C57BL/6J, Thy1.1 C57BL/6J, NFkBp50^-/-, IFNAR^-/- male mice were provided by the NIAID Taconic Farms Animal supply Contract (Hudson, NY, USA). TLR-2^-/-, TLR-4^-/- and TLR-9^-/- mice were kindly provided by Dr. Giorgio Trinchieri and CARD9^-/-, Sting^-/- and IFNγR^-/- mice by Drs. Michail Lionakis, Mahtab Moayeri and Dragana Jankovic (NIAID), respectively. Irg1-GFP mice were developed by CRISPR/Cas9 mediated generation of an Acod1-C-terminal GFP fusion mouse. The genomic sequence corresponding to the Acod1 locus (Accession number: NM_008392.1) was downloaded from the UCSC Genome Browser. The sequence surrounding the C-terminal exon was used to identify putative guide RNA sequences using sgRNA Scorer 2.0 (PMID: 28146356). Candidate sequences were identified and tested in P19 cultured cells for editing efficiency. Two candidate sequences (protospacer adjacent motif in bold); (CACGGTGGAAAGCCTTATAACGG) and (GAAACAGAGACAAGCGTATATGG) were used for mouse model generation. A long ssDNA donor containing approximately 500 bp of homology on each side of the stop codon, linker, GFP sequence and silent mutations in the sgRNA target sites was created using the Guide-it production kit (Clontech), which was complexed to Cas9 protein. Cas9 protein was expressed from the SP-Cas9 expression plasmid (Addgene #62731) (PMID: 25910214) by the Protein Expression Laboratory at the Frederick National Lab for Cancer Research. SP-Cas9 was a gift from Niels Geijsen (Addgene plasmid # 62731; http://n2t.net/addgene:62731; RRID:Addgene_62731). Embryos were injected with the ssDNA:Cas9 complex and surgically transferred to recipient mice within 24 hr. Tail biopsies from the recipient mice were used for PCR genotyping using these primers: Left forward (5’-AACAAACAGATCTTGGACCTG-3’), Left reverse (5’-CGTCGTCCTTGAAGAAGATG-3’), Right forward (5’-CATCTTCTTCAAGGACGACG-3’), Right reverse (5’-GAAGACATTCAGCGTGGTTG-3’).

All mice were on a C57BL6 genetic background and bred and housed in NIH animal biosafety level (ABSL) 2 or 3 facilities and used at 8-11-weeks old. Male mice were employed in all experiments. All studies were conducted in accordance with protocol LPD-99E approved by the NIAID Animal Care and Use Committee. All animals were maintained under specific pathogen-free conditions with ad libitum access to water and food. C57BL/6J or Thy1.1 C57BL/6J mice were used interchangeably in different experiments with indistinguishable results.

The H37Rv strain of Mtb and BCG (Pasteur strain) were grown in Middlebrook 7H9 plus Tween 80 (0.05%) enriched with 10% ADC (albumin, dextrose and catalase). The genetically modified fluorescent reporter strain H37Rv-RFP was originally provided by Dr. Joel Ernst (University of California, USA) and grown in Middlebrook 7H9 enriched with 0,05% tween 80, 10% ADC and 30 µg/ml kanamycin (Sigma-Aldrich, USA). All bacterial strains were maintained in culture at 37°C until the midlog phase (OD 0.6-0.9) on day 7.

BMDMs were prepared according to previously published protocols (Amaral et al., 2019). Briefly, bone marrow cells were harvested from femurs and tibia of mice by flushing with 1X PBS. A single-cell preparation was obtained by careful cycling through a 26-gauge needle. Cells were then cultivated in petri dishes (100 x 15 mm) with DMEM/F-12 (Gibco, USA) supplemented with sodium pyruvate (Gibco, USA; 1 mM), L-glutamine (Gibco, USA; 2 mM), 1% Hepes (Life Technologies, USA), gentamicin (Gibco, USA; 25μg/ml) and 10% fetal bovine serum (FBS) and 20% L929 supernatant media and incubated at 37°C and 5% CO₂. L929 cell-conditioned medium (10 ml) was added on day 4 and the cells were cultured for another 3 days to achieve their full differentiation into macrophages.

