Pharmacological Poly (ADP-Ribose) Polymerase Inhibitors Decrease Mycobacterium tuberculosis Survival in Human Macrophages

Diabetes mellites (DM) is correlated with increased susceptibility to and disease progression of tuberculosis (TB), and strongly impairs effective global TB control measures. To better control the TB-DM co-epidemic, unravelling the bidirectional interactivity between DM-associated molecular processes and immune responses to Mycobacterium tuberculosis (Mtb) is urgently required. Since poly (ADP-ribose) polymerase (PARP) activation has been associated with DM and with Mtb infection in mouse models, we have investigated whether PARP inhibition by pharmacological compounds can interfere with host protection against Mtb in human macrophage subsets, the predominant target cell of Mtb. Pharmacological inhibition of PARP decreased intracellular Mtb and MDR-Mtb levels in human macrophages, identifying PARP as a potential target for host-directed therapy against Mtb. PARP inhibition was associated with modified chemokine secretion and upregulation of cell surface activation markers by human macrophages. Targeting LDH, a secondary target of the PARP inhibitor rucaparib, resulted in decreased intracellular Mtb, suggesting a metabolic role in rucaparib-induced control of Mtb. We conclude that pharmacological inhibition of PARP is a potential novel strategy in developing innovative host-directed therapies against intracellular bacterial infections.


INTRODUCTION
Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis (Mtb). Around 10 million new TB cases and 1.4 million deaths are reported annually and one quarter of the world's population is latently infected with Mtb (1). Different risk factors have been identified for progression of latent TB infection (LTBI) to active disease, including age, malnutrition, coinfection with human immunodeficiency virus (HIV) or cytomegalovirus (CMV), and diabetes mellitus (DM) (1)(2)(3). In addition, recent studies have shown that DM is also associated with poorer clinical TB outcome following anti-bacterial treatment. Current estimates are that 15% of the global TB burden is now associated with DM (4,5). Host-directed therapy (HDT) offers the potential for better treatment of Multi-Drug-Resistant (MDR)-TB, as well as shorten current standard (6-9 month) treatment regimens, thereby reducing toxicity, enhancing treatment compliance and as a result reducing emergence of de novo drug resistance. To better control the TB-DM co-epidemic, unravelling the bidirectional interactions between DM-dysregulated molecular processes and immune responses to Mtb is essential.
Macrophages (MFs) are known for their dual role in TB pathogenesis, as they provide a primary host niche for Mtb during infection while also being key host effector cells eliminating Mtb, and therefore represent key target cells in developing and evaluating novel therapeutic strategies for TB/ DM, including innovative HDT. MFs are classically subdivided into pro-inflammatory macrophages (M1) and antiinflammatory macrophages (M2), representing the polar ends of the macrophage spectrum. Both M1 and M2 have been reported in tuberculous granulomas, with M2 predominating in granulomas from patients with active TB (6).
Poly (ADP-ribose) polymerase (PARP) activation has been associated with cancer, diabetes and endothelial dysfunction in experimental mouse models and in humans (7)(8)(9)(10). PARP-deficient mice and mice treated with PARP inhibitor PJ34 were resistant to streptozocin-induced diabetes, and pharmacological inhibition of PARP has been proposed for the treatment of DM (11)(12)(13). Interestingly, in addition, PARP inhibitors have recently been proposed as HDT compounds for reducing TB-induced inflammation (14), thereby providing a potential mechanistic link between the TB-DM co-morbidity at the molecular level. Several PARP inhibitors (PARPi) have been clinically approved for the treatment of cancer or are currently being evaluated in clinical trials for the treatment of advanced BRCA1/2 mutant ovarian and breast cancers, which can accelerate translation to HDTs against Mtb. PARP encompasses a family of enzymes involved in different cellular processes such as control of genomic stability, programmed cell death and DNA repair. When assisting in the repair of single-strand DNA nicks, PARP enzymes bind to single-stranded DNA breaks (SSB) or double-stranded DNA breaks (DSB) to generate PAR polymers on itself (auto-PARylation), histones and chromatin-associated proteins, which together leads to chromatin relaxation and recruitment of repair proteins. Inhibition of PARP results in accumulation of SSB due to delayed DNA repair. Given the dual role of PARP activation on diabetes and tuberculosis infection, we hypothesized that small molecule inhibitors that interfere with PARP activation at the cellular level may be able to redirect macrophage function in response to Mtb infection. Such chemical compounds could be valuable tools as part of immunomodulatory HDT regimens.
In the current study, we explore the potential of PARP inhibitors in the treatment of intracellular Mtb infections in human macrophages. We demonstrate that pharmacological inhibition of PARP decreases intracellular Mtb in human macrophages and modulates the immune response of Mtbinfected human macrophages. To our knowledge, the potential of targeting PARP for HDT in a human model of Mtb-infection has not been reported yet, and our work identifies PARPi as novel HDT compounds against Mtb and possibly other intracellular infectious diseases.

