Определение чувствительности Mycobacterium tuberculosis к противотуберкулёзным препаратам с помощью полногеномного секвенирования и программного обеспечения «Mykrobe»

Полный текст

Introduction

Drug resistance is a serious problem challenging antimicrobial therapy around the world¹. Treatment of tuberculosis patients with antituberculosis drugs is one of the main strategies for disease control in Bulgaria and worldwide [1–4]. Mycobacterium tuberculosis infection and resistance in some cases is associated with coinfection like HIV [5–7], and this fact should be taken into account when therapy is subscribed. Antimicrobial drugs gradually lose activity against pathogens as a result of increasing microbial resistance. Along with classical methods of determination of M. tuberculosis complex drug resistance, new ones related to DNA sequence assays were recently introduced². PCR and targeted sequencing of genes causing antimicrobial resistance are widely used and became common. The introduction of next-generation sequencing technologies gave to researchers new opportunities for more powerful and detailed analysis [8–11]. Another advantage of NGS analysis is that when a strain once being sequenced, the information could be stored in formats such as FASTQ, FAST5 or others, depending on the technology, this strain can be analyzed in further studies with different software tools for different genetic, phylogenetic and epidemiological investigations. Illumina sequencing technology is widely applied. Soon application of next-generation sequencing technology will be mandatory in the description of bacterial or viral strains and will be widely used in other fields of medicine and will displace many phenotypic and current molecular genetic methods. Software tools were developed for drug resistance determination using data of whole-genome sequenced microorganisms, such as ABRicate, ARIBA [12], ARGs-OAP [12], ARG-ANNOT [14], CASTB [15], KvarQ [16], MTBseq [17], PhyResSe [18], RAST [19], ResFinderST [20], RGI [21], SRST2 [22], SSTAR [23], TB Profiler [24] and Mykrobe tested by us [25]. Mykrobe has several advantages in comparison with previous software:

1) it has an updated catalogue, increasing the sensitivity for determination of pyrazinamide resistance;

2) it allows the users to add their catalogues;

3) it has improved identification for non-tuberculosis mycobacterial species.

The Mykrobe software specificity and sensitivity were estimated in a previous study comparing phenotypic and whole-genome sequencing results obtained for 4362 isolates of M. tuberculosis. The estimated sensitivity of Mykrobe was 100, 95, 82, 99%, and the specificity is 99, 100 99, 99% respectively for rifampicin, isoniazid, pyrazinamide, and ethambutol [26]. Mykrobe software is not popular in Bulgaria and has not been applied. The present work aimed to determine the phenotypic and genotypic susceptibility of M. tuberculosis to antituberculosis drugs. Tasks of present work included determination of phenotypic susceptibility of M. tuberculosis strains isolated from Bulgarian patients, whole-genome sequencing of the strains and application of the software tool “Mykrobe” for drug-susceptibility testing.

Materials and Methods

This study includes 34 multi-resistant M. tuberculosis (MDR-TB) strains isolated from 34 Bulgarian patients’ samples; out of them 9 were isolated in 2009, 23 — in 2010, and 2 strains were isolated in 2011. Their resistance has been phenotypically confirmed by BACTEC MGIT 960 System. The genotypic resistance for most of them was determined by the line probe assay (LPA) (71% of them were tested for first-line drugs and 21% were tested for second-line drugs) at the National Reference Laboratory of tuberculosis, NCIPD.

Isolation of M. tuberculosis strains from clinical samples

The studied clinical materials were processed by homogenization and decontamination, in accordance with the standard operating procedures described in the Methodological instructions for microbiological diagnosis and treatment of tuberculosis (Bulgarian Ministry of Health, 2009). Each sample was inoculated in two tubes Löwenstein–Jensen solid media and one tube with liquid media (Mycobacteria Growth Indicator Tube — MGIT). We used the products of “Becton Dickinson”. The tubes were cultivated at 37°C. The result of cultivation on solid media was evaluated by the scale of semi-quantitative assessment of growth according to the above-mentioned guidelines.

