ABSTRACT
Mycobacterium tuberculosis (Mtb) genetic micro-diversity in clinical isolates may underline mycobacterial adaptive response during tuberculosis (TB) development and provide insights to anti-TB treatment response. Here, we investigated the genetic micro-diversity of persistent clinical Mtb isolates and its dynamics during controlled stress experiments chosen to mimic the course of TB disease. A minor variant characterized by a modified lipid profile, became enriched during stress experiments. The variant was found to be associated with drug-tolerance involving efflux pumps deregulation and with intracellular persistence subsequent to phagolysosome avoidance and autophagy decrease. Furthermore, pharmacological modeling predicted increased risk of treatment failure for isolates enriched in the former variant. Even under extreme experimental conditions, the diversity in variants was always maintained, suggesting a strong functional complementarity between variants of the Mtb bacterial population during disease dynamics. Such diversity may be responsible of Mtb adaptation to its host and play a role along the successive steps leading to multi-drug resistant TB. Altogether, our findings support the fact that Mtb micro-diversity should be further explored as a promising tool to detect patients at risk of poorly responding to anti-TB treatment, ultimately allowing improved and personalized TB management.
INTRODUCTION
Tuberculosis (TB) caused by Mycobacterium tuberculosis (Mtb) complex remains one of the most prevalent and deadly infectious diseases, responsible for 10 million cases and 1.45 million deaths worldwide in 2018 [1]. Among the reasons behind this dramatic situation stands the chronicity of Mtb infection, characterized by long periods of latency, linked to the ability of the tubercle bacilli to persist in the host tissues. TB disease therefore requires a long duration of antibiotic treatment to achieve sterilization of intracellular and intracavitary bacilli. Anti-TB drug resistance detection is mandatory upon TB diagnosis, as it is known to hamper treatment efficacy. However, persistent infections with delayed response to treatment may be observed without any in vitro proven antibiotic resistance. Hypothetically, pre-existing sub-populations enclosed within Mtb clinical isolates that are antibiotic tolerant could be responsible for such persistent infections [2, 3], but evidence of this is still lacking.
Since the introduction of next generation sequencing (NGS) enabling whole genome sequencing (WGS), WGS analysis performed on clinical Mtb isolates allowed to unmask the genetic particularities of Mtb sub-populations or variants. Minor variants are therefore frequently present within the diversity of Mtb population before treatment onset [4–9] and may also reveal during TB treatment. Some of these variants harbor drug-resistance mutations, whilst others carry SNP in loci involved in modulation of innate immunity and in Mtb cell envelop lipids [10–13]. Although the emergence of variants occurring in loci not directly related to drug-resistance is still poorly understood [14, 15], these could be the key to understanding early mechanisms leading to Mtb drug-resistance.
Genetic micro-diversity has been observed in other bacteria responsible for chronic infections, such as Helicobacter pylori, Pseudomonas aeruginosa and Staphyloccocus aureus, and has been linked to host-adaptation as well as disease outcome [16–18]. This raises the question of the influence of Mtb clinical isolates’ micro-diversity on patients’ response to anti-TB treatment, beyond the canonical Mtb drug resistance phenomenon. We hypothesized that Mtb micro-diversity may be involved in bacilli persistence and in drug tolerance, both possibly favoring treatment failure. Thus, to better understand the evolution of Mtb micro-diversity during infection and treatment and its role in the pathophysiology of TB, we investigated variant composition and dynamics of persistent Mtb clinical isolates. Thereafter, we focused on 1 Mtb clinical isolate to capture its genetic diversity before and after controlled stress experiments chosen to mimic the course of TB disease. The identified variants stemmed from the initial clinical isolate were then assessed for their ability to tolerate the major anti-TB drugs and to persist during macrophage infection. Importantly, initial Mtb micro-diversity was found to be associated with both drug-tolerance and intra-cellular persistence and should thus be further considered to improve TB management and prevent treatment failure.
METHODS
Bacterial strains
Mtb strains used in this study were isolated from pulmonary specimens during routine care in the Lyon University Hospital, France. “Initial clones” IC0 and IC100 were obtained by cloning of “initial isolate” (II). Clones were submitted to WGS to confirm the absence (IC0) or presence (IC100) of the variant of interest (C 3280555 G) and to confirm the absence of other mutations.
