Host phagocytes respond to infections by innate defense mechanisms through metabolic shuffling to restrict the invading pathogen. However, this very plasticity of the host provides an ideal platform for pathogen-mediated manipulation. In the human (THP1/THP1 dual/PBMC-derived monocyte-derived macrophages) and mouse (RAW264.7 and C57BL/6 bone marrow–derived) macrophage models of Mycobacterium tuberculosis infection, we have identified an important strategy employed by clinical lineages in regulating the host immune–metabolism axis. We show greater transit via the macrophage phagosomal compartments by Mycobacterium tuberculosis strains of lineage: M. tuberculosis lineage 3 is associated with an ability to elicit a strong and early type I IFN response dependent on DNA (in contrast with the protracted response to lineage: M. tuberculosis lineage 1). This augmented IFN signaling supported a positive regulatory loop for the enhanced expression of IL-6 consequent to an increase in the expression of 25-hydroxycholesterol in macrophages. This amplification of the macrophage innate response-metabolic axis incumbent on a heightened and early type I IFN signaling portrays yet another novel aspect of improved intracellular survival of clinical M. tuberculosis strains.
Along-standing association with the human population has been critical to the development of host-specific immune-modulatory mechanisms by Mycobacterium tuberculosis. The era of the advent of high-throughput genome analysis has facilitated the differentiation of M. tuberculosis strains into seven distinct lineages that can be further classified into phylogenetically ancient (restricted geographic penetration) and modern (widespread) lineages with distinct mechanisms of virulence and disease manifestation (1–3). The frontline contact of M. tuberculosis activates several immune/transcriptional response relays in the host macrophages. M. tuberculosis lineages vary significantly in their inflammatory response induction and immune-modulatory functions (4–6).
Type I IFN expression, consequent to the cGAS–STING–nucleic acid signaling pathway activation, represents one such response of the host macrophages (7, 8). Although the pathogen beneficial properties of type I IFN response in intracellular bacterial infections are well established (9), its role in modulating inflammatory responses in M. tuberculosis infections is conflicting: on the one hand, a protective positive regulation of mycobacteria induced type I IFN signaling in IL-12 production and Ag presentation (10) and a definite detrimental pathogenic effect on macrophage activation and expression of cytokines such as TNF-α, IL-12, and IL-10 in macrophages (11–15). Recent studies have elucidated multiple mechanisms of M. tuberculosis–driven activation of this response pathway involving host factors such as reactive oxygen species (ROS)-induced mitochondrial damage and pathogen-associated factors such as ESX-1–mediated phagosomal escape and mycobacterial cyclic dinucleotides (16–19). Differences in type I IFN response induction by diverse M. tuberculosis strains/lineages are well recognized (20). However, mechanisms underlying strain-specific type I IFN still remain to be elucidated in detail. Given the predominance of two distinct lineages in the Indian subcontinent with differing inflammatory properties (4), we questioned whether the M. tuberculosis strains of lineages (lineage 1 [L1] and L3) differ in IFN1-inducing capacities. We demonstrate that the higher type I IFN induction (IFN-β) by modern lineages (L3, L4, and L2) is dependent on bacterial DNA presentation to the cGAS–STING signaling pathway as a result of an early and increased bacterial trafficking via the endolysosomal pathway. In line with this observation, we show an associated enhancement of proinflammatory IL-6 in macrophages infected with these M. tuberculosis. We further attribute an important role for 25-hydroxy cholesterol in regulation of IL-6 expression in these infected macrophages. This study highlights a novel mechanism of the plasticity of clinical M. tuberculosis by coupling type I IFN with macrophage proinflammatory response. Understanding the physiological consequence of this phenomenon would provide important insights into the successful adaptation of M. tuberculosis to the host environment.
Materials and Methods
Reagent and chemicals
LPS (tlrl-smlps) and Polyinosinic:polycytidylic acid [poly(I:C)] (tlrl-pic) were purchased from InvivoGen (Toulouse, France). Bafilomycin A, chloroquine, and 25-hydroxycholesterol (25-HC) were from Sigma-Aldrich. Lipofectamine LTX (A12621) and IFN-γ (PMC4033) were from Invitrogen-Thermo Fisher Scientific Inc.