Pharmacological Inhibition of PARP Reduces Mtb Survival in Human Macrophages in a Host-Directed Manner
To study whether PARP may play a role during Mtb infections in human macrophages, we first evaluated intracellular Mtb survival after treatment with four clinically relevant PARPi ( Figures 1A, B). We selected 10 µM as a relatively high PARPi concentration, which has been shown to target PARP in vitro (15)(16)(17) and which allowed us to directly compare drug efficacy of PARPi with each other and with control kinase inhibitor H-89 (18,19). Rucaparib significantly decreased intracellular Mtb CFUs in M1 and M2 more efficiently than H-89 ( Figure 1C, n=6). Additionally, niraparib and pamiparib decreased intracellular Mtb in M1 and A-966492 decreased intracellular Mtb in M2, suggesting cell-specific effects of PARP inhibition on bacterial load for these compounds. Furthermore, we assessed the effect of PARPi on cytotoxicity using uninfected macrophages from four different donors, to ascertain that PARPi was not detrimental to healthy human macrophages. PARPi did not affect host-cell integrity and was not cytotoxic ( Figure 1D). Next, we used a flow cytometry-based method (18) to evaluate whether the inhibition of Mtb outgrowth by PARPi was accompanied by a decrease in the percentage of cells harboring Venus-expressing Mtb. Rucaparib, but not other PARPi, significantly reduced the percentage of Mtb-infected cells both in M1 and M2 ( Figure 1E, n=8), further corroborating the Mtb-inhibiting effect of rucaparib in human macrophages. To exclude that PARPi had direct microbicidal effects, liquid Mtb cultures were treated with PARPi in the absence of macrophages in two independent experiments ( Figure 1F). None of the PARPi limited bacterial growth directly, indicating that PARPi restrict intracellular outgrowth of Mtb by modulating host pathways.
To investigate whether PARPi-induced inhibition of intracellular bacterial growth was specific for Mtb, we evaluated whether pharmacological inhibition of PARP also restricted growth of other intracellular pathogenic bacteria, including Mycobacterium avium (Mav), Salmonella enterica serovar Typhimurium (Stm) and methicillin-resistant Staphylococcus aureus (MRSA). Interestingly, niraparib, but not other PARPi, significantly decreased Stm CFUs in M1 and M2 (Figure 2A, n=6) without exhibiting direct microbicidal effects in two independent experiments ( Figure 2B). Rucaparib did not significantly diminish Mav, Stm or MRSA CFUs, suggesting that the effect of rucaparib is specific for Mtb. These results imply that both rucaparib and niraparib inhibit specific species of bacteria through modeling of host signaling pathways.
Taken together, these data suggest that PARP activity is involved in Mtb survival in human macrophages, and that although there is some difference in their efficacy against Mtbinfected M1 compared to Mtb-infected M2, pharmacological inhibition of PARP is a promising strategy to target intracellular Mtb. Moreover, the effects of different PARP inhibitors are highly specific to certain pathogen species.

The Translational Potential of PARPi for Clinical Application
To evaluate the translational potential of PARPi for the development of HDT, we first tested whether PARPi displayed activity against macrophages infected with two different multidrug resistant (MDR)-Mtb strains: an MDR-Mtb strain belonging to the Beijing genotype (strain 16319) and an MDR- Dutch outbreak strain ( Figures 3A, B, n=6), highlighting the potential of PARPi to control intracellular outgrowth not only of drug-sensitive but also MDR-Mtb strains. Next, the interaction of PARPi with first-line antibiotics was studied. Macrophages from five donors were infected with Mtb and treated with PARPi in the presence or absence of a nontoxic, suboptimal dose of RIF or INH ( Figure 3C) (20,21). As expected, PARPi significantly decreased Mtb CFUs in M1 ( Figure 3D). In M2, however, several HDT compounds, including rucaparib, niraparib and positive control H-89 increased Mtb CFUs in two donors, illustrating the typically high variation in the response of donors to HDT compounds. Niraparib, pamiparib and rucaparib significantly enhanced the activity of INH against intracellular Mtb in M1. The coefficient of drug interaction (CID) of niraparib, pamiparib and rucaparib with RIF was close to 1 in M1, suggesting that these PARPi had an additive effect to the antibiotic treatment and not a synergistic effect (Supplementary Table 1). Furthermore, compared with other PARPi, pamiparib seemed to enhance the activity of RIF in M2 and of INH in M1 and M2, albeit not significant.