MGIT liquid media tubes had a bar code and a fluorescence sensor on the bottom. The result was automatically generated by BACTEC MGIT 960 system and the Becton Dickinson's software. From the positive test tubes we performed the immunochromatographic test BD MGIT TB Identification Test (“Becton Dickinson”) in order to detect Mycobacterium tuberculosis complex. The Drug Sensitivity Tests (DST) was done using BACTEC and Geno Type MTBDR plus v.1.0 and Geno Type MTBDR sl v.1.0 (“Hain Lifescience”).

Phenotypic DST of M. tuberculosis complex to first- and second-line anti-TB drugs

The phenotypic susceptibility of the strains to the following anti-tuberculosis drugs: streptomycin, isoniazid, rifampicin, ethambutol, ofloxacin, amikacin, kanamycin and capreomycin was evaluated by the proportion method using the fully automated BACTEC MGIT 960 system, according the manufacturer's instructions (BACTEC System User Guide MGIT 960, 2004). The tested critical concentrations were as follows: streptomycin — 1.0 μg/ml; isoniazid — 0.1 μg/ ml; rifampicin — 1.0 μg/ml; ethambutol — 5.0 μg/ml; ofloxacin — 2.0 μg/ml; amikacin — 1.0 μg/ml; kanamycin — 5.0 μg/ml; capreomycin — 2.5 μg/ml.

Molecular genetic methods for detecting resistance to the first and second line anti-tuberculosis drugs We used line probe assays (LPA) Geno Type MTBDR plus v.1.0 and Geno Type MTBDR sl v.1.0 (“Hain Lifescience”) according the manufacturer's instructions. The tests were based on DNA-STRIP technology. Rifampicin resistance was found by detection of a mutation in the rpoB gene encoding the β-subunit of RNA polymerase. The resistance to isoniazid was searched in two genes: a mutation in the katG gene encoding peroxidase, which causes a high level of resistance, and a mutation in the inhA gene encoding enoyl- (acyl-protein carrier) reductase (NADH), causing a low level resistance. Resistance to the fluoroquinolones was detected by GenoType MTBDR sl, v.1.0, scoping for mutation in the gyrA gene encoding DNA gyrase. The resistance to aminoglycosides and cyclic peptides was detected by proved mutation in the rrs gene encoding the 16S rRNA.

DNA isolation

Phenotypically proven multidrug-resistant M. tuberculosis strains were grown on Löwenstein–Jensen medium for 35–42 days at 37°C. A full inoculation loop of fresh culture was resuspended in 400 μl TE buffer [ 10 mM Tris-HCl, 1 mM EDTA (pH 7.0)] in a 1.5–2.0 ml screw cap tube. The samples were incubated for 20 min at 80°C for inactivation of mycobacterial culture. DNA was isolated performing method described by van Soolingen and modified by us. [27]. In each tube to the lysozyme was added 1U RNase H followed by incubation for 4 h at 37°C than 70 μl of 10% sodium dodecyl sulfate (SDS) and 10 μl proteinase K at a concentration of 10 mg/ml were added and tubes were incubated for 24 h at 65°C. All other steps were performed following the original methodology. DNA quality was checked spectrophotometrically at optical density 1.8–1.9 OD measured on 260/280 nm.

Strain sequencing and bioinformatic analysis

Whole genomes of 34 M. tuberculosis resistant strains collected by the Bulgarian National Reference Laboratory of Tuberculosis were sequenced at the Supranational Reference Laboratory of Tuberculosis, San Raffaele Institute, Milan, Italy. Sequencing results were provided as FASTQ files written in fasta.qz format. Information about the whole genome of each strain was stored in two files: one with the amplicons ordered in 5 ' → 3 ' direction, and a second ordered in 3 ' → 5. The majority of the amplicons were 117 bp in size without adapter regions on both ends. Bacterial genome size was about 4.5 million base pairs. The size of each file containing genome data was about 250 MB. We installed Mykrobe software (www.mykrobe.com), version v.0.8.1 for desktop under Windows 10 [28] which is freely available for non-commercial use to analyze genome sequences. We loaded pair of readings sequenced in both directions. After overlaying of fragments on the catalogues used by Mykrobe offline, containing information about the genes determining drug resistance, drug resistance and related gene mutations were presented in a tabular form. The agreement between resistant phenotypic cultural method and Mykrobe software prediction tool was illustrated by Venn diagram³ for each tuberculostatic. Reference vaccine strain BCG SL222 Sofia [29], originating from Russian vaccine strains BCG-I seed lot 374(a) was used as a negative reference control.