Bacterial strains and growth conditions
The mycobacterial strains used in the study are detailed in Table I. Mycobacterial and Escherichia coli strains were cultured as per standard protocols with the appropriate supplements and selection antibiotics as required. The M6 promoter fragment was amplified as described earlier (21) and cloned upstream of GFP in the mycobacterial expression vector, pMV261, and electroplated into mycobacteria. Expression of ESAT6 in the culture filtrate and secreted M. tuberculosis cultures was performed by immunoblotting with anti-ESAT6 mAb HYB76-8 as described earlier (3). The mmpL8 null mutant (ΔM8) was created by homologous recombination using specialized transduction and recombineering as described earlier (22). Transducing phages generated by recombineering-mediated generation of phages were used for transduction of wild-type (WT) M. tuberculosis Erdman strain (Erd) cultures, and hygR colonies were screened for deletion by quantitative PCR (qPCR) and checking expression of sulfolipid in apolar lipid fractions by Thin-layer chromatography as per recommended protocols (23). For lipid analysis, log-phase cultures were labeled with 0.5 µCi of 1-14C propionic acid (American Radiolabeled Chemicals) for 16 h. The cell pellets were collected and subjected to lipid according to standard protocols (23). Apolar lipids were then separated in 1–D Thin-layer chromatography in petroleum ether:diethyl ether::90:10 (24) and developed by phosphor imaging.
Macrophage culture and maintenance
All the cell lines used were tested for mycoplasma contamination at regular intervals and cultured according to the manufacturer’s recommendations. THP1-Dual cells, RAW-Lucia ISG, ISG-KO-cGAS, ISG-KO-STING, ISG-KO-TRIF, ISG-KO-MAVS, ISG-KO-MDA5, and ISG-KO-RIG-I cells were obtained from InvivoGen (Toulouse, France). THP1 cells were grown in HiglutaXL RPMI-1640 with 10% FBS and 1 mM sodium pyruvate (Himedia Laboratories, Mumbai, India). THP1 cells differentiated into macrophages using PMA as described earlier (22). RAW264.7 cells were grown in DMEM high glucose (Himedia Laboratories) with 10% FBS. Human monocyte-derived macrophages (MDMs) were isolated from 20 ml as described earlier (21). For primary mouse macrophages, bone marrow cells were harvested from the hind limbs of C57Bl6 mice and differentiated into macrophages in 20% L cell supernatant for 7 d as described earlier (25). For ELISA, standard protocols were used according to the manufacturer’s recommendations: mouse TNF (88732488) and IL-6 (88706488) from eBioscience, Thermo Fisher Scientific Inc. Analysis of expression of ESAT6, CFP10 was done by using the Abs ab26246 and ab45074 (Abcam, Cambridge, MA) as per standard protocols.
Silencing cholesterol 25-hydroxylase in RAW macrophages
For silencing cholesterol 25-hydroxylase (CH25H), lentiviral short hairpin RNA (shRNA; pZIP-mEF1-α-RFP-Puro, shERWOOD UltramiR Lentiviral shRNA target gene set for gene Ch25h; Transomic Technologies) was used to silence Ch25h in RAW 264.7 macrophages according to recommended protocols. Lentiviral particles were generated in Lenti-X 293T cells and used to transduce macrophages followed by selection of stable clones in 4 µg/ml puromycin (Sigma-Aldrich). Scrambled shRNA was used as control. Knockdown efficiency was confirmed by qRT-PCR.
Preparation of M. tuberculosis fractions for stimulation of macrophages
M. tuberculosis crude extracts were prepared from logarithmic liquid cultures with similar bacterial densities for the different strains. Cells were lysed with a mixture of 0.1- and 0.7-µm zirconia beads in PBS using a bead beater. The resultant lysates were cleared off the cell debris by high-speed centrifugation and sterilized by filtration and used to transfect THP1 dual cells using Lipofectamine LTX (Invitrogen-Thermo Fisher Scientific Inc.) reagent according to the recommended protocol. Genomic DNA was isolated according to the standard protocols and used for transfection. Extracts were normalized to total protein concentrations before use for stimulation of macrophages. Enzymatic treatment of lysates was performed as (1) Proteinase K at 55°C for 1 h; or (2) DNase, RNase at 37°C for 1 h and thermal denaturation at 96°C for 30 min.
Infection of macrophages with M. tuberculosis
Macrophage cultures were seeded at the requisite density and infected with single-cell suspensions as described earlier (22). At different time intervals, the cells/RNA/supernatants were harvested and used for analysis. Bacterial numbers were counted from individual wells by standard dilution plating of the lysate. For analysis of expression by qRT-PCR, 1 μg of RNA was used for cDNA synthesis with Verso cDNA synthesis kit (Thermo Fisher Scientific Inc.) followed by qRT by using the DyNAmo Flash SYBR green qPCR kit (Thermo Fisher Scientific Inc.) in a Roche LC480 thermocycler. Primers sequences used for the study are detailed in Table II. Macrophages infected with fluorescent M. tuberculosis strains were used for lysosomal localization by using 100 nM LysoTracker Red DND-99 (Invitrogen-Thermo Fisher Scientific Inc.), according to the manufacturer’s recommendations. Macrophages transfected with mito-DsRed (26) were used for analysis of mitochondrial architecture. The cells were fixed in 4% paraformaldehyde for 30 min, mounted on slides with Prolong diamond anti-fade (Invitrogen- Thermo Fisher Scientific Inc.), and subjected to microscopy in the Leica TCS SP8 Confocal Microscope (Leica Microsystems). Slides were imaged in confocal mode with 63× oil immersion objective in Leica TCS SP8 confocal microscope. Twelve-bit images with pixel size between 90 and 110 nm were acquired. Argon laser (488 nM) was used for capturing GFP-expressing M. tuberculosis and laser 561 nM for LysoTracker-positive (lyso+) lysosomes. GFP and RFP images were acquired using photomultiplier tube and hybrid detector with bidirectional xyz sequential scanning. For estimating mitochondrial ROS by MitoSOX staining, macrophages were washed thrice with PBS and stained with 5 µM MitoSOX (Invitrogen-Thermo Fisher Scientific Inc.) in HBSS for 30 min at 37°C, fixed, and subjected to high-content microscopy in a IN Cell Analyzer 6000 (GE Healthcare Life Sciences). For analyzing the effect of 25-HC on inflammation, bone marrow–derived macrophages (BMDMs) were infected with M. tuberculosis at a multiplicity of infection (MOI) of 5 along with different concentrations of 25-HC.
