Mycobacterium tuberculosis associated with severe tuberculosis evades cytosolic surveillance systems and modulates IL-1β production

Abstract

Genetic diversity of Mycobacterium tuberculosis affects immune responses and clinical outcomes of tuberculosis (TB). However, how bacterial diversity orchestrates immune responses to direct distinct TB severities is unknown. Here we study 681 patients with pulmonary TB and show that M. tuberculosis isolates from cases with mild disease consistently induce robust cytokine responses in macrophages across multiple donors. By contrast, bacteria from patients with severe TB do not do so. Secretion of IL-1β is a good surrogate of the differences observed, and thus to classify strains as probable drivers of different TB severities. Furthermore, we demonstrate that M. tuberculosis isolates that induce low levels of IL-1β production can evade macrophage cytosolic surveillance systems, including cGAS and the inflammasome. Isolates exhibiting this evasion strategy carry candidate mutations, generating sigA recognition boxes or affecting components of the ESX-1 secretion system. Therefore, we provide evidence that M. tuberculosis strains manipulate host-pathogen interactions to drive variable TB severities.

Fig. 1. M. tuberculosis clinical isolates associated with severe TB induce lower cytokine responses.

a The clinical data for 681 adult pulmonary TB (pTB) cases were reviewed and classified according to the absence or presence of comorbidities. b Top panel: genetic makeup through a validated panel of autosomal ancestry-informative markers of reference populations of African, European, East Asian, and Native American biogeographical origin. Bottom panel: the genetic ancestry for a group of 60 TB patients was determined, together with TB contacts from the same area and a reference Portuguese population. Represented are European ancestry in blue, African in orange, East Asian in pink and Native American in purple. c Clinical decision system developed to classify the severity of TB at presentation in mild, moderate or severe. d Patients who had no known comorbidities (n = 133) were classified into mild, moderate, or severe TB cases at presentation. e M. tuberculosis isolates recovered for 117 of the 133 selected TB patients were classified into lineages and those belonging to lineage 4 (L4) were further classified into L4.3/LAM, L4.1.1/X, L4.1.2/Haarlem or other lineage 4 sublineages. f Mild, moderate, and severe TB cases, as determined in d, were stratified according to the sublineage, the infecting bacteria belonged to. Statistical analysis was performed with Fisher’s exact test. g PBMC from 8-10 IGRA + donors were infected with L4 M. tuberculosis isolated from mild (n = 10), moderate (n = 9), or severe (n = 7) TB patients. An MOI of 1 was used and 24 h post-infection the amounts of secreted IL-1β, IL-12p40, IL-10, and IFN-β quantified. Represented are bean plots for each cytokine. Black lines show the medians; white lines represent individual data points; polygons represent the estimated density of the data. IFN-β was undetected for all conditions. Statistical analysis was performed with Pearson correlation; R and p-values were: IL-1β R = –0.3471; R2 = 0.1205; p < 0.00001; IL-12 R = –0.3419; R2 = 0.1169; p < 0.00001; and IL-10 R = –0.1918; R2 = 0.0368; p = 0.002473. CRP C-reactive protein, Hb hemoglobin, LAM Latin-American and Mediterranean, N no, TB tuberculosis, Y yes.

Fig. 2. Common polymorphisms in groups of low-cytokine-inducing M. tuberculosis clinical isolates.

Genome-wide analysis was performed for 26 M. tuberculosis clinical isolates with known IL-1β inducing profiles in human PBMCs. a Phylogenetic tree of the isolates constructed with genome-wide SNPs and using maximum likelihood. The color gradient between dark blue (low) and dark red (high) indicates the level of IL-1β induced by each strain. b Genes with (white) or without (black) polymorphisms exclusively present in >30% of the M. tuberculosis isolates classified as low or high IL-1β inducers. c ESX-1 secretion system and associated regulators showing polymorphisms exclusively present in the low or high IL-1β inducing strains.

Fig. 3. M. tuberculosis manipulate IL-1β induction independent of host TB status.

M. tuberculosis isolates 4I2 and 6C4 were selected based on the TB presentation, MTBC sublineage and intensity of cytokine induction in infected PBMCs. PBMCs from IGRA +  (a) or IGRA– (b) donors; CD14 + monocytes purified from IGRA– donors (c); PMA-differentiated THP-1 cells (d) or C57BL/6 mouse BMDMs (e) were infected with M. tuberculosis isolate 4I2 (solid circles or bars) or 6C4 (open squares or bars) for 24 h and the amount of secreted IL-1β quantified in the culture supernatants by immunoassay. MOIs of 1 (a–d) or 2 (e) were used for infection. Represented is the mean ± SEM; n = 10 donors (a, b), or n = 3 donors (c) or n = 3 wells from one (d) or four (e) independent experiments. Undetected values are not represented. Statistical analysis was performed using two-tailed unpaired Student’s t-test (*p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001).

Fig. 4. M. tuberculosis isolate associated with severe TB modulates the macrophage transcriptome.

