, vagCD, pndCA, ccdAB, relE, hoksok, srnBC).
Comparison of wild type and cured derivatives using the AMR08 microarray identified a range of AMR genes missing in the plasmid-cured strain and hence was associated with the cured IncF plasmids. IncF curing of ARD289 and ARD1053 resulted in loss of the 150kb plasmid. In the case of ARD289, all the resistance genes (aadA1, aadA2, blaSHV1, blaOXA2, intI1, ereA, qnr, sul1, sul2 & tetD) were lost. The cured IncF plasmid in ARD1053 was only associated with sul2 and tetA, while the remaining resistance genes (aadA1, aadA2, strAB, tetB & blaTEM1) were either located on the 3 remaining plasmids or on the chromosome. Both ARD289 and ARD1053 cured plasmid carried the class 1 integrase. However, in one isolate, ARD120, IncF plasmid curing of a 150 kb plasmid did not change the resistance genes (strAB, sul2 & blaTEM1) present, indicating that they were not carried on the IncF plasmid and may either be carried in the remaining 120kb plasmid or on the chromosome.
Analysis of E. coli from the Minocycline treatment group identified 6 blaCTX-M14 positive isolates from one visit from one participant. Analysis of their plasmid content showed that 1 isolate (ARD1257) lacked a 140kb plasmid present in the other isolates. To ascertain whether this plasmid carried the resistance genes (tetB, strAB, blaTEM1, aadA4, sul1, and dfrA17/19) missing from ARD1257, an IncF cured derivative of ARD1258 was constructed using pCURE2, which confirmed that the 140kb plasmid absent in ARD1257 harboured the resistance genes. In addition, 14 multiple resistant isolates shown to harbour multiple plasmids were selected to determine plasmid-mediated conjugal transfer of resistance genes to other hosts such as Salmonella. In only 6 of the 14 strains were we able to transfer a MDR plasmid to Salmonella using the methods employed in the study.
To determine the fitness costs associated with multiple resistance, four E. coli isolates (ARD120; ARD289; ARD1053; ARD1258) which were resistant to amoxicillin, ampicillin, sulphonamide and aminoglycoside, were investigated for their ability to adapt to different conditions. The wild type isolates and their plasmid-cured derivatives were characterised for their ability to survive in different environments harbouring a range of toxic compounds and metabolic effectors by utilizing the phenotype microarray (PM). In all isolates, the presence of resistance genes could be attributed to the corresponding phenotype seen using the PM plates. Similarly, curing of the plasmid in most instances resulted in a sensitive phenotype for the respective antimicrobial. The presence of these plasmids did not have a significant effect on growth in rich or minimal media. However, additional phenotypes could be attributed to the IncF plasmids harbored by each E. coli isolate. Loss of the plasmids resulted in the ability of several isolates to survive in a range of stress/toxic conditions such as tellurite, cadmium and colistin. In addition, it was demonstrated using a Galleria mellonella in vivo infection model that the natural loss of a large multi-drug resistant plasmid could result in enhanced virulence with respect to both an isolate harboring this plasmid and one artificially cured of the plasmid.
A chick colonisation model was used to study the effect of the presence of transferable antibiotic resistance elements in bacteria, on colonisation and survival in the chick gastrointestinal tract. For Enterococcus faecalis, the presence of a transposon (Tn916) harbouring an antibiotic resistance gene, somewhat increased fitness as these strains were detected up to two weeks after inoculation. In contrast, the wild type strain was lost immediately after inoculation. In contrast, E. coli human isolates were able to colonise the chick gut to high levels. The presence of a plasmid harbouring antibiotic resistance genes had little effect on colonisation. However, one strain (ARD120) was more prevalent in the chick liver and spleen than the cured derivative, indicating a possible role of the plasmid in invasion. There was also some indication of the possible transfer of resistance genes to the indigenous microbiota from the challenged strains.
A set of probes and primers were developed in silico which can be used in a microarray format to detect mobile genetic elements (MGEs). The microarray (MGE01) was printed and included the Gram-negative associated MGE probes. A total of 41 MGE probes were validated targeting genes for toxin/antitoxin addiction systems, replication and relaxases using known isolates.
In the context of the ANTIRESDEV project, “pathogens of major clinical importance” are defined as those pathogens which account for the vast majority of severe or life-threatening infections in hospitals, especially on intensive care units or in immunocompromised patients after transplantation, respectively. Deep respiratory infections (e.g. pneumonia) and bloodstream infections (e.g. sepsis) represent the entities which account for the majority of infection-related deaths, still showing mortality rates of 30-60% despite all technical and medical progress. Also, the direct and indirect economic burdens of pneumonia and sepsis are immense, accounting for approximately 50% of all costs on intensive care units. The pathogen panel of pneumonia or sepsis is well defined (15-20 different bacterial pathogens) and has been quite stable over the last decades. Nevertheless, although dealing with well-known pathogens like Staphylococcus aureus or Pseudomonas aeruginosa, antibiotic treatment, especially when empirically initiated in live-threatening conditions, turns out to become increasingly ineffective for two main reasons: i) the pharmaceutical industry has basically stopped the development of new antibiotics for economic reasons, we therefore have to deal with the antibiotics we have available and cannot expect any new drugs in the near future, ii) selection pressure as a result of extensive, and very often inappropriate, usage of antibiotics, has resulted in bacteria having acquired and accumulated multiple antibiotic resistances up to the point of pan-drug resistance with no effective treatment options left. Due to the short generation time of bacteria, the impressive repertoire of existing and continuously evolving new resistance mechanisms, as well as the ongoing worldwide spread of multidrug-resistant organisms, WE ARE LOSING THE RACE AGAINST ANTIBIOTIC RESISTANCE at the moment.
