Study setting and isolates

This retrospective cross-sectional study was conducted on stored TB cultures at Makerere University College of Health Sciences in Kampala, Uganda, between 24^th June, 2019, and 17^th April, 2020. A total of 52 M. tuberculosis clinical isolates were studied, of which 20 were drug-susceptible and 32 were MDR-TB isolates, i.e., rifampicin-resistant and isoniazid-resistant. The clinical and demographic characteristics of the patients, sputum sampling and TB culturing procedures were described previously in the national and Kampala drug-resistance surveys [3,16]; briefly, the isolates were collected from a nationally representative cohort of new TB cases and previously treated sputum smear-positive patients registered at the TB diagnostic and treatment centres during 2009−2011 [3]. Similarly, phenotypic drug susceptibility testing for sensitivity to anti-TB drugs was described in the drug resistance surveys [3,16] however, repeat culturing and susceptibility testing using the Löwenstein-Jensen (L-J) proportional method was done to validate the susceptibility patterns. M. tuberculosis culturing and drug sensitivity testing were performed in a BSL-3 Mycobacteriology Laboratory at the Department of Medical Microbiology, Makerere University School of Health Sciences.

Extraction of chromosomal DNA

Stored bacterial isolates were accessed for sub-culturing on 1^st July, 2019; isolates were recovered by sub-culturing on Middlebrook 7H10 agar (Becton and Dickson, USA), incubating at 37ºC in a carbon dioxide incubator (Thermal Scientific, USA), and observing daily for 28 days for growth of the bacilli. The cells were harvested and suspended in absolute ethanol (Sigma scientific, USA) to kill them by suffocation. The suspension was centrifuged at 16,000 g to obtain the cell pellet that was later re-suspended in 0.25X Tris-EDTA (TE) buffer. High-quality bacterial chromosomal DNA was extracted from M. tuberculosis by following the CTAB/chloroform extraction method [17]. Briefly, the cells, suspended in 0.25X TE buffer, were centrifuged at 3,000 g to wash off the media salts and residual ethanol. To lyse the cells, the cell pellet was re-suspended in 400 µL of fresh 0.25X TE buffer, followed by adding 50 µL of lysozyme (40 mg/mL) and incubating at 37^ºC, overnight. Then, we added 150 µL of 10% Sodium Dodecyl Sulphate (SDS)/ proteinase K solution to the mixture and incubated at 65^ºC for 1 hour to ascertain complete cell lysis, precipitation of proteins and other cell debris. After this, 100 µL of 5M NaCl was added to each tube, followed by 100 µL of CTAB/NaCl, and the mixture was inverted several times until the contents turned milky. To ensure complete precipitation of all cell debris the solution was incubated briefly at 65^ºC for 10 minutes. To purify the DNA, 750 µL of chloroform/isoamyl alcohol mixture was added to the samples and centrifuged at 16,000 g for 10 minutes. After this, the aqueous phase that contained the DNA was carefully transferred to a pre-labelled sterile 1.5 ml microfuge tube, followed by adding 600 µL of absolute ice-cold isopropanol to precipitate the DNA. After incubating at −20^ºC for 2 hours, the mixture was centrifuged for 10 minutes at 16,000 g to pellet the DNA. The pellet was washed with 1 ml of 70% ice-cold ethanol and dried at room temperature for 1 hour. The DNA was eluted in 50 µL of 0.25X TE buffer. Before use in PCRs the quality and quantity of the extracted DNA were determined by electrophoresis on a 1% (w/v) agarose gel and a NanoDrop spectrophotometer (ThermoFisher Scientific).