BMDMs generated from C57BL/6 mice were infected with M. tuberculosis isolates 4I2 or 6C4 at a MOI of 2, or kept non-infected (NI). Six hours post-infection cell cultures were lysed, RNA extracted and subjected to targeted RNA-Seq. a Principal component (PC) analysis for data obtained from M. tuberculosis 4I2-infected (red), M. tuberculosis 6C4-infected (blue) and non-infected (NI, yellow) cells. Venn diagram (b) and volcano plots (c) representing the significantly differentially expressed genes (adjusted p-value ≤ 0.05 and log(fold-change) ≥2 or ≤ –2) when comparing BMDMs infected with M. tuberculosis isolate 4I2 or 6C4 with NI cells. In c, significantly differentially expressed genes (NI versus infected cells; adjusted p-value ≤ 0.05 and log(fold-change) ≥2 or ≤ –2) are shown in orange. Significant genes (adjusted p-value ≤0.05 and log(fold-change) between 2 and –2) are shown in black. Non-significant genes are displayed in light gray. Pathway analysis for significantly expressed genes detected in BMDMs infected with M. tuberculosis isolate 4I2 (d) or 6C4 (e) versus NI cells. Bar plots represent the number of genes present in each identified pathway and the color codes reflect adjusted p-values. In both differential expression analysis and pathway analysis, multiple testing adjustments were performed using the Benjamini–Hochberg (BH) procedure. The 20 most affected pathways are represented for each condition.

Fig. 5. Induction of high IL-1β by M. tuberculosis 4I2 results from enhanced inflammasome activation.

a Gene expression heatmap representing cytokine expression for non-infected BMDMs (NI, yellow) and BMDMs infected with M. tuberculosis isolate 4I2 (dark red) or M. tuberculosis isolate 6C4 (blue). z-scores are color coded as shown. b C57BL/6 WT BMDMs were generated and infected with M. tuberculosis isolates 4I2 (solid circles) or 6C4 (open squares), and at the indicated time points, the cell cultures were lysed, RNA extracted, converted to cDNA and subjected to real-time PCR. c C57BL/6 WT or TLR2 deficient BMDMs were generated and infected with M. tuberculosis isolates 4I2 (solid circles) or 6C4 (open squares). Twenty-four hours post-infection the culture supernatants were harvested and the amount of IL-1β quantified by ELISA. d THP-1-ASC-GFP monocytes were PMA-differentiated and left uninfected (NI) or infected with M. tuberculosis isolates 4I2 (solid bars) or 6C4 (open bars) at different MOI, as indicated. Six hours post-infection, cells were fixed and stained with DAPI prior to image acquisition. The percentage of Speck-positive cells was determined for the different conditions. C57BL/6 WT BMDMs were generated and infected with M. tuberculosis isolates 4I2 (e, f) or 6C4 (g, h) in the absence (0) or presence of increasing doses of NLRP3 (MCC950; e, g) or Caspase-1 (VX-765; f, h) inhibitors. BMDMs were infected with increasing MOI of M. tuberculosis isolate 6C4 (I), with a combination of both isolates at different doses (j) or with M. tuberculosis isolate 6C4 in the absence (6C4) or presence (6C4 + ATP) of 5 mM of ATP (k). e–k Twenty-four hours post-infection the culture supernatants were harvested and the amount of IL-1β quantified by ELISA. Unless indicated, an MOI of 2 was used for infections. Represented is the mean ± SEM of triplicate wells for one experiment representative of two (b–j); or for one experiment (k). Statistical analysis was performed using two-tailed unpaired Student’s t-test (c, k) or One-way ANOVA (d–i) (* p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001).

Fig. 6. M. tuberculosis isolate 6C4 evades macrophage cytosolic surveillance pathways.

a Heatmap representing the expression of genes associated with the cGAS pathway in non-infected BMDMs (NI; yellow) and BMDMs infected with M. tuberculosis isolate 4I2 (dark red) or M. tuberculosis isolate 6C4 (blue). z-scores are color coded as shown. C57BL/6 WT BMDMs were generated and infected with b M. tuberculosis isolates 4I2 (solid bars) or 6C4 (open bars) or c with M. tuberculosis isolate 4I2 in the absence (0) or presence of different doses of the cGAS inhibitor RU521. Six hours post-infection, the cell cultures were lysed, RNA extracted, converted to cDNA, and subjected to real-time PCR to quantify the expression of the indicated genes. d C57BL/6 WT or AIM2 deficient (-/-) BMDMs were generated and infected with M. tuberculosis isolates 4I2 (solid bars) or 6C4 (open bars). Infection of C57BL/6 WT BMDMs in the absence (-) or presence (+) of (e) a K⁺ channel inhibitor (glybenclamide; 25 µg mL^–1), f a cathepsin B (CatB) inhibitor (ZRLR; 10 µM) or g a phagocytosis inhibitor (cytochalasin D; 5 µM). h–i C57BL/6 WT BMDMs were generated and infected with live or rifampicin (RIF)-treated M. tuberculosis isolates 4I2 (solid bars) or 6C4 (open bars). d–h Twenty-four hours post-infection, the culture supernatants were harvested and the amount of IL-1β quantified by ELISA. i Six hours post-infection the cell cultures were lysed, RNA extracted, converted to cDNA, and subjected to real-time PCR to quantify the expression of Ifnb. Represented is the mean ± SEM of triplicate wells from one experiment representative of two (b, c, g); from two (d, e, h) or one (f, i) independent experiments. c Represented is the % of inhibition for the expression of each gene in BMDMs infected with M. tuberculosis isolate 4I2 in the presence of RU.521 relatively to those infected in the absence of the cGAS inhibitor. An MOI of 2 was used for all infections. Statistical analysis was performed using two-tailed unpaired Student’s t-test (*p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001).