Antibiotics exposure using the BACTEC 960 system
Antibiotic exposure experiments were performed in Mycobacterial Growth Indicator Tube (MGIT) using the BACTEC 960 system (Becton Dickinson, Sparks, MD). Antibiotic solutions were added to MGIT at the required concentration and inoculated with 105 CFU/mL. Tolerance to antibiotics was also assessed in presence of efflux pump inhibitors reserpine (100µM, Sigma-Aldrich, Saint-Quentin-Fallavier, France) and valinomycin (0.1 µM, Sigma-Aldrich). Results were expressed using MGIT time to positivity (TTP) system, reflecting the bacterial growth [19].
MIC Determination
Minimum inhibitory concentrations (MIC) were determined for rifampicin (RIF) and isoniazid (INH, Sigma-Aldrich) by a standard microdilution method as previously described [20].
Mtb RNA extraction and reverse transcription
Mycobacteria were incubated with RNAprotect Bacteria Reagent (Qiagen, Valencia, CA), lysed and Mtb total RNA was isolated by using RNeasy Plus mini kit according to the manufacturer’s instruction (Qiagen). Reverse transcription was performed by using Reverse Transcription System according to the manufacturer’s instruction (Promega, Fitchburg, USA).
Droplet digital PCR
ddPCR procedure was conducted using QX100 Droplet Digital PCR System (Bio-Rad, California, USA). Total reaction volume was 22 µL, containing 11 µL of 2X ddPCR Supermix for Probes (No dUTPs), 8.9 μL of template DNA, 10 U of HindIII restriction enzyme (GmbH) and 1.1 µL of a 20X primers-probes mix (ddPCR mutant assay) provided by Bio-Rad and guaranteed following MIQE guidelines. The sequence studied was CGACACCGTTCGTGACCTCATCGCCCGTTGGGAGCAGCGGGACGTGATGGCGCGC GAGGTG[G/C]CCGTCGACGTGGCGTCGCACTCGCCTCAAGTCGATCCGATACTCG ACGATTTGGCCGCGGC and the product was 70 bp. A no-template control and a positive control were used in every ddPCR batch. PCR amplification was conducted after droplet generation. Analysis were done with QuantaSoft Analysis Pro software version 1.0.596 (Bio-Rad). Positive droplets were counted and percentages of wild-type and mutant templates were calculated (% mutant =[(n mixed population/2)+n mutant]/total n amplified).
DNA extraction, targeted NGS and WGS
Genomic DNA was purified from cleared lysate using a QIAamp DNA mini Kit (Qiagen). For targeted NGS, PCR were performed using high fidelity DNA polymerase with primers indicated in Table S1. DNA libraries were prepared with Nextera XT kit (Illumina, San Diego, USA). Samples were sequenced on NextSeq or MiSeq system (Illumina) to produce 150 or 300 base-pair paired-end reads, as previously described [21].
Bioinformatic analysis of Illumina data
Reads were mapped with BOWTIE2 to the H37Rv reference genome (Genbank NC000962.2) and variant calling was made with SAMtools mpileup, as previously described [21]. A valid nucleotide variant was called if the position was covered by a depth of at least 10 reads and supported by a minimum threshold rate of 10%. Regions with repetitive or similar sequences were excluded, i.e. PE, PPE, PKS, PPS, ESX. The WGS reference genome coverage ranges 96.5% to 99.0%, with an average depth of coverage of 46x to 185x.
Sequences have been submitted to European Nucleotide Archive (ENA) under accession number PRJEB37306.
Apolar lipid extraction and analysis by GC-MS
Mycobacterial apolar lipids were extracted as previously described [22, 23], hydrolyzed with KOH 5% in methanol at 100°C for 90 min and then acidified. Fatty acids released were extracted with chloroform-ethanol (3:1). The extracted fatty acids from lower phase were derivatized with pentafluorobenzyl bromide (PFB) and analyzed in the negative ion mode by gas chromatography-mass spectrometry (GC-MS/MS Agilent HP7890B/7000C). The gas chromatograph was operated at the Functional Lipidomics platform, acknowledged by IBiSA (Infrastructure in Biology, Health and Agronomy).
Killing kinetics of RIF and INH alone and combination and mathematical modeling
In vitro experiments of the antibacterial effect of INH and RIF alone and in combination, as well as the mathematical modeling of the results were performed as described elsewhere [20].
