Author response: The antibiotic bedaquiline activates host macrophage innate immune resistance to bacterial infection

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Antibiotics are widely used in the treatment of bacterial infections. Although known for their microbicidal activity, antibiotics may also interfere with the host’s immune system. Here, we analyzed the effects of bedaquiline (BDQ), an inhibitor of the mycobacterial ATP synthase, on human macrophages. Genome-wide gene expression analysis revealed that BDQ reprogramed cells into potent bactericidal phagocytes. We found that 579 and 1,495 genes were respectively differentially expressed in naive- and M. tuberculosis-infected macrophages incubated with the drug, with an over-representation of lysosome-associated genes. BDQ treatment triggered a variety of antimicrobial defense mechanisms, including phagosome-lysosome fusion, and autophagy. These effects were associated with activation of transcription factor EB, involved in the transcription of lysosomal genes, resulting in enhanced intracellular killing of different bacterial species that were naturally insensitive to BDQ. Thus, BDQ could be used as a host-directed therapy against a wide range of bacterial infections. eLife digest The discovery of antibiotic drugs, which treat diseases caused by bacteria, has been a hugely valuable advance in modern medicine. They work by targeting specific cellular processes in bacteria, ultimately stopping them from multiplying or killing them outright. Antibiotics sometimes also affect their human hosts and can cause side-effects, such as gut problems or skin reactions. Recent evidence suggests that antibiotics also have an impact on the human immune system. This may happen either indirectly, by affecting ‘friendly’ bacteria normally present in the body, or through direct effects on immune cells. In turn, this could change the effectiveness of drug treatments. For example, if an antibiotic weakens immune cells, the body could have difficulty fighting off the existing infection – or become more vulnerable to new ones. However, even though new drugs are being introduced to combat the worldwide rise of antibiotic-resistant bacteria, their effects on immunity are still not well understood. For example, bedaquiline is an antibiotic recently developed to treat tuberculosis infections that are resistant to several drugs. Giraud-Gatineau et al. wanted to determine if bedaquiline altered the human immune response to bacterial infection independently from its direct anti-microbial effects. Macrophages engulf foreign particles like bacteria and break them down using enzymes stored within small internal compartments, or ‘lysosomes’. Initial experiments using human macrophages, grown both with and without bedaquiline, showed that the drug did not harm the cells and that they grew normally. A combination of microscope imaging and genetic analysis revealed that exposure to bedaquiline not only increased the number of lysosomes within macrophage cells, but also the activity of genes and proteins that increase lysosomes’ ability to break down foreign particles. These results suggested that bedaquiline treatment might make macrophages better at fighting infection, even if the drug itself had no direct effect on bacterial cells. Further studies, where macrophages were first treated with bedaquiline and then exposed to different types of bacteria known to be resistant to the drug, confirmed this hypothesis: in every case, the treated macrophages became efficient bacterial killers. In contrast, older anti-tuberculosis drugs did not have any such potentiating effect on the macrophages. This work sheds new light on our how antibiotic drugs can interact with the cells of the human immune system, and can sometimes even boost our innate defences. Such immune-boosting effects could one day be exploited to make more effective treatments against bacterial infections. Introduction Antibiotics are commonly used in the treatment of bacterial infections, and, in effectively combating such diseases, have substantially increased human life expectancy. As with most drugs, antibiotic treatment can also alter host metabolism, leading to adverse side-effects, including nausea, hepatotoxicity, skin reactions, and gastrointestinal and neurological disorders. Such side-effects can become