With regard to “pathogens of major clinical importance”, emphasis was placed on the following tasks:
1) A pre-developed microarray for Pseudomonas aeruginosa was updated during the project. The number of detectable resistance determinants was updated (from 148 to 350), new multiplex-PCRs established, and the whole assay optimized. The complete procedure can now be performed within 4 hours.
2) For Acinetobacter baumannii, the second important non-fermenter next to Pseudomonas, no pre-developed or commercial microarray was available. Therefore, a completely new DNA-microarray was developed within the project. The array is capable of detecting a wide variety of beta-lactamases, carbapenemases, aminoglycoside-modifying enzymes, mutations in gyrA and parC, as well as many other resistance determinants. A total of 91 different resistance genes can be amplified and analyzed in parallel.
Both, the Pseudomonas and Acinetobacter array, were validated with reference strains and compared to the results provided by the German Reference Center for Gram-Negative Pathogens, especially with respect to carbapenemases. All isolates obtained during the study were analyzed and genotyped, however, the number of isolates was rather small for these two species, and the test collective was extended with isolates provided by the ANTIRESDEV partners or strain collections for validation and testing. Due to the broad detection spectrum and low turn-around time, these arrays influenced the development of two diagnostic devices from commercial companies, which resulted in the participation of our study group in two multicenter studies with over 800 included intensive care unit patients each. Since new drugs are not in the treatment pipeline, multiplexed molecular diagnostics providing robust and fast treatment relevant pathogen and resistance information are the only way at the moment to improve patient outcome.
3) All Staphylococcus spp. obtained during the studies were genotyped using a commercial microarray (~ 300 isolates). The array covers over 350 different resistance and virulence factors, allowing a deep insight into the genetic background. In all cases, the determined phenotype could be explained by the underlying genotype detected by microarray. The most prevalent genes found were small spectrum beta-lactamases (blaZ), macrolide resistance conferring ermC, tetracycline resistance mediating tetK, and fosB1. However, none of the above mentioned antibiotics is used in the treatment of critically-ill patients. No MRSA were detected.
4) Extended-Spectrum-Beta-Lactamases (ESBLs) were found in two patients, one in the treatment and one in the placebo group. Microarray genotyping and sequencing analysis showed that these responsible genes belonged to the CTX-M15 group, which is consistent with the current prevalence in Western Europe, showing that ctx-m15 is the most prevalent ESBL gene in Europe at the moment.
A very interesting outcome of the study was the finding that very few “pathogens of major clinical importance” were isolated from the healthy volunteers investigated. Consequently, resources were devoted to the following complementary activities:
a) High-throughput species identification (ID) of the bacterial isolates obtained by the four clinical trials using state-of-the-art matrix-assisted laser deionization/ionization time-of-flight mass-spectrometry (MALDI-TOF-MS) technology; approximately 2,000 identifications were carried out
b) High-throughput antibiotic susceptibility testing (AST) of the bacterial isolates obtained by the four clinical trials using state-of-the-art semi-automated microdilution technology (VITEK2 instrument). The phenotype for at least 20 different antibiotics was determined for each isolate using the EUCAST breakpoint criteria; approximately 2,000 isolates were phenotyped for antibiotic susceptibility.
Species identification and antibiotic susceptibility testing were the basis for phenotype-genotype correlations, epidemiologic investigations as well as statistical analyzes needed for comparison with the culture-independent results obtained by next generation sequencing.
Since multidrug resistance represents the main challenge in modern infectious disease management, a great effort was made to identify the genetic resistance mechanisms underlying an observed phenotype. This constituted a crucial point in the study analysis since different genetic determinants can lead to the same phenotype, e.g. fluoroquinolone resistance can be mediated either by chromosomal point mutations in the gyrase (gyrA gene) and/or topoisomerase IV (parC gene), by efflux systems (e.g. mexAB-oprM in Pseudomonas), or plasmid-encoded (qnr or aac(6′)-Ib-cr genes). Whether the first mechanisms are vertical gene/mutation transfer from one generation to the next generation of the same species, needing the clonal spread of a resistant organism, plasmid-encoded horizontal gene transfer can spread from one species to another, resulting in different hygiene measures and epidemiologic surveillance, respectively. Therefore, all Staphylococcus spp. including MRSA, ESBL-producing Enterobacteriaceae, Pseudomonas aeruginosa, and Acinetobacter baumannii isolates were genotyped for antibiotic resistance-mediating genes using DNA-microarray technology. DNA-microarrays allow the simultaneously detection of hundreds to thousands of different genes and subsequently represent the method of choice for resistance genotyping of the study isolates. Due to the constantly growing number of newly described mutations or genes conferring resistance, DNA-microarrays need to be updated from time to time.
c) Fluoroquinolones are an important group of antibiotics. Point mutations in the gyrase gyrA and topoisomerase IV parC genes are the most common reason for fluoroquinolone resistance. Around 4% of the aerobic Gram-negative isolates showed a resistance phenotype, microarray analysis and sequencing results demonstrated that all our resistant isolates contained mutations at one of the well-defined hot-spot sites in the gyrase gene.