Cell Culture and Infection protocol
U937 cells were differentiated with PMA and infected with bacterial suspension at a multiplicity of infection (MOI) of 10:1 (bacteria to cells). Invasiveness and intracellular multiplication were evaluated as described elsewhere [24].
Immunofluorescence and image acquisition
Infected cells were stained for confocal microscopy analysis, 6 h and 24 h post-infection (hpi), as described elsewhere [24]. For the study of autophagic flux, 3 h before staining, cells were incubated with chloroquine (Sigma) 40µM.
CCF-4 assay for translocation of Mtb in host cell cytosol
Infected cells were stained with 50 μM CCF4 (Invitrogen, Carlsbad, USA) in buffer containing anion transport inhibitor (Invitrogen) for 2 h at room temperature, then were washed and fixed with PFA 4% for 30min, as recommended by the manufacturer, before performing directly fluorescence acquisition [25, 26]. The results obtained were normalized on intracellular bacterial load.
Measurement of lactate dehydrogenase (LDH) in cell culture supernatants
LDH release of U937 cell culture supernatant was measured at 24 and 96 hpi by using the colorimetric kit (Roche, Mannheim, Germany) and following the manufacturer’s instructions.
Quantitative determination of apoptosis
At 24 and 96 hpi, infected cells were stained for apoptosis detection with Annexin V FITC (Thermo Fisher Scientific Inc, Waltham, USA) for 15min at room temperature. After fixing with PFA 4% for 30min, cells were analyzed using fluorescent microscopy.
Infected macrophage RNA extraction
Total macrophage RNA was obtained from Mtb-infected cells at 24 hpi using the Trizol method as previously described [27] and total RNA was isolated with miRNeasy mini kit according to the manufacturer’s instruction (Qiagen).
Transcriptomic analysis
Transcriptomic analysis was performed at ViroScan3D/ProfileXpert platform (www.viroscan3d.com). Briefly, RNA from each sample was quantified using Quantus HSRNA method (Promega) and qualified using Fragment analyzer HSRNA (AATI). After quality control, total RNA were submitted to polyA capture using NextFlex poly(A) Beads (PerkinElmer, Waltham, USA), then mRNA were submitted to library preparation using NextFlex Rapid Directional mRNA (PerkinElmer) and 100 ng input. As recommended by the ENCODE consortium ERCC RNA spike-In (Invitrogen) were added to samples in order to ensure reproducibility of the experiments. Single-read sequencing with 75 bp read length was performed on NextSeq 500 high output flowcell (Illumina). Mapped reads for each samples were counted and normalized using FPKM method [28]. Fold change between the different groups were calculated using median of groups and p-value of difference were calculated using t-test with equal variance without p-value correction in the R software. Population clustering analysis from transcriptomic data was conducted using principal component analysis (PCA) with the R package ade4 [29]. Heatmaps were performed thanks to Heatmapper [30], using average linkage clustering method and Pearson distance measurement method.
Quantitative real time RT-PCR (qRT-PCR)
Differential expression of selected genes was determined by qRT-PCR using FastStart Essential DNA Green Master kit (Roche) on the LightCycler® Nano Instrument (Roche), according to the manufacturer’s instruction. The cDNA was amplified with gene specific primers (Table S2 (Mtb primers) and S3 (human primers)). The transcripts of 16S RNA and glyceraldehyde phosphate dehydrogenase (GAPDH) were used to normalize the Ct values of the Mtb and human target genes respectively. Fold change in gene expression was calculated using the formula 2-ΔΔCt.
Cytokine production assay
Cell culture supernatants were screened at 6, 24 and 96 hpi for the presence of 27 human cytokines and chemokines using the Bio-Plex Pro Human Cytokine Standard 27-Plex kit (Bio-Rad Laboratory, Hercules, USA) on a FLEXMAP 3D® analyzer (Luminex, Austin, USA). Data were analyzed using Bio-Plex Manager software version 6.1 (Bio-Rad Laboratory).