DNA sequencing and sequence analysis

The RNAP encoding genes (rpoA, rpoB, rpoC, rpoD) were PCR-amplified with previously published gene-specific primer sequences and procedures [7] for the rpoB/RRDR, and in-house primer sequences for rpoA (5’-CCGGTCACCATGTACCTACG-3’ forward, 5’-GGATGTCAAGCAGGTCGGAT-3’ reverse), rpoC (5’-CTACGTGATCACCTCGGTCG-3’ forward, 5’-GTTGACGATGATTTCCGGCG-3’ reverse), and rpoD (5’-CGATCGCGCGAAAAACCATCT-3’ forward, 5’-CACCGACTGCAGTTGATCCT-3’reverse). The targeted gene segments were successfully PCR-amplified from all the study isolates. The total reaction volume for all PCRs was 60 µL, prepared according to the HotStar PCR kit (QIAGEN, Hilden, Germany). Briefly, each reaction contained 27.5 µL nuclease-free water, 6 µL 10x PCR buffer, 12 µL Q-solution, 3 µL 10 mM MgCl₂, 3 µL of 10 mM dNTPs, 1.5 µL each of reverse and forward primers, 0.5 µL High-fidelity Taq DNA polymerase (5U/µL) (Sigma-Aldrich, USA), and 10 ng/µL (in 5 µL volume) of the chromosomal DNA template. Amplification was achieved in a Thermocycler (Bio-Rad Laboratories Inc., Singapore) using the following program: initial denaturation at 95°C, 5 minutes, followed by 35 cycles each consisting of 95°C, 45 seconds, 60°C, 45 seconds, and 72°C, 50 seconds, with a final extension step of 72°C, 10 minutes. Five microliters each of the PCR product was analysed using 1% agarose gel electrophoresis with Ethidium bromide 5 mg/ml staining. Gels were run at 120 V for 1 hour and visualized using a UVP Gel documentation (Benchtop Trans-illuminator System_BioDoc-it, CA, USA). Then, 50 µL each of the PCR products was purified using the QIAmp DNA purification minikit (QIAGEN, Hilden, Germany), and the pure amplicons were sequenced at ACGT Inc. (Wheeling IL, USA).

The sequences obtained were analysed first through BLAST searches at NCBI https://blast.ncbi.nlm.nih.gov/Blast.cgi to confirm they significantly align to the expected gene/protein sequences in M. tuberculosis. To determine the amino acid substitutions, the DNA sequences were translated into amino acid sequences using the Molecular Evolutionary Genetics Analysis Version 6.0 (MEGA6.06) software [7] or the ExPASy online database http://web.expasy.org/translate/. To identify mutations, the sequenced amplicons were aligned in MEGA6.06 and BioEdit v7.2.5.0 to the rpoA, rpoB, rpoC and rpoD genes sequences from M. tuberculosis H37Rv (NC_000962), a reference strain pan-susceptible to anti-TB drugs. To determine whether the identified mutations were resistance-conferring, we followed the tbdream database [18] https://tbdreamdb.ki.se/Info/ and published literature [1,2,6–9,11,13,14]. All the data was curated, compiled and presented as tables or percentages depending on the frequency of mutations that occurred in the examined genes.

Whole genome sequencing of M. tuberculosis genomic DNA

Since some of the clinical isolates of rifampicin-resistant M. tuberculosis lacked rifampicin-resistance-conferring mutations in the rpoB/RRDR upon analysis of the Sanger sequenced DNA amplicons, as a validation step, we performed whole genome sequencing (WGS) of all the 52 M. tuberculosis clinical isolates and the control strain H37Rv. WGS was performed on a MiSeq platform (Illumina) at the Genomics Unit of the Department of Immunology and Molecular Biology at Makerere University College of Health Sciences, by following the protocols published by Illumina Inc. Briefly, 20 indexed paired-end libraries were prepared starting with 0.2 ng/µL of input M. tuberculosis chromosomal DNA using the Nextera® XT DNA sample preparation guide (Illumina Inc., CA USA). This protocol included stages of tagmentation, indexing, normalization and library pooling before loading the pooled library into a MiSeq Reagent Kit v2 (MS-102–2003, 500 cycle) (Illumina). This kit offers an improved chemistry to increase cluster density with a maximum output of 8.5 gigabytes (GB) of data. The cartridge was loaded in the MiSeq instrument (Illumina inc.) to run a resequencing workflow for ~72 hours. The DNA library for WGS was spiked with PhiX v3 control (Illumina) to a final concentration of 5% recommended for low diversity libraries. The generated FASTQ files were analysed and high-quality files (Phred score of ≥30) uploaded on PhyResSE, a web-based tool that delineates M. tuberculosis antibiotic resistance markers and their lineages using WGS data (https://bioinf.fz-borstel.de/mchips/phyresse/).

Quality control

PCR-amplicons from M. tuberculosis H37Rv, a reference strain that is susceptible to all anti-TB drugs, were sequenced and included as controls. Likewise, a known MDR-TB strain that contained well-known rifampicin and isoniazid resistance-conferring mutations, earlier confirmed by the Genotype MTBDRplus assay (Hain Life Sciences) was included in the study as the positive control for drug resistance mutations. Negative controls for DNA extraction, PCRs, and DNA sequencing were included to eliminate contamination. High-Fidelity Taq DNA polymerase (Sigma-Aldrich, USA) which delivers superior results with a twofold higher yield and threefold greater fidelity compared to regular Taq DNA Polymerase was used in the PCRs. Sanger DNA sequencing-identified mutations were revalidated through WGS.

