Bacterial strains.

Mtb strains used in this study were isolated from pulmonary specimens during routine care in the Lyon University Hospital, France. The IMV and the 4MBE variant were obtained by cloning single-colony variants of the P1/L2 clinical isolate (Table 1). Briefly, bacterial suspension of the last isolate of P1/L2 patient was homogenized by vortexing in the presence of glass beads and Tween 80, and was plated on 7H10 supplemented with OADC to obtain single colonies. Clones were submitted to WGS to confirm the absence (IMV) or presence (4MBE) of the polymorphism of interest (C 3280555 G) and to confirm the absence of other mutations within the limits of the interpretation of short-read sequencing. Strains are stored at the Lyon University Hospital BSL3 laboratory and are available for research use if requested.

The Bandim TB score.

The modified Bandim TB score considers 5 symptoms (cough, hemoptysis, dyspnea, chest pain, night sweats) and 5 clinical findings (anemia, tachycardia, positive finding at lung auscultation, fever, BMI [<18 and <16]), with one point for each. One clinical finding was excluded, the mid upper arm circumference as this data was not available in the Lyon University Hospital. Accordingly, patients were stratified into two severity classes, mild (Bandim score ≤4) and moderate or severe (≥5) [45–47].

The MUST.

The nutritional status of TB patients was evaluated thanks to the Malnutrition Universal Screening Tool (MUST), which includes three variables: unintentional weight loss score (weight loss < 5%  =  0, weight loss 5–10%  =  1, weight loss > 10%  =  2), BMI score (BMI > 20.0  =  0, BMI 18.5–20.0  =  1, BMI < 18.5  =  2) and anorexia (if yes = 2). A MUST ≥ 4 is associated with poor TB prognosis [48].

MIC determination.

Minimum inhibitory concentrations (MIC) were determined for RIF and INH (Sigma-Aldrich, Saint Louis, MO, USA) by a standard microdilution method as previously described [49]. Briefly, the assay was performed in a 96-well microtiter plate, with the concentrations ranging from 8 to 0.0008 mg/L, with a final inoculum of approximately 1.10⁵ CFU/mL in each well. The plates were incubated at 37° C for 12 to 18 days, and the wells were assessed for visible turbidity. The lowest concentration at which there was no visible turbidity was defined as the MIC.

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, USA). Antibiotic solutions were added to MGIT at the required concentration and inoculated with 10⁵ CFU/mL for RIF exposure at 1 x MIC and with 2.10⁶ CFU/mL for RIF exposure at 4 x MIC. For the drug-free growth control, MGIT were inoculated with 10³ CFU/mL or 2.10⁴ CFU/mL, respectively. Analysis of the fluorescence was used to determine whether the tube was instrument positive; i.e. the test sample contains viable organisms, and results were expressed using MGIT time to positivity (TTP) system, reflecting bacterial growth [23,24]. The whole bacterial cultures were recovered to proceed to DNA or RNA extraction and purification.

Droplet digital PCR.

ddPCR procedure was conducted using QX100 Droplet Digital PCR System (Bio-Rad Laboratories, Hercules, CA, 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 and 1.1 μL of a 20X primers-probes mix (ddPCR mutant assay) provided by Bio-Rad (concentrations and sequences of probes are not provided by the manufacturer but they guaranty the validation by Minimum Information for Publication of Quantitative Real-Time PCR Experiments [MIQE] guidelines). The sequence studied was CGACACCGTTCGTGACCTCATCGCCCGTTGGGAGCAGCGGGACGTGATGGCGCGCGAGGTG[G/C]CCGTCGACGTGGCGTCGCACTCGCCTCAAGTCGATCCGATACTCGACGATTTGGCCGCGGC and the product was 70 bp. A no-template control and a positive control were used in each ddPCR batch. The procedure was carried out with droplet generation then PCR amplification with the following conditions: 95°C for 10 min followed by 40 cycles of 94°C for 30 s and 55°C for 1 min, and 98°C for 10 min, all steps with a ramp rate of 2°C/s. Analysis were done with QuantaSoft Analysis Pro software version 1.0.596 (Bio-Rad Laboratories). 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, obtained after lysozyme and proteinase K digestion, using a QIAamp DNA mini Kit (Qiagen). For targeted NGS, we performed PCR using Platinum SuperFi DNA polymerase (Life Technologies SAS, Carlsbad, CA, USA) and primers indicated in S3 Table. DNA libraries were prepared with Nextera XT kit (Illumina, San Diego, CA, USA). Samples were sequenced on NextSeq or MiSeq system (Illumina) to produce 150 or 300 base-pair paired-end reads at the Bio-Genet NGS facility of Lyon University Hospital, as previously described [50].