BMDMs were detached by flushing with cold PBS and seeded in 24-well plates at 10⁶ cells/well. In some experiments, cells were treated with phagocytosis inhibitors [1 µM mycalolide B (Enzo Life Sciences, USA) (Tosh et al., 2016); or 40 µM dynasore (Sigma Aldrich, USA)] for 1 h prior to bacterial infection, or with inhibitors of gene transcription [1 μg/ml actinomycin D (Sigma Aldrich, USA)] or protein synthesis [10 μg/ml cycloheximide (Sigma Aldrich, USA)] for 2 h before infection. In other experiments BMDMs were treated with recombinant IFN-β (R&D Systems – USA) or IFN-γ (R&D Systems – USA) concomitant with infection. BMDMs were infected with H37Rv, H37RV-RFP or BCG strains routinely at a MOI of 1:1 for 3 h, washed and then cultivated for additional 3 h, or stimulated with LPS (10 ng/mL). In some experiments, Mtb was opsonized with fresh or heat-inactivated (30 min, 56°C) C57BL/6 mouse sera for 1 h prior to macrophage exposure.

Human elutriated monocytes were obtained from peripheral blood of healthy donors from the NIH blood bank under a protocol approved by NIAID and the Department of Transfusion Medicine, and were differentiated into macrophages as previously described (Amaral et al., 2019). Briefly, CD14⁺ monocytes were purified by magnetic column, plated in 96-well plates (Corning, USA) containing RPMI 1640 medium (Life Technologies, USA) and supplemented with 10% human AB serum (Corning) and human rM-CSF (50 ng/ml; Peprotech, USA) for 7 days with addition of fresh media with the indicated growth factors every 48 h. Differentiated macrophages were then infected with H37Rv bacilli (routinely at MOI=1:1).

C57BL/6 (BL/6) mice were infected via aerosol with approximately 100 H37Rv bacilli, in a whole body inhalation system (Glas Col, USA). Mouse infection was confirmed by plating lung homogenates obtained at day 0 of infection on 7H11 agar medium (Sigma-Aldrich) enriched with 0.5% glycerol (Mallinckrodt Pharmaceuticals, USA) and 10% OADC (BD Biosciences, USA) media.

Total RNA was extracted from cells or lung homogenates using RNA isolation kits (Zymo Research, USA) and RT-PCR performed as previously described (Mayer-Barber et al., 2014). Briefly, the cells or the post-caval lobe of mice lung were lysed in 300 µl of TRIZOL (Life Technologies) and passed through columns (Zymo Research, USA) according to the manufacturer’s protocol. Total RNA was extracted by disrupting the pulmonary tissue in 2 ml tubes containing 2.7 mm glass bead and TRIZOL, using RNA isolation kit. The isolated RNA was quantified and diluted to 100 ng/µl. For cDNA conversion, dNTP mixture, oligo (dT), random primers and superscript II reverse transcriptase (Thermo Fisher Scientific, USA) were used. Real time PCR reaction was performed on an 790HT Fast Real Time PCR device with Quant-Studio 7 Real-Time PCR Systems (Thermo Fisher Scientific, USA). Irg1, Tnf, Il12p35, Ifnb1, Oas1a and Mx2 gene expression were determined by Sybr Green qRT-PCR (Applied Biosystems, USA), according to the manufacturer’s instructions. GAPDH was used as housekeeping gene and changes in mRNA expression between control and infected macrophages were calculated using the 2^^-ΔΔCT method (Fold change). The sequences of the specific primers employed are listed in Table 1.

BMDMs (10⁶) were stained using fluorochrome-labeled monoclonal antibodies to CD11b and TLR2. In some experiments, cells were infected with H37Rv bacteria expressing red fluorescent protein (RFP). The samples were acquired using a FACS Fortesa flow cytometer (BD Biosciences, USA) and analyzed by FlowJo software (FlowJo LLC, USA).

Student’s t tests and One-Way ANOVA followed by Tukey’s post-test were performed using Prism 7 software (GraphPad Software Incorporated, USA). Data are shown as mean ± SEM values. Differences between groups were considered significant when the p-value was <0.05, with asterisks denoting the degree of significance (*P < 0.05; **P < 0.01 and ***P < 0.001).