Estimation of 25-HC in cell supernatants by ELISA
The levels of 25-HC were measured in infected macrophage lysate using 25-HC ELISA kit (MyBioSource). At indicated time points postinfection (p.i.), the cells were washed three times in PBS, and lysate was made in PBS resuspended cells subjected to freeze/thaw cycle at −20°C for three times. The lysate was centrifuged at 1000 × g for 15 min at 4°C, and the supernatant collected was used for the estimation of 25-HC. A total of 100 μl of standards/culture lysate was added to the coated wells, and PBS was used in the blank control well. A total of 10 μl of balance solution was used in the sample wells only and mixed well. A total of 50 μl of conjugate was added to each well except the blank control well and mixed well. The wells were incubated at 37°C for an hour and washed with buffer five times. A total of 50 μl substrate A and 50 μl substrate B were added to each well and incubated for 15–20 min at 37°C. A total of 50 μl of stop solution was added to each well to stop the reaction, and a reading was taken at OD 450 nm using a microplate reader.
Transmission electron microscopy
THP1 cells were infected with mycobacteria at an MOI of 5 for 24 h, fixed in 2.5% glutaraldehyde and 4% paraformaldehyde, dehydrated in graded series of alcohol, and embedded in Epon 812 resin. Ultrathin sections were cut and stained with uranyl acetate and lead citrate and imaged by using Tecnai G2 20 twin (FEI) transmission electron microscope (FEI, Thermo Fisher Scientific Inc, USA).
All experiments were performed in multiple biological replicates (n = 2 or 3). Statistical analysis was done by using Student t test and corresponds to *p < 0.05, **p < 0.01, and ***p < 0.001, respectively.
Six- to ten-week-old mice were housed in standard aseptic conditions and used for BMDM isolation according to approved guidelines of the institutional animal ethics committee (IGIB/IAEC/10/29). The protocol for isolation of human blood PBMCs was approved by the Institutional Human Ethics Committee (Reference no. CSIR-IGIB/IHEC/2017-18 Dt, 08.02.2018). Blood was collected from people with their informed consent as per the institutional guidelines.
M. tuberculosis lineages induce differential type I IFN response in macrophages
The type I IFN response of macrophages has generated interest as a propathogenic response induced by M. tuberculosis in macrophages (27–29). We employed the THP1-Dual cells and established commercial reporter lines for induction of type I IFN expression by quantitative measurement of ISG54 (+5 IFN-stimulated response elements)-regulated expression of secreted luciferase to quantitate M. tuberculosis–induced IFN1 expression. Infection with Erd at an MOI of 5 elicited 10- to 15-fold higher levels by 24 and 48 h in comparison with uninfected cells (Fig. 1A). Clinical M. tuberculosis strains of different phylogenetic lineages (Table I) elaborated a distinct pattern of response in these macrophages: modern M. tuberculosis lineages (L2, HN878; L3, N4, N24; L4, H37Rv, Erdman, H37Ra) elaborating 8- to 20-fold higher luminescence activity at 24 h of infection contrasting with the significantly lower (∼2.5-fold) activity in macrophages infected with the ancient lineage L1(T83, N72, N73) M. tuberculosis. In line with previous reports of Bacille Calmette–Guérin (BCG) strains failing to induce this response in macrophages, we did not detect any reporter induction in the cell supernatants (Fig. 1B, Supplemental Fig. 1A). This higher activation potential of L3 M. tuberculosis maintained across different infective doses with L3 M. tuberculosis strains (N4 and N24) could induce 5- to 10-fold higher response than a corresponding dose of L1 M. tuberculosis strains (N72 and N73) by 24 and 48 h of infection (Fig. 1C). Even at the highest MOI of 10, the response against L1 M. tuberculosis strains was significantly lower than the response elicited by equivalent dose or by the 10-fold lower dose of L3 M. tuberculosis. Temporal analysis of IFNB1 gene expression (Table II) revealed a gradual increase between 1 and 3 h (3.5-fold) to ∼30-fold by 12 h of infection, which was maintained at ∼16-fold even by 24 h of infection with the L3 M. tuberculosis strain N24 (Fig. 1D). In contrast, the L1 strain (N73) induced this response after 6 h of infection, attaining peak values of ∼3-fold by 12 h and remaining at 5- to 7-fold lower levels compared with L3 by 24 h of infection. This protracted IFN1 activation by L1 M. tuberculosis was evident even in primary human MDMs (Fig. 1E) and the mouse macrophage line RAW264.7 (Fig. 1F). Consistently, L3 M. tuberculosis induced a significantly higher (4- to 10-fold) IFN response than L1 M. tuberculosis at any point in infection, implying the universality of this phenomenon. Surprisingly, this response required active infection, because macrophages infected with nonviable M. tuberculosis failed to initiate this response (Fig. 1G).
