Have compensatory mutations facilitated the current epidemic of multidrug-resistant tuberculosis?

Have compensatory mutations facilitated the current epidemic of multidrug-resistant tuberculosis? Compensatory mutations have been suggested to promote multidrug-resistant tuberculosis (MDR-TB) transmission, but their role in facilitating the recent transmission of MDR-TB is unclear. To investigate the epidemiological significance of compensatory mutations, we analyzed a four-year population-based collection of MDR-TB strains from Shanghai (the most populous city in China) and 1346 published global MDR-TB strains. We report that MDR-TB strains with compensatory mutations in the rpoA, rpoB,or rpoC genes were neither more frequently clustered nor found in larger clusters than those without compensatory mutations. Our results suggest that compensatory mutations are not a major contributor to the current epidemic of MDR-TB. Introduction outbreaks in HIV-negative patients were regarded as It had been thought that drug-resistance mutations successful examples of compensatory evolution . would introduce a fitness cost into resistant Mycobacter- Recent studies have investigated the role of compensa- ium tuberculosis (M. tb) tory mutations in MDR-TB transmission by comparing , and the resulting reductions in virulence and transmissibility would prevent multidrug- the presence of compensatory mutations in clustered or 8,9 resistant tuberculosis (MDR-TB) strains from dis- nonclustered strains and found that putative compen- 2,3 seminating widely . Early mathematical models pre- satory mutations in rpoC and rpoA were more common in dicted that MDR-TB should remain a local problem , but VNTR-clustered strains. However, these studies did not the steady growth of the MDR-TB epidemic worldwide determine whether the putative compensatory mutations has contradicted these early expectations. A subsequent accumulated before or after the MDR-TB strains were study suggested that even if the average fitness of MDR- transmitted. If these compensatory mutations accumu- TB strains is low, a small proportion of resistant strains lated after transmission, they should not be considered as that are relatively more fit will outcompete the less fit and factors that promote or facilitate transmission. Moreover, drug-susceptible strains . These more fit, resistant strains the sampling methods used in these studies were not were thought to contain compensatory mutations that population based, and the IS6110/MIRU-VNTR defined would restore fitness and thus constitute an important clusters could have perhaps been further separated by 5,6 factor in the spread of MDR-TB strains , and MDR-TB whole-genome sequencing (WGS). To avoid these pitfalls, we examined the role of compensatory mutations in MDR-TB transmission by using WGS to determine Correspondence: Qian Gao (qgao99@yahoo.com) transmission clusters in the MDR-TB strains collected in Shenzhen Center for Chronic Disease Control, Shenzhen, China Key Laboratory of Characteristic Infectious Disease & Bio-safety Development Shanghai over a 4-year period (2009–2012). Our findings of Guizhou Province Education Department, Zunyi Medical University, Zunyi, contradict the previous inferences and do not support a Guizhou, China Full list of author information is available at the end of the article. These authors contributed equally: Qingyun Liu, Tianyu Zuo. © The Author(s) 2018 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. 1234567890():,; 1234567890():,; Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 2 of 8 significant role of compensatory mutations in promoting (compensated)-type clusters, in which all strains harbor the ongoing MDR-TB epidemic. the same compensatory mutation—indicating transmis- sion of a compensated MDR-TB strain (compensatory Results mutations occurring before transmission); (2) N (non- Collection of MDR-TB isolates compensated)-type clusters, containing only strains with From 2009 to 2012, a total of 324 MDR-TB isolates no putative compensatory mutation—indicating trans- mission of a noncompensated MDR-TB strain; (3) M were collected from 31 tuberculosis (TB) hospitals in Shanghai, China. All of these isolates were MIRU-VNTR (mixed)-type clusters, in which MDR-TB strains harbored genotyped, and 122 clustered MDR-TB strains had different compensatory mutations or only a proportion of undergone WGS previously . Here, to determine the the strains in the cluster contained compensatory muta- presence of compensatory mutations in VNTR non- tions, indicating that the compensatory mutations had clustered MDR-TB strains, we randomly selected 105 occurred after transmission (see more details about isolates from the remaining 202 VNTR nonclustered M-type clusters in the discussion). The groups so defined MDR-TB isolates to sequence the full-length genes rpoA, are theoretically similar to the three prototypical types of rpoB, and rpoC. The combined results of the 227 clusters used to evaluate the role of drug resistance in Shanghai MDR-TB strains were termed the “Shanghai transmission . In the Shanghai dataset, 12 transmission dataset”. Meanwhile, we obtained WGS records of 8331 clusters were determined as C-type, 18 as N-type, and six M. tb isolates of global origin from the European as M-type (Fig. 2a); in the Global dataset, 26 transmission Nucleotide Archive (ENA) (Supplementary Table 1), and clusters were determined as C-type, 84 as N-type, and 1346 of these strains were identified as MDR-TB (Sup- seven as M-type. plementary Table 3). Of these MDR-TB strains, 602 were collected through retrospective cohorts or a population- Compensated MDR-TB strains did not cause larger based approach (Supplementary Table 1) and thus were transmission clusters appropriate for the subsequent analysis. The data from If the putative compensatory mutations stimulated the the global MDR-TB isolates were termed the “Global transmission of MDR-TB, then C-type clusters would be dataset”. expected to be larger than N-type clusters, but in the Shanghai strains, the C-type clusters were not larger than Identification of putative compensatory mutations and either the N-type or the M-type clusters (Wilcoxon rank- sum test, P = 0.4053 and 0.9587, respectively) (Fig. 2b). transmission clusters Through phylogenetic reconstruction of the MDR-TB An analysis of the size of the clusters in the global MDR- strains obtained above, we found 60 nonsynonymous TB strains yielded similar results (Wilcoxon rank-sum mutations in the rpoA, rpoB,or rpoC genes that had arisen test, P = 0.3816 and 0.2277, respectively) (Fig. 2c). Nota- at least twice in parallel and were identified as putative bly, the largest clusters observed in both the Shanghai and compensatory mutations (Fig. 1, Supplementary Table 4). Global datasets (with seven and 23 linked cases, respec- Of these mutations, six were in rpoA,16in rpoB, and 38 in tively) were caused by MDR-TB strains without com- rpoC. We set “mutation parallelism” as a criterion to pensatory mutations (Fig. 2b, c). Moreover, we did not exclude lineage mutations or neutral polymorphisms that find any particular compensatory mutation that was were fixed due to genetic drift. Although this filter might associated with larger transmission clusters in either also exclude some uncommon compensatory mutations, dataset. These results suggest that compensated MDR-TB it increased the level of confidence that the mutations strains are not prone to generate more secondary cases identified were truly compensatory mutations. We inclu- than noncompensated MDR-TB strains do. ded rpoB in the search for putative compensatory muta- tions because additional mutations in rpoB have been Compensated MDR-TB strains were not more likely to be reported to both restore fitness and increase the level of clustered rifampicin resistance . MDR-TB strains with less than 12 Previous studies suggested that MDR-TB strains with Single-nucleotide polymorphisms (SNPs) difference were compensatory mutations were more frequently clus- 8,9 considered to constitute a transmission cluster. A total of tered . However, these studies treated “compensatory 36 such clusters were found in the Shanghai dataset, and mutations” as a static feature and ignored the possibility 117 clusters were identified in the Global dataset. that compensatory mutations could occur in secondary cases that were initially infected with noncompensated Three types of transmission clusters MDR-TB strains. This is probably what has occurred in To distinguish whether the compensatory mutations the