Ethics statement

This study was approved by the Makerere University School of Biomedical Sciences Research and Ethics Committee (approval # SBS-HDREC-667). This committee approved the use of archived Mycobacterium tuberculosis isolates previously collected from patients who participated in the National Drug Resistance Survey that investigated the phenotypic levels and patterns of resistance to first- and second-line anti-TB drugs among new and previously treated sputum smear-positive TB cases in Uganda [3,16]. The parent studies stated herein had previously obtained written informed consent from the participants for sample storage and use of stored samples in further studies. The data were analysed anonymously and the authors did not have access to information that could identify individual participants during or after data collection.

rpoB mutations

Fig 1 and Table 1 summarize the patterns of mutations and SNPs we found in rpoB. A total of 24 MDR-TB isolates (75%, 24/32) had mutations in the rpoB/RRDR, i.e., Lys1025Glu (3%, 1/32); Ser531Leu (31%, 10/32); His526Asp (6%, 2/32); His526Leu (3%, 1/32); His526Tyr (3%, 1/32); His526Arg (3%, 1/32); His526Gly (3%, 1/32); Asp516Tyr (13%, 4/32); Asp516Val (6%, 2/32); Glu513Lys (3%, 1/32); Leu511Pro (3%, 1/32); Leu492Leu (3%, 1/32); and Gln490Arg (3%, 1/32), and these are similar to mutations that have been reported before [2,7,10,15] (note that three isolates had double mutations – see ahead). With the exception of Lys1025Glu, Leu492Leu and Gln490Arg, all mutations occurred in the rpoB/RRDR and confer high-level resistance to rifampicin [2,8] apart from Leu511Pro, Asp516Tyr, and His526Gln that are categorized as low-level rifampicin-resistance-conferring mutations [2,8]. Therefore, apart from two mutations (i.e., Glu513Lys and Leu511P), rifampicin-resistance in this study is attributed to mutations at three notable codons – 516, 526, and 531, which is in line with reports from many settings around the world [2,7]; notably, there is high polymorphism at codon 526 compared to the other two codons (516 and 531), Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Mutations in the rpoB and rpoC genes of the MDR-TB clinical isolates.

https://doi.org/10.1371/journal.pone.0328957.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Rifampicin-resistance-conferring mutations in rpoB (panel A) and putative compensatory mutations in rpoB (panel A) and rpoC (panel B) of MDR M. tuberculosis clinical isolates.

https://doi.org/10.1371/journal.pone.0328957.g001

Furthermore, three isolates (#s 3, 19, 20) were double mutants (Table 1, Fig 1) in that each harboured two mutations i.e., Asp516Val & Lys1025Glu; Leu492Leu & Ser531Leu; and Gln490Arg & Ser 531Leu, respectively. Asp516Val and Ser531Leu are high-level rifampicin-resistance-conferring mutations (see above) while the other three mutations in the double mutants occurred outside of the rpoB/RRDR and are not likely to be resistance-conferring [2]; non-synonymous rpoB polymorphisms like Gln490Arg and Lys1025Glu that are outside of the rpoB/RRDR and are not resistance-conferring have been reported to have compensatory effects [6] hence, are putative compensatory mutations.