We used a two-pronged approach to confirm whether the lower IFN1 activation by L1 M. tuberculosis strains was due to active suppression of IFN1 response in macrophages. In the first approach, we checked IFN1 expression in a coinfection assay of N24 and N73 in different ratios (Fig. 1H). At all doses, N24 induced significantly higher response than N73; however, in mixed infections, presence of 4-fold excess N73 (N73:N24::4:1) could not diminish the response induced by N24 with comparable levels in macrophages infected with N24 alone, even at the overall higher MOI of 10. The second approach involved a transactivation/suppression assay testing whether N73 could suppress the IFN1 response of macrophages to a strong inducer-like poly(I:C). Again, N73 failed to diminish poly(I:C)-mediated IFN response (Fig. 1I). In most cases, response to poly(I:C) was clearly augmented in conditions of costimulation with either N24 or N73.
L1 M. tuberculosis are delayed in inducing type I IFN response in macrophages despite the presence of an active ligand
The delayed kinetics of type I IFN induction by the L1 lineages implies one of two possibilities: (1) absence/decreased expression of the active ligand, or (2) failure of the ligand to activate the signaling complex. This prompted our search for the unique molecular constituent responsible for the differential IFN1 activation by the two lineages. Previous studies have elucidated several pathogen–host cell–associated factors as determinants of this response pathway. The mycobacterial ESX-1 secretion system aiding cytosolic egress of ligands and activation of the cytosolic surveillance pathway results in type I IFN activation (8, 20). In divergence with these findings, we could not attribute the ESX1 in the refractory properties of L1 M. tuberculosis on account of the following: (1) despite the reported differences in the levels of the ESX-1 proteins between H37Ra and H37Rv (30), we did not observe any difference in IFN1 activation; and (2) we observed comparable expression of ESAT6 a key component of the M. tuberculosis ESX1 system, in our clinical M. tuberculosis strains (L1 and L3) (Supplemental Fig. 1B). We also did not observe any difference in MitoSOX staining of macrophages infected with either M. tuberculosis strain, contrasting with the previously identified mode of mitochondrial ROS-mediated type I IFN signaling (20), hinting at alternative early events associated with recognition of bacterial ligands (Supplemental Fig. 1C).
Lipofectamine-based transfection of DNA results in delivery of DNA to the endosomal system and eventually to the cytosol via formation of multiple transient pores in the endosomal membrane (31). As a first step, we checked whether the crude cell extracts delivered by transfection would initiate IFN1 signaling, and we established the kinetics of this response with Erd extract (cell lysates) in cells; IFN1 activation was increased from ∼10-fold at 3 h to ∼75-fold by 6 h and then declining to ∼10-fold expression by 24 h of stimulation with different doses (Fig. 2A, Supplemental Fig. 1D). Neither denaturing the extracts with heat nor treatment with protease or RNase altered the IFN induction property; treatment with DNase completely abrogated type I IFN induction by the extracts, implicating an important role for the M. tuberculosis DNA (Fig. 2B). Interestingly, we observed that bacterial extracts from L1 and L3 M. tuberculosis or BCG were equally proficient in IFN1 induction (Fig. 2C). Coupled to our observation and the previously reported importance of DNA-sensing and cytosolic sensors such as STING (20, 32, 33), we questioned whether the quality/quantity of DNA defined the differential IFN1 response. Macrophages transfected with 125 ng Erd DNA efficiently elicited the type I IFN response by 3 h; this response was completely abolished in macrophages stimulated with DNase-treated DNA (Fig. 2D). More importantly, genomic DNAs from N73, as well as BCG, were similar to Erdman and N24 in their IFN-inducing capacities (Fig. 2E), hinting at the absence of appropriate ligand presentation to the cellular-sensing machinery in the case of N73 or BCG infections.