critical when antibiotic treatment is long and involves several drugs, as in the treatment of tuberculosis (TB), where 2–28% of patients develop mild liver injury during treatment with first-line drugs (Agal et al., 2005). Antibiotics can interfere with the immune system, indirectly through the disturbance of the body’s microbiota (Ubeda and Pamer, 2012), or directly by modulating the functions of immune cells. Such interactions may impact treatment efficacy or the susceptibility of the host to concomitant infection. For example, after treatment completion, TB patients are more vulnerable to reactivation and reinfection of the disease, suggesting therapy-related immune impairment (Cox et al., 2008). Drug-sensitive TB can be cured by combining up to four antibiotics in a 6-month treatment; specifically, isoniazid (INH), rifampicin (RIF), ethambutol (EMB) and pyrazinamide (PZA) for 2 months, and INH and RIF for additional 4 months. INH induces apoptosis of activated CD4+ T cells in Mycobacterium tuberculosis (MTB)-infected mice (Tousif et al., 2014) and leads to a decrease in Th1 cytokine production in household contacts with latent TB under preventive INH therapy (Biraro et al., 2015). RIF has immunomodulatory properties and acts as a mild immunosuppressive agent in psoriasis (Tsankov and Grozdev, 2011). RIF reduces inflammation by inhibiting IκBα degradation, mitogen-activated protein kinase (MAPK) phosphorylation (Bi et al., 2011), and Toll-like receptor 4 signaling (Wang et al., 2013). PZA treatment of MTB-infected human monocytes and mice significantly reduces the release of pro-inflammatory cytokines and chemokines (Manca et al., 2013). Recently, Puyskens et al. showed that several anti-TB drugs bind to the aryl hydrocarbon receptor and may impact host defense (Puyskens et al., 2020). It is therefore necessary to understand how antibiotic treatment modulates macrophage functions, and more generally, how it impacts the host immune response. The worldwide rise in antibiotic resistance is a major threat to global health care. A growing number of bacterial infections, such as pneumonia, salmonellosis, and TB, are becoming harder to treat as the antibiotics used to treat them become less effective. While new antibiotics are being developed and brought to the clinic, their effects on the human immune system are not being studied in-depth. Here, we have investigated the impact of a recently approved anti-TB drug, bedaquiline (BDQ), on the transcriptional responses of human macrophages infected with MTB. Macrophages are the primary cell target of MTB, which has evolved several strategies to survive and multiply inside the macrophage phagosome, including prevention of phagosome acidification (Sturgill-Koszycki et al., 1994), inhibition of phagolysosomal fusion (Armstrong and Hart, 1975) and phagosomal rupture (Simeone et al., 2012; van der Wel et al., 2007). They play a central role in the host response to TB pathogenesis, by orchestrating the formation of granulomas, presenting mycobacterial antigens to T cells, and killing the bacillus upon IFN-γ activation (Cambier et al., 2014). BDQ is a diarylquinoline that specifically inhibits a subunit of the bacterial adenosine triphosphate (ATP) synthase, decreasing intracellular ATP levels (Andries et al., 2005; Koul et al., 2007). It has 20,000 times less affinity for human ATP synthase (Haagsma et al., 2009). The most common side effects of BDQ are nausea, joint and chest pain, headache, and arrhythmias (Diacon et al., 2012; TMC207-C208 Study Group et al., 2014). However, possible interactions between BDQ and the host immune response have not been studied in detail. Understanding the impact of BDQ on the host immune response may help to develop strategies aiming at improving drug efficacy and limiting side-effects, including cytotoxicity, alteration of cell metabolism, and immunomodulation. Results BDQ modulates the response of naïve and MTB-infected macrophages We treated human monocyte-derived macrophages from four healthy donors with BDQ at 5 µg/mL, which corresponds to the concentration detected in the plasma of TB patients treated with BDQ (Andries et al., 2005). This concentration did not affect cell viability over an incubation period of 7 days (Figure 1—figure supplement 1). After 18 hr of treatment, we characterized the genome-wide gene expression