Results and Discussion

We determined the drug-susceptibility of 34 strains of M. tuberculosis collected at the National Reference Laboratory of Tuberculosis, NCIPD, Sofia, Bulgaria. All investigated strains were MDR, and in two of them, we proved extensive drug resistance (XDR). The phenotypic and genotypic drug resistance of strains is shown in Table.

 

Designations of strains are shown in the first column of Table. All phenotypically proven rifampicin and isoniazid-resistant strains were 100% confirmed by Mykrobe software. In thirty-one strains, rifampicin resistance was caused by S450L mutation — following the M. tuberculosis H37R nomenclature (or S531L according to the Escherichia coli nomenclature used by Hain Lifescience). One strain demonstrated H445Y mutation — according to M. tuberculosis H37Rv (H526), one was with H445D (H526D), and one with N432L (N513L) mutations in the rpoB gene.

S315T mutation responsible for isoniazid resistance was observed in the katG gene in strain N:22_09. In strain, N:60_10 double mutation in the inhA gene was detected. I194T mutation in the inhA gene was proven in combination with C-15T mutation in the promoter region of the inhA gene. C-15T mutation in the promoter region of the inhA gene was found in 33 strains, which is responsible for isoniazid resistance. C-15T mutation is prevalent and most widespread (> 50%) among the MDR strains of M. tuberculosis isolated in Bulgaria with spoligotype SIT41 (TUR) [30]. These results were consistent with previous studies [30–33].

We can conclude that the C-15T mutation itself is a marker for isoniazid resistance in investigated strains with SIT41 spoligotype and is predictive for MDR. We observed a 75% (21/28) overlapping rate between BACTEC and Mykrobe for ethambutol resistance. In 4 strains (13%), M306V mutation in the embB gene was proven, which was expected to cause ethambutol resistance, but these strains were phenotypically susceptible. We can suggest that in these four strains M306V mutation is a polymorphism unrelated to phenotypic ethambutol resistance. Strain 22_09 was phenotypically identified as resistant, but mutation M306V or another related was not identified. This strain might have other, unknown mutations leading to ethambutol resistance

The sensitivity of some strains to streptomycin was very interesting. Phenotypically 87% (n = 27) of streptomycin-resistant strains were proven and only 59% (n = 19) were detected by Mykrobe. In 8 strains streptomycin sensitivity was detected by analysis of FASTQ files with Mykrobe, while BACTEC showed phenotypic resistance to this antituberculosis drug. For 19 strains proven by Mykrobe streptomycin resistance, we identified in 2 of them that streptomycin resistance was caused by A450X nucleotide mutation, in 9 by A514X, and in 8 by C517X nucleotide mutation in rrs gene encoding 16S rRNA which is responsible for streptomycin resistance. We can suggest that other unknown mutations, different from the described above in rpsL, rrs, and gidB genes, are responsible for streptomycin resistance. In other cases, streptomycin resistance can be caused by efflux or change in streptomycin targeting. Our investigation showed that the Mykrobe software tool has limited capability to test streptomycin resistance, giving a 30% error rate.

Different mechanisms of fluoroquinolone resistance were found in different strains [34]. In strains 21_10, 23_10, 24_10, 52_10, 60_10, and 62_10, resistance to ofloxacin, moxifloxacin and ciprofloxacin was caused by A90V mutation in gyrA gene and in strains 38_10, 60_10, 32_11 and 62_10, by D94H mutation in the same gene. Comparing phenotypic and genotypic resistance to ofloxacin, amikacin and kanamycin, we observed 100% coincidence of results. Six strains were phenotypically and genotypically proven as pre-XDR with 100% coincidence to fluoroquinolone resistance. A disadvantage of our study is the small number of proven MDR strains resistant to this drug. In the studied group of MDR strains, we observed two that were phenotypically and genotypically confirmed as XDR.