The inherent ability of DNA from the different mycobacterial strains to induce type I IFN advocated strongly for the inability of this component to reach the cytoplasmic signaling apparatus as the likely reason for the differential response of macrophages. We hypothesized that delivery of M. tuberculosis L1 and BCG to the cytosol would overcome the inherent IFN attenuation. In fact, transfection with lipofectamine of BCG or L1 M. tuberculosis significantly augmented IFN1 induction in these cells by 6 h of stimulation (Fig. 2F, 2G), establishing the importance of ligand/bacterial intracellular localization in this response.
|Strain Name .||Lineage .||Origin .|
|N72||1||San Francisco, CA|
|N73||1||San Francisco, CA|
|T83||1||San Francisco, CA|
|N04||3||San Francisco, CA|
|N24||3||San Francisco, CA|
|Erdman||4||Jeffery S. Cox, UC Berkley, CA|
|H37Rv||4||NIMR, London, UK|
|Beijing HN878||2||NIMR, London, UK|
|M. bovis BCG (Tokyo)||Not applicable||NIMR, London, UK|
|Strain Name .||Lineage .||Origin .|
|N72||1||San Francisco, CA|
|N73||1||San Francisco, CA|
|T83||1||San Francisco, CA|
|N04||3||San Francisco, CA|
|N24||3||San Francisco, CA|
|Erdman||4||Jeffery S. Cox, UC Berkley, CA|
|H37Rv||4||NIMR, London, UK|
|Beijing HN878||2||NIMR, London, UK|
|M. bovis BCG (Tokyo)||Not applicable||NIMR, London, UK|
The importance of DNA-mediated activation of cytoplasmic signaling was further substantiated when we analyzed the individual contribution of DNA- or RNA-sensing pathways. Because RAW264.7 lines devoid of specific nucleic acid–sensing components were commercially available, we first checked whether the differential IFN-inducing capacity was also seen in the murine macrophage cell lines. We used RAW-Lucia (ISG), a modified version of RAW264.7 murine macrophage lines with capacity to report with luminescence on type I IFN induction similar to THP1 Dual cells. These cells also phenocopied the refractory response of L1 M. tuberculosis–infected macrophages: 5- to 10-fold higher IFN by L3 M. tuberculosis in comparison with L1 M. tuberculosis strains (Fig. 2H). Although the majority of RNA-sensing receptors and adaptors showed a negligible effect (Supplemental Fig. 1E), the loss of TRIF (ISG-KO-TRIF) hampered M. tuberculosis–mediated type I IFN response in macrophages (∼60% decrease from WT). Most importantly, macrophages defective in DNA sensing and signaling, RAW-Lucia ISG-KO-cGAS and ISG-KO-STING cells, failed to elaborate any response to any of the M. tuberculosis strains, again stressing on the role of DNA signaling in type I IFN response. In line with this observation, we observed that loss of type I IFN signaling was sufficient to restrict growth of M. tuberculosis in macrophages lacking the key cytosolic signaling adaptor, STING (Fig. 2I). Over 3 d of culture, WT macrophages supported M. tuberculosis N24 growth, in contrast with the restricted growth to initial levels at all times (bacteriostasis), despite similar levels of uptake in the deficient macrophages.
|Gene .||Forward Primer (5′–3′) .||Reverse Primer (5′–3′) .|
|Gene .||Forward Primer (5′–3′) .||Reverse Primer (5′–3′) .|
Early localization to acidified compartments is important for M. tuberculosis–induced type I IFN induction
Given the importance of ligand presentation to the cytosolic signaling complex for type I IFN induction, we hypothesized that molecular events immediately after uptake that result in cytoplasmic delivery of bacterial components would be different in the two M. tuberculosis lineages. Moreover, the early induction of type I IFN in L3-infected macrophages advocated for events that are immediate to infection as causative, specifically, the transit of M. tuberculosis via the host cell phagosomal-lysosomal machinery for breakdown and Ag presentation. Not surprisingly, manipulation of cellular trafficking to hinder maturation of the endosomal compartments by M. tuberculosis has been recognized as one of the well-recognized contributors of successful residence in host phagocytes (34). To test whether the strains followed altered intracellular trafficking, we checked the intracellular residence of these bacteria p.i. and observed distinct intracellular localization of M. tuberculosis strains in THP1 cells by transmission electron microscopy: the L3 M. tuberculosis strain (N24) localizing primarily in compact double-membranous vesicles at 24 h as opposed to the larger less defined vesicular structures interspersed with intracellular particles for the L1 M. tuberculosis strain, N73 (Fig. 3A). Analysis of cellular portioning of the two M. tuberculosis strains at 24 h also hinted at major differences in subcellular localization of these strains: significantly higher numbers of L3 M. tuberculosis (N24) localizing to lyso+ vesicles than the L1 M. tuberculosis (N73) p.i. (Fig. 3B). This differential partitioning to lyso+ vesicles for the M. tuberculosis strains was also observed in RAW264.7 macrophages (Fig. 3C). In a temporal analysis of M. tuberculosis transit via the endosomal pathway, as early as 1 h of infection, L3 M. tuberculosis (N24) preferentially associated with lyso+ vesicles (>2-fold) as compared with N73. Despite a gradual decrease of N24 in lyso+ vesicles with time over the course of 24 h, consistently 2-fold higher numbers of N24 were restricted to lyso+ vesicles in comparison with N73 M. tuberculosis.