PARPi Induce Immunomodulatory Effects in Mtb-Infected Human M1 and M2
To investigate whether the effect of PARPi against Mtb correlated with macrophage activation during Mtb infection, cytokine and chemokine production was quantified in the supernatants of Mtb-infected macrophages that were treated with PARPi or vehicle control DMSO overnight. Mtb-infection induced cytokine and chemokine production in M1 and M2 and tended to induce tumor necrosis factor (TNF)-a, macrophage inflammatory protein (MIP)-1a,MIP-1b and IP-10 and fractalkine in M1 and M2 ( Figure 4A  1% v/v). Data represent the means ± S.D. of triplicate wells from a representative experiment out of two independent experiments. Differences were significant by RM two-way ANOVA with Dunnett's multiple test correction against DMSO. *p < 0.05, **p < 0.01, ***, ****p < 0.0001. and significantly decreased IFN-a2 secretion by M2 compared to DMSO. Correlation plots between the cytokines and chemokines that were significantly modified by PARPi and Mtb CFUs did not identify a positive correlation between secretion profiles of macrophages and Mtb loads (Supplemental Figure 1). Collectively, these data suggest that pamiparib and rucaparib can modify the cytokine and chemokine response to Mtb infection. Next, we investigated the expression levels of activation markers on the surface of Mtb-infected M1 and M2 that were treated with PARPi overnight. Mtb-infection tended to increase macrophage activation markers in M1 and M2 ( Figure 4C, n=4). Surprisingly, rucaparib, but not other PARPi, tended to decrease expression of CD192 and CD209 and expression of human leucocyte antigen (HLA)-DR on the surface of Mtb-infected M1 and M2 ( Figure 4D and Supplemental Figure 2, n=4). Furthermore, the expression of costimulatory molecule CD86 on the surface of M2 tended to be increased by rucaparib, and was significantly increased by pamiparib. These data suggest that rucaparib increases macrophage activation and possibly antigen presentation.
Given that several HDT compounds with demonstrated efficacy against intracellular Mtb in macrophages exert their activity by activating autophagy (22)(23)(24)(25), we next assessed whether treatment with PARPi induced autophagy and lysosomal maturation. Accumulation of autophagic and lysosomal vesicles was quantified by confocal imaging on Mtbinfected and PARPi-treated macrophages from four donors that were stained with CYTO-ID or Lysotracker, respectively.  Formation of autophagic or lysosomal vesicles could not be detected after 4h treatment with PARPi compared to DMSO (Supplemental Figures 3A, B). To examine the effect of PARPi on the autophagic flux, macrophages from three donors were treated with PARPi in the presence or absence of bafilomycin A1 (Baf) which is an autophagy inhibitor that prevents fusion between autophagosomes and lysosomes. As expected, treatment with Baf induced microtubule-associated protein light chain 3 (LC3)-II accumulation in M1 and M2 (Supplemental Figure 3C). However, PARPi did not increase LC3-II protein levels, regardless of the presence of Baf. Also, PARPi did not increase LAMP1 protein levels, suggesting that PARPi do not act via induction of autophagic and/or lysosomal pathways to control intracellular Mtb in human macrophages. Collectively, our data demonstrate that PARPi modulate the immune response of Mtb-infected human macrophages via cytokine/chemokine expression and surface markers, which play a role in antigen presentation and chemotaxis. Our data also suggest that a role of induction of autophagy and phagosome maturation in PARP-induced Mtb control is not likely mechanistically involved in the mode of action.