Antibiotic resistance genes that are capable of spread between bacteria can either be contained on a plasmid (these elements can replicate independently of the bacterial chromosome) or on integrative conjugative elements (ICE), also called conjugative transposons, which cannot replicate independently of a host replicon. The integrative elements tend to be more stable and have a broader host range than plasmids so it is important to determine what type of genetic elements act as vectors for antibiotic resistance genes.
Plasmids and ICE are made up of modules, in the case of plasmids these consist of a replication region, sometimes a transfer region, a regulatory region and an accessory region (often antibiotic resistance) that is not concerned with plasmid maintenance or transfer. ICE have similar modules but do not usually have a replication region but do have a recombination region so they can enter host replicons. In order to determine what genetic elements were present in the clinical isolates we constructed a gene array which contained genes from the most common mobile genetic elements (i.e. genes encoding, replication regions of plasmids, mobilization and conjugation regions of plasmids and ICE and integrases) and antibiotic resistance genes. As this array was designed to screen for mobile genetic elements it was called the “mobilome array”. It allowed the clinical isolates to be screened for genes involved in resistance and transfer. The major finding from this part of the work was that there were no completely novel mobile genetic elements or antibiotic resistance genes found although novel associations of different modules of mobile genetic elements and resistance genes were found and some of these were investigated in more detail.
Four isolates were chosen from the clinical studies and screened for plasmid content. To determine the location of resistance genes in these isolates, a directed plasmid curing (i.e. forcing the bacterial host to lose the plasmid) approach was used. Comparison of wild type and cured derivatives using the mobilome array identified a range of antimicrobial resistance (AMR) genes that were missing in the plasmid cured strains implying a plasmid location for these genes. The resistances included aminoglycoside (strAB, aadA1 & aadA2), sulphonamide (sul1, sul2) and β-lactamase (blaTEM1, blaOXA2, blaSHV and blaOXA) genes. However, in one isolate ARD120, curing of a 150 kb plasmid did not change the resistance genes (strAB, sul2 & blaTEM1) present, indicating that they were not carried on this plasmid and may either be carried in the remaining 120kb plasmid or on the chromosome.
In one strain, loss of a 150kb plasmid resulted in loss of the resistance genes (aadA1, aadA2, blaSHV1, blaOXA2, intI1, ereA, qnr, sul1, sul2 & tetD). In another strain, loss of a plasmid was associated with loss of sul2 and tetA, while the remaining resistance genes (aadA1, aadA2, strAB, tetB & blaTEM1) were either located on the 3 remaining plasmids or on the chromosome. A type 1 integrase was also found to be contained on a plasmid using these methods. These genes are commonly associated with cassettes of antibiotic resistance genes called integrons.
As there was a delay in receiving the clinical isolates, we tested pooled saliva samples for organisms resistant to minocycline and screened them for the presence of Tn916-like elements by PCR (the Tn916 family is a diverse group of ICE responsible for the spread of resistance to clinically-relevant antibiotics). A variety of Tn916-like elements were found to be responsible for minocycline resistance. Some of these have been previously characterized but a novel element that had not previously been characterized was identified, this was designated Tn6087. This element encodes both antiseptic and antibiotic resistance, and could transfer by transformation to Streptococcus australis. Other Tn916¬-like elements were investigated including two new Streptococcus mitis elements which encode minocycline resistance and could also transfer to other streptococcal species by transformation but not conjugation. These new elements were designated Tn6225 and Tn6227. A putative aminoglycoside resistance gene (aph) is also present in Tn6225.
A mef(E) gene (encoding resistance to erythromycin) carried by a macrolide efflux genetic assembly (mega) was found integrated into the new genetic element, Tn6226, present in the genome of an erythromycin-resistant S. mitis strain. This element was not capable of transfer under the conditions we tested.
A novel plasmid, pSI01, was identified. This confers minocycline resistance on S. infantis and could be transformed into S. australis but could not transfer by conjugation.
As the elements Tn6087, pSI01, Tn6225, Tn6226 and Tn6227 are novel, their complete DNA sequences were obtained. The main finding from the sequence analysis was that all the mobile elements consisted of different modules that allowed different antibiotic and antiseptic resistance genes to be linked on the same genetic element. The sequence analysis also showed that the majority of integrative elements are related to the prototype genetic element Tn916, even though there are numerous deletions, insertions and rearrangements. The sequencing of both the plasmids and the integrative elements also shows that mobile elements can associate with each other in different ways to generate novel genetic elements, underscoring the fact that bacteria have access to a plethora of different genetic elements which allows them to resist antibiotics and antiseptics.