Statistical analysis
Statistical analyses were performed with Graph Pad Prism 5. Data obtained from lipidomic analysis, RIF exposure experiments, variant frequency analysis by ddPCR, quantitative RT-PCR, intracellular CFU counting, macrophage cell death and cytokines and chemokines production were expressed as mean ± standard deviation (SD). Means were compared using One-Way ANOVA with Bonferroni’s correction or t-test. Parameter values obtained from pharmacological models were given as point estimate [95% confidence interval (CI)]. Confocal microscopy data were expressed as median values [interquartile range (IQR)]. Results were compared by using Kruskal-Wallis analysis, using Dunn’s Multiple Comparison Test. *p<0.05, ** p <0.01, *** p <0.001. $ Significantly different from condition without chloroquine; $ p<0.05, $$ p <0.01, $$$ p <0.001.
RESULTS
A minor variant of Mtb clinical isolate is selected in response to both rifampicin exposure and macrophage infection
Sixteen clinical pan-susceptible strains isolated from the sputum of patients undergoing anti-TB treatment for 4 to 29 weeks were exposed for 4 weeks to rifampicin (RIF), in vitro, at 1xMinimum Inhibitory Concentration (MIC) to investigate the emergence of variants. A total of 32 SNPs emerging in 12 out of 16 Mtb strains after RIF exposure were identified (Table S4). Following RIF exposure, one of these 12 isolates showed dramatic increase of the frequency of a single variant, harboring 1 non-synonymous SNP C 3280555 G, in a locus involved in cell wall and cell processes (Rv2940c mas, Table S1). To decipher the dynamics of the variant composition of this clinical isolate we first performed WGS on 20 grown colonies plated before in vitro RIF exposure. Apart from the predominant clone, 4 variants were identified and their presence in the initial isolate was confirmed by targeted next generation sequencing (NGS), at frequencies between 0.5% and 9% (Fig. 1A). Six loci differed amongst the variants, with a maximum pairwise distance of 4 SNPs. Independently from RIF exposure, the initial isolate was submitted to 7 days of macrophage infection. Interestingly, NGS analysis of these two final isolates found that the C 3280555 G variant was the most enriched during both RIF exposure (from 1.5% to 45%) and macrophage infection (from 1.5% to 74%), whilst the relative frequency of the others 3 variants varied within a range of 10% (Fig. 1A).
To further study the emerging C 3280555 G variant, 5 strains containing the C 3280555 G variant at various frequencies were compared: the initial clinical isolate named “initial isolate” (II); 2 “evolved isolates” (EI) enriched in the C 3280555 G variant at a mean frequency of either 40% (EI40) after RIF exposure or 80% (EI80) after macrophage infection; 1 clone representing the predominant clone of the II, not carrying the C 3280555 G polymorphism (Initial Clone, IC0); and 1 clone of the II carrying the fixed C 3280555 G polymorphism (IC100; Fig. 1B).
The C 3280555 G harboring variant overexpresses mycocerosates and more particularly the tetramethyl-branched components of mycocerosates
The SNP C 3280555 G results in an amino-acid change, A721P, of mycocerosic acid synthase (Mas, Rv2940c). It is located in Mas acyltransferase domain, involved in multimethylated mycocerosates synthesis by iterative condensation with methyl-malonyl-CoA units [31, 32]. Mycocerosates are components of the phthiocerol dimycocerosate (PDIM) family of lipids of Mtb, strongly involved in host-pathogen interaction. Because the mycocerosate components are expected to be altered in the C 3280555 G variant, the lipid profile of IC100 and IC0 and a half-half mixture (50:50) was explored, focusing on the production of the 4 main components of mycocerosates (Fig. 2A): the two trimethyl-branched, C27 and C29, and the two tetramethyl-branched, C30 and C32 [22, 33]. When comparing to IC0, IC100 profile showed significantly lower trimethyl-branched C27 component (11.4% ± 2.2%) than tetramethyl-branched C30 component (80.1% ± 2.4%; Fig. 2B), consistent with a higher affinity of Mas enzyme for methyl-malonyl-CoA, to promote tretramethyl mycocerosate synthesis in IC100. Moreover, the relative proportion of mycocerosates produced by IC100 was approximately twice as high compared to IC0 (Fig. 2C). Given this lipid profile, the variant is called hereafter the 4MBE (tetra-methyl branched enriched) variant.