Lactate Dehydrogenase, a Secondary Target of Rucaparib, Induces Mtb Control in Human Macrophages
Hexose-6-phosphate dehydrogenase (H6PD) and lactate dehydrogenase (LDH) were recently identified as additional target molecules of rucaparib (26). We therefore hypothesized that these metabolic targets might play a role in the more profound inhibitory effect of rucaparib on Mtb survival compared to other PARPi ( Figures 1C, E, 3A, 5A). Lactate levels were significantly decreased in the supernatants of Mtbinfected M1 and M2 that had been treated with rucaparib compared to DMSO, suggesting that rucaparib indeed impaired LDH activity ( Figure 5B, n=4). Interestingly, none of the other selected PARPi significantly affected lactate levels, compatible with the fact that none of these have been reported to modulate LDH activity. To study whether inhibition of H6PD or LDH could have contributed to the effect of rucaparib on Mtb control, we treated Mtb-infected macrophages with the specific LDH-inhibitor FX-11 or the G6PD-inhibitor 6-aminonicotinamide (6-AN) and determined intracellular Mtb levels in a flow cytometry-based assay, which allows a more rapid quantification of intracellular Mtb levels compared to classical CFU assays and generally showed a greater effect window of rucaparib compared to CFU assay ( Figure 1C versus 1E). Interestingly, FX-11 significantly decreased the percentage of Mtb-infected M1 (n=7) and M2 (n=8), whereas no effect of 6-AN on intracellular Mtb could be detected ( Figure 5C). FX-11 did not affect the percentage of live M1 cells (n=6), but did decrease the percentage of live cells in a subset of M2 cultures (n=5). Together, these data suggest that inhibition of LDH, but not G6PD, likely contributed to the Mtb-decreasing effect of rucaparib. Corroborating this hypothesis, FX-11 treatment significantly decreased lactate levels in the supernatant of Mtb-infected M1, corresponding with decreased LDH activity ( Figure 5E, n=6). FX-11 treatment did not result in decreased lactate levels in M2, which is likely a result of cellular toxicity in M2 ( Figures 5D, E). Collectively, these data suggest that LDH could be involved in rucaparibinduced inhibition of Mtb and identifies LDH as potential host target during Mtb infections.