We have previously demonstrated that L1 M. tuberculosis strains harbor a mutation in papA2, rendering them incapable of sulfolipid synthesis (35). A dominant role for the M. tuberculosis sulfolipids has also been demonstrated in modulation of the cellular phagosome maturation and acidification (36–38). We questioned whether the lack of sulfolipids renders M. tuberculosis incapable of inducing the type I IFN response. Our observation of the absence of IFN1 response in macrophages infected with T83 M. tuberculosis strain (Fig. 1B), a member of the L1 lineage belonging to the Vietnam strain, but capable of expressing sulfolipids, did not favor the sulfolipid difference hypothesis. Alternatively, the M. tuberculosis mutant ΔmmpL8, lacking mature sulfolipids in M. tuberculosis cell surface, was similar to the WT M. tuberculosis strain in its IFN1-inducing ability, confirming that the difference in sulfolipids was not a determinant of IFN1-inducing capacity (Supplemental Fig. 1F). With phthiocerol dimycocerosates (PDIM) expression correlating to type I IFN (39), we analyzed whether the expression of this glycolipid was different in the two lineages. Analysis of radiolabeled lipids from the extracts of L3 and L1 showed comparable levels of PDIM similar to the levels in the Erd strain, thereby advocating for a completely novel mechanism of type I IFN induction difference between the strains (Supplemental Fig. 1G).
Given the high numbers of N24 localization to lyso+ compartments and comparable intracellular growth rates of N24 and N73 (Supplemental Fig. 1H), it was reasonable to assume an enhanced propensity of N24 to be better equipped for survival in this subcellular niche. The ability of M. tuberculosis to counter phagosomal-lysosomal fusion and acidification events is well recognized (40, 41). To understand whether phagosomal acidification was important for type I IFN induction, we treated macrophages with an established phagocytosis activation signal (LPS+ IFN-γ) or phagosomal acidification blockers, bafilomycin A and chloroquine, before infection with M. tuberculosis. Activation of macrophages with IFN-γ and LPS enhanced the presence of M. tuberculosis in lyso+ vesicles (Fig. 3D). Importantly, the numbers of lyso+ N73 M. tuberculosis increased to comparable levels as N24 M. tuberculosis by 6 h of infection. Consequently, a 10-fold higher type I IFN response was observed in N73-infected IFN-γ+ LPS-pretreated macrophages, leading to loss of the differential IFN response between strains (Fig. 3E and inset panel). Although addition of IFN-γ and LPS significantly boosted type I IFN induction even in N24-infected macrophages, treatment with phagosome acidification blockers such as bafilomycin A or chloroquine, on the contrary, stunted (∼5- and 2-fold, respectively) the IFN response in these macrophages (Fig. 3F), emphasizing the importance of phagosomal trafficking and acidification in M. tuberculosis–mediated IFN induction in macrophages.
The proinflammatory response of macrophages to M. tuberculosis lineages correlates with IFN1-inducing capacity
Infection of macrophages with M. tuberculosis induces transcriptional programs and elaboration of high levels of proinflammatory cytokines TNF and IL-6. To test whether L1 M. tuberculosis strains were attenuated in activating macrophages overall, we analyzed TNF and IL-6 expression in macrophages infected with L1 and L3 M. tuberculosis. Similar to the situation with IFN1 expression, we observed lower levels of TNF and IL-6 in both RAW264.7 macrophage lines and in primary BMDMs from C57BL/6 mice (Fig. 4A, 4B). Contrasting with the high levels of IL-6 secretion in L3-infected RAW264.7 cells, L1 M. tuberculosis failed to elaborate this cytokine. Even with TNF, the levels in L1-infected macrophages were significantly lower than in the L3-infected cells. This refractory pattern of L1 M. tuberculosis was also observed in BMDMs with significantly lower levels of IL-6 cytokine in the cell supernatants by 24 h of infection. In fact, over a period of 72 h of infection of macrophages, L3 M. tuberculosis strains induced significantly higher levels of IL-6 from as early as 12 h p.i. and sustained higher levels thereafter. In contrast, minimal IL-6 was observed in the supernatant from macrophages infected with L1 M. tuberculosis strains at all the time points of infection (Fig. 4C). A similar pattern of low IL-6 induction by L1 M. tuberculosis was also observed in human MDMs of four healthy individuals at 6 and 24 h of infection (Fig. 4D). Further, analysis of transcript levels by qPCR mirrored the pattern of secreted IL-6 in infected BMDMs with ∼30-fold higher il6 expression in L3 M. tuberculosis infection as early as 6 h p.i. Expression levels declined by 24 h in L3-infected macrophages, although remaining at significantly higher levels than in the N73-infected cells at this time of infection (Fig. 4E), suggestive of a strong correlation between type I IFN induction and IL-6 induction by L3 M. tuberculosis strains in macrophages.