profiles of BDQ-treated macrophages by RNAseq, with DMSO-treated cells serving as a control. The expression of 579 genes was affected by BDQ (FDR < 0.05, Figure 1—source datas 1 and 2), with 186 being upregulated and 393 being downregulated. We classified all 579 genes by performing gene-set enrichment analysis using ClueGO cluster analysis (Bindea et al., 2009). The gene set upregulated by BDQ was significantly enriched for genes associated with lysosome, phagocytic vesicle membrane, vacuolar lumen, hydrolase activity and lipid homeostasis (Figure 1A). Figure 1 with 2 supplements see all Download asset Open asset BDQ modulates the response of human macrophages. Cells from four individual donors were treated with BDQ (5 μg/mL) for 18 hr. Differentially expressed genes were identified by mRNAseq. (A) Gene ontology enrichment analysis of genes whose expression is upregulated by BDQ treatment, using the Cytoscape app ClueGO (FDR < 0.05; LogFC >0.5). (B) Cells were infected with BDQ-resistant MTB for 24 hr and then treated with BDQ (5 μg/mL) for an additional 18 hr. Gene ontology enrichment analysis of genes whose expression is up-regulated by BDQ treatment in BDQr-MTB-infected cells, using the Cytoscape app ClueGO (FDR < 0.05; LogFC >0.5). (C) Venn diagram showing the number of genes regulated by BDQ treatment in naive and BDQr-MTB-infected macrophages, relative to untreated controls. (D) Heatmap showing differential expression of genes differentially expressed by BDQ in naive and BDQr-MTB-infected cells. Each column corresponds to one donor. Data were normalized to determine the log ratio with respect to the median expression of each gene. Figure 1—source data 1 Genes whose expression is upregulated in naive macrophages upon BDQ treatment. FDR < 0.05. https://cdn.elifesciences.org/articles/55692/elife-55692-fig1-data1-v1.xlsx Download elife-55692-fig1-data1-v1.xlsx Figure 1—source data 2 Genes whose expression is downregulated in naive macrophages upon BDQ treatment. FDR < 0.05. https://cdn.elifesciences.org/articles/55692/elife-55692-fig1-data2-v1.xlsx Download elife-55692-fig1-data2-v1.xlsx Figure 1—source data 3 Genes whose expression is upregulated in BDQr-MTB-infected macrophages upon BDQ treatment. FDR < 0.05. https://cdn.elifesciences.org/articles/55692/elife-55692-fig1-data3-v1.xlsx Download elife-55692-fig1-data3-v1.xlsx Figure 1—source data 4 Genes whose expression is downregulated in BDQr-MTB-infected macrophages upon BDQ treatment. FDR < 0.05. https://cdn.elifesciences.org/articles/55692/elife-55692-fig1-data4-v1.xlsx Download elife-55692-fig1-data4-v1.xlsx Figure 1—source data 5 Genes differentially expressed in BDQr-MTB infected macrophages by BDQ. FDR < 0.05. https://cdn.elifesciences.org/articles/55692/elife-55692-fig1-data5-v1.xlsx Download elife-55692-fig1-data5-v1.xlsx Figure 1—source data 6 Genes differentially expressed in naive macrophages by BDQ. FDR < 0.05. https://cdn.elifesciences.org/articles/55692/elife-55692-fig1-data6-v1.xlsx Download elife-55692-fig1-data6-v1.xlsx Figure 1—source data 7 Differentially expressed genes both in naive and in BDQr-MTB infected macrophages upon BDQ treatment. FDR < 0.05. https://cdn.elifesciences.org/articles/55692/elife-55692-fig1-data7-v1.xlsx Download elife-55692-fig1-data7-v1.xlsx Table 1 Gene Ontology (GO) functional annotation of genes differentially expressed by BDQ only in naïve- and BDQr-MTB-infected macrophages. Specific NAIVE BDQ genesGO categoryavg. LogFCp-valueCell division-0.518.34E-05Sphingolipid metabolic process0.331.42E-04Angiogenesis0.675.16E-04Spindle-0.635.46E-04Lysosomal lumen0.351.21E-04Glycosphingolipid metabolic process0.351.21E-03Response to oxidative stress0.491.26E-03Mitotic cell cycle-0.581.29E-03Specific INFECTED BDQ genesGO categoryavg. LogFCp-valueEndoplasmatic reticulum-Golgi intermediate compartment-0.382.90E-07Membrane raft-0.344.93E-05Cellular protein metabolic process-0.353.32E-04Lipid binding-0.364.88E-04Ribonucleoprotein complex binding-0.375.17E-04Protein dephosphorylation-0.345.43E-04Lysosomal membrane-0.346.31E-04Ubiquitin-dependent protein catabolic process0.746.48E-04 We next evaluated if BDQ could modulate gene expression in MTB-infected cells. In order to exclude potential differences due to the MTB bacillary load between treated and untreated cells, we generated a virulent BDQ-resistant strain of