DISCUSSION
In search of new potential treatments against TB, we have studied a series of PARPi, which are being considered for treatment of type 2 diabetes, for their ability to improve intracellular growth control by infected human M1 and M2 as novel candidate HDT therapeutics. Here, we showed that clinical compounds targeting PARP were able to decrease the intracellular survival of Mtb, suggesting that pharmacological inhibition of PARP can be a novel tool for TB treatment. PARPi had no direct antimycobacterial activity, suggesting that PARPi act in a HDT like fashion. In line with this, macrophage treatment with PARPi was associated with the modulation of macrophage function at the cytokine level and surface marker level. Lactate dehydrogenase (LDH) was identified as an additional target molecule for rucaparib's activity against intracellular Mtb which might explain the more profound inhibitory effect of rucaparib on Mtb survival compared to other PARPi. To our knowledge, the potential of PARPi as HDTs against Mtb has not been demonstrated before.
PARPs are host proteins involved in cellular processes acting via PARylation of their targets. It has recently been shown that PARP1 activation is triggered in mouse lungs upon Mtb infection (27). Importantly, genetic PARP1 depletion resulted in decreased lung bacterial burden in female mice. Here, we studied the role of PARP inhibition in a human and clinically relevant context. All clinical PARPi used in this study are known PARP1 and PARP2 inhibitors and have the ability to inhibit multiple proteins of the PARP family (28)(29)(30)(31). Our data suggest that the efficacy of PARPi on bacterial clearance is specific for Mtb and does not translate to other intracellular bacteria, including the closely related species Mav. While not investigated here, this suggests that PARP is not a key target for MRSA, Stm and Mav, although the exact role of PARP during other bacterial infections remains to be resolved.
Several PARP inhibitors, such as olaparib and talazoparib, can be metabolized by the hepatic enzyme CYP3A4, which is induced by the first-line TB antibiotic RIF (32)(33)(34). Rucaparib, however, is less sensitive to metabolization by CYP3A4 and the sensitivity of other PARPi remains to be determined. In our in vitro infection model, pamiparib had a significant additive effect to RIF in M1 and seemed to have an additive effect to RIF in M2 and to INH in both M1 and M2. This possible additive effect of pamiparib to RIF and INH in M2 was unexpected, since pamiparib did not show activity against Mtb in the absence of antibiotics in these cells. Although we observed differences in the cellular response to pamiparib in M1 versus M2 based on cytokine/chemokine secretion, it remains currently undefined In addition to their role in DNA repair, mammalian PARPs display properties that are associated with and are likely to impact host-pathogen interactions. Firstly, PARPs are induced by the production of interferon-stimulated genes (ISGs), implicating a role during virus infection (35)(36)(37)(38). The impact of PARP activity on virus replication remains controversial, as both promotion and reduction of viruses by PARP have both been described, which is likely dependent on the virus (as reviewed by (39). Moreover, an elevated type I IFN response correlates with progression to active TB in humans (40)(41)(42) and has been implicated in susceptibility to Mtb in mouse models (43,44). Secondly, PARPs are involved in pro-inflammatory gene expression. Macrophages from PARP-deficient mice were unable to induce nuclear factor kappa B (NF-kB)-mediated activation in response to lipopolysaccharide (LPS), limiting inducible nitric oxide synthase (iNOS), tumor necrosis factor alpha (TNF-a) and interferon gamma (IFN-g) levels (45,46). Here, we show that the inflammatory response of Mtb-infected macrophages is modified by pharmacological inhibition of PARP. At the cytokine level, pamiparib and rucaparib induced expression of chemokines MIP-1a and MIP-1b in M1. We furthermore show that the higher MIP-1a and MIP-1b levels did not correlate directly with increased inhibition of intracellular Mtb, compatible with their primary role in chemotactic cellular recruitment. Decreased MIP-1a expression is associated with diabetes in Mtb-infected human monocyte-derived macrophages and in Mtb-infected diabetic mice (47)(48)(49). This suggests that macrophages obtained from diabetic patients or mice may be impaired in the recruitment of immune cells to Mtb sites of infection. Additionally, alveolar macrophages of diabetic mice feature decreased TNF-a secretion (47). Mice that have impaired TNF-a secretion and mice and humans treated with anti-TNF-a antibodies are more susceptible to Mtb infection, highlighting a direct protective effect of TNF-a against Mtb (50-52). Here, TNF-a was significantly induced by pamiparib treatment and it tended to be induced by A-966492 and rucaparib in M1. However, like for MIP-1a, higher TNF-a levels did not correlate directly with decreased intracellular Mtb levels. Together, these data demonstrate that PARPi treatment alters the cytokine/chemokine response of human macrophages during infection with Mtb. These effects are not directly correlated with intracellular Mtb control by macrophages, but these cytokines/chemokines could exert effects in vivo via other cells such as recruitment of immune cells. The potential of PARPi to improve recruitment of immune cells to sites of infection remains to be elucidated and could play an important role in the in vivo activity of PARPi against Mtb. When evaluating the activation status of Mtb-infected macrophages in response to PARPi treatment, rucaparib appeared to increase the expression of HLA-DR and CD86, in contrast to other PARPi. As previously published, monocytederived macrophages (MDMs) obtained from type 2 diabetes (T2D) patients displayed lower levels of HLA-DR and CD86 compared to MDMs obtained from healthy individuals upon n vitro Mtb infection (48). Collectively, these observations support the hypothesis that rucaparib is able to reverse diabetes-induced functional effects of macrophages and suggest that rucaparib could increase antigen presentation by human macrophages to specific T-cells.
Several studies have shown that diabetes was associated with a decrease in Mtb phagocytosis capacity in human and murine macrophages (47,48,53), possibly resulting from highly glycosylated proteins that impair recognition of bacterial components. Phagocytosis as mechanism of action of PARPi can be excluded in our current study, since PARPi were added well after phagocytosis, and further cell-to-cell exchange of Mtb bacteria was prevented by supplementing the culture medium with low dose gentamicin.
Multiple PARPi, including niraparib, rucaparib and talazoparib have recently been described to induce autophagy in cancer cell lines, as detected by increased LC3-I to LC3-II conversion on western blot analysis (54,55). In our current study, induction of autophagy could not be detected, however, perhaps because non replicating primary cells (M1, M2) were used instead of cell lines.
Proteome and metabolome profiling have recently revealed additional targets for PARPi which could contribute to Mtb control in our human macrophage infection model. In vitro kinase profiling showed that olaparib and rucaparib have a relatively similar affinity for members of the PARP family, but differ substantially in their affinity for protein kinases (29,56). Our observation that rucaparib performed significantly better against Mtb than other PARPi suggests that unique kinases targeted by rucaparib might contribute to its effect against Mtb. The reported unique additional targets of rucaparib include cyclin dependent kinase 1 and 16 (CDK1, CDK16), pim-3 oncogene (PIM3), hexose-6-phosphate dehydrogenase (H6PD) and lactate dehydrogenase (LDH) (26,28,56,57). Here we show that rucaparib decreased lactate production in Mtb-infected M1 and M2. Moreover, the specific LDH inhibitor FX-11 had a similar effect as rucaparib on intracellular Mtb load. This suggests that LDH inhibition could be the major mechanism of action of rucaparib, and as such could be a promising target pathway in developing HDT against Mtb. In agreement with our findings, oral administration of FX-11 was recently described to decrease intracellular Mtb levels in mouse BMDM in a host-directed manner (58). Based on our data, we cannot exclude additional effects of rucaparib on further (secondary) targets during Mtb infection in human macrophages. Similarly, studying the potentially additive effects of the already described unique additional targets of niraparib are interesting in the development of novel HDT against Mav and Stm. Deoxynucleoside kinase (DCK) is a reported additional target of niraparib, and analysis of a published dataset of infected M2 showed that DCK is significantly upregulated 4h post Stm, but not Mtb infection (56,59). Taken together, these data suggest that DCK is an interesting candidate for HDT against Stm.
There are several limitations of the current study. First, the current study has the limitation of small sample size. Although a small sample is common among studies with (primary) cells due to the laborious nature of these experiments, this could have resulted in insufficient statistical power between groups. Furthermore, we applied a single-dose administration of HDT compounds with a relatively short effect time (i.e. after overnight infection), which is an experimental setup that we generally use for the identification of novel HDT compounds (18,20,60). Although applying other compound concentrations, repeated doses or a longer treatment exposure could have resulted in a bigger effect size, the host targets (PARP and LDH) that we identified in our in vitro human macrophage model were recently published as host targets using in vivo mouse models as discussed above (27,58). This agreeing set of independent findings underscores the power and validity of our approach, which aims to decipher novel pathways involved in host defense against intracellular bacterial infections in humans, and develop corresponding novel HDT.
In summary, we identified PARP as a host target for HDT during Mtb infections, and confirmed that PARP inhibition is a promising avenue in the development of HDT against Mtb. Furthermore, our data show that also off-target kinase pharmacology of PARP inhibitors may expand the current clinical scope of PARP inhibitors and that these molecules deserve serious consideration in the development of repurposed HDT against intracellular bacterial infections.