As the DNA sequencing analysis indicated that there were a limited number of “core” mobile genetic elements, particularly those related toTn916 and that the roles of the various modules in these elements are not particularly well understood, we undertook a mutational analysis of Tn916 itself and some of the Tn916-like elements. Mutations have been made in the potential regulatory regions of Tn916-like genetic elements; the Tn916 orf9 mutant has decreased ability to excise and transfer. This is a surprising result which does not fit the current model for Tn916 transfer, illustrating the need for further work on these important genetic elements. Mutational analysis of the putative conjugation region of the Tn916-like conjugative transposon Tn5251 proved that orfs 13 to 19 in this region of the element are required for conjugation. The mutational analysis of Tn6087 proved that this element contained an antiseptic resistance gene and showed where this gene was located on the element.
Further work was carried out to investigate the function of the putative aph gene found in S. mitis. MICs for 10 aminoglycoside antibiotics were determined for streptococcal strains carrying aph and for Escherichia coli BL21 strain expressing high levels of Aph protein. MIC values were not significantly different compared to control strains. A 6-HIS-tagged Aph protein was purified and challenged against aminoglycosides but no activity was detected, indicating that the aph gene is not involved in resistance to the aminoglycosides tested.
The DNA sequence analysis of plasmid pSI01 showed that it contained a toxin/anti-toxin module which in other mobile elements has been shown to increase plasmid stability. In order to test this hypothesis we deleted this module. The hypothesis predicts that in the mutants the plasmids should become unstable due to loss of the addiction module. However the plasmid did not become unstable. Another possibility is that the toxins encoded by this system will kill competing bacteria; this will be tested in future work.
In conclusion the main finding from this work package was that there seems to be a limited number of “core” genetic elements that are responsible for the spread of antibiotic resistance but that these elements are capable of associating with each other in very many different ways. The mutational analysis has shown that these elements are regulated in rather unexpected ways.
Isolation of Staphylococcus aureus from the clindamycin and ciprofloxacin studies was very infrequent; only four S. aureus isolates were obtained from these particular study volunteers. Consequently, it was not possible to compare the properties of the strains in terms of resistance development or acquisition, neither regarding changing resistance levels or growth rates. In contrast, the prevalence of S. aureus in the volunteers who participated in the minocycline and amoxicillin studies was 45.4 %. During the course of the ANTIRESDEV project, no development of resistance or acquisition of resistance markers was observed in those S. aureus isolates obtained from the nose, face, neck or trunk. Furthermore, the administration of amoxicillin or minocycline did not have an effect on the growth rate or resistance levels, and no correlation between these two parameters could be found. Typing of the isolates revealed the presence of the most common clones usually found in Europe. In addition, new strain types were identified which were submitted to the S. aureus MLST database. The majority of the strains were amoxicillin resistant due to the presence of β-lactamase. No methicillin-resistant S. aureus (MRSA) strains were isolated. All volunteers were found to carry just one particular S. aureus strain, with two exceptions carrying several strain types. Three unrelated isolates sampled from volunteer V021 were characterised in more detail: Isolates ARD147 and ARD148, unrelated strains isolated before amoxicillin treatment, were compared to isolate ARD152, which was the only strain found one week after treatment and over the following two months, suggesting that it had replaced ARD147 and ARD148. Growth rate, survival under different culture conditions, including competition experiments, survival, adherence and internalisation using non-professional phagocytes, was better in strains ARD147 and ARD148. However, strain ARD152 produced slightly more biofilm and had higher beta-lactamase levels, even in the absence of antibiotics. Thus, ARD152 might have had an advantage compared to ARD147 and ARD148 during the amoxicillin treatment, which could have led to the replacement of strains ARD147 and ARD148.
Two plasmids harbouring either the resistance determinants tetK or mupA, conferring tetracycline and mupirocin resistance, respectively, were characterised with regard to their fitness cost. Plasmid stability was determined in the ANTIRESDEV isolates containing these plasmids during one week: No major changes in the proportion of bacteria positive for the plasmid were observed. No reduction in growth rate could be found in the sensitive laboratory strain RN4220 that had been transformed with the tetK and mupA plasmids, suggesting that these resistances were well adapted to minimise any fitness cost.