Bioinformatic analysis of Illumina data.

Reads were mapped with BOWTIE2 to the Mtb H37Rv reference genome (Genbank NC000962.2) and variant calling was made with SAMtools mpileup, as previously described [50]. For WGS, 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. Regarding targeted NGS, a valid variant was called if the position was covered by a depth of at least 20 reads and supported by a minimum threshold rate of 0.1%. Depth of coverage of position of interest ranged from 10000 to 57000x.

Sequences have been submitted to European Nucleotide Archive (ENA) under accession number PRJEB37306.

Apolar lipid extraction and analysis by gas chromatography–mass spectrometry (GC-MS).

Mycobacteria were grown in Middlebrooks 7H9 medium supplemented with 10% OADC (oleic acid, albumin, dextrose, catalase) until mid-log phase and pelleted before being washed twice in sterile water. The pellet was resuspended in methanol: 0.3% aqueous NaCl (10:1) and petroleum ether was added to separate the liquid layers. The non-aqueous petroleum ether extracts containing apolar lipids were dried under nitrogen. Apolar lipids were hydrolyzed with KOH 5% in methanol at 100°C for 90 min. The reaction mixture was acidified with 3N HCl and acidic water was added. Fatty acids released were extracted with chloroform-ethanol (3:1). The upper phase was removed and lower phase stored. The extracted fatty acids 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 performed at the Functional Lipidomics platform, acknowledged by Infrastructure in Biology, Health and Agronomy (IBiSA).

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 in a previous publication [49].

Killing kinetics of single drug. For each strain, an inoculum of approximately 5.10⁵ CFU/mL was prepared in 7H9 Middlebrook medium. The inoculum was incubated without any antibiotic (growth control) and with RIF, at concentrations of 1/32, 1/16, 1/8, 1/4, 1/2, 1, 2, 4, 16, 64, and 256 multiples of MIC (xMIC). Incubation was performed in 96 deepwell plates (1 ml working volume) at 37°C. Each well was homogenized by pipetting and vortexing, serially diluted, and plated on 7H10 agar. The numbers of CFU per milliliter was determined at day 7 of drug exposure. Each experiment was performed twice.

For drug tolerance experiments, an inoculum of approximately 2.10⁷ CFU/mL was prepared in 7H9 Middlebrook medium. The inoculum was incubated without any antibiotic (growth control) or with RIF, at concentrations of 2, 4, 8, and 16xMIC. At days 3, 5, 7, and 10 of drug exposure bacteria were processed as described above for CFU counting to determine subsequently the MDK₉₉ of each RIF doses [51,52]. Each experiment was performed thrice.

Killing kinetics of INH + RIF combination. For each strain, a bacterial inoculum was prepared as described above. Twelve concentrations of IHN (0, 1/32, 1/16, 1/8, 1/4, 1/2, 1, 2, 4, 8, 16, and 64 multiples of MIC) were combined with 12 possible concentrations of RIF (same multiples of MIC as INH) in an incomplete checkerboard design. The incubation was performed as for single drug experiments. The numbers of CFU per milliliter was determined after day 7 of drug exposure. A total of 142 bacterial counts were available for each strain.

Mathematical modelling of single drug effect. A Hill pharmacodynamic equation was used to describe the antibacterial effect of single drugs [53]. The model was fit to single drug log-transformed data, as follows: (1)

Where E is the drug effect, C is the drug concentration, E_max is the maximal effect EC₅₀ is the median effect concentration, and H is the Hill coefficient of sigmoidicity. GraphPad Prism for Windows version 5.02 (GraphPad Software, La Jolla, CA, USA) was used for model fitting and parameter estimation.

The antibacterial effect E considered was the net difference between the bacterial count measured with no drug (C = 0) and the count for a given drug concentration at the same time-point. Drug concentrations were normalized to the MIC, so C is expressed as the ratio of the actual concentration of drug in culture divided by the MIC.

Response-surface modeling of INH and RIF combined effect. We used the Minto model [49,54] to describe the combined antimicrobial effect of INH and RIF based on data measured after 7 days of therapy.