Cell Culture
Human peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats of healthy anonymous donors (Dutch, adults) after written informed consent (Sanquin Blood Bank, Amsterdam, The Netherlands) by density gradient centrifugation over Ficoll/amidotrizoaat as described earlier (61). The use of buffy coats for research purposes was approved by the Institutional Review Board of the Leiden University Medical Center, The Netherlands. Magnetic cell sorting using anti-CD14-coated microbeads (Miltenyi Biotec, Auburn, CA) was used to isolate CD14+ monocytes. CD14+ cells were cultured for six days at 37°C/5% CO 2 in Gibco Roswell Park Memorial Institute (RPMI) 1640 medium (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS), 2 mM L-Alanyl-L-Glutamine (PAA, Linz, Austria), 100 units/ml penicillin (Thermo Fisher Scientific), 100 µg/ml streptomycin (Thermo Fisher Scientific) and either 5 ng/ml granulocytemacrophage colony-stimulating factor (GM-CSF, Thermo Fisher Scientific) to promote M1 differentiation or 50 ng/ml macrophage colony-stimulating factor (M-CSF, R&D Systems, Abingdon, United Kingdom) to promote M2 differentiation. Cytokines were added again at day 3 of differentiation in equal concentrations. The M1 and M2 macrophage phenotypes were subsequently validated by flow cytometry based surface marker expression (M1: CD14 low , CD163 low , CD11b high ; M2: CD14 high , CD163 high , CD11b low ) and by quantifying IL-10 and IL-12p40 production by Enzyme-Linked Immuno Sorbent Assay (ELISA) following stimulation of cells in the presence or absence of 100 ng/ml lipopolysaccharide (LPS) for 24h (In vivoGen, San Diego, United States) as described before (60).