had occurred before or after transmission, we divided the M-type clusters (Fig. 2a), and therefore, it should MDR-TB clusters into three types (Fig. 2a): (1) C not be considered as representing the transmission of Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 3 of 8 Fig. 1 Putative compensatory mutations in the rpoA, rpoB, and rpoC genes identified in this study. Each putative compensatory mutation was supported by at least two independent evolution events Fig. 2 Compensated MDR-TB strains were not associated with larger transmission clusters. a A maximum likelihood phylogenetic tree showing genomic clusters in the Shanghai dataset. The strain identifiers were “year, strain number, and compensatory mutation type”. The three genomic cluster types are illustrated with different colors, as indicated. The isolate names in gray represent the VNTR-clustered strains that were separated by WGS. Comparison of cluster sizes in C-type, M-type, and N-type clusters in the Shanghai MDR-TB dataset (b) and the Global MDR-TB dataset (c); each dot represents a genomic cluster identified by WGS Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 4 of 8 Table 1 Ratios of compensated strains in clustered and nonclustered MDR-TB groups a a 2 Groups Total With CMs Without CMs χ P value Clustered MDR-TB (%) 78 32 (41.0%) 46 (59.0%) 2.260 0.133 Nonclustered MDR-TB (%) 133 41 (30.8%) 92 (69.2%) CMs compensatory mutations Excluded M-type clusters Fig. 3 Recently formed MDR-TB strains would decrease the ratio of compensatory mutations in the nonclustered MDR-TB group. a A theoretical scheme shows that recently formed MDR-TB strains would be included, while the secondary clustered strains resulting from transmission would occur outside of the study’s observation period. Thus, newly formed MDR-TB strains would be assigned to the nonclustered group. This schematic diagram also shows that transmission that occurred within the first 2 years will be captured in our study, while the secondary cases resulting from transmission that occurred during the last 2 years could be beyond the study’s observation period. b Comparison of the number of drugs to which strains are resistant in clustered and nonclustered MDR-TB strains. c Comparison of the number of drug-resistance mutations in clustered and nonclustered MDR-TB strains. d Comparison of collection time distribution between clustered and nonclustered MDR-TB groups; the collection time of each isolate was counted as “days to the end of the study duration (31 December 2012)” compensated MDR-TB strains. Accordingly, we excluded mutations that occurred after transmission result in a M-type clusters from an analysis of the ratios of com- higher ratio of compensatory mutations in clustered than pensatory mutations in the clustered and nonclustered in nonclustered MDR-TB strains. groups, and compared only C-type and N-type clusters. Surprisingly, no significant difference was observed in the The dilution effect of recently formed MDR-TB strains ratios of compensatory mutations in clustered compared Even when the M-type clusters were excluded, the to nonclustered groups, indicating that compensated clustered group had a nonsignificantly higher ratio of MDR-TB strains were not more frequently clustered compensatory mutations than that of the nonclustered (Table 1). group (41.0% versus 30.8%). One explanation might be To allow a comparison with previous studies, we repe- that new MDR-TB strains are continuously emerging ated the analysis without excluding the M-type clusters. during antibiotic treatment, and these new MDR-TB When the M-type clusters were included, the ratio of strains might not have had enough time to accumulate compensatory mutations in the clustered group was sig- compensatory mutations. For the same reason, within the nificantly higher than that in the nonclustered group 4-year duration of the study, they might not have had (Supplementary Table 5), similar to the results of previous enough time to cause secondary TB cases (Fig. 3a). Thus, studies. In our M clusters, 62.5% (10/16) of MDR-TB these recently formed “young” MDR-TB strains would be strains harbored compensatory mutations, which was assigned to the nonclustered group and thereby decrease much higher than that of either clustered (41.0%) or the ratio of compensatory mutations in this group. In nonclustered (30.8%) groups. M-type clusters are those in contrast, all strains in MDR-TB clusters are relatively which compensatory mutations are present in only some “older” (had become MDR-TB prior to their transmis- of the strains in a cluster, presumably because the sion), and thus would have had a longer period to accu- mutations have occurred during the transmission of the mulate additional drug resistance and compensatory clustered strain. Therefore, when the M-type clusters are mutations than strains in the nonclustered group. If this is included with the clustered MDR-TB strains, the true, then clustered MDR-TB strains should have wider Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 5 of 8 drug-resistance spectra (number of drugs to which they We believe in the accuracy of the results of our study are resistant) than those of nonclustered MDR-TB strains. because it incorporated several methodological improve- To test this inference, we compared the drug-resistance ments. First, we used WGS to identify transmission spectra of the clustered versus nonclustered MDR-TB clusters, which is more precise than IS6110 or MIRU- strains in the Global dataset. This analysis demonstrated VNTR . In our data, a total of 28 VNTR-clustered strains that the median number of antibiotics to which (23% of the VNTR-clustered strains) were further sepa- nonclustered MDR-TB strains were resistant was 4.22 rated by WGS. Second, we set strict criteria to identify (4.10 ~ 4.34), while for clustered MDR-TB strains it was putative compensatory mutations and excluded phyloge- 5.13 (5.05 ~ 5.21) (Student’s t-test, P < 0.0001, Fig. 3b). In netic/lineage mutations or neutral polymorphisms. Third, addition, as a single M. tb strain can accumulate our analysis discriminated between the accumulation of multiple mutations that confer resistance to a single compensatory mutations before or after the transmission 12–14 antibiotic (evolution of high-level resistance) ,we of MDR-TB strains (i.e., C- versus M-type clusters) in the further compared the numbers of drug-resistance- analysis of cluster size and the ratio of compensatory conferring mutations in these two groups. Consistent mutations. with our prediction, the clustered MDR-TB strains had, The effect of drug resistance on transmission has been 4,17 on average, more drug-resistance mutations than the long debated , and different studies have reported nonclustered MDR-TB strains did (Student’s t-test, heterogeneous results suggesting that MDR-TB strains P < 0.0001, Fig. 3c). can be ten times more or ten times less transmissible than 11,18 Finally, to verify whether nonclustered MDR-TB strains drug-susceptible strains . However, the reasons tended to occur toward the end of the study period, i.e., underlying these variations could simply reflect the dif- were indeed younger, we compared the distributions of ferences in study settings and methodologies 4,18 the collection times of the initial cultured clinical speci- employed; the drug-resistant strains reported to have mens (counted as days to the end of the study duration) reduced transmissibility were found in settings with for the isolates in the clustered versus nonclustered effective TB control , while strains with increased groups. We found that the isolates in the nonclustered transmissibility were reported from regions with high TB 4,18 group had a shorter average time since collection than the burdens . Thus, it seems more likely that the hetero- isolates in the clustered group did (mean values: 762.08 geneous and discordant published results serve to versus 903.14, P < 0.0001, Wilcoxon rank-sum test, demonstrate that TB transmission is a process that is Fig. 3d), which suggests that the strains in the non- primarily influenced by environmental factors such as TB clustered group tended to be isolated toward the end of control policy, time with the illness before diagnosis, the study period. These analyses support our inference treatment efficacy, and general quality of the health-care that clustered MDR-TB strains differ from nonclustered system . MDR-TB strains, suggesting a dilution effect from We consider that there are at least four factors that can recently formed MDR-TB