rpoC mutations and co-occurrence with mutations in rpoB/RRDR

A total of 16 MDR-TB isolates (50%, 16/32) had mutations in rpoC – fifteen non-synonymous i.e., Leu39Phe (3%, 1/32); Tyr61His (3%, 1/32); Asp271Gly (3%, 1/32); Ser377Ala (3%, 1/32); Pro481Thr (3%, 1/32); Val483Ala (6%, 2/32); Leu516Pro (3%, 1/32); Ala521Asp (3%, 1/32); Gly594Glu (13%, 4/32); Asn698Ser (3%, 1/32); Leu823Pro (3%, 1/32) and four synonymous, i.e., Val68Val (3%, 1/32); Arg84Arg (3%, 1/32); Val197Val (3%, 1/32); and Ala542Ala (3%, 1/32). Hence, the most frequent rpoC mutations were Gly594Glu (13%) and Val483Ala (6%), Fig 1 & Table 1. Of note, mutations Pro481Thr, Val483Ala, Leu516Pro, Ala521Asp, Gly594Glu and Asn698Ser are the most frequently reported rpoC compensatory mutations in rifampicin-resistant M. tuberculosis clinical isolates in many settings around the world [1,6,8–14]. While most of the mutations have been previously reported, we identified four new polymorphisms i.e., Leu39Phe, Tyr61His, Asp271Gly, and Ser377Ala, Fig 1 & Table 1, which are likely to be compensatory mutations since the RNAP encoding genes are highly conserved in bacteria particularly M. tuberculosis [1,9,19]. Furthermore, 15 (63%) of the 24 MDR-TB isolates with rifampicin-resistance conferring mutations in rpoB/RRDR had non-synonymous mutations in rpoC; in line with current evidence that compensatory mutations do occur only in rifampicin-resistant clinical isolates with rifampicin-resistance conferring mutations in rpoB/RRDR [1,9,12–14], we did not find mutations/SNPs in rpoC of the eight MDR-TB isolates that lacked rifampicin-resistance-conferring mutations, Table 1.

Overall, in this study, the proportion of MDR-TB isolates with putative compensatory mutations in rpoC i.e., 66.7% (16/24) is high compared to rates from other countries [1,8,9,12–14], and the fact that MDR-TB prevalence is low in Uganda [20,21]; however, other investigators reported comparable or higher rates, e.g., Wang et al in China [13] and Vargas et al in Peru [12] reported that 98.2% (54/55) and 54% (95/175) respectively, of rifampicin-resistant M. tuberculosis isolates with resistance-conferring mutations in rpoB/RRDR had putative compensatory nonsynonymous mutations in rpoA/rpoC.

Furthermore, almost all isolates (90%, 9/10) with the rpoB/Ser531Leu resistance-conferring mutation harboured putative compensatory mutations in rpoC (Table 1), and similar observations for this mutation have been reported [8–11,13,14]. Furthermore, Gly594Glu, the most prevalent rpoC mutation in our study (Table 1) and reported before as a putative compensatory mutation [1,6,8–14], occurred among both rifampicin-resistant and rifampicin-susceptible isolates in Peru [12]. This discrepancy could be attributed to issues related to phenotypic drug susceptibility testing approaches i.e., the L-J proportional method used in the current study [3,16] vs. slide drug susceptibility testing (SDST) in the Peruvian study [12]. Generally, phenotypic susceptibility testing methods, especially SDST, could yield false-negative results [22] for isolates with low-level resistance-conferring mutations leading to mis-identification of resistant isolates as susceptible. Furthermore, the synonymous mutation Ala542Ala was previously reported as a lineage-defining polymorphism (i.e., phylogenetic marker) for the Latin American Mediterranean (LAM) sub-lineage of the Euro-American M. tuberculosis super-family [9]; however, in this study, we did not detect Ala542Ala in LAM strains but instead was detected in M. tuberculosis sub-lineage Uganda, Table 1. This discrepancy could be related to the sensitivity of genotyping approaches used i.e., Spoligotyping/IS6110 RFLP fingerprinting in the South African study [9], which is inferior and has a low discriminatory power compared to DNA sequencing used in the current study [23].

One limitation in our study is that we had fewer rifampicin-resistant TB isolates, since Uganda as a country has low MDR-TB rates. However, our findings have implications for the understanding of fitness, transmissibility and antimicrobial resistance mechanisms in a TB endemic setting with high HIV-TB coinfection rates. Further studies in Uganda could consider compensatory mutations as potential factors contributing to strain fitness or increased transmission of resistant TB strains. This is particularly important in the context of strain genetic background, given that Uganda has a predominant strain causing majority of TB disease in the country, i.e., M. tuberculosis sub-lineage Uganda, a Lineage 4 (EAM) sub-clade [24–26]. In addition to Lineage 4, M. tuberculosis comprises seven other major genetic clades, i.e., Lineages 1, 2, 3, 5, 6, 7 and 8, which are distributed differently around the world, and this has led to local adaptation of the pathogen [27] and infecting one quarter of the global human population (approximately 2 billion people) despite evidence of the host mounting a strong immune response to TB infection [28,29]. Little is known about the mechanisms underlying the predominance of the M. tuberculosis sub-lineage Uganda strains in Uganda. Compensatory mutations could play a role, occurring more selectively in rifampicin-resistant sub-lineage Uganda strains compared to other genotypes, conferring more fitness hence, improved host adaptability and/or transmissibility.