The aminoglycoside-modifying enzyme Eis2 represents a new potential in vivo target for reducing antimicrobial drug resistance in Mycobacterium abscessus complex

To the Editor:

Mycobacterium abscessus complex (MABSC) is an emerging opportunistic pathogen complex responsible for lung infections after lung colonisation in people with pulmonary disorders, such as bronchiectasis or cystic fibrosis [1, 2]. It is becoming one of the most clinically relevant nontuberculous mycobacteria because of the severity of infection and poor response to antibiotic treatment. The MABSC includes three subspecies: M. abscessus abscessus, bolletii and massiliense [3–5]. MABSC pulmonary disease is characterised by the presence of specific microbiological, clinical and radiological features described in the ATS/ERS/ESCMID/ IDSA (American Thoracic Society/European Respiratory Society/European Society of Clinical Microbiology and Infectious Diseases/Infectious Diseases Society of America) consensus statement [1]. Infections are difficult to treat due to the high level of antibiotic resistance conferred by an almost impermeable cell wall, drug efflux pumps, or drug-modifying enzymes [2, 6]. Antibiotic regimens recommended for treatment of MABSC pulmonary infections generally combine 3–4 antibiotics including clarithromycin, amikacin, cefoxitin, imipenem or tigecycline for 12–24 months [1]. Despite toxicity, the aminoglycoside amikacin remains a key component in the regimen. Resistance to aminoglycosides is mediated by several mechanisms: 1) modifying enzymes, such as aac(2′), aph(3″) and eis2; 2) increased efflux; 3) decreased cell wall permeability; and 4) modifications of the 30S ribosomal subunit (e.g. RpsL protein). In a recent study, the deletion of eis2, encoding a promiscuous N-acetyltransferase, from the genome of M. abscessus ATCC19977, increased in vitro susceptibility to amikacin, kanamycin and capreomycin [7, 8]. The findings highlight that drug-modifying enzymes may provide new complementary therapeutic targets to fight antimicrobial resistance, through increasing the bacterial susceptibility to specific approved antibiotics thereby improving treatment options.

We aimed to determine whether Eis2 is a relevant drug-modifying enzyme for bacterial resistance in a preclinical model of MABSC infection and if it could be considered a clinically relevant target to reinforce existing antibiotic treatments in MABSC clinical isolates.

To understand the Eis2 relevance as MABSC drug resistance determinant in vivo we tested M. abscessus (Mabs) Δeis2 mutant and its parental wildtype (wt) strain (ATCC19977) in a preclinical model of lung infection. As mentioned before, in vitro Δeis2 showed a four-fold increase in the susceptibility of minimal inhibitory concentration (MIC) to amikacin in comparison to the isogenic wt strain [7]. We used a recently refined mouse model of Mabs respiratory infection exploiting C57BL/6N mice and agar beads [9–11]. Mice were infected with an intratracheal injection of 10⁵ colony-forming units (CFU) embedded in agar beads, allowing persistence in the lung of C57BL/6N immunocompetent mice with a sustained bacterial load. Mice were treated daily by intraperitoneal injection of amikacin (200 mg·kg⁻¹) starting from 2 h before bacterial challenge (day 0) for the next 6 consecutive days (figure 1a). 7 days after challenge, mice were sacrificed and their general health status, and the local (lung homogenates) and systemic (spleen homogenates) bacterial burdens were evaluated. We observed that mice infected with Δeis2 mutant strain displayed a similar bacterial load as those challenged with wt strain ATCC19977, suggesting a similar bacterial capacity to multiply and persist in our model of infection and disproving the role of Eis2 as a virulence factor (figure 1b). We confirmed that amikacin treatment has antibacterial activity (∼1 log CFU reduction) against the wt strain (ATCC19977) in our model of respiratory infection, as previously described. Here, we determined that the bacterial burden in the lung of mice infected with Δeis2 mutant strain and treated with amikacin was significantly reduced in comparison to 1) lungs of mice infected by Δeis2 mutant strain and treated with vehicle (∼3–4 log CFU reduction, p<0.0001; Δeis2+AMK 1.805×10³ median CFUs versus Δeis2 7.195×10⁶ median CFUs); and 2) lungs of mice infected with wt strain and treated with amikacin (∼3 log CFU reduction, p<0.0001; Δeis2+AMK 1.805×10³ median CFUs versus Δeis2 1.52×10⁶ median CFUs) (figure 1b). The same trend was also observed at the systemic level. In the spleen, mice infected with Δeis2 mutant strain and treated with amikacin displayed a significantly reduced bacterial burden in comparison to mice infected by wt strain and treated with the antibiotic (figure 1c). Overall, these data prove that genetic inactivation of eis2 increases amikacin susceptibility at local and systemic compartments during lung Mabs respiratory infection in an in vivo preclinical model. We analysed haematoxylin and eosin stained lung tissue slides at day 7 post challenge (three mice per group) to evaluate the inflammatory foci in mice infected by Δeis2 strain without (Δeis2) or with (Δeis2+AMK) amikacin treatment. As shown in figure 1d, the lung infected with Δeis2 strain and treated with vehicle displayed more extended inflammatory foci (indicated by arrows), mainly composed by monocytes, than the lung infected with Δeis2 strain and treated with amikacin. This analysis was performed in three different mice per group (data not shown), confirming beneficial treatment of amikacin in limiting inflammatory foci in the lungs and potentially limiting the risk of the bacterial spreading at the systemic level. Overall in vivo data suggest that the deletion of eis2 strongly enhances amikacin antibacterial activity in the treatment of MABSC infections.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Preclinical murine model to study the lack of Mycobacterium abscessus eis2 gene during amikacin treatment and analysis of eis2 genomic sequence among 727 publicly available genomes of M. abscessus complex (MABSC). a) Experimental design. Mice were infected with an intratracheal injection of 1×10⁵ colony-forming units (CFU) of M. abscessus (Mabs) wildtype (wt) and Δeis2 embedded in agar beads at day 0. Amikacin (AMK) or vehicle (saline solution) was administered daily through intraperitoneal (IP) injection until day 6. At day 7, mice were sacrificed, and lungs and spleens collected. b) Mabs burden (lines represent median values of each group) in the lungs of mice treated with vehicle or amikacin and infected with Mabs wt or Δeis2. ****: p<0.0001 (Mann–Whitney t-test, two-tailed). c) Mabs burden in the spleen of mice treated with vehicle or amikacin and infected with Mabs wt or Δeis2. *: p<0.05; **: p<0.01 (Mann–Whitney t-test, two-tailed); ns: nonsignificant. d) Lung tissue slides were stained with haematoxylin and eosin at day 7 after challenge. This panel compares lungs of mice infected with Mabs Δeis2 and treated either with amikacin or with vehicle. Arrows indicate extended inflammatory foci mainly composed of monocytes. e) Number of overall MABSC genomes considered to determine the number and percentage of MABSC subspecies strains carrying eis2 non-synonymous mutation as compared to Mabs ATCC19977, among clinical MABSC abscessus, bolletii and massiliense strains. f) The Eis2 protein (PDB ID 6RFT) is shown in cartoon/surface representation. The MABSC abscessus, bolletii and massiliense subspecies mutations are shown in green, red and blue spheres, respectively. The three Eis2 domains Ct GNAT domain, C GNAT domain and Nt GNAT domain are shown in pink, dark grey and orange, respectively. The cofactor acetyl-CoA is shown in yellow.