strains that could in turn pro- obscure the function of compensatory mutations in MDR- vide an explanation for the relatively lower ratio of TB transmission. First, the efficacy of the TB control compensatory mutations in the nonclustered MDR-TB program. In countries or regions with high MDR-TB group. prevalence and poor management of MDR-TB patients, the transmission of MDR-TB strains is highly likely, even Discussion in the absence of compensatory mutations. In such cir- Our findings suggest that, in contrast to published cumstances, outbreaks and epidemics of MDR-TB are 5,15 20,21 reports , MDR-TB strains with compensatory muta- mainly driven by environmental factors . Second, tions are not more frequently found in clusters, nor are MDR-TB patients are, on average, infectious for longer they more likely to belong to larger transmission clusters periods due to the long course of treatment and high 22,23 than the MDR-TB strains without these mutations do. treatment failure rates . Thus, there might be more Previous studies have made direct comparisons of the chances for MDR-TB patients to generate secondary number of strains with putative compensatory mutations cases, which could conceivably compensate for a modest in clustered versus nonclustered MDR-TB strains and decrease in fitness. Third, the most common drug- reported an enrichment of compensatory mutations in the resistance mutations carried by clinical isolates are asso- 8,9 clustered group . However, we observed that M-type ciated with the lowest fitness cost in vitro, and such slight clusters had an impact on this comparison because decreases in fitness may not be sufficient to affect trans- 1,6 inclusion or exclusion of these clusters would dramati- mission . Moreover, positive epistatic effects between cally alter the results. Hence, we infer that the previous different drug-resistance mutations could further ame- observations would probably change if they excluded liorate the fitness cost imposed by individual drug- 17,24 M-type clusters. resistance mutations . Fourth, it is still possible that Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 6 of 8 the fitness cost measured in vitro might not reduce the Analysis of WGS data ability of M. tb to transmit and establish in a new host We used a validated pipeline for the mapping of 25,26 10,29 in vivo . This idea was supported by a previous study sequencing reads to the reference genome . In brief, using tuberculin skin testing to trace infection that found the Sickle tool was used for trimming WGS data. an equal prevalence of infection among contacts exposed Sequencing reads with Phred base quality above 20 and to patients harboring both drug-resistant and drug- reads length longer than 30 were retained for analysis. susceptible strains . The whole-genome sequence of the M. tb H37Rv strain The objective of this study was to investigate the (NC_000962.2) was used as the reference template for effects of compensatory mutations on MDR-TB trans- mapping reads. Sequencing reads were mapped to the missions that occurred within three or fewer years (recent reference genome using Bowtie2 (v2.2.9) . SAMtools 28 32 transmission) , and the 4-year observation window (v1.3.1) was used for SNP calling with mapping quality should have captured most of the transmissions that greater than 30. Fixed mutations (frequency ≥75%) were occurred within the first 2 years of the study. For identified using VarScan (v2.3.9) with at least ten reads transmissions that occurred during the last 2 years supporting and strand bias filter option on. We excluded of the study, the secondary cases might have developed all SNPs located in noise or repetitive regions of the TB disease after our observation window. (Fig. 3a). genome (e.g., PPE/PE-PGRS family genes, phage sequen- However, we believe that the transmission pattern ces, insertions or mobile genetic elements). in the last 2 years should not substantially differ from that in the first 2 years, and therefore, further extending the Identification of putative compensatory mutations time frame of our study would not change its major The maximum likelihood method was used to recon- conclusion. struct the phylogenetic trees with 500 bootstrap repeats In conclusion, we did not observe a promoting influence using MEGA (v6.06) , and the phylogenetic tree was of compensatory mutations on the transmission of MDR- further visualized by FigTree (http://tree.bio.ed.ac.uk/ TB, and we suggest that the putative contribution of software/figtree/). Putative compensatory mutations in compensatory mutations may be overwhelmed by the the rpoA, rpoB,or rpoC genes were identified through the complex and more powerful effects of environmental following criteria: (1) nonsynonymous mutations in the factors. rpoA, rpoB or rpoC genes that were present in only rifampicin-resistant strains; (2) each putative compensa- tory mutation must have arisen at least twice indepen- Methods MDR-TB datasets dently (parallel selection). We wrote a Python script to A total of 324 MDR-TB isolates were collected from 1 implement these criteria with the phylogenetic tree and a January 2009 to 31 December 2012, in Shanghai (the most mutation matrix, and the Python script was uploaded to populous city in China) during a population-based GitHub. observational study . MIRU-VNTR genotyping (12 loci) was performed for all MDR-TB strains, and the clustered Genotypic drug resistance detection and identification of strains were further subjected to WGS . In the present transmission clusters study, we further sequenced the entire gene sequences of Drug-resistance-associated mutations in M. tb were the rpoA, rpoB, and rpoC genes in 105 nonclustered obtained from the database TBDReaMDB and articles MDR-TB strains to detect putative compensatory muta- reporting rifampicin-resistance mutations (Supplemen- tions. The PCR amplicons were sent for Sanger sequen- tary Table 2). M. tb isolates carrying any characterized cing, and the results were analyzed to detect putative rifampicin-resistance mutation were identified as compensatory mutations using Geneious (http://www. rifampicin-resistant strains, whereas those isolates with geneious.com/). additional isoniazid-resistance mutations were deter- A total of 8331 published whole-genome sequences mined as MDR-TB strains. To determine the drug- of M. tb strains were downloaded from the ENA resistance spectra of the MDR-TB isolates in the Global (http://www.ebi.ac.uk/ena). The quality criteria for dataset, we additionally identified drug-resistance- data inclusion were set as follows: (1) the average associated mutations for nine drugs (amikacin, capreo- sequencing depth should be above 10-fold; (2) the mycin, ethambutol, ethionamide, fluoroquinolones, genome coverage rate should be above 95%. To avoid isoniazid, kanamycin, para-aminosalicylic acid, pyr- false transmission clustering, we included only the last azinamide, streptomycin) by inspection of their whole- isolate of longitudinal isolates collected from the same genome sequences. Any isolates with genetic distances of patient. Information about the geographic isolation of less than 12 SNPs were classified into a cluster .To these isolates was obtained from the articles or the compare the collection time distributions between the authors. isolates in the clustered and nonclustered groups, we Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 7 of 8 Publisher’s note determined the collection time of each isolate as “days to Springer Nature remains neutral with regard to jurisdictional claims in the end of the study duration (31 December 2012)”. For published maps and institutional affiliations. example, if an isolate was collected on 1 January 2011, Supplementary Information accompanies this paper at (https://doi.org/ then its collection would be counted as “365 (days in 10.1038/s41426-018-0101-6). 2010) + 366 (days in 2012) = 731 days”. A Python script was written for screening transmission Received: 24 February 2018 Revised: 26 April 2018 Accepted: 29 April 2018 clusters in both the Shanghai and Global datasets, and this Python script was uploaded to GitHub. Statistical analysis References The Wilcoxon nonparametric rank-sum test and Stu- 1. Gagneux, S. et al. The competitive cost of antibiotic resistance in Myco- dent’s t-test were used to compare discrete variables bacterium tuberculosis. Science 312,1944–1946 (2006). 