The 4MBE variant is tolerant to RIF alone and in combination with isoniazid
As micro-diversity patterns changed upon RIF exposure, the susceptibility of the 4MBE variant to this drug was explored. Although there were no significant difference regarding the MIC values (Table S5), the tolerance of the 4MBE variant to sub-MIC doses of RIF was investigated. The kinetics of the bacterial growth explored by the time to positivity (TTP) approach, revealed a significantly shorter TTP for IC100, EI80, and EI40 (from 6.9 to 9.7 days) compared to II (15.6 ± 1.6 days) upon antibiotic treatment with 1xMIC of RIF (Fig. 3A). Controlled bacterial assembly experiments, mixing IC0 and IC100 at various ratios, yielded similar results (Fig. S1A). Thus TTP values decreased as 4MBE variant frequency increased in the starting inoculum, suggesting a better ability of the 4MBE variant to persist upon RIF exposure.
To further explore RIF tolerance of the 4MBE variant, concentration-effect experiments were performed, and results yielded were then implemented in a pharmacological efficacy prediction model. There was a decrease in RIF susceptibility for EI40, EI80, and IC100 at sub-MIC doses (Fig. 3B). EI40, EI80, and IC100 exhibited higher EC50 values (from 2.3 to 5.9) than II (1.0, 95% CI [0.6; 1.6]), suggesting a lower potency of this antibiotic on 4MBE variant. Regarding IC100, there was a significant decrease in the Hill coefficient value (0.61 [0.49; 0.74]) compared to EI40 (0.96 [0.79; 1.13]) and EI80 (0.89 [0.72; 1.06]), indicating also a better tolerance to supra-MIC concentrations of RIF (Table S6). Overall, these results indicate that 4MBE variant is tolerant to sub-MIC doses of RIF compared to II, despite the absence of canonical RIF resistance.
As RIF is part of a complex regimen during TB treatment, the antibacterial effect of the combination of both first-line anti-TB drugs (isoniazid (INH) and RIF) was investigated on IC0 and IC100 (Fig. 3C). The Minto model described the combined antibacterial effect very well, R2 values ≥ 0.90 and low bias and imprecision values. Importantly, the interaction parameter βU50 was significantly greater than 0 for IC0 (1.43 [0.94; 1.91]), which means that the combined effect of INH and RIF was synergistic. Conversely, it was not significantly different from 0 for IC100 (0.68 [-0.38; 1.74]), indicating that the combined effect was just additive. Taken together, these results indicate that isolates enriched in the 4MBE variant have increased tolerance to RIF alone but also in combination with INH, possibly favoring treatment failure.
To explore a possible selective advantage of the 4MBE variant upon treatment, the dynamics of variant assemblies upon RIF exposure was monitored. The variant frequency thus increased for II (from 1.5% to 30% ±13%) and for EI40 (from 39 ±5.0% to 55 ±6.7%) (Fig. 3D), consistent with the previous NGS experiments (Fig. 1A). Controlled bacterial assembly experiments, mixing IC0 and IC100 at various ratios, yielded similar results (Fig. S1B). Interestingly, the variant frequency never reached 100%, as ddPCR allowed to detect at least 0.6% of the variant lacking C 3280555 G (IC0) (Fig. S1B).
Efflux pumps are involved in RIF tolerance of the 4MBE variant
Given the involvement of efflux pumps in Mtb antibiotic tolerance [34, 35], the impact of efflux pump inhibitors (reserpine and valinomycin) on RIF tolerance of IC0 and IC100 was explored. Whereas in the presence of valinomycin, there was a RIF dose-dependent increase in TTP for both IC0 and IC100, in the presence of reserpine a significant increase in TTP was found for IC100 only (ΔTTP 2.05 ± 0.29 days; Fig. 3E). These results suggest that efflux pumps are involved in RIF tolerance of the 4MBE variant, with a predominant role of reserpine-sensitive efflux pumps rather than valinomycin-sensitive ones. Consistently, quantitative RT-PCR experiments found higher drrABC and p55 mRNA levels for IC100 compared to IC0 (Fig. 3F). The DrrABC pump is known to display higher sensitivity to reserpine than valinomycin [36] and has been previously linked to RIF low-level resistance in Mtb without rpoB mutation [37].
4MBE variant frequency correlates with improved Mtb intra-macrophagic survival
As Mtb micro-diversity was also found to evolve upon macrophage infection (Fig. 1A), the abilities of intra-macrophage survival of II, EI40, EI80, and IC100 were explored at 6 h and 96 h post-infection (hpi).