To identify genetic factors influencing fitness and virulence in MRSA strains, whole genome sequencing of a near-isogenic strain set differing in the type of the SCCmec element was performed. Representative SCCmec elements from community-acquired MRSA (CA-MRSA) and hospital-acquired (HA-MRSA) had been transferred into the sensitive S. aureus strain BB255. Phenotypic differences of the CA- and the HA-MRSA strains were identified by determining their resistance levels and resistance profiles, as well as their growth rate, using phenotype microarrays. RA2, a fast-growing, low-resistant CA-MRSA strain harbouring a type IV SCCmec element was found to be generally more resistant towards a broad range of substances than its HA-MRSA counterpart, RA120. These included disinfectants, antibiotics such as tetracycline, streptomycin and chloramphenicol, but also high urea concentration and high pH. Unexpectedly, whole genome sequencing revealed that, apart from the SCCmec element (25 kb), flanking regions from the donor strain had been transferred as well into BB255 during the construction of RA2. Thus, RA2 contains a total of 67 kb foreign DNA, which complicated the identification of genetic factors responsible for the observed phenotypes. So far, no additional known resistance determinant has been found in the RA2 SCCmec element. Whole genome sequencing did identify the presence of an additional isolated single-nucleotide polymorphism (SNP), not in the vicinity of the SCCmec element, in RA2. However, this SNP, located in the phosphate transport factor pstS was neither involved in resistance nor fitness of RA2, as determined by testing a RA2 derivative where a wild type variant of pstS was introduced.
Comparison of the HA-MRSA RA120 to a faster-growing, but less-resistant variant, ME51, by whole genome sequencing revealed only one SNP in the diadenylate cyclase gene dacA in the fitter but less-resistant strain. Diadenylate cyclases were recently discovered to synthesize the new second messenger cyclic diadenosine monophosphate (c-di-AMP). Re-introduction of the mutation into the highly-resistant but slower-growing strain reduced resistance and increased the growth rate, suggesting a direct connection between the dacA mutation with the phenotypic differences of these strains. Quantification of cellular c-di-AMP revealed a decrease in the c-di-AMP level caused by the mutation in dacA. Phenotypic characterisation of the strains with decreased c-di-AMP levels showed reduced autolysis, increased salt tolerance and a reduction in the basal expression of the cell wall stress stimulon. These results indicate that c-di-AMP performs a cell envelope-related signalling function in S. aureus. The influence of c-di-AMP on growth rate and methicillin resistance levels in MRSA indicate that altering c-di-AMP levels could be a mechanism by which MRSA strains can increase their fitness levels while reducing their methicillin resistance levels.
Analyses of defined model strain sets led to the identification of several new factors affecting resistance, fitness and virulence in S. aureus: i) The new regulator XdrA, influencing methicillin resistance levels and homogeneity, as well as growth rate, was shown by transcriptomic analysis to also affect gene expression, including transcription of one of the most prominent immunomodulatory S. aureus proteins, SpA. ii) Factors involved in stress response to cell wall active antibiotics and influencing resistance towards these antibiotics, such as the two-component system VraSR, were further characterised. A new player, YvqF, was identified to physically interact with VraS and to influence resistance. iii) Extensive studies on the LCP proteins of S. aureus showed the importance of these factors for resistance, fitness and virulence. Furthermore, experiments supported a role of these proteins in the synthesis of wall teichoic acids, cell envelope constituents influencing a vast array of bacterial processes, including cation homeostasis, cell separation and colonisation of the human host. Thereby, a contribution has been made to explain how these factors can influence as different tasks as reducing sensitivity towards antibiotics, maintenance of fitness and expression of virulence. iv) Analysis of a mutant library identified 17 new genes influencing beta-lactam resistance and pointed to physiological processes previously unknown to be important for resistance. Characterisation of these mutants showed that also growth rate, cell envelope stability, response to cell wall stress and virulence factor expression could be affected. Importantly, competition of these mutants in co-culture experiments was generally reduced, indicating the occurrence of secondary effects upon deletion of these genes that affect those cellular processes required for fitness. For one of the factors, SecDF, involved in protein secretion, reduced virulence was also shown.
Taken together, numerous new factors influencing resistance and vital properties of S. aureus were described, making this work a comprehensive study of genes having the potential to reduce the biological cost inflicted by certain antibiotic resistances and allowing resistant strains to maintain considerable fitness and virulence.
Twenty one strains of enterococci obtained during the clinical studies were investigated for the presence of Tn916 by microarray and PCR. One isolate was found to contain an active copy of wild-type Tn916 and was used as a donor in filter-mating experiments to a Tn916-free isolate of E. faecalis. Four of the resulting transconjugants were assayed for fitness compared to the original recipient in comparative growth assays. Each transconjugant had different biological costs. Southern Blot analysis showed that the transconjugant with the highest biological cost (largest reduction in fitness), designated T7, contained two elements whereas the other three all contained one. T7 was used in a pairwise competition assay with the recipient and was shown to have a relative fitness value of 0.8 (20% fitness cost). T7 was then allowed to evolve for approximately 200 generations in order to see if the fitness cost could be ameliorated, either by loss of the element or by compensatory mutations. At approximately 43 generations, the growth rate had returned to a similar level as the recipient (without the element). The strain was still resistant to tetracycline but had lost one of the copies of Tn916 and the remaining copy had moved location. These strains had their genomes sequenced. The fitness cost may have been ameliorated by the translocation of the element into a more neutral insertion site. T7 was then used in the in vivo fitness experiments and was shown to be more able to colonise the chicken model. The strains isolated from the faeces of the chickens have also been sequenced to determine if the genetic changes that have been observed by RFLP analysis could explain their increased fitness in the animal model compared to the original recipient strain (analysis is ongoing). In conclusion, acquisition of Tn916 by E. faecalis was found to have a biological cost which is likely to be dependent on the insertion site and that this cost can be quickly compensated for during experimental evolution. Paradoxically the acquisition of Tn916 appears to give E. faecalis a competitive advantage in terms of colonising the animal model.