2. Luciani, F., Sisson, S. A., Jiang, H., Francis, A. R. & Tanaka, M. M. The epide- between groups. Chi-square analysis was performed to miological fitness cost of drug resistance in Mycobacterium tuberculosis. Proc. compare the ratios of compensatory mutations in differ- Natl Acad.Sci.USA 106,14711–14715 (2009). ent groups. All statistical analyses were performed and 3. Dye, C., Williams, B. G., Espinal, M. A. & Raviglione, M. C. Erasing the world’s slow stain: strategies to beat multidrug-resistant tuberculosis. Science 295, visualized in RStudio (https://www.rstudio.com/). 2042–2046 (2002). 4. Cohen, T., Sommers, B. & Murray, M. The effect of drug resistance on the Ethical approval fitness of Mycobacterium tuberculosis. Lancet Infect. Dis. 3,13–21 (2003). 5. Handel, A., Regoes, R. R. & Antia, R. The role of compensatory mutations in the The institutional review board of the Shanghai CDC emergence of drug resistance. PLoS Comput. Biol. 2, e137 (2006). approved the analysis with the anonymous dataset. 6. Comas, I. et al. Whole-genome sequencing of rifampicin-resistant Myco- bacterium tuberculosis strains identifies compensatory mutations in RNA polymerase genes. Nat. Genet. 44,106–110 (2012). Data availability 7. Meftahi, N. et al. Evidence for the critical role of a secondary site rpoB The analysis scripts that were written and used in this mutation in the compensatory evolution and successful transmission of an MDR tuberculosis outbreak strain. J. Antimicrob. Chemother. 71,324–332 study are available online at GitHub (https://github.com/ (2016). StopTB/Compen_Muta_Transmit). 8. de Vos, M. et al. Putative compensatory mutations in the rpoC gene of rifampin-resistant Mycobacterium tuberculosis are associated with ongoing transmission. Antimicrob. Agents Chemother. 57,827–832 (2013). Acknowledgements 9. Li,Q.J.etal. Compensatory mutationsofrifampinresistanceare associated We are grateful to Dr. Ted Cohen for the helpful suggestions and discussion of with transmission of multidrug-resistant Mycobacterium tuberculosis Beijing this work. This study was funded by the National Science and Technology Genotype Strains in China. Antimicrob. Agents Chemother. 60,2807–2812 Major Project of China (2017ZX10201302 and 2018ZX10715012 to Q.G.), (2016). National Natural Science Foundation of China (91631301 & 81661128043 to Q. 10. Yang, C. et al. Transmission of multidrug-resistant Mycobacterium tuberculosis G., 81701975 to Q.L., and 81402727 to C.Y.), NSFC-BRICS grant (81761148027), in Shanghai, China: A retrospective observational study using whole-genome Sanming project of Medicine in Shenzhen (SZSM201611030) and sequencing and epidemiological investigation. Lancet Infect. Dis. 17, 275–284 JSGG20170413142559220. (2017). 11. Burgos, M.,DeRiemer, K., Small,P.M., Hopewell,P.C.& Daley, C. L. Effect of drug resistance on the generation of secondary cases of tuberculosis. J. Infect. Authors’ contributions Dis. 188, 1878–1884 (2003). Q.L., T.Z., and Q.G. designed and implemented the study. J.W. and G.Z. 12. Bahrmand, A. R.,Titov,L. P., Tasbiti, A. H.,Yari, S. &Graviss,E.A. High-level contributed the clinical M. tb isolates’ DNA. C.Y. conducted the MIRU-VNTR rifampin resistance correlates with multiple mutations in the rpoB gene of typing and analysis. P.X. conducted the Sanger sequencing. Q.L., T.Z., and M.G. pulmonary tuberculosis isolates from the Afghanistan border of Iran. J. Clin. analyzed the sequencing reads and performed the cluster analysis. Q.J. Microbiol. 47,2744–2750 (2009). performed the statistical analysis. Q.L., T.Z., R.P., H.T., and Q.G. drafted the 13. Machado, D. et al. High-level resistance to isoniazid and ethionamide in manuscript. All authors critically reviewed and approved the final version of multidrug-resistant Mycobacterium tuberculosis of the Lisboa family is asso- the manuscript. ciated with inhA double mutations. J. Antimicrob. Chemother. 68,1728–1732 (2013). 14. Safi, H. et al. Evolution of high-level ethambutol-resistant tuberculosis through interacting mutations in decaprenylphosphoryl-beta-D-arabinose biosynthetic Author details and utilization pathway genes. Nat. Genet. 45, 1190–1197 (2013). 1 2 Shenzhen Center for Chronic Disease Control, Shenzhen, China. Key 15. Huo,F.etal. Increasing prevalence of rifampicin-resistant Mycobacterium Laboratory of Characteristic Infectious Disease & Bio-safety Development of tuberculosis is associated with the transmission of strains harboring com- Guizhou Province Education Department, Zunyi Medical University, Zunyi, pensatory mutations in China: a 10-year comparative analysis. Antimicrob. Guizhou, China. Shanghai Municipal Center for Disease Control and Agents Chemother. https://doi.org/10.1128/AAC.02303-17 (2018). Prevention, Shanghai, China. Department of Epidemiology of Microbial 16. Pan, X. L. et al. A quantitative and efficient approach to select MIRU-VNTR loci Diseases, School of Public Health, Yale University, New Haven, CT, USA. Unité based on accumulation of the percentage differences of strains for dis- de Génétique Mycobacterienne, Institut Pasteur, Paris, France. Key Laboratory criminating divergent Mycobacterium tuberculosis sublineages. Emerg. of Medical Molecular Virology at the School of Basic Medical Sciences, Microbes Infect. 6, e68 (2017). Shanghai Medical College, Shanghai Public Health Clinical Center, Fudan 17. Salvatore, P.P.etal. Fitness costs ofdrug resistance mutations in multidrug- University, Shanghai, China resistant Mycobacterium tuberculosis: a household-based case-control study. J. Infect. Dis. 213,149–155 (2016). 18. Borrell, S. & Gagneux, S. Infectiousness, reproductive fitness and evolution of Conflict of interest drug-resistant Mycobacterium tuberculosis. Int. J. Tuberc. Lung. Dis. 13, The authors declare that they have no conflict of interest. 1456–1466 (2009). Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 8 of 8 19. Blower, S. M. & Chou, T. Modeling the emergence of the ‘hot zones’:tuber- 26. Bjorkman, J., Nagaev, I., Berg, O. G., Hughes, D. & Andersson, D. I. Effects of culosis and the amplification dynamics of drug resistance. Nat. Med. 10, environment on compensatory mutations to ameliorate costs of antibiotic 1111–1116 (2004). resistance. Science 287,1479–1482 (2000). 20. Casali, N. et al. Microevolution of extensively drug-resistant tuberculosis in 27. Snider, D. E. Jr, Kelly, G. D., Cauthen, G. M., Thompson, N. J. & Kilburn, J. O. Russia. Genome Res. 22, 735–745 (2012). Infection and disease among contacts of tuberculosis cases with drug- 21. Casali, N. et al. Evolution and transmission of drug-resistant tuberculosis in a resistant and drug-susceptible bacilli. Am.Rev.Respir. Dis. 132,125–132 (1985). Russian population. Nat. Genet. 46,279–286 (2014). 28. Teeter, L. D. et al. Evaluation of large genotypic Mycobacterium tuberculosis 22. Lange, C. et al. Management of patients with multidrug-resistant/extensively clusters: contributions from remote and recent transmission. Tuberculosis drug-resistant tuberculosis in Europe: a TBNET consensus statement. Eur. (Edinb.) 93,(Suppl.), S38–S46 (2013). Respir. J. 44,23–63 (2014). 29. Liu, Q. et al. Genetic features of Mycobacterium tuberculosis modern Beijing 23. Wu, W. et al. Improvement in clinical outcome and infection control using sublineage. Emerg. Microbes Infect. 5, e14 (2016). molecular diagnostic techniques for early detection of MDR tuberculous 30. Joshi, N. & Fass, J. Sickle: A sliding-window, adaptive, quality-based trimming spondylitis: a multicenter retrospective study. Emerg. Microbes Infect. 6,e97 tool for FastQ files (Version 1.33)[Software]. (2011). (2017). 31. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. 24. Shcherbakov, D. et al. Directed mutagenesis of Mycobacterium Methods 9,357–359 (2012). smegmatis 16S rRNA to reconstruct the in vivo evolution of aminoglycoside 32. Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics resistance in Mycobacterium tuberculosis. Mol. Microbiol. 77,830–840 25,2078–2079 (2009). (2010). 33. Koboldt,D.C.etal. VarScan2:somatic mutation andcopy numberalteration 25. MacLean, R. C. & Vogwill, T. Limits to compensatory adaptation and the discovery in cancer by exome sequencing. Genome Res. 22,568–576 (2012). persistence of antibiotic resistance in pathogenic bacteria. Evol. Med. Public 34. Tamura, K., Stecher,G., Peterson,D., Filipski,A.&Kumar, S. MEGA6: Molecular Health 2015,4–12 (2014). Evolutionary Genetics Analysis version 6.0. Mol. Biol. Evol. 30,2725–2729 (2013). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Emerging Microbes & Infections Springer Journals

Have compensatory mutations facilitated the current epidemic of multidrug-resistant tuberculosis?

Free
8 pages
Loading next page...
