ABSTRACT
Lactate dehydrogenase A (LDHA) mediates interconversion of pyruvate and lactate. Increased lactate turnover is shared by malignant and immune cells. Hypoxic lung granuloma in Mycobacterium tuberculosis-infected animals present elevated levels of Ldha and lactate. Such alteration in metabolic milieu could influence the outcome of interactions between M. tuberculosis and its infected immune cells. Given the central role of LDHA for tumorigenicity, targeting lactate metabolism is a promising approach for cancer therapy. Here, we sought to determine the importance of LDHA for Tuberculosis (TB) disease progression and its potential as a host-directed therapeutic target. To this end, we administered FX11, a small-molecule NADH-competitive LDHA inhibitor, to M. tuberculosis infected C57BL/6J mice and Nos2−/− mice with hypoxic necrotizing lung TB lesions mimicking human pathology more closely. FX11 did not inhibit M. tuberculosis growth in aerobic/hypoxic liquid culture, but modestly reduced the pulmonary bacterial burden in C57BL/6J mice. Intriguingly, FX11 administration limited M. tuberculosis replication and onset of necrotic lesions in Nos2−/− mice. In this model, Isoniazid (INH) monotherapy has been known to exhibit biphasic killing kinetics owing to the probable selection of an INH-tolerant subpopulation. This adverse effect was corrected by adjunct FX11 treatment and augmented the INH-derived bactericidal effect against M. tuberculosis. Our findings therefore support LDHA as a potential target for host-directed adjunctive TB therapy and encourage further investigations into the underlying mechanism.
IMPORTANCE Tuberculosis (TB) continues to be a global health threat of critical dimension. Standard TB drug treatment is prolonged and cumbersome. Inappropriate treatment or non-compliance results in emergence of drug-resistant Mycobacterium tuberculosis strains (MDR-TB) that render current treatment options ineffective. Targeting the host immune system as adjunct therapy to augment bacterial clearance is attractive as it is also expected to be effective against MDR-TB. Here, we provide evidence that pharmaceutical blockade of host lactate dehydrogenase A (LDHA) by a small-molecule limits M. tuberculosis growth and reduces pathology. Notably, LDHA inhibition potentiates the effect of Isoniazid, a first-line anti-TB drug. Hence, its implications of our findings for short-term TB treatment are profound. In sum, our findings establish murine LDHA as a potential target for host-directed TB therapy.
INTRODUCTION
Tuberculosis (TB) is the leading cause of mortality from an infectious agent globally (1) and its treatment includes six-month long therapy with combinations of drugs. Development of newer drugs with superior efficacy and safety is urgently required to shorten the treatment duration as well as to manage drug-resistant TB effectively. Pathogen-targeted treatment is the preferred choice, however, host-directed approaches are being increasingly recognized for adjunct therapy to reduce pathogen load and ameliorate exacerbated organ damage during TB granuloma progression (2, 3). Radiotracer imaging of M. tuberculosis-infected lungs has revealed heterogeneity – in size, metabolism, and infection – within and between granulomas in a single host (4, 5). In general, the significance of metabolic processes on immune functions is increasingly accepted (6–8). Heterogeneous responses in granuloma, therefore, could partly be attributed to metabolic state(s)/energy phenotype(s) of different immune cells (e.g., macrophages, neutrophils, lymphocytes) that are influenced by their microenvironment and local infection dynamics. Understanding of pathogen-induced immunometabolic dysregulation in granuloma can provide insights into the vital pathways in the infected host and thereby reveal novel therapeutic target candidates.