To determine the fitness costs associated with multiple resistance, four E. coli isolates resistant to amoxicillin, ampicillin, sulphonamide and aminoglycoside, were investigated further. The wild type isolates and their plasmid-cured derivatives were characterised for their ability to survive in environments harbouring a range of toxic compounds and metabolic effectors by utilizing the phenotype microarray. The presence of these plasmids did not have a significant effect on growth in rich or minimal media. However, additional phenotypes could be attributed to the IncF plasmids harbored by each isolate. Loss of the plasmids resulted in the ability of several isolates to survive in a range of stress/toxic conditions. Using a Galleria mellonella in vivo infection model it was found that the loss of a large multi-drug resistant plasmid could result in enhanced virulence. A chick colonisation model was used to study the effect of the presence of transferable antibiotic resistance elements on colonisation and survival in the chick gastrointestinal tract. For E. faecalis, the presence of a transposon (Tn916) harbouring an antibiotic resistance gene increased fitness as these strains were detected up to two weeks after inoculation whereas the wild type strain was lost immediately after inoculation. In contrast, E. coli isolates were able to colonise the chick gut to high levels. The presence of a plasmid harbouring antibiotic resistance genes had little effect on colonisation. However, one strain (ARD120) was more prevalent in the chick liver and spleen than the cured derivative, indicating a possible role of the plasmid in invasion.
Antibiotic resistance is now considered to be a problem as important as climate change and terrorism. It is claimed that modern medicine is likely to end if antibiotic resistance is not adequately addressed. The results from our studies seem to contradict these predictions and suggest that clinicians in the community, at least in London, will be able to continue to use minocycline and amoxicillin for some time, without precipitating an antibiotic resistance crisis. Our findings show that in a small number of volunteers, there is already a high background of resistance in some species of bacteria, but a low level of resistance in others. Treatment with a single course of these antibiotics led to no increase in resistant bacteria in the case of amoxicillin, and to a low level, short-lived increase in resistant bacteria for minocycline. The interpretation of these findings is that in some places in some countries, and in certain situations, for example the community, treatment with some antibiotics such as amoxicillin and minocycline is unlikely to increase the number of antibiotic resistant bacteria. So these antibiotics can continue to be used carefully in these situations. Our findings do not diminish the threat of antibiotic resistance to modern medicine in other places such as hospital intensive care units, in other countries and with the use of other antibiotics.
These findings will be disseminated as follows:
a) Summary of the findings will be shown on the ANTIRESDEV website to disseminate to the general public.
b) Paper describing the findings of the amoxicillin study will be published in a peer-reviewed journal.
c) Paper describing the findings of the minocycline study will be published in a peer-reviewed journal.
d) Review paper describing the results, together with those of other partners in this project, will be published in a peer-reviewed journal.
e) Ensuring access via the website to the ANTIRESDEV databases showing the amoxicillin and minocycline data to enable future studies by other groups.
f) To use the amoxicillin and minocycline data to inform health care decision makers of some of the factors influencing the emergence and persistence of antibiotic-resistant bacteria following the administration of particular antibiotics and thereby provide opportunities for them to implement appropriate policies concerning antibiotic use
g) The data will be presented at a symposium held during the 36th International Congress of the Society for Microbial Ecology and Disease in September 2013
The overall results from studies of the administration of ciprofloxacin or clindamycin to healthy volunteers are that single standard treatments with either of these antibiotics have a major impact on selected species in the oral and intestinal microbiotas. The effects are both quantitative and qualitative and are seen immediately after treatment but may be present for longer time i.e. up to 1 year (the duration of the study). Only minor changes were observed in the development of resistance during the study period. The respective antibiotic was detected in faeces only and only on day 11, i.e. directly after the treatment. No obvious correlation between the resistance pattern between the isolates from any of the body sites and the three different volunteer groups could be observed. The results from this study will be widely distributed through publications and oral presentations at appropriate conferences. The final leaflet, containing the results of the whole project, will be distributed both internally within the Karolinska Institutet and externally at national and international meetings/conferences.
The results obtained using next generation sequencing techniques clearly showed a profound impact of antibiotics on the composition of the oral and faecal microbiomes and resistomes. These effects were so broad, large and in many cases long-lasting (depending on the antibiotic and the body site, the impact may vary in its duration of impact from less than 1 month to more than 1 year) that they, when disseminated as planned may lead to a re-evaluation of antibiotic use in Europe.