 
/lp/springer_journal/have-compensatory-mutations-facilitated-the-current-epidemic-of-hYPTcDfA81
Publisher
Nature Publishing Group UK
Copyright
Copyright © 2018 by The Author(s)
Subject
Biomedicine; Biomedicine, general; Immunology; Medical Microbiology; Microbiology; Antibodies; Vaccine
eISSN
2222-1751
D.O.I.
10.1038/s41426-018-0101-6
Publisher site
See Article on Publisher Site

Abstract

Compensatory mutations have been suggested to promote multidrug-resistant tuberculosis (MDR-TB) transmission, but their role in facilitating the recent transmission of MDR-TB is unclear. To investigate the epidemiological significance of compensatory mutations, we analyzed a four-year population-based collection of MDR-TB strains from Shanghai (the most populous city in China) and 1346 published global MDR-TB strains. We report that MDR-TB strains with compensatory mutations in the rpoA, rpoB,or rpoC genes were neither more frequently clustered nor found in larger clusters than those without compensatory mutations. Our results suggest that compensatory mutations are not a major contributor to the current epidemic of MDR-TB. Introduction outbreaks in HIV-negative patients were regarded as It had been thought that drug-resistance mutations successful examples of compensatory evolution . would introduce a fitness cost into resistant Mycobacter- Recent studies have investigated the role of compensa- ium tuberculosis (M. tb) tory mutations in MDR-TB transmission by comparing , and the resulting reductions in virulence and transmissibility would prevent multidrug- the presence of compensatory mutations in clustered or 8,9 resistant tuberculosis (MDR-TB) strains from dis- nonclustered strains and found that putative compen- 2,3 seminating widely . Early mathematical models pre- satory mutations in rpoC and rpoA were more common in dicted that MDR-TB should remain a local problem , but VNTR-clustered strains. However, these studies did not the steady growth of the MDR-TB epidemic worldwide determine whether the putative compensatory mutations has contradicted these early expectations. A subsequent accumulated before or after the MDR-TB strains were study suggested that even if the average fitness of MDR- transmitted. If these compensatory mutations accumu- TB strains is low, a small proportion of resistant strains lated after transmission, they should not be considered as that are relatively more fit will outcompete the less fit and factors that promote or facilitate transmission. Moreover, drug-susceptible strains . These more fit, resistant strains the sampling methods used in these studies were not were thought to contain compensatory mutations that population based, and the IS6110/MIRU-VNTR defined would restore fitness and thus constitute an important clusters could have perhaps been further separated by 5,6 factor in the spread of MDR-TB strains , and MDR-TB whole-genome sequencing (WGS). To avoid these pitfalls, we examined the role of compensatory mutations in MDR-TB transmission by using WGS to determine Correspondence: Qian Gao (qgao99@yahoo.com) transmission clusters in the MDR-TB strains collected in Shenzhen Center for Chronic Disease Control, Shenzhen, China Key Laboratory of Characteristic Infectious Disease & Bio-safety Development Shanghai over a 4-year period (2009–2012). Our findings of Guizhou Province Education Department, Zunyi Medical University, Zunyi, contradict the previous inferences and do not support a Guizhou, China Full list of author information is available at the end of the article. These authors contributed equally: Qingyun Liu, Tianyu Zuo. © The Author(s) 2018 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. 1234567890():,; 1234567890():,; Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 2 of 8 significant role of compensatory mutations in promoting (compensated)-type clusters, in which all strains harbor the ongoing MDR-TB epidemic. the same compensatory mutation—indicating transmis- sion of a compensated MDR-TB strain (compensatory Results mutations occurring before transmission); (2) N (non- Collection of MDR-TB isolates compensated)-type clusters, containing only strains with From 2009 to 2012, a total of 324 MDR-TB isolates no putative compensatory mutation—indicating trans- mission of a noncompensated MDR-TB strain; (3) M were collected from 31 tuberculosis (TB) hospitals in Shanghai, China. All of these isolates were MIRU-VNTR (mixed)-type clusters, in which MDR-TB strains harbored genotyped, and 122 clustered MDR-TB strains had different compensatory mutations or only a proportion of undergone WGS previously . Here, to determine the the strains in the cluster contained compensatory muta- presence of compensatory mutations in VNTR non- tions, indicating that the compensatory mutations had clustered MDR-TB strains, we randomly selected 105 occurred after transmission (see more details about isolates from the remaining 202 VNTR nonclustered M-type clusters in the discussion). The groups so defined MDR-TB isolates to sequence the full-length genes rpoA, are theoretically similar to the three prototypical types of rpoB, and rpoC. The combined results of the 227 clusters used to evaluate the role of drug resistance in Shanghai MDR-TB strains were termed the “Shanghai transmission . In the Shanghai dataset, 12 transmission dataset”. Meanwhile, we obtained WGS records of 8331 clusters were determined as C-type, 18 as N-type, and six M. tb isolates of global origin from the European as M-type (Fig. 2a); in the Global dataset, 26 transmission Nucleotide Archive (ENA) (Supplementary Table 1), and clusters were determined as C-type, 84 as N-type, and 1346 of these strains were identified as MDR-TB (Sup- seven as M-type. plementary Table 3). Of these MDR-TB strains, 602 were collected through retrospective cohorts or a population- Compensated MDR-TB strains did not cause larger based approach (Supplementary Table 1) and thus were transmission clusters appropriate for the subsequent analysis. The data from If the putative compensatory mutations stimulated the the global MDR-TB isolates were termed the “Global transmission of MDR-TB, then C-type clusters would be dataset”. expected to be larger than N-type clusters, but in the Shanghai strains, the C-type clusters were not larger than Identification of putative compensatory mutations and either the N-type or the M-type clusters (Wilcoxon rank- sum test, P = 0.4053 and 0.9587, respectively) (Fig. 2b). transmission clusters Through phylogenetic reconstruction of the MDR-TB An analysis of the size of the clusters in the global MDR- strains obtained above, we found 60 nonsynonymous TB strains yielded similar results (Wilcoxon rank-sum mutations in the rpoA, rpoB,or rpoC genes that had arisen test, P = 0.3816 and 0.2277, respectively) (Fig. 2c). Nota- at least twice in parallel and were identified as putative bly, the largest clusters observed in both the Shanghai and compensatory mutations (Fig. 1, Supplementary Table 4). Global datasets (with seven and 23 linked cases, respec- Of these mutations, six were in rpoA,16in rpoB, and 38 in tively) were caused by MDR-TB strains without com- rpoC. We set “mutation parallelism” as a criterion to pensatory mutations (Fig. 2b, c). Moreover, we did not exclude lineage mutations or neutral polymorphisms that find any particular compensatory mutation that was were fixed due to genetic drift. Although this filter might associated with larger transmission clusters in either also exclude some uncommon compensatory mutations, dataset. These results suggest that compensated MDR-TB it increased the level of confidence that the mutations strains are not prone to generate more secondary cases identified were truly compensatory mutations. We inclu- than noncompensated MDR-TB strains do. ded rpoB in the search for putative compensatory muta- tions because additional mutations in rpoB have been Compensated MDR-TB strains were not more likely to be reported to both restore fitness and increase the level of clustered rifampicin resistance . MDR-TB strains with less than 12 Previous studies suggested that MDR-TB strains with Single-nucleotide polymorphisms (SNPs) difference were compensatory mutations were more frequently clus- 8,9 considered to constitute a transmission cluster. A total of tered . However, these studies treated “compensatory 36 such clusters were found in the Shanghai dataset, and mutations” as a static feature and ignored the possibility 117 clusters were identified in the Global dataset. that compensatory mutations could occur in secondary cases that were initially