Untargeted metabolite analysis has identified elevated levels of lactate in necrotic granuloma of M. tuberculosis-infected guinea pigs (9). Generation of lactate from pyruvate, a terminal glycolytic step, is catalyzed by lactate dehydrogenase A (LDHA), whose functions depend on hypoxia-inducible factors (HIFs) (10). Both LDHA and HIF1-α transcripts have been found to be significantly induced in M. tuberculosis-infected mouse lungs (11, 12), and the essential function of HIF1-α in controlling TB progression has already been recognized (10). Although metabolic phenotypes of malignant and immune cells show some critical differences, they present many similarities (13). In most cancer cells, aerobic glycolysis (Warburg effect) or hypoxia adaptation requires LDHA, and its inactivation using the NADH competitive inhibitor, FX11 (3-dihydroxy-6-methyl-7-(phenylmethyl)-4-propylnaphthalene-1-carboxylic acid; PubChem CID: 10498042), has been shown to regress lymphoma and pancreatic cancer (14). In this report, we interrogate whether FX11-mediated LDHA inhibition could result in host-beneficial and pathogen-detrimental outcome in murine TB models and its relevance to host-directed therapy.
FINDINGS
FX11 affects bioenergetics and glycolysis in human Panc (P) 493 B-lymphoid cells (14). Here, we assessed the FX11-induced response in interferon-gamma (IFN-γ) stimulated but uninfected murine bone marrow derived macrophages (BMDMs) (Methods in Text S1). FX11 addition increased the oxygen consumption rate (OCR), but decreased the respiratory capacity, membrane potential, and ATP synthesis in a concentration-dependent manner (Fig. 1A and B; Text S2). Essentially, FX11 (at 14.3 µM) uncoupled the mitochondrial respiratory chain and phosphorylation system. Likewise, FX11-mediated LDHA inhibition increased the extracellular acidification rate (ECAR) (implying increased glycolysis) but depleted the cellular glycolytic reserve (Fig. 1C and D). Such FX11-dependent glycolytic induction could be argued, in part, as a measure to compensate the reduced mitochondrial energy generation. Nevertheless, these observations establish that FX11-mediated LDHA inhibition profoundly affects bioenergetics and glycolysis in BMDMs. Intriguingly, recent studies have demonstrated that energy-flux changes in macrophages depend on viability and virulence of M. tuberculosis (15, 16). Upon virulent M. tuberculosis infection, human monocyte-derived macrophages shift their energy generation to mitochondrial fatty acid oxidation with concomitant decrease in glycolysis (16). We interrogated whether FX11-mediated impairment of respiratory/glycolytic function directly affects the intramacrophage M. tuberculosis survival. Because high concentration of FX11 affected viability of BMDMs, we tested 1.43 μM concentration and the bacterial survival remained identical between untreated and FX11-treated conditions (Fig. 1E).
Although FX11 is an analog of anti-bacterial gossypol (17), we found that FX11 is non-toxic to M. tuberculosis under the tested conditions. The aerobic growth of M. tuberculosis in glycerol or sodium L-lactate was comparable between FX11-treated and untreated growth control (Fig. 1F and G). Similarly, fluorescence measurement of green fluorescent protein expressing M. tuberculosis, as a function of growth, under 1% O2 hypoxia revealed that FX11 did not affect the bacterial viability, albeit a minor decrement in fluorescence was noted (Fig. S1A and B). Moreover, development of a pale brownish color in FX11-supplemented hypoxic culture only was noted suggesting that this small-molecule is differentially metabolized under such condition. Finally, the respiratory functions in M. tuberculosis also remained unperturbed when FX11 was added (Fig. 1H). We conclude that the bioenergetics effects of FX11 are highly host-specific.
Subsequently, the effect of FX11 was evaluated in two murine TB models. In a first experiment, C57BL/6J mice were aerosol-infected with 100 CFU of M. tuberculosis H37Rv. At 4 weeks post-infection, mice received either FX11 (2 mg/kg) or 2% (final concentration) dimethyl sulfoxide (DMSO) as placebo by oral gavage (6 days/week) for further 4 weeks. Post-treatment effect was monitored at 2 and 4 weeks by enumerating CFU from excised lungs and spleens of euthanized animals. FX11 administration resulted in approximately 0.5 log10 reduction in pulmonary M. tuberculosis counts (Fig. 2A; Fig. S2A) with less apparent effect on splenic CFU. Administered dose of FX11 is similar to those in a previous study (14), and further dose increment is restricted due to poor compound solubility. Furthermore, complete inhibition of LDHA could result in adverse events as it is essential for cellular homeostasis.