Salivary microbiomes appeared to be more stable and resilient than faecal microbiomes. Salivary microbiomes of healthy individuals were stable in species richness and diversity throughout one year without any intervention. Irrespective of the antibiotic, 30 days after an intervention, salivary microbiomes showed recovery with respect to both diversity and phylogenetic measures. This extreme resilience indicates the importance of a healthy oral microbiome and underlines the recent views on maintaining a healthy balance in the oral cavity and the effects that this microbiome has on e.g. vascular integrity.
In the majority of cases, the faecal microbiome also returned to its original and natural composition. For several antibiotics, the butyrate-producing bacteria were impacted temporarily by the antibiotics. Butyrate is an important substrate for the colonocytes and is important in the maintenance of the gastrointestinal barrier function. It is important to direct future studies on the possible effects of antibiotics on colonocyte physiology and gut barrier function. If needed, strategies for treatment can be adjusted to minimize the effects on butyrate production during treatment or to complement this after treatment.
The approach applied to the analysis of the resistome is new and may have great potential for application. Discussions have been initiated with several Dutch farming companies on the application of the tool to establish biologically relevant conditions for the application of antibiotics in farming. This may enable a further reduction of antibiotic use in farming.
These findings will be disseminated towards key players in the Netherlands (National Institute for Public Health and the Environment; Royal Dutch Society for Microbiology etc.), via personal contacts and lectures at conferences. More importantly, specifically information the profound impact of antibiotics on the composition of the oral and faecal microbiomes, and the subsequent responsibilities that clinicians have in providing antibiotics have already been made part of the Dentistry curriculum of the Academic Centre for Dentistry Amsterdam (the largest dental school in the Netherlands). Also this information has been shared among the Dentistry researchers of the Netherlands in their 2013 annual meeting as well as among the dental professionals that were present at the “life-long-learning day’s” “Modern Microbiology” of the ACTA-Quality Practice program.
Analysis of the Gram-positive bacteria isolated from healthy volunteers indicated that treatment with amoxicillin, minocycline and clindamycin for a short period (7-10 days) did not significantly alter the abundance of antibiotic resistance genes within Staphylococcus and Streptococcus with the exception of tet(M) in staphylococci and erm(B) in streptococci. The development of a new microarray and new labelling system enables the screening of Gram-positive bacteria for the presence of up to 116 different genes within one working day. This system has already found application in research and has been used to screen starter cultures and probiotic bacteria. The DNA labelling and amplification is independent of the input DNA and the microarray system used, which enables it to be adapted for Gram-negative bacteria as well as for virulence factors or other comparable applications. This straight-forward and simple workflow can easily be designed for use as a commercially-available detection kit. Such a test would offer a rapid, sensitive and cost-effective approach for applications in basic research, food safety, and surveillance programs for antimicrobial resistance.
The final leaflet which summarizes the overall results of the ANTIRESDEV project will be distributed to the Federal Office of Public Health (FOPH), swissmedic (Swiss Agency for Therapeutic Products), as well as to the Department for communication of the University of Berne. The leaflet will also be included into the annual report of the Institute of Veterinary Bacteriology, University of Berne, which is send to up to 200 persons including research scientists and members of Federal Offices. The final leaflet will be published on the homepage of the Institute of Veterinary Bacteriology.
The effect over one year of four different antibiotic treatments on aerobic and anaerobic Gram-negative bacteria, were monitored using an antimicrobial resistance (AMR) gene microarray. The results contribute to our understanding of how these treatments affect those Gram-negative bacteria that are human symbionts. The results showed that there may be a transient effect on aerobic Gram-negative antibiotic-resistant strains, especially Escherichia coli, immediately after treatment with Amoxicillin, Ciprofloxacin and Clindamycin. In contrast, for Minocycline treatment, a transient effect on resistance was seen only in Gram-negative anaerobes such as Bacteroides spp. Gram-negative isolates from the placebo group were also found to harbour many resistance genes which fluctuated over time, indicating that the gain and loss of AMR genes is a natural feature of the human microbiota. Additionally, it was found that it is not uncommon for multi-resistant E. coli isolates to persist in the human gut for up to one year, during which time they may retain all, or a core subset, of their resistance genes. The resistances present in Gram-negative bacteria are most likely carried by mobile elements such as plasmids, which were found to bestow other fitness properties on the bacteria such as survival in the presence of heavy metals (e.g. tellurite or cadmium). It was also shown that some of these human-adapted multi-resistant E. coli strains were able to colonise the chick gastrointestinal tract to high levels. It is already well know that animals can be a reservoir for transmission of antibiotic-resistant bacteria to humans; these results indicate that transmission can also potentially occur in the opposite direction which will contribute to exacerbating the problem of antibiotic resistance. The results from this study will be widely distributed through publications, oral presentations and distribution of the final leaflets. This will be done both internally within our institute and externally at national and international meetings/conferences, and to stake holders such as the UK National Government.