infected with noncompensated Three types of transmission clusters MDR-TB strains. This is probably what has occurred in To distinguish whether the compensatory mutations the had occurred before or after transmission, we divided the M-type clusters (Fig. 2a), and therefore, it should MDR-TB clusters into three types (Fig. 2a): (1) C not be considered as representing the transmission of Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 3 of 8 Fig. 1 Putative compensatory mutations in the rpoA, rpoB, and rpoC genes identified in this study. Each putative compensatory mutation was supported by at least two independent evolution events Fig. 2 Compensated MDR-TB strains were not associated with larger transmission clusters. a A maximum likelihood phylogenetic tree showing genomic clusters in the Shanghai dataset. The strain identifiers were “year, strain number, and compensatory mutation type”. The three genomic cluster types are illustrated with different colors, as indicated. The isolate names in gray represent the VNTR-clustered strains that were separated by WGS. Comparison of cluster sizes in C-type, M-type, and N-type clusters in the Shanghai MDR-TB dataset (b) and the Global MDR-TB dataset (c); each dot represents a genomic cluster identified by WGS Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 4 of 8 Table 1 Ratios of compensated strains in clustered and nonclustered MDR-TB groups a a 2 Groups Total With CMs Without CMs χ P value Clustered MDR-TB (%) 78 32 (41.0%) 46 (59.0%) 2.260 0.133 Nonclustered MDR-TB (%) 133 41 (30.8%) 92 (69.2%) CMs compensatory mutations Excluded M-type clusters Fig. 3 Recently formed MDR-TB strains would decrease the ratio of compensatory mutations in the nonclustered MDR-TB group. a A theoretical scheme shows that recently formed MDR-TB strains would be included, while the secondary clustered strains resulting from transmission would occur outside of the study’s observation period. Thus, newly formed MDR-TB strains would be assigned to the nonclustered group. This schematic diagram also shows that transmission that occurred within the first 2 years will be captured in our study, while the secondary cases resulting from transmission that occurred during the last 2 years could be beyond the study’s observation period. b Comparison of the number of drugs to which strains are resistant in clustered and nonclustered MDR-TB strains. c Comparison of the number of drug-resistance mutations in clustered and nonclustered MDR-TB strains. d Comparison of collection time distribution between clustered and nonclustered MDR-TB groups; the collection time of each isolate was counted as “days to the end of the study duration (31 December 2012)” compensated MDR-TB strains. Accordingly, we excluded mutations that occurred after transmission result in a M-type clusters from an analysis of the ratios of com- higher ratio of compensatory mutations in clustered than pensatory mutations in the clustered and nonclustered in nonclustered MDR-TB strains. groups, and compared only C-type and N-type clusters. Surprisingly, no significant difference was observed in the The dilution effect of recently formed MDR-TB strains ratios of compensatory mutations in clustered compared Even when the M-type clusters were excluded, the to nonclustered groups, indicating that compensated clustered group had a nonsignificantly higher ratio of MDR-TB strains were not more frequently clustered compensatory mutations than that of the nonclustered (Table 1). group (41.0% versus 30.8%). One explanation might be To allow a comparison with previous studies, we repe- that new MDR-TB strains are continuously emerging ated the analysis without excluding the M-type clusters. during antibiotic treatment, and these new MDR-TB When the M-type clusters were included, the ratio of strains might not have had enough time to accumulate compensatory mutations in the clustered group was sig- compensatory mutations. For the same reason, within the nificantly higher than that in the nonclustered group 4-year duration of the study, they might not have had (Supplementary Table 5), similar to the results of previous enough time to cause secondary TB cases (Fig. 3a). Thus, studies. In our M clusters, 62.5% (10/16) of MDR-TB these recently formed “young” MDR-TB strains would be strains harbored compensatory mutations, which was assigned to the nonclustered group and thereby decrease much higher than that of either clustered (41.0%) or the ratio of compensatory mutations in this group. In nonclustered (30.8%) groups. M-type clusters are those in contrast, all strains in MDR-TB clusters are relatively which compensatory mutations are present in only some “older” (had become MDR-TB prior to their transmis- of the strains in a cluster, presumably because the sion), and thus would have had a longer period to accu- mutations have occurred during the transmission of the mulate additional drug resistance and compensatory clustered strain. Therefore, when the M-type clusters are mutations than strains in the nonclustered group. If this is included with the clustered MDR-TB strains, the true, then clustered MDR-TB strains should have wider Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 5 of 8 drug-resistance spectra (number of drugs to which they We believe in the accuracy of the results of our study are resistant) than those of nonclustered MDR-TB strains. because it incorporated several methodological improve- To test this inference, we compared the drug-resistance ments. First, we used WGS to identify transmission spectra of the clustered versus nonclustered MDR-TB clusters, which is more precise than IS6110 or MIRU- strains in the Global dataset. This analysis demonstrated VNTR . In our data, a total of 28 VNTR-clustered strains that the median number of antibiotics to which (23% of the VNTR-clustered strains) were further sepa- nonclustered MDR-TB strains were resistant was 4.22 rated by WGS. Second, we set strict criteria to identify (4.10 ~ 4.34), while for clustered MDR-TB strains it was putative compensatory mutations and excluded phyloge- 5.13 (5.05 ~ 5.21) (Student’s t-test, P < 0.0001, Fig. 3b). In netic/lineage mutations or neutral polymorphisms. Third, addition, as a single M. tb strain can accumulate our analysis discriminated between the accumulation of multiple mutations that confer resistance to a single compensatory mutations before or after the transmission 12–14 antibiotic (evolution of high-level resistance) ,we of MDR-TB strains (i.e., C- versus M-type clusters) in the further compared the numbers of drug-resistance- analysis of cluster size and the ratio of compensatory conferring mutations in these two groups. Consistent mutations. with our prediction, the clustered MDR-TB strains had, The effect of drug resistance on transmission has been 4,17 on average, more drug-resistance mutations than the long debated , and different studies have reported nonclustered MDR-TB strains did (Student’s t-test, heterogeneous results suggesting that MDR-TB strains P < 0.0001, Fig. 3c). can be ten times more or ten times less transmissible than 11,18 Finally, to verify whether nonclustered MDR-TB strains drug-susceptible strains . However, the reasons tended to occur toward the end of the study period, i.e., underlying these variations could simply reflect the dif- were indeed younger, we compared the distributions of ferences in study settings and methodologies 4,18 the collection times of the initial cultured clinical speci- employed; the drug-resistant strains reported to have mens (counted as days to the end of the study duration) reduced transmissibility were found in settings with for the isolates in the clustered versus nonclustered effective TB control , while strains with increased groups. We found that the isolates in the nonclustered transmissibility were reported from regions with high TB 4,18 group had a shorter average time since collection than the burdens . Thus, it seems more likely that the hetero- isolates in the clustered group did (mean values: 762.08 geneous and discordant published results serve to versus 903.14, P < 0.0001, Wilcoxon rank-sum test, demonstrate that TB transmission is a process that is Fig. 3d), which suggests that the strains in the non- primarily influenced by environmental factors such as TB clustered group tended to be isolated toward the end of control policy, time with the illness before diagnosis, the study period. These analyses support our inference treatment efficacy, and general quality of the health-care that clustered MDR-TB strains differ from