Decreased Susceptibility to Tigecycline Mediated by a Mutation in mlaA in Escherichia coli Strains

LETTER

Tigecycline, a type of glycylcycline, is a novel expanded-spectrum antibiotic that is active against most Gram-negative and Gram-positive bacteria, including antibiotic-resistant strains such as carbapenem-resistant members of the family Enterobacteriaceae (1–3). However, tigecycline-resistant strains have emerged since tigecycline was approved for clinical use (4, 5). It has been reported that overexpression of the AcrAB efflux pump results in decreased susceptibility to tigecycline in Enterobacteriaceae (6–8). Ribosomal protein S10 is also a general target of tigecycline adaptation (9–11). In addition to the efflux pumps and ribosomal S10 protein described previously, other possible mechanisms of tigecycline resistance require further investigation. In the present study, two pairs of Escherichia coli strains were studied by whole-genome sequencing to identify mutants under selective pressure from tigecycline.

E. coli ATCC 25922 was used as the initiating strain (12). The AcrAB efflux pump of E. coli ATCC 25922 was inactivated by deletion of the acrAB genes as previously described (13), with the hybrid primers listed in Table S1 in the supplemental material. The resulting strain was named 25922ΔacrAB. Tigecycline-resistant mutants were selected by successive passages through Luria-Bertani broth containing increasing concentrations of tigecycline in an induction experiment. E. coli ATCC 25922 and 25922ΔacrAB were used as parental strains. The selective tigecycline concentration began with 0.0625 μg/ml and doubled every 24 h until the mutants grew at a concentration of 32 μg/ml. The overnight cultures at each step were stored at −80°C in the presence of 20% glycerol. Two tigecycline-resistant mutants, namely, 25922-TGC8 (tigecycline MIC, 8 μg/ml) and 25922ΔacrAB-TGC8 (tigecycline MIC, 8 μg/ml), were obtained from ATCC 25922 (tigecycline MIC, 0.125 μg/ml) and 25922ΔacrAB (tigecycline MIC, 0.0625 μg/ml), respectively. Both mutants exhibited nonsusceptibility to tetracycline and minocycline (see Table S2 in the supplemental material). Genomic DNA from 25922-TGC8 and 25922ΔacrAB-TGC8 was sequenced with Illumina HiSeq 2000 (Illumina Inc., San Diego, CA) following a paired-end 2 × 100-bp protocol. The reads were mapped against the reference genome of E. coli ATCC 25922 (CP009072) with the CLC Genomics Workbench 9 software (Qiagen, Valencia, CA). The putative single nucleotide polymorphisms and deletion mutations in 25922-TGC8 and 25922ΔacrAB-TGC8 were also predicted (14). Twenty-three putative mutation sites in 25922ΔacrAB-TGC8 and 42 in 25922-TGC8 were found by bioinformatic analysis. Of these, two mutations in 25922ΔacrAB-TGC8 and six in 25922-TGC8 were confirmed by PCR and Sanger sequencing of the putative mutation sites (Table 1). One mutation in mlaA was found in both 25922-TGC8 and 25922ΔacrAB-TGC8. It was a deletion mutation of six nucleotides that resulted in the deletion of two amino acids and thus a truncated protein. Mutations in rpsJ (G169C) and marR (G311A) were also found in 25922-TGC8.

TABLE 1.

Mutations in 25922ΔacrAB-TGC8 and 25922-TGC8 compared with parental strains

Strain and reference position	Gene and product	Change
		Nucleotide sequence	Amino acid sequence
25922ΔacrAB-TGC8
286573	mlaA, putative phospholipid-binding lipoprotein	CTTCAA deletion at 129–134	NF deletion at 43–44
1435949	infB, translation initiation factor IF-2	G793C	G265R
25922-TGC8
286573	mlaA, putative phospholipid-binding lipoprotein	CTTCAA deletion at 129–134	NF deletion at 43–44
1300420	rpsJ, ribosomal protein S10	G169C	G57L
1534008	tsaD, tRNA threonylcarbamoyladenosine modification protein	TTG insertion at 346–347	L insertion at 116
3541486	marR, multiple antibiotic resistance protein	G311A	G104D
3865514	fadR, fatty acid metabolism transcriptional regulator	G397T	133 stop
4621400	entC, isochorismate synthase	C915A	S305R

The mlaA gene was deleted from ATCC 25922 and 25922ΔacrAB with the λ Red recombinase system (for the primers used, see Table S1 in the supplemental material). The isolates obtained were named 25922ΔmlaA and 25922ΔacrABΔmlaA. DNA fragments carrying the wild-type mlaA gene and a mutated mlaA gene were amplified from ATCC 25922 and 25922ΔacrAB-TGC8, respectively. After amplification, the amplimer was cloned into plasmid pCR2.1. mlaA deletion-carrying strains 25922ΔmlaA and 25922ΔacrABΔmlaA were used for transformation. The tigecycline MICs for 25922ΔmlaA and 25922ΔacrABΔmlaA were the same as those for their parental strains. However, when 25922ΔmlaA and 25922ΔacrABΔmlaA were complemented with mutational mlaA (named mlaA+), the MICs of tigecycline were 8-fold higher than those for the parental strains (see Tables S3 and S4 in the supplemental material), while no change was noted when the bacteria were transformed with the empty pCR2.1 vector and wild-type mlaA. In addition, we detected three mutated loci (mlaA, marR, and rpsJ) in the genomes of the series of isolates of 25922-TGC8 recovered at successive steps of the induction experiment (tigecycline MICs of 0.25 to 8 μg/ml). mlaA was the first mutated gene that appeared in the successive-passage experiment, and the MIC of tigecycline increased to 1 μg/ml. The marR mutation was the second mutation detected, and the MIC increased to 4 μg/ml. rpsJ was the last mutation that appeared, and the MIC increased to 8 μg/ml (see Table S5 in the supplemental material). Interestingly, it seems that multiple mechanisms (Mla system, efflux pump, and ribosomal S10 protein) can accumulate gradually in the development of tigecycline resistance.

In this study, we constructed an AcrAB efflux pump deletion strain (25922ΔacrAB) and induced resistance to tigecycline. According to the whole-genome sequencing data, only two mutations could be verified in 25922ΔacrAB-TGC8: mlaA and infB. It seems that tigecycline resistance can occur without the AcrAB efflux pump, and a mutation in ribosomal protein S10 was also not mandatory. Because the mlaA mutation was found in both 25922ΔacrAB-TGC8 and 25922-TGC8, it is reasonable to postulate that this mlaA mutation may play an important role in tigecycline resistance. This hypothesis was confirmed by the deletion and complementation experiments, in which the mlaA mutation led to an 8-fold increase in the tigecycline MIC (see Table S4 in the supplemental material). The Mla system is an ABC transport system that can transfer phospholipids from the outer membrane (OM) to the inner membrane to maintain OM lipid asymmetry (15). We propose that mutation of mlaA may increase the efficiency of this transfer and thus enhance the barrier function of the OM. Because of the widely distributed Mla system in Gram-negative bacteria, it may be easy to induce mlaA mutations under the stress of tigecycline. Our study contributes to the comprehensive understanding of tigecycline resistance mechanisms in Enterobacteriaceae.

Accession number(s).

The genome sequences of 25922-TGC8 and 25922ΔacrAB-TGC8 have been deposited in the NCBI SRA database and assigned accession numbers SRR3744959 and SRR3744956.