Re-evaluation of a novel resistance mutation in eccC₅ of the ESX-5 secretion system in ofloxacin-resistant Mycobacterium tuberculosis

Sir,

We previously reported that the V762G mutation in Rv1783 (subsequently numbered V1221G;^1,2eccC₅, encoding a gene in the ESX-5 membrane complex secretion system) of Mycobacterium tuberculosis was associated with fluoroquinolone resistance.³ However, subsequent work has not confirmed these findings.

We performed WGS using 150 bp paired-end reads (NextSeq500) on the eight fluoroquinolone-resistant M. tuberculosis isolates in which the V1221G mutation in Rv1783 (position 2022918) was originally identified using 36 bp paired-end reads.³ We also sequenced the M. tuberculosis H37Rv isolate that had undergone allelic exchange and site-directed mutagenesis to introduce the Rv1783: V1221G mutation, and the fluoroquinolone-susceptible laboratory strain M. tuberculosis H37Rv.

Confirmatory Sanger sequencing of mutations identified by WGS was performed as previously described.⁴ We repeated MIC testing on the eight clinical isolates, H37Rv, and the mutant M. tuberculosis isolate. We used the resazurin microtitre assay (REMA) method, using an ofloxacin critical concentration of 2 mg/L.^5,6 Site-directed mutagenesis and allelic exchange were performed in the original study,³ but not repeated.

WGS of the eight fluoroquinolone-resistant M. tuberculosis isolates using 150 bp paired-end reads revealed the same DNA gyrase mutations as with the 36 bp paired-end reads: three isolates had mutations associated with fluoroquinolone resistance (A90V, D94G in gyrA and N538I in gyrB) and five did not.

The eight fluoroquinolone-resistant M. tuberculosis isolates in which the V1221G polymorphism at position 2022918 in Rv1783 was present on 36 bp paired-end reads did not have this polymorphism identified when 150 bp paired-end reads were used. The average depth of coverage was 868 reads. Results did not differ according to strain lineage.

The original Sanger DNA Baser alignment and reads confirmed the presence of the V1221G mutation at position 2022918. However, there were several additional flanking polymorphisms surrounding this region. When alignment was repeated, flanking polymorphisms were removed; no polymorphism was identified at 2022918.

WGS of the mutant strain revealed that, rather than introducing the full Rv1783 gene (with the V1221G mutation at position 2022918) into M. tuberculosis H37Rv, an ∼900 bp fragment that included the V1221G mutation was introduced; the promoter region was not included. WGS revealed ∼10× higher read depth across the ∼900 bp fragment including the V1221G mutation, consistent with ∼10× additional copies of this sequence per cell (Figure 1a). The V1221G mutation was not identified in the M. tuberculosis H37Rv chromosome; there was no evidence of insertion at the chromosomal attB site, where an integrating plasmid would be expected to be found. WGS of the mutant strain did not identify DNA gyrase mutations. Given the findings on WGS, the likely actual events of the experiments are shown in Figure 1(b).

Figure 1.

(a) WGS results of the mutant strain revealed ∼10× greater sequence depth for an ∼900 bp fragment of Rv1783 that contains the V1221G mutation. Poorly aligned reads at the borders of the fragment indicate clusters of misaligned bases. No other mutations were identified in this strain. (b) The proposed site-directed mutagenesis and allelic exchange experiments, and the likely actual events and issues identified. Based on WGS of the mutant M. tuberculosis strain that was generated. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

On repeat testing of the eight clinical isolates, ofloxacin MICs were the same as originally reported.³ The ofloxacin MIC for H37Rv and the mutant M. tuberculosis strain with the ∼900 bp fragment (which included the V1221G mutation in Rv1783) also remained the same: 1 and 16–32 mg/L, respectively.

We were not able to confirm the presence of the V1221G mutation in Rv1783 (eccC₅) in eight fluoroquinolone-resistant M. tuberculosis isolates using WGS with 150 bp paired-end reads. Subsequent to our initial publication, it was reported that the short 36 bp paired-end reads can falsely call polymorphisms at the end of the read.⁷ This likely explains the false-positive finding of the V1221G polymorphism. We initially identified this polymorphism in M. tuberculosis isolates of both European–American and East Asian lineages. However, all of these strains had undergone WGS using 36 bp paired-end reads, thus negating what we thought was additional evidence of a true mutation. In addition, the initial Sanger sequencing incorrectly confirmed the presence of the mutation, possibly due to the presence of flanking polymorphisms surrounding the region.

Although the V1221G polymorphism in Rv1783 was a false positive, our introduction of it via site-directed mutagenesis and allelic exchange increased the ofloxacin MIC from 1 to 32 mg/L, which we again confirmed. However, following attempted allelic exchange, the WT Rv1783 gene was still present in the chromosome, and only an ∼900 bp fragment with the V1221G polymorphism was introduced. Sequence depth of this fragment was ∼10× greater than the rest of the chromosome, suggesting that ∼10 copies were present. However, the location of the ∼900 bp fragment was unclear. Although there was no evidence of incorporation into the genome (e.g. at attB where integrating plasmids are normally located), the allelic exchange plasmid did not contain a mycobacterial origin and thus should not replicate extra-chromosomally. Of note, WGS is unable to distinguish between a true copy-number duplication event of this size in the M. tuberculosis genome versus the presence of plasmids in the cytosol.

The mechanism of fluoroquinolone resistance of the mutant M. tuberculosis strain was therefore unclear. There were no DNA gyrase mutations present. Given the importance of Rv1783 (eccC₅) and ESX-5 in cell wall stability, virulence and uptake of nutrients across the outer membrane of M. tuberculosis,^8,9 it is possible that alterations of this gene could affect fluoroquinolone resistance in M. tuberculosis. Our WGS of the mutant revealed no other mutations. Nonetheless there could have also been coincidental introduction of another (unknown) mechanism of fluoroquinolone resistance.