nonclustered system . MDR-TB strains, suggesting a dilution effect from We consider that there are at least four factors that can recently formed MDR-TB strains that could in turn pro- obscure the function of compensatory mutations in MDR- vide an explanation for the relatively lower ratio of TB transmission. First, the efficacy of the TB control compensatory mutations in the nonclustered MDR-TB program. In countries or regions with high MDR-TB group. prevalence and poor management of MDR-TB patients, the transmission of MDR-TB strains is highly likely, even Discussion in the absence of compensatory mutations. In such cir- Our findings suggest that, in contrast to published cumstances, outbreaks and epidemics of MDR-TB are 5,15 20,21 reports , MDR-TB strains with compensatory muta- mainly driven by environmental factors . Second, tions are not more frequently found in clusters, nor are MDR-TB patients are, on average, infectious for longer they more likely to belong to larger transmission clusters periods due to the long course of treatment and high 22,23 than the MDR-TB strains without these mutations do. treatment failure rates . Thus, there might be more Previous studies have made direct comparisons of the chances for MDR-TB patients to generate secondary number of strains with putative compensatory mutations cases, which could conceivably compensate for a modest in clustered versus nonclustered MDR-TB strains and decrease in fitness. Third, the most common drug- reported an enrichment of compensatory mutations in the resistance mutations carried by clinical isolates are asso- 8,9 clustered group . However, we observed that M-type ciated with the lowest fitness cost in vitro, and such slight clusters had an impact on this comparison because decreases in fitness may not be sufficient to affect trans- 1,6 inclusion or exclusion of these clusters would dramati- mission . Moreover, positive epistatic effects between cally alter the results. Hence, we infer that the previous different drug-resistance mutations could further ame- observations would probably change if they excluded liorate the fitness cost imposed by individual drug- 17,24 M-type clusters. resistance mutations . Fourth, it is still possible that Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 6 of 8 the fitness cost measured in vitro might not reduce the Analysis of WGS data ability of M. tb to transmit and establish in a new host We used a validated pipeline for the mapping of 25,26 10,29 in vivo . This idea was supported by a previous study sequencing reads to the reference genome . In brief, using tuberculin skin testing to trace infection that found the Sickle tool was used for trimming WGS data. an equal prevalence of infection among contacts exposed Sequencing reads with Phred base quality above 20 and to patients harboring both drug-resistant and drug- reads length longer than 30 were retained for analysis. susceptible strains . The whole-genome sequence of the M. tb H37Rv strain The objective of this study was to investigate the (NC_000962.2) was used as the reference template for effects of compensatory mutations on MDR-TB trans- mapping reads. Sequencing reads were mapped to the missions that occurred within three or fewer years (recent reference genome using Bowtie2 (v2.2.9) . SAMtools 28 32 transmission) , and the 4-year observation window (v1.3.1) was used for SNP calling with mapping quality should have captured most of the transmissions that greater than 30. Fixed mutations (frequency ≥75%) were occurred within the first 2 years of the study. For identified using VarScan (v2.3.9) with at least ten reads transmissions that occurred during the last 2 years supporting and strand bias filter option on. We excluded of the study, the secondary cases might have developed all SNPs located in noise or repetitive regions of the TB disease after our observation window. (Fig. 3a). genome (e.g., PPE/PE-PGRS family genes, phage sequen- However, we believe that the transmission pattern ces, insertions or mobile genetic elements). in the last 2 years should not substantially differ from that in the first 2 years, and therefore, further extending the Identification of putative compensatory mutations time frame of our study would not change its major The maximum likelihood method was used to recon- conclusion. struct the phylogenetic trees with 500 bootstrap repeats In conclusion, we did not observe a promoting influence using MEGA (v6.06) , and the phylogenetic tree was of compensatory mutations on the transmission of MDR- further visualized by FigTree (http://tree.bio.ed.ac.uk/ TB, and we suggest that the putative contribution of software/figtree/). Putative compensatory mutations in compensatory mutations may be overwhelmed by the the rpoA, rpoB,or rpoC genes were identified through the complex and more powerful effects of environmental following criteria: (1) nonsynonymous mutations in the factors. rpoA, rpoB or rpoC genes that were present in only rifampicin-resistant strains; (2) each putative compensa- tory mutation must have arisen at least twice indepen- Methods MDR-TB datasets dently (parallel selection). We wrote a Python script to A total of 324 MDR-TB isolates were collected from 1 implement these criteria with the phylogenetic tree and a January 2009 to 31 December 2012, in Shanghai (the most mutation matrix, and the Python script was uploaded to populous city in China) during a population-based GitHub. observational study . MIRU-VNTR genotyping (12 loci) was performed for all MDR-TB strains, and the clustered Genotypic drug resistance detection and identification of strains were further subjected to WGS . In the present transmission clusters study, we further sequenced the entire gene sequences of Drug-resistance-associated mutations in M. tb were the rpoA, rpoB, and rpoC genes in 105 nonclustered obtained from the database TBDReaMDB and articles MDR-TB strains to detect putative compensatory muta- reporting rifampicin-resistance mutations (Supplemen- tions. The PCR amplicons were sent for Sanger sequen- tary Table 2). M. tb isolates carrying any characterized cing, and the results were analyzed to detect putative rifampicin-resistance mutation were identified as compensatory mutations using Geneious (http://www. rifampicin-resistant strains, whereas those isolates with geneious.com/). additional isoniazid-resistance mutations were deter- A total of 8331 published whole-genome sequences mined as MDR-TB strains. To determine the drug- of M. tb strains were downloaded from the ENA resistance spectra of the MDR-TB isolates in the Global (http://www.ebi.ac.uk/ena). The quality criteria for dataset, we additionally identified drug-resistance- data inclusion were set as follows: (1) the average associated mutations for nine drugs (amikacin, capreo- sequencing depth should be above 10-fold; (2) the mycin, ethambutol, ethionamide, fluoroquinolones, genome coverage rate should be above 95%. To avoid isoniazid, kanamycin, para-aminosalicylic acid, pyr- false transmission clustering, we included only the last azinamide, streptomycin) by inspection of their whole- isolate of longitudinal isolates collected from the same genome sequences. Any isolates with genetic distances of patient. Information about the geographic isolation of less than 12 SNPs were classified into a cluster .To these isolates was obtained from the articles or the compare the collection time distributions between the authors. isolates in the clustered and nonclustered groups, we Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 7 of 8 Publisher’s note determined the collection time of each isolate as “days to Springer Nature remains neutral with regard to jurisdictional claims in the end of the study duration (31 December 2012)”. For published maps and institutional affiliations. example, if an isolate was collected on 1 January 2011, Supplementary Information accompanies this paper at (https://doi.org/ then its collection would be counted as “365 (days in 10.1038/s41426-018-0101-6). 2010) + 366 (days in 2012) = 731 days”. A Python script was written for screening transmission Received: 24 February 2018 Revised: 26 April 2018 Accepted: 29 April 2018 clusters in both the Shanghai and Global datasets, and this Python script was uploaded to GitHub. Statistical analysis References The Wilcoxon nonparametric rank-sum test and Stu- 1. Gagneux, S. et al. The competitive cost of antibiotic resistance in Myco- dent’s t-test were used to compare discrete variables bacterium tuberculosis. Science 312,1944–1946 (2006). 