Acquired macrolide resistance genes in nontypeable Haemophilus influenzae

Nontypeable Haemophilus influenzae (NTHi) is an opportunistic pathogen that is associated with a range of respiratory infections, including acute exacerbations of chronic obstructive pulmonary disease (COPD), and community acquired pneumonia (CAP). Macrolide antibiotics such as azithromycin are being more frequently used to manage these conditions, including those where NTHi may be involved, despite macrolides having relatively poor antibiotic activity against H. influenzae (azithromycin MICs of wild-type strains typically cover the range of 0.25-4 ˜í¬¿g/mL). The efficacy of these antibiotics in managing these conditions is further threatened by the emergence of strains of NTHi exhibiting high-level macrolide resistance. A range of different mechanisms of macrolide resistance are recognised broadly within bacteria. Resistance has been attributed to the presence of mutations in structural components of the ribosomal binding site of macrolides. Such mutations can occur in the 23S rRNA gene, as well as in the L4 and L22 structural proteins. Increased expression of inherent macrolide efflux mechanisms such as acrAB resulting from mutations in regulatory regions of these efflux pumps have also been recognised as a potential cause of macrolide resistance. Finally, resistance has also been attributed to the acquisition of macrolide resistance genes (AMRGs), which are readily disseminated among species on mobile genetic elements. There are a large number of different AMRGs and associated proteins, and the mechanisms by which they produce resistance vary. For example, the erm genes encode enzymes which modify the ribosomal binding site, the mef genes encode alternate efflux systems, and an additional group of genes encode enzymes which directly deactivate the macrolide. High-level macrolide resistance in NTHi is uncommon and has historically been attributed to chromosomal mutations in ribosomal structural elements or regulatory regions controlling acrAB. In contrast, AMRGs have not been widely associated with NTHi and a number of studies have failed to detect these genes in this particular species. There has been a single study in which these genes have been reported to be highly prevalent in NTHi. In that study, among a collection of 106 NTHi isolates from children with cystic fibrosis enrolled in a placebo-controlled azithromycin trial, all isolates had at least one AMRG, with many of the isolates carrying two or three of these genes. However, the phenotypic effect of these genes was not consistent; only 27 of the AMRG-carrying isolates exhibited phenotypic resistance to macrolides. While the findings of that study identify the emergence and potential spread of AMRGs in NTHi, it raises a number of questions regarding the prevalence of these genes within NTHi isolates more broadly and the role they play in generating a resistant phenotype. In Chapter 4 of this thesis, a collection of 186 NTHi respiratory isolates of variable azithromycin resistance phenotype (azithromycin MIC range: 0.09 to >256 ˜í¬¿g/mL; MIC50: 1.5 ˜í¬¿g/mL; MIC90: 3 ˜í¬¿g/mL) derived from both cystic fibrosis and non-cystic fibrosis patients was established and analysed for; 1) the presence of macrolide resistance-associated L4, L22 and 23S rRNA mutations, and 2) the presence of the AMRGs erm(A), erm(B), erm(C), erm(F), mef(A) and mef(E). For the detection of the AMRGs, two methods were used; 1) a novel PCR using locked nucleic acid dual-labelled hydrolysis probes, and 2) the original primer set used by the authors of the previous study. While L22 and 23S rRNA alterations were detected in 2 isolates with high-level macrolide resistance (azithromycin MIC ‚Äöv¢‚Ä¢ 256 ˜í¬¿g/mL), none of the isolates were found to carry any of the AMRGs using the novel PCR detection method. When using the primers described in the original study, mef(A) and erm(A) were detected in a number of isolates. Subsequent analysis of these amplicons. revealed them to be false positive results, raising questions as to the possibility of false positive results in the original study. Over 100 different AMRGs have been recognised, and the development and increasing availability of whole genome sequencing (WGS) techniques now allows for efficient and thorough analysis of sequences for the detection of antibiotic resistance mechanisms. In Chapter 5 of this thesis, WGS was utilised to further investigate the presence of a broad selection of other AMRGs (n=72) in NTHi, using the SPANDx pipeline. WGSs of two isolates of NTHi exhibiting high-level macrolide resistance obtained from the study of Chapter 4, as well as an additional 89 publically available WGSs of NTHi isolates of variable resistance phenotype, were examined in the study. None of the specified AMRGs were detected among this collection of WGSs. In addition, the WGSs of the 2 isolates from Chapter 4 with high-level macrolide resistance were interrogated for any AMRGs using the Comprehensive