Before dissecting the signaling pathways involved in Irg1 expression, we confirmed the ability of Mtb (H37Rv strain) to induce this response in vitro and vivo. High levels of Irg1 gene message were observed in the lungs of Mtb-infected mice beginning at 3 weeks after aerosol infection (Figure S1A). Consistent with previous findings (Shi et al., 2005; Nair et al., 2018; Hoffmann et al., 2019), Mtb also induced potent Irg1 gene expression in both human monocyte-derived macrophages (Figure S1B) and mouse BMDMs (Figure S1C). In the case of the murine cells, high Irg1 transcript levels were observed as early as 3 h, peaking at 6-9 h and declining at 12 h after bacterial exposure. Mtb also induced Tnf and Il12 gene expression with similar kinetics in the same macrophages, but the highest response of these cytokine messages proceeded the peak of Irg1 expression by approximately 3h (Figures S1D, E). Remarkably, significant Irg1 induction was observed at multiplicities of infection as low as 0.1:1 illustrating the potency of Mtb as a trigger of this gene (Figure S1F).

To demonstrate that Mtb-induced Irg1 gene expression results in production of the Irg1 protein, we employed a recently developed GFP reporter mouse in which Irg1 production is associated with a fluorescent signal. BMDMs from these mice were exposed to H37Rv bacilli expressing RFP reporter. When examined at 6 h post-infection (p.i.), Irg1-GFP production by macrophages dramatically increased at MOIs as low as 1:1 and at 24 h most cells expressed the protein (Figures 1A–C). Interestingly, a sizeable fraction of Irg1-GFP⁺ BMDMs showed an uninfected Mtb-RFP^- phenotype, suggesting the existence of paracrine Irg1 induction (Figures 1A, B). Nevertheless, in infected BMDMs, the induced Irg1 protein levels directly correlated with the MFI of H37Rv-RFP expression (Figure 1D).

The cellular response to Mtb has been shown to depend on recognition of the bacillus by pattern recognition receptors (PRR) located in the plasma membrane, endosomes and host cell cytoplasm (Means et al., 1999; Underhill et al., 1999; Schoenen et al., 2010; Collins et al., 2015; Basu et al., 2018; Wagener et al., 2018; Chai et al., 2020). The Toll-like receptor family plays a major role in this process triggering host gene expression largely through MyD88 dependent NFκB mediated signaling (Underhill et al., 1999; Scanga et al., 2004; Stamm et al., 2015). To assess the involvement of TLR signaling in Mtb-induced Irg1 expression, we compared BMDMs from MyD88 or NFκBp50-deficient mice with BL/6 macrophages. We observed at 6 h p.i. a partial but significant decrease in Irg1 gene expression in the absence of MyD88 and NFκBp50 (Figures 2A, B). Analysis of the macrophage response at 12 h p.i. showed that this reduction in Irg1 levels was not due to a delay in gene expression.

A major MyD88 dependent PRR triggered by Mtb is TLR-2 (Yu et al., 2014; Stamm et al., 2015) and consistent with its well documented importance (Underhill et al., 1999), BMDMs from TLR-2^-/- mice, but not from TLR-4^-/- or TLR-9^-/- mice, displayed a defective Irg1 response to Mtb similar to MyD88^-/- and NFκB-p50^-/- macrophages (Figures 2A, B). In control experiments we showed that BMDMs produce significant Irg1 responses to LPS and (Figure 2C) CpG oligonucleotides (Figure 2D) confirming that the failure of TLR-4 and TLR-9 deficiency to inhibit Mtb induced Irg1 is not due to a generalized defect in the function of these TLR under the conditions employed (Figure 2D). In addition to TLR, a second major PRR family involved in Mtb innate recognition are the c-type lectins (CLR), and in particular dectin 1 and mincle, that signal through the adaptor caspase recruitment domain-containing protein 9 (CARD9) (Wagener et al., 2018). Nevertheless, CARD 9^-/- BMDMs exposed to mycobacteria expressed Irg1 at levels similar to control BL/6 macrophages, demonstrating that these CARD9-dependent CLR pathways are not required for the Mtb-induced Irg1 response (Figure 2E, left panel). In addition, trehalose dimycolate the major Mtb ligand for mincle stimulated only marginal Irg1 expression (Figure 2E, right panel). Together, these experiments revealed a substantial and highly specific contribution of the TLR2-MyD88-NFκB pathway for Irg1 induction in Mtb-infected macrophages, presumably occurring during initial contact of the pathogen with the host cell surface and/or endosomal membranes.