2. Luciani, F., Sisson, S. A., Jiang, H., Francis, A. R. & Tanaka, M. M. The epide- between groups. Chi-square analysis was performed to miological fitness cost of drug resistance in Mycobacterium tuberculosis. Proc. compare the ratios of compensatory mutations in differ- Natl Acad.Sci.USA 106,14711–14715 (2009). ent groups. All statistical analyses were performed and 3. Dye, C., Williams, B. G., Espinal, M. A. & Raviglione, M. C. Erasing the world’s slow stain: strategies to beat multidrug-resistant tuberculosis. Science 295, visualized in RStudio (https://www.rstudio.com/). 2042–2046 (2002). 4. Cohen, T., Sommers, B. & Murray, M. The effect of drug resistance on the Ethical approval fitness of Mycobacterium tuberculosis. Lancet Infect. Dis. 3,13–21 (2003). 5. Handel, A., Regoes, R. R. & Antia, R. The role of compensatory mutations in the The institutional review board of the Shanghai CDC emergence of drug resistance. PLoS Comput. Biol. 2, e137 (2006). approved the analysis with the anonymous dataset. 6. Comas, I. et al. Whole-genome sequencing of rifampicin-resistant Myco- bacterium tuberculosis strains identifies compensatory mutations in RNA polymerase genes. Nat. Genet. 44,106–110 (2012). Data availability 7. Meftahi, N. et al. Evidence for the critical role of a secondary site rpoB The analysis scripts that were written and used in this mutation in the compensatory evolution and successful transmission of an MDR tuberculosis outbreak strain. J. Antimicrob. Chemother. 71,324–332 study are available online at GitHub (https://github.com/ (2016). StopTB/Compen_Muta_Transmit). 8. de Vos, M. et al. Putative compensatory mutations in the rpoC gene of rifampin-resistant Mycobacterium tuberculosis are associated with ongoing transmission. Antimicrob. Agents Chemother. 57,827–832 (2013). Acknowledgements 9. Li,Q.J.etal. Compensatory mutationsofrifampinresistanceare associated We are grateful to Dr. Ted Cohen for the helpful suggestions and discussion of with transmission of multidrug-resistant Mycobacterium tuberculosis Beijing this work. This study was funded by the National Science and Technology Genotype Strains in China. Antimicrob. Agents Chemother. 60,2807–2812 Major Project of China (2017ZX10201302 and 2018ZX10715012 to Q.G.), (2016). National Natural Science Foundation of China (91631301 & 81661128043 to Q. 10. Yang, C. et al. Transmission of multidrug-resistant Mycobacterium tuberculosis G., 81701975 to Q.L., and 81402727 to C.Y.), NSFC-BRICS grant (81761148027), in Shanghai, China: A retrospective observational study using whole-genome Sanming project of Medicine in Shenzhen (SZSM201611030) and sequencing and epidemiological investigation. Lancet Infect. Dis. 17, 275–284 JSGG20170413142559220. (2017). 11. Burgos, M.,DeRiemer, K., Small,P.M., Hopewell,P.C.& Daley, C. L. Effect of drug resistance on the generation of secondary cases of tuberculosis. J. Infect. Authors’ contributions Dis. 188, 1878–1884 (2003). Q.L., T.Z., and Q.G. designed and implemented the study. J.W. and G.Z. 12. Bahrmand, A. R.,Titov,L. P., Tasbiti, A. H.,Yari, S. &Graviss,E.A. High-level contributed the clinical M. tb isolates’ DNA. C.Y. conducted the MIRU-VNTR rifampin resistance correlates with multiple mutations in the rpoB gene of typing and analysis. P.X. conducted the Sanger sequencing. Q.L., T.Z., and M.G. pulmonary tuberculosis isolates from the Afghanistan border of Iran. J. Clin. analyzed the sequencing reads and performed the cluster analysis. Q.J. Microbiol. 47,2744–2750 (2009). performed the statistical analysis. Q.L., T.Z., R.P., H.T., and Q.G. drafted the 13. Machado, D. et al. High-level resistance to isoniazid and ethionamide in manuscript. All authors critically reviewed and approved the final version of multidrug-resistant Mycobacterium tuberculosis of the Lisboa family is asso- the manuscript. ciated with inhA double mutations. J. Antimicrob. Chemother. 68,1728–1732 (2013). 14. Safi, H. et al. Evolution of high-level ethambutol-resistant tuberculosis through interacting mutations in decaprenylphosphoryl-beta-D-arabinose biosynthetic Author details and utilization pathway genes. Nat. Genet. 45, 1190–1197 (2013). 1 2 Shenzhen Center for Chronic Disease Control, Shenzhen, China. Key 15. Huo,F.etal. Increasing prevalence of rifampicin-resistant Mycobacterium Laboratory of Characteristic Infectious Disease & Bio-safety Development of tuberculosis is associated with the transmission of strains harboring com- Guizhou Province Education Department, Zunyi Medical University, Zunyi, pensatory mutations in China: a 10-year comparative analysis. Antimicrob. Guizhou, China. Shanghai Municipal Center for Disease Control and Agents Chemother. https://doi.org/10.1128/AAC.02303-17 (2018). Prevention, Shanghai, China. Department of Epidemiology of Microbial 16. Pan, X. L. et al. A quantitative and efficient approach to select MIRU-VNTR loci Diseases, School of Public Health, Yale University, New Haven, CT, USA. Unité based on accumulation of the percentage differences of strains for dis- de Génétique Mycobacterienne, Institut Pasteur, Paris, France. Key Laboratory criminating divergent Mycobacterium tuberculosis sublineages. Emerg. of Medical Molecular Virology at the School of Basic Medical Sciences, Microbes Infect. 6, e68 (2017). Shanghai Medical College, Shanghai Public Health Clinical Center, Fudan 17. Salvatore, P.P.etal. Fitness costs ofdrug resistance mutations in multidrug- University, Shanghai, China resistant Mycobacterium tuberculosis: a household-based case-control study. J. Infect. Dis. 213,149–155 (2016). 18. Borrell, S. & Gagneux, S. Infectiousness, reproductive fitness and evolution of Conflict of interest drug-resistant Mycobacterium tuberculosis. Int. J. Tuberc. Lung. Dis. 13, The authors declare that they have no conflict of interest. 1456–1466 (2009). Liu et al. Emerging Microbes & Infections (2018) 7:98 Page 8 of 8 19. Blower, S. M. & Chou, T. Modeling the emergence of the ‘hot zones’:tuber- 26. Bjorkman, J., Nagaev, I., Berg, O. G., Hughes, D. & Andersson, D. I. Effects of culosis and the amplification dynamics of drug resistance. Nat. Med. 10, environment on compensatory mutations to ameliorate costs of antibiotic 1111–1116 (2004). resistance. Science 287,1479–1482 (2000). 20. Casali, N. et al. Microevolution of extensively drug-resistant tuberculosis in 27. Snider, D. E. Jr, Kelly, G. D., Cauthen, G. M., Thompson, N. J. & Kilburn, J. O. Russia. Genome Res. 22, 735–745 (2012). Infection and disease among contacts of tuberculosis cases with drug- 21. Casali, N. et al. Evolution and transmission of drug-resistant tuberculosis in a resistant and drug-susceptible bacilli. Am.Rev.Respir. Dis. 132,125–132 (1985). Russian population. Nat. Genet. 46,279–286 (2014). 28. Teeter, L. D. et al. Evaluation of large genotypic Mycobacterium tuberculosis 22. Lange, C. et al. Management of patients with multidrug-resistant/extensively clusters: contributions from remote and recent transmission. Tuberculosis drug-resistant tuberculosis in Europe: a TBNET consensus statement. Eur. (Edinb.) 93,(Suppl.), S38–S46 (2013). Respir. J. 44,23–63 (2014). 29. Liu, Q. et al. Genetic features of Mycobacterium tuberculosis modern Beijing 23. Wu, W. et al. Improvement in clinical outcome and infection control using sublineage. Emerg. Microbes Infect. 5, e14 (2016). molecular diagnostic techniques for early detection of MDR tuberculous 30. Joshi, N. & Fass, J. Sickle: A sliding-window, adaptive, quality-based trimming spondylitis: a multicenter retrospective study. Emerg. Microbes Infect. 6,e97 tool for FastQ files (Version 1.33)[Software]. (2011). (2017). 31. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. 24. Shcherbakov, D. et al. Directed mutagenesis of Mycobacterium Methods 9,357–359 (2012). smegmatis 16S rRNA to reconstruct the in vivo evolution of aminoglycoside 32. Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics resistance in Mycobacterium tuberculosis. Mol. Microbiol. 77,830–840 25,2078–2079 (2009). (2010). 33. Koboldt,D.C.etal. VarScan2:somatic mutation andcopy numberalteration 25. MacLean, R. C. & Vogwill, T. Limits to compensatory adaptation and the discovery in cancer by exome sequencing. Genome Res. 22,568–576 (2012). persistence of antibiotic resistance in pathogenic bacteria. Evol. Med. Public 34. Tamura, K., Stecher,G., Peterson,D., Filipski,A.&Kumar, S. MEGA6: Molecular Health 2015,4–12 (2014). Evolutionary Genetics Analysis version 6.0. Mol. Biol. Evol. 30,2725–2729 (2013).

Journal

Emerging Microbes & InfectionsSpringer Journals

Published: Jun 6, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off