The susceptibility of the strains to pyrazinamide was evaluated only with Mykrobe. Seventeen (50%) pyrazinamide resistance strains were observed. Resistance was caused by six different mutations in the pncA gene. The most often detected mutation was P69L in 6 strains. G97C was the second most common mutation proven in 4 strains. Based on these results, we can conclude that resistance to pyrazinamide is a result of several mutations in the pncA gene.

All studied by us M. tuberculosis strains with Geno Type test and Mykrobe tool showed identical results for antituberculosis drug resistance. Mykrobe gave more full and detailed information (Figure). The disadvantage of Geno Type is that it does not cover all mutations responsible for the resistance of M. tuberculosis to the currently applied anti-TB drugs.

Conclusion

Currently, different methods of identifying drug resistance have been developed and introduced into practice. Next-generation sequencing and bioinformatics data analysis are fast developing technologies and they will be used more widely soon. Next-generation sequencing technologies will be mandatory in the characterization and registration of new strains. Comparison of different methods showed that in some cases one could identify a mismatch between expected and observed phenotypes and genotypes. The genotypes include marker genes or other genome regions involved in drug resistance. Resistance can be caused by other unknown markers. This fact does not allow us to ignore phenotypic methods for the determination of antimicrobial resistance, and to prescribe drugs based on DNA analyses, despite enormous possibilities provided by whole-genome sequencing and bioinformatics. However, phenotypic methods do not determine the mechanism of resistance. Application of whole genome sequencing assay allows observing different genetic modifications associated with different mechanisms causing drug-resistance of M. tuberculosis complex. Different mutations cause different levels of resistance, and this will be the subject of our future investigations. Next-generation sequencing allows not only to compare data with phenotypically detected resistance but also to find relations between a mutation(s) and level of resistance.

Об авторах

V. Tolchkov

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Автор, ответственный за переписку.
Email: tolchkov@gmail.com
ORCID iD: 0000-0001-8005-6036

Vladimir Tolchkov - PhD, assistant professor, National Reference Laboratory (NRL) for high risk bacterial infections

Sofia

Болгария

Y. Hodzhev

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Email: fake@neicon.ru
ORCID iD: 0000-0002-7770-3316

Yordan Hodzhev - PhD student, NRL for high-risk bacterial infections

Sofia

Болгария

B. Tsafarova

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Email: fake@neicon.ru
ORCID iD: 0000-0002-8183-3869

Borislava Tsafarova - PhD student, NRL for high-risk bacterial infections

Sofia

Болгария

E. Bachiyska

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Email: fake@neicon.ru
ORCID iD: 0000-0003-2949-5011

Elizabeta Bachiyska - PhD, Associate Professor, Head, NRL for tuberculosis

Sofia

Болгария

Yu. Atanasova

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Email: fake@neicon.ru
ORCID iD: 0000-0001-6255-375X

Yuliana Atanasova - PhD student, biologist, NRL for tuberculosis

Sofia

Болгария

A. Baykova

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Email: fake@neicon.ru
ORCID iD: 0000-0002-4596-0270

Ana Baykova - biologist, NRL for tuberculosis

Sofia

Болгария

S. Yordanova

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Email: fake@neicon.ru
ORCID iD: 0000-0003-4856-3665

Stanislava Yordanova - PhD, Assistant Professor, NRL for tuberculosis

Sofia

Болгария

A. Trovato

WHO Collaborating Centre and TB Supranational Reference Laboratory San Raffaele Scientific Institute

Email: fake@neicon.ru
ORCID iD: 0000-0003-4494-5762

Alberto Trovato - PhD, senior researcher

Milano

Италия

D. Cirillo

WHO Collaborating Centre and TB Supranational Reference Laboratory San Raffaele Scientific Institute

Email: fake@neicon.ru
ORCID iD: 0000-0001-6415-1535

Daniela Cirillo - PhD, Head

Italy

Италия

S. Panaiotov

Department of Microbiology, National Centre of Infectious and Parasitic Diseases

Email: fake@neicon.ru
ORCID iD: 0000-0001-9997-5198

Stefan Panaiotov - DSc, professor

Sofia

Болгария

Список литературы

Дополнительные файлы