Mtb infection of macrophages is initiated by phagocytosis of the bacteria that are able to survive by inhibiting phagosome-lysosome fusion and/or by escaping into the cytosol (Scherr et al., 2009; Zhai et al., 2019). Signaling by PRR present in endosomal membranes or in the cytoplasm thus could also be involved in stimulating Irg1 expression. To determine whether Irg1 induction requires phagocytic uptake of Mtb, BMDMs were treated with increasing concentrations of an actin polymerization inhibitor (mycalolide B) or a dynamin inhibitor (dynasore), or cells were maintained at 4°C before and during bacterial infection. All treatments markedly inhibited phagocytosis (Figure 3A), without effecting cell viability (Figure S2A). Importantly, phagocytosis inhibition was associated with a sharp decrease in Irg1 gene expression induced by Mtb infection, whereas no change was observed for LPS-stimulated Irg1 responses (Figure 3B). Moreover, phagocytosis of inert latex beads did not induce Irg1 expression or enhance the response triggered by either Mtb or LPS (Figure 3B), demonstrating that phagocytosis in itself is not sufficient to trigger Irg1. Furthermore, phagocytosis inhibition did not affect cell surface TLR2 levels (Figure 3C), ruling out this possible explanation of the data. Finally, Irg1 expression was increased by opsonization of Mtb with fresh mouse serum, but not with heat-inactivated serum (Figure 3D). The opsonization-enhanced Irg1 response was seen at 3 h p.i. but not at 6 h p.i., suggesting that Irg1 expression was accelerated by promoting mycobacterial phagocytosis through macrophage complement receptors. Indeed, phagocytosis assays performed with RFP-labelled Mtb showed that opsonization with fresh serum increased the frequency of infected macrophages in comparison with unopsonized bacteria or Mtb pre-incubated with heat inactivated serum when examined at 1h and 3h p.i. (Supplementary Figure 2B).

We next evaluated whether the presence of Mtb products in the cytosol is required to induce Irg1 gene expression. First, we observed that the inhibition of phagosome acidification with bafilomycin reduced Irg1 expression in Mtb-infected BMDMs, but not in LPS-stimulated BMDMs (Figure 4A). Tnfa gene expression was not affected by this treatment, revealing a preferential role for phagosome acidification in the induction of Irg1 expression in Mtb-infected BMDMs (Figure 4A).

It is well established that Mtb from virulent strains, such as H37Rv, release molecules into the cytoplasm that activate cytosolic sensors by a mechanism dependent on the bacterial virulence factor ESAT-6 (Brodin et al., 2006; Groschel et al., 2016; Conrad et al., 2017). Therefore, we hypothesized that translocation of mycobacterial antigen from phagocytic vesicles into the cytosol and activation of cytosolic sensors could trigger Irg1 expression in Mtb-infected macrophages. To investigate this possibility, Irg1 responses were evaluated in Mtb-infected or LPS-treated BMDMs deficient in STING, a signaling adaptor activated by cGAS, a cytosolic DNA sensor stimulated during intracellular Mtb infection as a result of phagosomal permeabilization and required for its induction of Type IFN (Manzanillo et al., 2012; Collins et al., 2015; Watson et al., 2015; Cheng and Schorey, 2018). We observed markedly reduced levels of Irg1 expression in Mtb-infected STING^-/- BMDMs compared to their control BL/6 counterparts (Figure 4B), indicating an important role for the STING activation pathway in this response. In contrast, the Irg1 response to LPS was not altered in STING^-/- macrophages (Figure 4C). Also, in agreement with previous studies showing the requirement of STING for IFN-β production by Mtb-infected macrophages (Manzanillo et al., 2012; Collins et al., 2015; Watson et al., 2015), no increase in Ifnb1 expression and, consequently, in interferon-stimulated genes (ISGs) Oas1a and Mx2 were observed after Mtb infection of STING^-/- BMDMs (Figure 4D).