The ability of IC100 to invade macrophages was significantly better (32.104 ± 11.104 CFU/mL) than that of II (6.104 ± 2.104 CFU/mL) at 6 hpi and this pattern was confirmed at 96 hpi, with 16 times more intracellular bacteria (Fig. 4A). Similarly, experiments conducted on the controlled bacterial assembly found that the Mtb intracellular survival was directly related to the frequency of the 4MBE variant in the infecting inoculum (Fig. S1C). To further monitor Mtb variant composition upon host-pathogen interaction, the DNA of plated bacteria at 6 hpi and 96 hpi were recovered to assess the variant frequency by ddPCR (Figs. 4B and S1D). Upon infection by II and EI40, there was an enrichment of the 4MBE variant, up to 36% and 64%, respectively, at 96 hpi, whereas variant frequency remained stable for EI80 (Fig. 4B). The same pattern was observed in controlled bacterial assembly (Fig. S1D). Interestingly, for infecting inocula containing 90% and 99% of the variant in the initial bacterial mix, the frequency remained stable and did not reach 100%, highlighting the role of Mtb diversity upon macrophage infection.
Improved intra-macrophagic survival of the 4MBE variant is due to phagolysosome avoidance and autophagic flux inhibition
Mtb intramacrophagic behavior of II, EI40, EI80, and IC100 was explored by confocal microscopy analysis at 24 hpi (Fig. 4). The improved Mtb intramacrophagic survival was confirmed by a significantly greater number of both infected cells and bacterial burden per cell, upon infection by EI40, EI80, and IC100 compared to II (Fig. S2).
To explore the mechanisms of Mtb intra-macrophage survival promoted by isolates enriched in 4MBE variant, phagolysosome activation and avoidance were first explored (Fig. 4C-E). Although no differences were observed in the number of cells with acidified compartments (Fig. 4C), upon infection by II, there were significantly more bacteria in acidified compartments (18.2% [7.1-38.3]) compared to the median values upon infection by EI40, EI80, and IC100 (Fig. 4D). Loading of infected macrophages with CCF4-AM, a fluorescence resonance energy transfer (FRET)-based assay, revealed that the blue/green signal was significantly lower upon infection by EI40, EI80, and IC100 compared to II (from 3.7 and 9.8 times less signal at 96 hpi; Fig. 4E). This supports the notion that isolates enriched in the 4MBE variant avoid phagolysosomal fusion.
The autophagy activation and outcome were then explored. Chloroquine pretreatment showed an abolition of the differences of the proportion of cells with autophagy compartments between EI40-, EI80-, and IC100-infected macrophages and II-infected macrophages. This suggests an inhibition of the autophagic flux for isolates enriched in the 4MBE variant (Fig. 4F). This was confirmed by the significant decrease in the proportion of acidified autophagy compartments upon infection by EI40, EI80, and IC100 compared to II (Fig. 4G). Furthermore, there was a decrease in autophagy magnitude upon infection by EI40, EI80, and IC100, represented by a decrease of approximately 1.5 fold in LC3B puncta per cells for these strains compared to II (Fig. 4H). This indicate a weaker autophagy activation upon infection by isolates enriched in the 4MBE variant.
Taken together, these results support that the better intracellular survival of isolates enriched in the 4MBE variant during macrophage infection is driven by the increase of phagolysosome avoidance and the decrease of both autophagic flux and autophagy activation.
Macrophage apoptosis and inflammatory response are exacerbated upon infection by isolates containing high relative frequency of the 4MBE variant
Macrophage transcriptome analysis showed that 677 genes were differentially expressed between macrophages infected by II and those infected by either EI40 or EI80. Three distinct population clusters, II, EI40, and EI80, were recognized by principal component analysis (PCA; Fig. S3). The first principal component (46.4% of the variance) allowed a discrimination of II-infected macrophages from EI-infected macrophages, while the second one (17.7% of the variance) allowed discrimination of EI40- and EI80-infected macrophages.