M. tuberculosis (BDQr-MTB). The selected clone, which carried a Ala63→Pro mutation in subunit c of the ATP synthase (Andries et al., 2005; Figure 1—figure supplement 2A), had a similar generation time to wild-type bacteria when cultured in 7H9 liquid medium, although we observed a slower growth after 7 days of treatment. (Figure 1—figure supplement 2B). We also noted no difference in virulence or in intracellular growth of both wild-type- and BDQ-resistant MTB (Figure 1—figure supplement 2C-E). As expected, the MIC99 (defined as the concentration required to prevent 99% growth) for susceptible MTB was 0.07 µg/mL, a value similar to previously published study (Andries et al., 2005), while the MIC99 of the BDQr-MTB was 36 µg/mL. We infected macrophages with BDQr-MTB. After 24 hr of infection, cells were incubated for an additional 18 hr with BDQ (5 µg/mL). The bacillary load of resistant MTB inside macrophages was the same in untreated cells as in cells after 18 hr of BDQ treatment. In contrast, in the same experiment using BDQ-susceptible MTB, there was a 70% decrease in the bacillary load (Figure 1—figure supplement 2C). Following treatment, we characterized the genome-wide gene expression profiles of MTB-infected macrophages, as described above. The expression of 1,495 genes was affected by BDQ (FDR < 0.05, Figure 1A, Figure 1—source datas 3 and 4), with 499 being upregulated and 996 being downregulated. More genes were thus affected by BDQ treatment in MTB-infected cells than in naive macrophages. This probably reflects the fact that MTB infection induces an extensive remodeling of the transcriptome (Barreiro et al., 2012; Tailleux et al., 2008). The genes differentially expressed by BDQ only in MTB-infected macrophages are enriched in genes related to assembly of the endoplasmic reticulum-Golgi intermediate compartment, membrane raft and cellular protein metabolic process (Table 1, Figure 1—source datas 5 and 6). This probably reflects the cell adaptation to infection. Functional annotation of the gene set upregulated by BDQ also revealed that similar pathways were affected by BDQ in naive and BDQr-MTB-infected macrophages, with an enrichment for genes associated with glucose/phospholipid metabolism and lysosome (Figure 1B, Figure 1—source datas 3 and 4). 452 genes were differentially expressed in both naive and MTB-infected cells upon BDQ treatment with an over-representation of lysosome-associated genes (Figure 1C–D, Figure 1—source data 7). BDQ affects host metabolism As metabolic pathways were over-represented in our RNAseq analysis, we investigated if glycolysis is affected by BDQ treatment using the Seahorse Extracellular Flux analyzer. This assay measures the rate of proton accumulation in the extracellular medium during glycolysis (glycoPER) and can discriminate between basal glycolysis, induced glycolytic capacity (by addition of rotenone/antimycin A (Rot/AA), an inhibitor of the mitochondrial electron transport chain), and non-glycolytic acidification (by addition of the glycolytic inhibitor 2-deoxy-D-glycose (2-DG)). After incubation with BDQ, we observed a 30% decrease in basal glycolysis and glycolytic capacity compared to untreated cells (Figure 2A–B, Figure 2—figure supplement 1A–B). Figure 2 with 1 supplement see all Download asset Open asset Modulation of host metabolism by BDQ. (A–B) The Glycolytic Rate Assay was performed in heat killed-MTB stimulated macrophages treated with BDQ, in the presence of rotenone/antimycin A (Rot/AA) and 2-deoxy-D-glycose (2-DG), inhibitors of the mitochondrial electron transport chain and glycolysis, respectively (one-way ANOVA test). One representative experiment (of two) is shown. (C) Lipid profile of BDQr-MTB infected cells treated with BDQ, by MALDI-TOF (unpaired two tailed Student’s t test). PI: Phosphotidylinositol; CL: Cardiolipids; PE: Phosphatidylethanolamine; PG: Phosphatidylglycerol. Numbers correspond to mass-to-charge ratio (m/z). Cells derived from three donors were analyzed. Error bars represent the mean ± SD and significant differences between treatments are indicated by an asterisk, in which *p<0.05, p<0.01, *p<0.001. We assessed phospholipid metabolism, a pathway also identified in our ClueGO cluster analysis (Figure 1B). Like glycolysis, lipid metabolism affects macrophage phenotype and function (Remmerie and Scott, 2018). We analyzed