The Minto model is an extension of the Hill equation to drug combination and is defined as follows as follows: (2)

With and

And

In those equations, X_INH and X_RIF are concentrations of INH and RIF normalized to the drug potency, U quantifies the ratio of each drug in the combination,

E_max(U) is the maximal effect at ratio U, U₅₀(U) is the number of units of X_RIF associated with 50% of the maximal effect at ratio U, and H(U) is the coefficient of sigmoidicity at ratio U. Each of the three parameters of the Hill equation are function of U, the drug ratio. The concentration-effect curve associated with each ratio defined the contour of a response-surface of the drug combination.

As suggested by Minto et al., we used a polynomial function to describe E_max(U), U₅₀(U), and H(U), as described below: (3) where E_max,INH and E_max,RIF are the maximal effect of INH and RIF alone, respectively, and β_Emax is the coefficient of the two-order polynomial for E_max(U); (4) where H_INH and H_RIF are the Hill coefficient of sigmoidicity for INH and RIF alone, respectively, and β_H is the coefficient of the two-order polynomial for H(U); and (5)

Where β_U50 is the coefficient of the two-order polynomial for U₅₀(U).

The last equation allows to interpret the response surface in terms of synergy/antagonism. If β_U50 is not different from 0, U₅₀ is equal to 1 for all values of U, and the interaction is additive. If β_U50 is significantly greater than 0, the curve shows an inward curvature and the normalized drug mixture, is more than additive, which means synergy. If β_U50 is significantly lower than 0, the curve displays an outward curvature, and the normalized drug mixture (X_INH + X_RIF)/U₅₀(U)) is less than additive, which means antagonism.

The Minto model was fit to data by using non-linear regression within the Matlab software (version 2011b; The Mathworks, Natick, MA, USA). Point estimates of model parameters along with their confidence intervals were obtained. Goodness-of-fit of the model was assessed by analysis of plots of observed versus predicted antibacterial effects.

Values and 95% confidence intervals of the β_U50 coefficient were examined to interpret the combined action in terms of synergy/antagonism. Synergy was confirmed when the lower bound of the 95% confidence interval was greater than 0. Antagonism was stated when the upper bound of the 95% confidence interval was lower than 0. When the 95% confidence interval included 0, the interaction was considered as additive.

Cell Culture and Infection protocol.

U937 cells, monocytic cell line from pleural effusion, cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum were differentiated with phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich) 100nM for 48h at a density of 2.10⁵ cells/mL. Adherent cells were then infected with bacterial suspension at a multiplicity of infection (MOI) of 10:1 (bacteria to cells) in antibiotic-free culture medium, at 37°C in 5% CO₂ atmosphere, as previously described [55].

Intracellular CFU counting.

U937 cells differentiated cells were infected by Mtb, as described above. After 5 h of culture, infected cells were incubated with amikacin 200 mg/L for 1 h to remove the extracellular Mtb. The cells were washed thrice to remove amikacin. To study Mtb uptake by macrophages, cells were immediately lysed with distilled water containing saponin 0.1% for 10 min and intracellular mycobacteria were plated on complete 7H10 agar and incubated for 3–4 weeks. To measure intra-cellular multiplication, separate wells with amikacin-treated inoculated cells were further incubated for 96 h at 37°C in 5% CO₂ atmosphere. After that, extracellular bacteria were removed by 1 h of incubation with amikacin 200 mg/L and intracellular bacteria were counted as describe above.

Immunofluorescence and image acquisition.

U937 differentiated cells were infected by Mtb, as described above, before staining for confocal microscopy analysis, 6 h and 24 h post-infection (hpi), as previously described [55]. For the study of autophagic flux, 3 h before staining, cells were incubated with chloroquine (Sigma-Aldrich) 40μM to avoid acidification of autophagy compartments. Infected cells were washed thrice and incubated with LysoTracker Red DND-99 500nM (Thermo Fisher Scientific Inc, Rockford, IL, US; L7528) for 1 h to detect acidify compartments. Following 4% paraformaldehyde fixation over 20 min, cells were incubated with 100 mM glycine for 20 min and washed with PBS. Cells were then permeabilized with 0.1% Triton X-100 for 10 min and saturated in PBS-3% bovine serum albumin (BSA) for 30 min. Cells were labelled with anti-Mtb coupled FITC antibody (Abcam, Cambridge, MA, US; ab20962, 1/100 dilution) and anti-LC3B coupled Dy-Light 650 antibody (to assess autophagy activation; Thermo Fisher Scientific, PA5-22937, 1/400 dilution) in PBS-3% BSA-0.05% Triton X100 for 1h at room temperature. Slides were mounted in Fluoromount medium (Sigma -Aldrich) and observed with a Leica confocal microscope SP5X. Images were acquired with Leica Application Suite software. Stacks of confocal images and quantification of percent of colocalization were performed with ImageJ software [56] and JACoP plugin.