Type I response–regulated oxysterol expression is important for macrophage IL-6 in M. tuberculosis infections
Given the importance of STING signaling on type I IFN response by M. tuberculosis in our study, we first checked whether the expression of IL-6 and TNF was dependent on this pathway, and we compared expression levels in WT and STING−/− RAW264.7 macrophages p.i. with M. tuberculosis N24 or N73. Even in these cells, Il6 expression in L3-infected cells was higher by ∼10-fold than in L1-infected WT cells at 24 h p.i., mirroring the situation in primary macrophages (Fig. 5A). Interestingly, absence of STING completely abolished this response on infection with M. tuberculosis, alluding to the importance of this signaling in expression of IL-6 in macrophages. In contrast, lack of STING signaling did not significantly alter the Tnf expression levels in these macrophages. Previous studies have implicated a positive regulatory loop of type I IFN and IL-6 in macrophages mediated by the expression of the oxysterol-25-hydroxyl cholesterol (42, 43). To evaluate the possibility of this metabolic circuit, we first compared the expression of the gene Ch25h in BMDMs p.i. with M. tuberculosis N24 or N73. Infection with N24 induced this gene strongly (∼70-fold) by 6 h p.i., as opposed to the significantly lower levels (∼2- to 3-fold) in cells infected with N73 (Fig. 5B). The levels remained higher for N24 infection despite a decline in expression levels to ∼20-fold by 24 h. Infection with N24 also induced this gene in RAW264.7 macrophages by 6 and 24 h p.i., although at lower levels than in BMDMs (Fig. 5C). In contrast, the complete loss of Ch25h expression in STING−/− cells in response to N24 infection established the central role of type I IFN signaling in regulating oxysterol expression in response to M. tuberculosis infection in macrophages. To verify the role of Ch25h in M. tuberculosis infection–induced IL-6, we silenced this gene in RAW264.7 macrophages by 50% (Fig. 5D) and observed a significant reduction in the levels of Il6 induction p.i. with N24, again supporting the criticality of the novel regulatory circuit of type I IFN, oxysterol, and IL-6 expression in macrophages in response to M. tuberculosis infection.
The working model
Clinical M. tuberculosis strains differ in their intracellular trafficking routes to arrive at or avoid acidified compartments; L3 M. tuberculosis traffic to arrive early into acidified compartments of host macrophages. The reduced movement of L1 M. tuberculosis strains likely hinders the access of M. tuberculosis DNA to cytosolic DNA sensors, thereby limiting the nucleic acid response pathway and leading to weaker type I IFN response. The higher activation of type I IFN by a subset of M. tuberculosis strains manifests as a substantial upregulation of the proinflammatory cytokine IL-6 via induction of CH25H in a positive feed-forward loop (Fig. 6).
The grand success of M. tuberculosis as a human pathogen stems from its extreme plasticity to suit the host cell niche, modulate the host response, and survive multiple intracellular stresses. Host phagocytes immediately on contact activate inflammatory responses aimed at controlling the growth of the invading pathogen. Typically, the type I IFN response is directed against viral infections, but recent studies have highlighted the macrophage expression of type I IFN-β in response to intracellular (pathogenic) and extracellular bacterial (protective) infections (44–49). We observed that modern M. tuberculosis strains are proficient in inducing this response compared with the evolutionarily older strains of M. tuberculosis. We also observed a strong correlation between the STING-mediated IFN induction and high levels of IL-6 response in macrophages, arguing for a novel mechanism of macrophage response regulation by clinical M. tuberculosis strains. Recent evidence also connects the type I IFN–induced CH25H with the expression of proinflammatory cytokines such as IL-6 in macrophages (42, 43, 50–52). Given the lower expression of genes for cholesterol biosynthesis (data not shown) in the late stage of infection with N24 M. tuberculosis in line with the earlier report (53), it was surprising that expression levels of CH25H were, on the contrary, increased, signifying its importance in bridging type I IFN and IL-6 expression in macrophages. Moreover, the complete loss of CH25H expression in the STING-deficient macrophages after N24 infection highlights this novel regulatory loop of type I IFN–CH25H-IL-6 regulation in M. tuberculosis infections. Overall, we define a novel pathway of proinflammatory cytokine expression in M. tuberculosis–induced macrophages involving the early innate type I IFN response and active modulation of the host cell oxysterol metabolism. A detailed evaluation of STING activation, oxysterol synthesis, and IL-6 expression in M. tuberculosis infections will form the basis of future studies. Although such associations need to be further explored in the alternatively activated alveolar/tissue macrophages (54, 55) in comparison with the in vitro–derived macrophage models used in this study, our results identify a putative association between known pathogenicity of modern lineages (4) and their ability to mount a feed-forward proinflammatory response via the type I IFN axis in macrophage cellular models.