Antibiotic Resistance Database (CARD), and none were detected. Both isolates underwent further WGS analysis to confirm the L22 and 23S findings in Chapter 4; one isolate carried R88P in L22 and C2611G in 23S rRNA and the other isolate carried A2058G in 23S rRNA, all previously associated with decreased macrolide susceptibility in H. influenzae. Alterations in regulatory regions of acrAB were also detected in both isolates. Finally, transformation studies using donor genomic DNA from these 2 isolates were performed on H. influenzae Rd KW20. While none of the transformants that were generated exhibited as high an azithromycin MIC as the donors, a number of different regions of donor origin were detected in various transformants. The role of these regions in generating resistance in individual transformants was not clear but, with respect to the lower MICs exhibited by these transformants, the findings suggested a multifactorial aetiology for the high-level macrolide resistance seen in the donor isolates. In the Roberts et al. study, the effect of the AMRGs appeared to be inconsistent, with some isolates not exhibiting increased MICs (compared to a typical wild-type strain). The effect of these AMRGs on macrolide susceptibility in H. influenzae remains to be established. As a result, the aim of Chapter 6 was to transfer select AMRGs (erm(A), erm(B), erm(C), mef(A) and mef(E)) to H. influenzae Rd KW20 and examine the phenotypic effect of expression of these genes. Initially, attempts were made to conjugatively transfer these AMRGs from select Gram positive donors to H. influenzae Rd KW20. These attempts were unsuccessful. As a result, the AMRGs were cloned into the shuttle vector pLS88 and H. influenzae Rd KW20 was transformed with these constructed plasmids by electroporation. Clones were generated with a range of approaches, including with and without the native regulatory regions of the AMRG inserts, and in the former, tested for expression with and without the presence of an inducing agent. High-level expression of erm(A), erm(B) and erm(C) was demonstrated in at least some of the various conditions and resulted in increased macrolide resistance in these transformants. In contrast, expression of mef(A) and mef(E) did not have an effect on macrolide resistance. In the Roberts et al. study, conjugative transfer of mef(A) to H. influenzae Rd KW20 resulted in a moderate increase in azithromycin and erythromycin MICs; our findings therefore suggest that mef only increases MICs in combination with other macrolide resistance mechanisms such as msr(D) (found downstream of mef and not covered by our cloned inserts) or underlying chromosomal alterations. The above AMRGs are commonly encountered among human pathogens which share a respiratory niche with NTHi, including Staphylococcus aureus and various respiratory Streptococcus spp.. A number of other AMRGs, such as erm(42), msr(E) and mph(E), are typically encountered among animal commensals and pathogens such as Pasteurella spp. and Mannheimia spp.. H. influenzae is closely related to these pathogens and previous studies have demonstrated that they are able to exchange resistance determinants, including beta-lactamases. Therefore, the work of Chapter 7 explored the potential for conjugative transfer of erm(42), msr(E) and mph(E) (carried on the mobile multiresistance integrative and conjugative element ICEPmu1) and associated macrolide resistance from a bovine Pasteurella multocida isolate to H. influenzae Rd KW20. Transconjugants generated in this study were found to carry a truncated form of ICEPmu1 that lacked msr(E) and mph(E) but carried erm(42); transconjugants were found to exhibit increased erythromycin and clindamycin resistance. This truncated ICEPmu1 was successfully transferred from primary transconjugants to secondary H. influenzae Rd KW20 recipients, indicating that transfer functions were retained during conjugation. The acquisition of ICEPmu1 did not appear to have an impact on the fitness of H. influenzae Rd KW20 and the ICE was found to be stable in the absence of antibiotic selective pressure. In summary, the major findings of this thesis are that AMRGs are not widespread in NTHi and that the high prevalence of a selected set of these genes described in one recent study is probably unique to the circumstances of that study. Although complex regulatory regions in many AMRGs mean that expression and associated resistance may be dependent on the specific genetic context, we have shown that under favourable conditions, erm(A), erm(B) and erm(C) can produce macrolide resistance in H. influenzae in their own right, but this is unlikely for the mef genes. Although we were unable to demonstrate conjugative transfer of common AMRGs from respiratory Gram positive organisms to H. influenzae, we were able to demonstrate conjugative transfer of an AMRG encoding multi-resistance replicon from a closely related organism of animal origin. This replicon produced macrolide resistance, was stab...