the lipid profile of BDQ-treated cells using MALDI-TOF mass spectrometry. We observed an increase of phosphatidylinositols upon incubation with BDQ (Figure 2C, Figure 2—figure supplement 1C). No significant changes were observed in the levels of phosphatidylethanolamines, phosphatidylglycerols, or cardiolipins. Taken together, these data show that BDQ induced a significant metabolic reprogramming of both MTB-infected and resting macrophages. BDQ increases macrophage lysosomal activity Macrophages are involved in innate immunity and tissue homeostasis through their detection and elimination of microbes, debris, and dead cells, which occurs in lysosomes (Wynn et al., 2013). Lysosomes are acidic and hydrolytic organelles responsible for the digestion of macromolecules. Recent work has shown that they are also signaling platforms, which respond to nutrient and cellular stress (Lawrence and Zoncu, 2019). Functional annotations based on the GO database of the differentially expressed genes suggested a substantial impact of BDQ treatment on lysosome function (Figure 1A–B). We identified 38 and 54 differentially expressed genes by BDQ, respectively in naïve- and BDQr-MTB infected cells (FDR < 0.05, Figure 3A). These genes are involved in lysosome biogenesis, transport and degradation of small molecules, and lysosomal acidification. They included genes coding for components of vacuolar ATPase (V-ATPase), hydrolases, and SLC11A1 (NRAMP1), a divalent transition metal transporter involved in host resistance to pathogens, including MTB (Meilang et al., 2012). Figure 3 with 2 supplements see all Download asset Open asset BDQ activates the lysosomal pathway in human MTB-infected macrophages. (A) Heatmap showing differential expression of genes included in the Lysosome KEGG category (p-value<0.05). Each column corresponds to one donor. Data were normalized to determine the log ratio with respect to the median expression of each gene. (B) Macrophages were infected with BDQr-MTB expressing the GFP protein and incubated with BDQ (5 μg/mL) for 3 hr, 18 hr and 48 hr. Acid organelles were then labeled with 100 nM LysoTracker DND-99 for 1 hr. The fluorescence intensity was quantified by flow cytometry. (C–E) Cells were infected with GFP expressing BDQr-MTB (green) and treated with BDQ (5 μg/mL). After 18 hr and 48 hr of treatment, cells were labelled with LysoTracker (red) and fluorescence was analyzed by confocal microscopy. DAPI (blue) was used to visualize nuclei (scale bar: 10 μm). The quantification of LysoTracker staining and the percentage of LysoTracker-positive MTB phagosomes were performed using Icy software. (F) Macrophages were activated with heat-killed MTB and treated with BDQ for 18 hr and 48 hr. Cells were then incubated with DQ-Green BSA. Fluorescence was quantified by flow cytometry. Significant differences between BDQ treatment and control (DMSO) are indicated by an asterisk. One representative experiment (of at least three) is shown. Error bars represent the mean ± SD. *p<0.05, p<0.01, *p<0.001. To validate our transcriptomic we incubated BDQr-MTB-infected cells with LysoTracker a that acidic and analyzed them using flow cytometry. No differences were observed between control and treatment after 3 hr of BDQ treatment (Figure However, at 18 hr and 48 hr fluorescence intensity was substantially increased in macrophages incubated with BDQ compared to DMSO-treated cells and times These results were by confocal which revealed the of acidic upon treatment (Figure up to times more in BDQ-treated macrophages than untreated cells at 48 hr Figure We also observed a number of MTB phagosomes with LysoTracker-positive (Figure As the expression of genes coding for was upregulated upon BDQ treatment (Figure we the effect of the drug on BDQ-treated macrophages were incubated with DQ-Green a that by lysosomal 18 hr and 48 hr we observed a increase in fluorescence intensity upon treatment with BDQ to times more than untreated cells, Figure results were when we incubated naive macrophages with BDQ (Figure supplement 1). these data that BDQ induces of We performed additional experiments to that the effects of BDQ on lysosome were of infection with cells were untreated or stimulated with heat-killed bacteria MTB or and treated with BDQ. After 18 hr, was and we performed on a of lysosomal genes. We also analyzed the