With regard to pathogens of major clinical importance such as MRSA, ESBL producing Enterobacteriaceae, Pseudomonas aeruginosa or Acinetobacter baumannii, it was found that the short-term administration of amoxicillin, minocycline, ciprofloxacin or clindamycin to healthy volunteers appears not to be a risk factor for the acquisition and selection of multidrug resistant organisms. Healthy volunteers mimic the prevalence of resistance in the environment in which they live. We found no selection for, or impact on, isolates of major clinical importance. The main risk factors for infections with multidrug resistant pathogens are frequent antibiotic exposure and hospitalisation. Short-term and single antibiotic exposure of healthy patients is no risk factor for multidrug resistance as long as the indication and choice of antibiotic is made correctly. This represents a very important finding to communicate since many people are afraid of antibiotic intake due to the striking media coverage. These findings support the concept that frequent antibiotic exposure and hospitalization are the key factors to keep an eye on, tight epidemiologic surveillance systems and antibiotic stewardship programs are needed in order to cope with this major threat in hospital settings.
Since no new antibiotics can be expected in the near future, we have to cope with our current antibiotic arsenal. Therefore, it is crucial that antibiotics are administered in an appropriate and targeted way. Faster and multiplexed molecular diagnostics are one way in which we can achieve this goal. The updated or completely new designed DNA-microarrays developed during the ANTIRESDEV project are perfect tools for the detection of hundreds of resistance determinants in parallel in a short time. The first microarrays have already been commercialized and parts of them were incorporated into new diagnostic devices for which the first clinical trials have already been conducted. One major output of the ANTIRESDEV project is the transfer of these basic research tools into new diagnostic devices for the future.
Overall, the ANTIRESDEV results will be of great interest to the medical expert audience with regard to antibiotic exposure in healthy patients, but also for the general (patient) population being confronted with “headline” media releases. Furthermore, the technological advances made during the project will be made available to the diagnostic industry. National and international conferences, but also local activities or lectures for the general public in our hospital will be used for the dissemination of these important project findings.
The fact that antiseptic and antibiotic resistance are contained on the same genetic element has implications for the management of infectious diseases, as it is possible that treatment with antiseptics can lead to the spread of antibiotic resistance.
The finding that there is a limited number of core genetic elements that can associate with each other in an almost limitless number of ways, indicates that it is of crucial importance that we understand in more detail how these elements are regulated. Furthermore, as there appear to be a limited number of vectors involved in the spread of antibiotic resistance, it should be feasible with a concerted research effort to understand how these genetic elements replicate and transfer in bacterial populations. As a Tn916-like core appears to be responsible for the spread of a large number of different antibiotic resistance genes it is important that we undertake a more detailed analysis of this element. Furthermore, our results show that mutation of a predicted negative regulator of Tn916 transfer actually seems to decrease the frequency of transfer, this shows that the previous assumptions about how Tn916 is regulated is not necessarily correct.
Our results and those of ANTIRESDEV as a whole will be disseminated at meetings within the UK particularly those of the Society for General Microbiology. The results will also be presented at international meetings within Europe and beyond. We plan to present this work at ICACC, the largest meeting on antibiotic resistance in the world. Finally this work will be used to as a basis for further grant proposals to both the Horizon 2020 and domestic funding bodies.
The characterisation of S. aureus strains indicated that the administration of minocycline or amoxicillin did not induce the acquisition or development of resistance. Instead, resistance levels and growth rates varied during the monitored time in several clonal lineages, not necessarily correlating the extent of resistance to the level of fitness. These findings are of interest as they suggest that i) S. aureus constantly undergoes changes in fitness and resistance levels, adapting to shifting conditions in the human host, and ii) many resistances and resistance combinations do not inflict a fitness cost. Still, resistance is a phenotypic trait relying on the unobstructed functioning of the cell. Any gene involved in crucial processes and affecting resistance has the potential to represent a novel target for the development of antimicrobial therapies. Extending the list of genetic factors that can influence resistance, fitness and virulence in S. aureus is, therefore, of utmost importance. Several of the genes identified as affecting resistance were shown to be required for full virulence in vivo, making them very promising targets for antimicrobial substances that could control one of the most notorious opportunistic human pathogens. This would be of enormous benefit to the general public.
Some of the human-adapted multi-resistant E. coli strains were able to colonise chick gastrointestinal tracts to high levels. It is already well know that animals can be a reservoir for transmission of antibiotic resistant bacteria to humans; these results indicate that transmission can also potentially occur in the opposite direction which will contribute to exacerbating the problem of antibiotic resistance.
All members of the team at the University of Zurich regularly presented and discussed the results of the ANTIRESDEV project at progress meetings within the institute, during courses and research projects, at workshops, and at national and international conferences. ANTIRESDEV leaflets were distributed at all attended conferences. Dissemination of our overall results will be performed by distributing the final ANTIRESDEV leaflet at coming national and international meetings, to selected contacts, and by forwarding the leaflet to the national press agency. Follow-up studies and publications will reference the ANTIRESDEV project and homepage, thereby continuing to the dissemination of the results among the scientific community.
List of Websites:
The ANTIRESDEV project public website can be found at the following link:
The main contact for the project is:
Professor Michael Wilson
Professor of Microbiology
Department of Microbial Diseases
UCL Eastman Dental Institute
University College London;
Tel +44 20 7915 1231;