A total of 337 differentially expressed genes between macrophages infected by II and EI40 or EI80 were selected to build a heatmap, showing clustering according to biological replicates and to the frequency of the 4MBE variant in the Mtb infecting inoculum (Fig. 5A). The differentially expressed genes were classified into nine functional categories involved in Mtb macrophage response. Consistent with the macrophage infection experiments, actin cytoskeleton, autophagy, and phagolysosome pathways were found to be dysregulated upon infection by EI40 and EI80 compared to II. Moreover, there was also a dysregulation of inflammatory response, apoptosis and cell cycle, mitochondrial metabolism, lipid metabolism, and T cell signaling pathways.
In order to explore cell death mechanisms, necrosis (Fig. 5B) and apoptosis (Fig. 5C) of macrophages were assessed upon infection by II, EI40, and EI80, and IC100 at 24 and 96 hpi. At 96 hpi, there was no significant difference between II- (48% ± 3.0%) and EI40-induced necrosis rates (51.6% ± 3.1%). By contrast, the necrotic effect upon EI80 (55.2% ± 6.2%) and IC100 (64.9% ± 3.5%) infections was significantly greater than that obtained upon II infections (Fig. 5B). Furthermore, apoptosis was also significantly higher upon infection by EI80 and IC100 compared to II (Fig. 5C). Thus, macrophage cell death level increased with the frequency of the 4MBE variant in the Mtb infecting inoculum.
Cytokines (TNF-α, IL-6, IL-1β, IFN-γ) and chemokines (RANTES (CCL5) and interferon gamma-induced protein 10 (IP-10)) production were then measured at 6 hpi (Fig. S4), 24 hpi (Fig. 5D-I), and 96 hpi (Fig. S5) to explore macrophage inflammatory response. At 24hpi, there was a significant increase in the inflammatory response of macrophages upon infection by EI40, EI80, and IC100 compared to II, and this effect was constantly more pronounced for IC100 (Fig. 5D-I). Upon infection by IC100 there was a significant increase in TNF-α production by macrophages from 6 to 96 hpi (Fig. 5D, S4A, and S5A) and overall a sustained inflammatory response at 96 hpi (Fig. S5). Finally, chemokine signals were significantly increased upon infection by IC100 (Fig. 7H-I). Overall, the macrophage pro-inflammatory response was stronger (in terms of intensity and duration) upon infection by EI40, EI80, and IC100 than by II, the higher the frequency of the 4MBE variant in the infecting inoculum, the higher the inflammatory response.
DISCUSSION
The objective of the present study was to address the role of Mtb micro-diversity and its possible link with anti-TB treatment response. We therefore monitored Mtb micro-diversity during the course of the adaptive response of clinical isolates to environment variations consistent with TB infection and treatment. We explored 16 clinical pan-susceptible isolates responsible for pulmonary TB with persistent positive culture upon treatment and our results showed that micro-diversity in Mtb is not restricted to the few previously reported cases [5, 10, 11, 13, 38, 39]. Interestingly, WGS of isolated colonies allowed detection of additional SNPs when compared to WGS performed on the whole bacterial population, suggesting that more knowledge concerning Mtb diversity will be enabled in the future through the use of deeper sequencing.
We then focused on a single clinical isolate and highlighted the plasticity of its genetic micro-diversity, which presented a changing pattern of variant assembly depending on the experimental setting (intra-macrophage infection, antibiotic stress). Interestingly, a variant (called hereafter 4MBE variant) which was present at low frequency in the initial Mtb clinical isolate, became predominant under the stress conditions tested. Using a set of complementary approaches we were able to show that this type of variant displayed tolerance to anti-TB treatment.
It was recently suggested that numerous variants emerge and then disappear during the course of infection and treatment within individual patients [38, 39]. The minor variant identified in the present study may be beneficial during some stages of the Mtb infection lifecycle and, in parallel, may be deleterious in other conditions, explaining its potentially transient emergence. We found that this variant features a particular lipid profile consisting of enriched tetramethyl-branched mycocerosates, a PDIM component. Being located at the interface between the bacterium and the host, PDIM play important roles in directing host-pathogen interactions and are involved at many pathophysiological steps [25, 40–42]. Nevertheless, studies showed that PDIM and mycolic acids were down-regulated during extended intracellular infection to the benefit of other multimethyl-branched fatty acids [43–45]. In addition, the results presented herein clearly demonstrated more intense macrophage inflammatory response during infection with isolates containing a high frequency of the 4MBE variant, which could be deleterious during human lung infection. Consistently, a significant increase in macrophage cell death was also found, particularly concerning apoptosis occurrence, upon infection by these isolates. Apoptosis is generally regarded as a protective response, whereas necrosis is thought to favor inflammation and disease progression [46–48]. In that respect, macrophage apoptosis could be considered a limitative mechanism, preventing the spread of this variant. Taken together, the results of the present study suggest that isolates containing the 4MBE variant could mainly favor infection initiation while other variants that have yet to be identified may be more involved in long-term infection.