In Vitro Infection and Compound Treatment
Adherent cells were harvested by trypsinization and gentle scraping in FBS and seeded in RPMI 1640 medium supplemented with 10% FBS and 2 mM L-Alanyl-L-Glutamine on Costar 96-well flat bottom culture plates (30,000 cells/well) or on Costar 24-well flat bottom culture plates (300,000 cells/well) (Corning, Amsterdam, The Netherlands) and incubated overnight at 37°C/5% CO 2 . Mycobacterial cultures were diluted to an early log-phase corresponding with an OD600 of 0.25 one day prior to infection in 7H9 broth containing 10% ADC, 0.5% Tween-80 and 2% Glycerol. DsRed-expressing Stm was recovered from frozen stock and cultured in Difco ™ Luria-Bertani (LB) broth (BD Bioscience) containing 100 µg/ml ampicillin (Sigma-Aldrich) overnight at 37°C. Bacterial suspension was diluted 1:33 in LB broth and grown for 3-4h to reach a log-phase with an OD600 between 0.4-0.6. GFPexpressing MRSA was recovered from frozen stock and cultured in Tryptic Soy (TS) broth (BD Bioscience) containing 5 µg/ml tetracycline hydrochlorine overnight at 37°C. Bacterial suspension was diluted 1:33 in TS broth and grown for 2-3h to reach a logphase with an OD600 between 0.4-0.6. MRSA was harvested by centrifugation and resuspended in cold PBS supplemented with 5 mM EDTA. The indicated bacteria were diluted in RPMI 1640 medium (10% FBS and 2 mM L-Alanyl-L-Glutamine) to reach a multiplicity of infection (MOI) of 10. Accuracy of the MOI was validated by plating a serial dilution of the mycobacterial inoculum on Difco ™ Middlebrook 7H10 agar (BD Bioscience) plates containing 10% oleic acid-albumin-dextrose-catalase (OADC, BD Bioscience) and 5% glycerol, by plating Stm on Difco ™ LB agar plates (BD Bioscience) or by plating MRSA on TS agar (BD Bioscience).

Cellular Toxicity Assay
The number and percentage of dead cells based on plasma membrane integrity of the adherent cell population was quantified by analysis of microscopy images. Cells in 96-well flat bottom plates (30,000 cells/well) were stained with 2 µg/ml propidium iodide (Sigma-Aldrich) and 2 µg/ml Hoechst 33342 (H3570, Sigma-Aldrich) in 40 µl/well phenol red-free RPMI (Sigma-Aldrich) supplemented with 10% FBS and 2 mM L-Alanyl-L-Glutamine and incubated for 5 min at room temperature (RT). Cells were imaged using a Leica AF6000 LC fluorescence microscope (Leica Microsystems, Wetzlar, Germany) combined with a 10x dry objective. Total and dead cell numbers were quantified by respectively counting the nuclei and the number of propidium iodide-positive cells using ImageJ software (64). STSP (2.5 µM) was included as a positive control for cell death (Supplemental Figure 4).

Western Immunoblot Analysis
Cells (300,000 cells/well in 24-wells plates) were lysed with 100 µl/ well EBSB buffer (10% v/v glycerol, 3% SDS, 100 mM Tris-HCl, pH 6.8) supplemented with one tablet of cOmplete ™ EDTA-free protease inhibitor cocktail (Sigma-Aldrich) and one tablet of phosphatase inhibitor cocktail (PhosSTOP EASYpack, Sigma-Aldrich) per 10 ml. Cell lysates were boiled for 10 minutes at 95°C and stored at -20°C until use. Total protein concentrations were determined using a Pierce ™ BCA protein assay kit (Thermo Fisher Scientific) according to manufacturer' instructions and protein concentrations were equalized and diluted in Laemmli sample buffer (Biorad) containing b-mercaptoethanol (Sigma-Aldrich). Samples were loaded on a 15-well 4-20% Mini-PROTEAN ® TGX ™ Precast Protein Gel (Bio-Rad Laboratories, Veenendaal, the Netherlands) and Amersham ECL Rainbow Molecular Weight Marker was added as reference (Sigma-Aldrich). Proteins were transferred to methanol-activated Immun-Blot PVDF membranes (Biorad) in Tris-glycine buffer (25 mM Tris, 192 mM glycine and 20% methanol). Membranes were blocked for 1h in polysorbate 20 tris-buffered saline (TTBS) supplemented with 5% w/v non-fat dry milk and incubated with the indicated antibodies in 5% non-fat dry milk/TTBS overnight at 4°C. Membranes were washed thrice for 15 min in TTBS and stained with secondary antibodies in 5% non-fat dry milk/TTBS for 2h at RT. Membranes were washed for 30 min with TTBS prior to revelation using enhanced chemiluminescence (ECL) ™ Prime Western Blotting System reagent (GE Healthcare, Hoevelaken, The Netherlands). Imaging was performed on ChemiDoc XRS+ (Bio-Rad) or on an iBright Imaging System (Invitrogen, Breda, The Netherlands). Protein bands were quantified using ImageJ/Fiji software (64) and normalized to actin.