intensity of the LysoTracker staining using flow (Figure supplement results show that the effects on lysosome with BDQ treatment and were not after infection with MTB. BDQ PZA activity The capacity of BDQ to acidic may the efficacy of drugs, whose activity is In have suggested a between BDQ and PZA et al., and it is commonly that a is required for PZA activity against MTB and We thus infected macrophages with BDQr-MTB and treated them with BDQ and After 7 days of treatment, cells were and bacteria PZA showed bactericidal activity, with PZA resulting in a decrease in bacterial compared to untreated cells (Figure We confirmed that the combination of PZA with BDQ was bactericidal on MTB, leading to a decrease in using This decrease was not a of an effect between the two drugs, as BDQ at 1 had no We also found no between BDQ and PZA on the BDQ-resistant in 7H9 liquid medium (Figure Thus, the of PZA activity by BDQ is most due to the effect of BDQ on the host and in on the increase of lysosomal acidification. Figure 4 with 1 supplement see all Download asset Open asset BDQ PZA (A) Macrophages were infected with BDQr-MTB and treated with BDQ μg/mL) and After 7 of days treatment, cells were and bacteria were by in (B) of bacterial growth of BDQr-MTB in the presence of BDQ μg/mL) and different of were cultured in 7H9 medium with enrichment the drugs. One representative experiment (of at least three) is shown. Error bars represent the mean ± SD. *p<0.05, p<0.01, *p<0.001. We next BDQ with the first-line anti-TB drugs in liquid or in BDQr-MTB-infected macrophages. We found that BDQ did not the activity of ethambutol isoniazid and rifampicin in either While we exclude the that BDQ may have effects with anti-TB drugs, as has been described et al., we no evidence of with any of the first-line (Figure supplement 1). anti-TB drugs did not the lysosomal pathway in human macrophages after treatment with INH or PZA can directly host cell et al., 2012). We thus antibiotics might have similar effects to BDQ. We the genome-wide gene expression profiles of naïve macrophages and macrophages stimulated with and treated with PZA or RIF for 18 hr. We drug based on the detected in the plasma of treated TB Following treatment, only RIF and PZA significantly modulate gene expression in macrophages. and genes were differentially expressed in cells stimulated with heat-killed bacteria and exposed to RIF and respectively (Figure Figure supplement 1, Figure datas We classified these genes by performing gene-set enrichment analysis and confirmed that the lysosomal pathway was not induced upon RIF or PZA treatment (Table The expression of only two genes to this pathway was upregulated by and only one by compared to whose expression was by BDQ (Figure with these of these antibiotics were to increase LysoTracker staining (Figure Figure 5 with 1 supplement see all Download asset Open asset anti-TB drugs did not the lysosomal pathway in human macrophages. (A) of genes upon treatment with commonly used anti-TB drugs relative to untreated control. naïve- and macrophages were treated with INH PZA or RIF µg/mL). After 18 hr, differentially expressed genes were identified by mRNAseq. Venn diagram showing the number of genes regulated by PZA and RIF in naive and macrophages, in with the number of lysosomal genes differentially expressed by BDQ (FDR < 0.05, (D) Macrophages were incubated for 48 hr with BDQ, PZA or and then with Fluorescence intensity was analyzed by flow cytometry. One representative experiment (of at least three) is shown. Error bars represent the mean ± SD. Figure data 1 Differentially expressed genes in stimulated macrophages upon treatment. FDR < 0.05. Download Figure data 2 Differentially expressed genes in stimulated macrophages upon RIF treatment. FDR < 0.05. Download Figure data 3 Differentially expressed genes in naive macrophages upon RIF treatment. FDR < 0.05. Download Figure data 4 Differentially expressed genes in stimulated macrophages upon PZA treatment. FDR < 0.05. Download Figure data 5 genes naive macrophages upon PZA treatment. FDR < 0.05. Download Table 2 Gene Ontology (GO) functional annotation of differentially expressed genes in naïve- or cells treated with PZA or categoryavg. to side of endoplasmic signaling signaling receptor small protein of immune of categoryavg. cell of cell involved in transition of cell