As demonstrated previously, loss of STING/cGAS completely abrogated M. tuberculosis type I IFN response in macrophages, implicating cellular DNA signaling as a predominant mechanism of this response. Given the early induction of type I IFN and the absence of this response despite the presence of active ligand, we argued that a novel mechanism that actively concealed the DNA from the cellular nucleic acid signaling mechanism was responsible for the lower IFN induction by the L1 M. tuberculosis strains. Interestingly, loss of TRIF also resulted in reduction of IFN response in N24-infected macrophages, suggesting a role for RNA signaling in this response. Surprisingly, we did not observe loss of IFN induction by cellular extracts treated with RNase. Recent reports have identified a TRIF-dependent activation of STING in the DNA-induced IFN1 response of macrophages (56); our results demonstrate a dominant role for this signaling pathway in M. tuberculosis–mediated type I IFN response.
Two separate observations supported our hypothesis that intracellular localization early in infection would be primarily accountable for this differential IFN signaling by M. tuberculosis strains: induction of IFN response by (1) transfected BCG and N73 as opposed to direct infection and (2) absolute requirement of DNA to be transfected. Bacterial lysis by the phagosome-lysosome pathway has been implicated in several of the innate immune response pathways in phagocytes (57, 58).
Our results provide insights into the importance of transit of M. tuberculosis strains via an acidified compartment as an important determinant of IFN1 induction in macrophages. Although on the one hand only live bacteria were able to stimulate the macrophage type I IFN response, dead M. tuberculosis, despite their well-documented higher rates of trafficking to acidified P-L compartments, were unable to induce this response in macrophages, pointing at the ability of L3 M. tuberculosis strains to use its early localization to an acidified compartment for active presentation of DNA (in contrast with dead bacteria that fail in this presentation) to the cytosolic signaling complex. In addition, the loss of M. tuberculosis–mediated IFN response in Baf A/chloroquine-treated macrophages and a significant increase by LPS+IFN-γ pretreatment reveals an important role of phagosomal acidification in type I IFN induction.
Although sulfolipids have been demonstrated in promoting phagosomal acidification and L1 strains lack sulfolipids, our results do not purport an active role for this lipid in the type I IFN response. Despite the presence of mature sulfolipids in the L1 subtype M. tuberculosis strains T83 (35), induction of type I IFN was relatively low in THP1 macrophages. Moreover, although differences in sulfolipid expression between H37Ra and H37Rv are well established (59, 60), there was no discernable alteration of IFN1 expression between the two M. tuberculosis strains. The complete lack of change in IFN1 induction ability of a sulfolipid-deficient Erd (ΔM8) further absolves the critical role of sulfolipid in type I IFN induction by M. tuberculosis. In addition, in contrast with the screen by Barczak et al. (39) that identified PDIM as a major contributor of type I IFN signaling in M. tuberculosis infections, we did not observe any change in expression levels of this glycolipid in the two lineages, again indicative of a new mechanism of type I IFN induction by these strains.
In summary, we identify a novel association of clinical strains of M. tuberculosis to mount a pathogenic immune response via their early passage through an acidified endosomal compartment. This passage allows presentation of their DNA to cytosolic sensors, which then triggers an oxysterol-driven feed-forward loop of proinflammatory responses. An understanding of these early trafficking events in the future will help identify means to counter this pathogenic immune response triggered by M. tuberculosis.
We thank Jeffery S. Cox, UC Berkeley, for Erd and National Institute of Medical Research (NIMR), London, for the clinical strains. We thank Manish Kumar for the confocal microscopy facility. The transmission electron microscopy, mass spectrometry, and BSL3 facilities are duly acknowledged.
This work was supported by the Council of Scientific and Industrial Research, India (CSIR) (Grants BSC0123, BSC0403, MLP2012, and STS0016) and Department of Biotechnology, Ministry of Science and Technology, India (DBT) (Grant GAP0096). The student fellowships from CSIR and DBT are acknowledged. The support provided by the Academy of Scientific and Innovative research is duly acknowledged.
D.S., P.A., A.B., S.G., and V.R. were involved in conceptualizing and design of the work; the work was performed by D.S., P.A., R.P.V., A.S., and A.B.; and V.R., P.A., and D.S. were involved in the manuscript preparation and proofreading.
The online version of this article contains supplemental material.
Abbreviations used in this article:
bone marrow–derived macrophage
M. tuberculosis strain Erdman
(Mycobacterium tuberculosis) lineage 1
multiplicity of infection
National Institute of Medical Research
reactive oxygen species
The authors have no financial conflicts of interest.