CCF-4 assay for translocation of Mtb in host cell cytosol.

To detect mycobacterial escape from the phagosome and translocation in host cell cytosol, the CCF4 FRET assay was performed [28,57]. Briefly, infected cells were stained with 50 μM CCF4 (Invitrogen, Carlsbad, CA, USA) in buffer containing anion transport inhibitor (Invitrogen) for 2 h at room temperature, as recommended by the manufacturer. Cells were washed with PBS containing anion transport inhibitor before fixing with paraformaldehyde (PFA, Sigma-Aldrich) 4% for 30 min at room temperature in the dark. Cells were washed before performing directly fluorescence acquisition. Intact CCF4-AM emits green fluorescence (520 nm) due to FRET between the fluorescent moieties, indicating that mycobacteria are in intracellular compartments. Cleavage of CCF4-AM, due to β-lactamase expressed by cytosolic Mtb, leads to a shift in the fluorescence emitted to bleu (450nm). The results obtained were normalized on intracellular bacterial load.

Measurement of lactate dehydrogenase (LDH) in cell culture supernatants.

U937 cells differentiated cells were infected by Mtb, as described above. To quantify necrosis of U937 cells, the LDH release in cell culture supernatant from infected cells was measured at 24 h and 96 hpi by using the colorimetric kit (Roche, Mannheim, Germany) and following the manufacturer’s instructions.

Quantitative determination of apoptosis.

U937 differentiated cells were infected by Mtb, as described above, before staining for cell death analysis, at 24 hpi and 96 hpi. Infected cells were washed thrice and apoptosis was detected by Annexin V FITC (Thermo Fisher Scientific) after 15min of incubation at room temperature. After fixing with PFA 4% for 30min, cells were analyzed by using fluorescent microscopy.

Infected macrophage RNA extraction.

U937 differentiated cells were infected by Mtb, as described above, before RNA extraction. Total macrophage RNA was obtained from Mtb-infected cells at 24 hpi using the Trizol method as previously described [58]. Briefly, macrophages were lysed with Trizol reagent (Qiagen) and total RNA was isolated with miRNeasy mini kit according to the manufacturer’s instruction (Qiagen). The macrophage RNA was subjected to DNaseI digestion before final purification.

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, MA, USA), then mRNA were submitted to library preparation using NextFlex Rapid Directional mRNA (PerkinElmer) and 100 ng input. As recommended by the ENCODE (Encyclopedia of DNA Elements) consortium ERCC (External RNA Control Consortium) 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 [59]. 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 RStudio v.0.99.893 (RStudio Inc., Boston, MA, USA). Heatmaps were performed thanks to Heatmapper [60], using average linkage clustering method and Pearson distance measurement method.

Cytokine production assay.

U937 differentiated cells were infected by Mtb, as described above, and cell supernatant was recovered 6 hpi, 24 hpi and 96 hpi. Cell culture supernatants were screened for the presence of 27 human cytokines and chemokines using the Bio-Plex Pro Human Cytokine Standard 27-Plex kit (Bio-Rad Laboratory) on a FLEXMAP 3D analyzer (Luminex, Austin, TX, USA). Data were analyzed using Bio-Plex Manager software version 6.1 (Bio-Rad Laboratory).

Statistical analysis.

Statistical analyses were performed using GraphPad Prism for Windows, version 5.02 (GraphPad Software, La Jolla, CA, USA). Data obtained from lipidomic analysis, RIF exposure experiments, variant frequency analysis by ddPCR, intracellular CFU counting, lytic cell death, as well as cytokine and chemokine production were expressed as mean ± standard deviation (SD). Means were compared using One-Way ANOVA or Repeated Measures ANOVA followed with Bonferroni correction, unpaired t-test or Student’s t-test, where appropriate. For data obtained from pharmacological models, parameter values were given as point estimate [95% confidence interval, CI]. For confocal microscopy analysis and macrophage apoptosis, data were expressed as median values [interquartile range, IQR]. Results were compared by using Kruskal-Wallis analysis, using Dunn’s Multiple Comparison Test or Mann Whitney test, where appropriate. *p<0.05, ** p <0.01, *** p <0.001.