The main feature of the variant detected and studied herein, which could be discriminated from the major one based on WGS data, was a non-synonymous polymorphism of mas, a gene belonging to the PDIM cluster, and was thus not related to a canonical drug-resistant locus. Although the search of public databases did not allow to identify this fixed polymorphism in other Mtb genomes, we were able to detect it, at a frequency (∼1%) undetectable by WGS, in another Mtb clinical isolate by performing ddPCR assay. Among the 140 clinical isolates tested by ddPCR screening, this particular isolate was multi-drug resistant (MDR) belonging to the Euro-American lineage (data not shown). It is thus possible that this type of variant may play a role along the successive steps leading to MDR-TB, as it also showed increased RIF tolerance. The antibacterial effect of RIF alone and in combination with INH was different on the isolates enriched in 4MBE variant, with a decreased efficacy of RIF alone and loss of synergistic activity when in combination with INH. This suggests that optimal management may require higher drug doses in cases of persistent infection with Mtb isolates containing such variants. When focusing on resistance emergence, current diagnostic tools are based on the detection of fixed or unfixed variants in loci involved in drug resistance, to adapt treatment as early as possible [49, 50]. Likewise, our results suggest that the emergence of unfixed variants in loci not necessarily linked to canonical drug-resistance could also help as diagnostic markers. Indeed, features associated with Mtb tolerance, possibly through mutations affecting efflux pump expression, have been suggested to be potential pre-resistant markers [51–53]. Interestingly, the tolerance exhibited by the isolates enriched in 4MBE variant was probably supported by the differential expression of efflux pumps which have been shown to be involved in both PDIM and RIF transport [35, 54]. Herein we show for the first time that such a minor variant may be associated with drug-tolerance, promote treatment failure as predicted by pharmacological modelling, and eventually lead to drug-resistance. Interestingly, even under extreme experimental conditions, the diversity in variants was always maintained, suggesting the possibility of a strong functional complementarity between variants of the Mtb bacterial population during disease dynamics. This emphasizes the importance of Mtb micro-diversity, which could become a useful diagnostic tool to predict which patients are at risk of poor response to anti-TB treatment. These patients may require personalized dosing and therapeutic management in order to improve the outcome and to effectively prevent MDR-TB emergence. In-depth analysis of clinical isolates coupled to the development of in vitro models to assess minor variants with increased fitness, antibiotic tolerance, or enhanced virulence, could help identify mutations responsible for treatment failure, better understand Mtb pathogenicity, and ultimately, through the use of fine-tuned diagnosis tools, allow personalized TB management.
Author Contributions
Conceptualization: C.Ge., S.G., O.D.; Methodology: C.Ge., C.Gi., S.G., O.D.; Software: C.Ge., J-L.B., J-B.C., E.W.; Investigation and validation: C.Ge., E.H., A.B., J.H., V.B., F.B., M.G.; Data analysis: C.Ge., A.B., E.H., F.B., M.G., C.Gi., S.G., O.D.; Data interpretation: C.Ge., A.B., E.H., J-L.B., J.H., S.V., C.Gi., S.G., O.D.; Visualization: C.Ge, E.H., S.G., O.D.; Writing – Original Draft: C.Ge., S.G., O.D.; Writing – Review & Editing: All authors; Funding Acquisition: C.Ge., J-L. B., S.V., G.L., S.G., O.D.
Compliance with ethical standards
Conflict of Interests
The authors declare that they have no conflict of interest.
ACKNOWLEDGMENTS
This work was supported by the LABEX ECOFECT (ANR-11-LABX-0048) of Université de Lyon, within the program “Investissements d’Avenir” (ANR-11-IDEX-0007) operated by the French national research agency (Agence nationale de la recherché, ANR).
The authors thank Philip Robinson and Véréna Landel (DRCI, Hospices Civils de Lyon) for help in manuscript preparation.