Cytokine and Chemokine Secretion by Multiplex Beads Assay
Human cytokine/chemokine levels were determined using the MilliPlex Human Cytokine/Chemokine magnetic bead premixed 41-plex kit (Millipore Billerica, MA, USA) as described before (60). Culture supernatants were collected and sterilized by using a 96-well filter plate containing a 0.

Flow Cytometry
Adherent and floating cells (300,000 cells/well in 24-wells plates) were washed in PBS/0.1% BSA (Merck, Darmstadt, Germany) and Fc receptors were blocked with 5% human serum (HS) in PBS for 10 min at RT. Cells were washed in PBS/0.1% BSA and stained with antibodies diluted in PBS/0.1% BSA for 30 min at 4°C. Cells were washed in PBS/0.1% BSA and fixed with 1% paraformaldehyde for 1h at 4°C. Fluorescence minus one (FMO) control samples were included to define background fluorescence for each stain. Fluorescence staining was measured using a FACSLyric ™ flow cytometer with FACSDiva software (BD Bioscience). Geometric mean fluorescence intensity (gMFIs) was recorded for each sample. Data were analyzed using FlowJo v10 software. Baseline NADH levels were measured using the SpectraMax i3x plate reader at OD340 before the addition of 2 µl of three times diluted LDH from rabbit muscle (Roche Applied Science) to each well to initiate the lactate to pyruvate conversion. Following the addition of LDH, plates were incubated on a shaker at RT for 90 mins and the OD340 was measured again using a SpectraMax i3x plate reader.

Immunostaining and Confocal Microscopy
For confocal microscopy, cells were cultured on pre-washed black glass bottom poly-d-lysine coated 96-well plates (no. 1.5, MatTek Corporation, Ashland, MA, USA) at a density of 30,000 cells/well in RPMI 1640 medium supplemented with 10% FBS and 2 mM L-Alanyl-L-Glutamine and incubated overnight at 37°C /5% CO 2 . Cells were infected with DsRed-expressing Mtb H37Rv as described above and treated with compound (10 µM) or an equal volume of DMSO (0.1% v/v) for 4h. Culture medium was replaced with a solution of 75 nM LysoTracker ® Deep Red (Thermo Fisher Scientific) and 1x CYTO-ID ® Green 2.0 (1:500, Enzo Life Sciences, Bruxelles, Belgium) for 30 min at 37°C/5% CO 2 to stain lysosomal and autophagic vesicles, respectively. Cells were washed twice with PBS and fixed with Pierce ™ 1% w/v formaldehyde (Thermo Fisher Scientific) for 1h at RT, washed twice with PBS and then stored at 4°C. Prior to imaging, plates were incubated with 2 µg/ml Hoechst 33342 for 5 min at RT and imaged using a SP8WLL confocal microscope (Leica) using a 63X oil immersion objective. Each treatment condition was performed in triplicate wells and three images were taken from each well. For image analysis, lysotracker and CYTO-ID channel background was subtracted using a rolling ball algorithm with a 10 pixel radius in Fiji/ImageJ (64). Lysotracker area and CYTO-ID area were specified for each image using Cellprofiler 3.1.9 (65) and were normalized to cell count based on Hoechst 33342 staining.

Statistical Analysis
All statistical analyses were carried out using GraphPad Prism 8 software (Graphpad Software, San Diego, CA, USA). Normal distribution of data sets was evaluated using the Shapiro-Wilk normality test. Correlation was evaluated with a Spearman's rank correlation test. Parametric paired data were tested with a paired t-test when comparing two groups and with RM one-way ANOVA or two-way ANOVA followed by Dunnett's multiple comparison test versus control when comparing three or more groups. Nonparametric paired data were tested with Wilcoxon matched-pairs signed rank test when comparing two groups and with Friedman's test followed by Dunn's multiple comparison test versus control when comparing three or more groups. Statistical tests were considered significant when p<0.05 at 95% confidence interval.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Institutional Review Board of the Leiden University Medical Center, The Netherlands. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
CD and TO designed the project and the main conceptual ideas. CD, SS, and KW performed the experiments and analyzed the experimental data. CD and TO wrote the manuscript. TO supervised the project. All authors contributed to the article and approved the submitted version.

FUNDING
This work received financial support from the Netherlands Organization for Health Research and Development (ZonMw-TOP grant 40-00812-98-14038). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. was kindly provided by Prof. dr. William R. Jacobs (Albert Einstein College of Medicine, Yeshiva University, New York City, United States). MDR-Mtb strains were kindly provided by Dick van Soolingen and Kristin Kremer (Rijksinstituut voor Volksgezondheid en Milieu, RIVM).