To confirm that cytosolic sensing of Mtb antigens is crucial to the Irg1 response in infected BMDMs, we used BCG, an avirulent strain of Mycobacterium bovis (Mbv) that lacks the genes encoding the ESX-1 secretion system which includes ESAT-6 as a major component, and consequently, is unable to release antigens into the cytosol and activate STING. We observed that Irg1 expression was lower in BCG-infected BL/6 BMDMs compared to H37Rv-infected BL/6 BMDMs (Figure 4E), as was the induction of Ifnb1, Oas1a and Mx2 expression (Figure 4F). Furthermore, Irg1 was expressed at comparable low levels in BCG-infected BL/6 and STING^-/- BMDMs, consistent with the inability of these mycobacteria to activate STING (Figure 4G). Interestingly, no induction of Irg1 expression was observed in BCG-infected MyD88^-/- BMDMs (Figure 4H), perhaps because TLR2 and STING signaling are both absent in this situation. In contrast, BL/6 and MyD88^-/- BMDMs expressed high Irg1 levels after LPS stimulation, suggesting the involvement of the alternative TRIF pathway of TLR4 signaling in the case of this agonist (Figure 4H).

As IFN-β responses are impaired in the absence of STING, BMDMs deficient in the type I interferon receptor (IFNAR^-/-) were evaluated to determine whether this cytokine is involved in Irg1 expression. In agreement with previously published observations (Shi et al., 2005), we observed that IFNAR^-/- BMDMs expressed substantially lower Irg1 levels compared to BL/6 BMDMs after Mtb infection, whereas the Tnf response was not significantly affected. In addition, LPS-treated IFNAR^-/- BMDMs also displayed a diminished Irg1 response (Figure 5A). This observation can be explained by the known ability of LPS to stimulate Type I IFNs via TRIF-induced triggering of IRF3 (Monroe et al., 2010; Hu et al., 2012). As expected, Mtb-infected IFNAR^-/- BMDMs also showed defective ISG (Oas1a and Mx2) expression along with unchanged Ifnb1 levels (Figure 5A). Furthermore, stimulation with recombinant IFN-β substantially increased Irg1 expression in Mtb-infected BL/6 BMDMs (Figure 5B) while at the concentration employed a minimal response was triggered in uninfected macrophages. Together these data reinforce and expand previous findings (Shi et al., 2005) indicating that Type I IFN signaling is an important regulator of Irg1 expression during Mtb infection. Consistent with a de novo synthesis requirement of Type I IFNs for Irg1 induction, addition of the protein synthesis inhibitor cycloheximide at the time of infection partially suppressed the Irg1 response to Mtb, which as expected was fully blocked by the transcription inhibitor actinomycin D (Figure 5C).

Because in addition to Type I IFN, Type II IFN is also a potent inducer of Irg1 in macrophages (Price et al., 2019), the response of Mtb-infected IFNγR^-/- BMDMs was also assessed. The results indicated that in contrast to Type I signalling, IFN-γ signalling does not contribute to the Irg1 response to Mtb (Figure 5D). That Mtb induces insufficient amounts of IFN-γ to trigger autocrine activation of BMDMs was suggested by the observation that stimulation with the recombinant cytokine enhanced Irg1 expression in both uninfected and Mtb-infected BMDMs (Figure 5D). Moreover, combined IFN-γ plus IFN-β treatment did not further potentiate Irg1 expression over stimulation with IFN-γ alone. Together, these data indicate that in vitro Type I but not Type II IFN, play a major role in driving the Irg1 response in Mtb-infected macrophages, although it remains possible that in vivo Irg1 expression can be be enhanced by paracrine IFN-γ produced by other cell types.

Our data indicated that two main phagocytosis-dependent signaling pathways involving the TLR-2/MyD88/NFκB and STING/IFN-I/IFNAR axes take part in the Irg1 response to Mtb infection. To investigate the interdependence of these signaling pathways, Ifnb1 expression was assessed in Mtb-infected TLR-2^-/- and control BL/6 BMDMs. Because no difference in Ifnb1 expression was observed between these two macrophage types (Figure 6A), we concluded that TLR-2 does not mediate its effects by triggering Type I IFN production in Mtb-infected BMDMs. Moreover, TLR2 signaling was not required for phagocytosis of Mtb, which was similar in TLR-2^-/- and BL/6 BMDMs (Figure 6B). Together, these findings demonstrate that TLR2 signaling by Mtb promotes Irg1 expression in macrophages through a STING-independent signaling pathway.