Previous studies focused their efforts in demonstrating eis2 relevance, as MABSC drug resistance gene in vitro, through elegant mechanistic approaches [7, 12]. To date, it is still unclear whether Eis2 target is conserved in MABSC clinical isolates. We determined the genomic conservation of eis2 in MABSC publicly available genomes [3–5]. Our analysis included clinical isolates from MABSC subsp. abscessus, massiliense or bolletii, and from different geographic regions (Europe and the USA). Non-synonymous mutations were found in 25 out of 535 MABSC subsp. abscessus strains (4.7%), in 112 out of 144 strains of MABSC subsp. massiliense (77%) and 48 out of 48 strains MABSC subsp. bolletii (100%) (figure 1e). These data highlight that eis2 genomic sequence is well conserved in MABSC subsp. abscessus strains, while the biological relevance of non-synonymous genomic variation observed in MABSC subsp. massiliense or bolletii remains unclear. We took advantage of the Eis2 crystal structure [13] to determine whether the genomic variation observed among the MABSC public genomes could be associated with functional alteration of the catalytic site of Eis2 (figure 1f). We highlighted the mutated residues in the different subspecies, indicating that the mutations occur in distinct protein domains. In particular, most of the mutations mapped in MABSC subsp. bolletii are located close to the acetyl-CoA binding site, differently from MABSC subsp. massiliense and abscessus.

Moreover, the R92W mutation in MABSC subsp. bolletii, could affect the interaction with acetyl-CoA as highlighted from the X-Ray visual analysis (6RFT, figure 1f). The other mutations, shown in figure 1e, are not directly involved in altering interaction with acetyl-CoA as presented by Ung et al. [13]. In general, the mutations are spread on the protein structure and any possible functional effect could arise from allosteric interactions. While MABSC subsp. abscessus Eis2 structure is highly conserved, the knowledge of the structural consequences of mutations in Mabs subsp. bolletii and massiliense may pave the way to the design of Eis2-subspecies-specific inhibitors with the potential to counteract the intrinsic amikacin resistance of MABSC.

Our work provides evidence that Eis2 represents a relevant drug-modifying enzyme for bacterial resistance in vivo. Our data demonstrate that deletion of eis2 strongly improves amikacin antibacterial activity in our preclinical model of MABSC lung infections. Moreover, genomic analysis of eis2 among clinical isolates suggest that sequencing approaches may promote future personalised medicine strategies based on the stratification of conserved drug-modifying enzymes of MABSC.

While the Eis2 crystal structure was already published, the relevance of non-synonymous mutations found in our analysis remains to be elucidated. Indeed, this highlights again the clinical relevance of genomic bacterial data for the design of future tailored personalised antimicrobial therapies. It is noteworthy that potent inhibitors with a strong translational relevance have been identified against Eis of Mycobacterium tuberculosis [14, 15]. However, amikacin is more important in the treatment of MABSC infection than in tuberculosis. Therefore, future identification of an Eis2 inhibitor could strongly enhance amikacin antibacterial activity at least in the treatment of infections by MABSC subsp. abscessus.

In conclusion, our data, obtained from a preclinical model of respiratory infection and from the genome of clinical bacterial isolates, demonstrated that the aminoglycoside-modifying enzyme Eis2 represents a new potential antimicrobial target for fighting multidrug resistance in M. abscessus infections.

Shareable PDF