TB granulomas in C57BL/6J mice rarely progress into necrosis, whereas Nos2−/− mice present hypoxic necrotizing lung lesions that are characteristics hallmarks of human TB (18–20). Therefore, in a second experiment, the effect of FX11 (2 mg/kg), either individually or in combination with isoniazid (INH, 25 mg/kg), was evaluated in Nos2−/− mice (Fig. 2B). Efficacy was determined by assessing histopathology and bacterial viability. FX11 administration was apparently well-tolerated because treated animals showed no increased distress or weight loss (Fig. S2B). As previously observed (20), onset of hypoxic and necrotic lesions became apparent at day 56 (at treatment start). Although the number and size of lesions were comparable, further development of necrotic lesions were ceased in the FX11-treated group (Fig. 2C,D; Fig. S2C,D). Likewise, 2 or 4 weeks of FX11 administration limited further bacterial growth in lungs and spleens. Immunofluorescence staining of paraffin-embedded lung sections revealed that LDHA expression co-localized with hypoxic lesions (Fig. 2C; Fig. S3). Nonetheless, FX11 administration had no apparent impact on LDHA immunofluorescence which is probably due to non-inhibitory effects of FX11 on transcription/translation. Moreover, enzymatic quantification of lactate from the excised lung tissues presented erroneous and irreproducible data (data not shown). Thus, no concrete evidence could be presented to corroborate the in vivo inhibitory effect of FX11 on LDHA.
Necrotic lesions in the Nos2−/− model have been correlated with the evolution of slow/non-growing INH-tolerant subpopulation (20). Accordingly, we interrogated whether FX11-mediated inhibition of progression to necrotic granuloma potentiates INH efficacy possibly by preventing the emergence of the drug-tolerant population. Indeed, the combination of FX11 and INH resulted in superior efficacy, and there was no further cessation of bactericidal activity of INH, when compared with monotherapy (Fig. 2B). While this observation requires further validation in other experimental models, it has immense implications for shortening TB treatment and minimizing the risk of emergence of drug resistance in M. tuberculosis.
Genetic ablation of LDHA in T cells has been found to protect mice from IFN-γ-mediated lethal pathology of autoimmune responses (21). Similarly, lactate accumulation has been indicated to severely impair IFN-γ-dependent tumor immunosurveillance (22). It is a well-established paradigm that IFN-γ has a central role in macrophage activation and tissue-protection in TB (23, 24). Besides, HIF-1α is not only a transcriptional regulator of LDHA, but also coordinates IFN-γ-dependent adaptive immunity to M. tuberculosis (10). It has been reported that IL-17 limits HIF1α expression (and lactate accumulation) and hypoxic necrotic granuloma development in C3HeB/FeJ mice infected with an M. tuberculosis clinical isolate (12). Thus, LDHA inhibition resulting in heightened IL-17 activity and/or reduced IFN-γ-dependent exacerbated inflammation could explain the FX11-limited necrotic granuloma progression in the Nos2−/− mouse model. However, the cause of reduction in M. tuberculosis burden upon FX11 treatment is difficult to explain. FX11-mediated LDHA inhibition perhaps alters the balance of pro- and anti-inflammatory cytokines thereby contributing to M. tuberculosis clearance (25). Observed FX11effects can also be linked to factors other than LDHA inhibition. E.g., reactive catechol moiety of FX11 or its drug-intermediates (under oxygen limiting conditions) could cause off-target effects. FX11 has been shown to induce oxidative stress (14), which could restrict bacterial growth and augment INH efficacy against M. tuberculosis. Finally, FX11 administration could deprive M. tuberculosis from utilizing host-derived lactate for energy generation (26). Therefore, in depth analysis of mechanism underlying LDHA inhibition and pathogen clearance is warranted as it is a promising host-directed therapy approach in adjunct to canonical drug treatment.
TEXT S2: Linear regression modelling analysis to determine the effect of FX11 on bone marrow derived macrophages bioenergetics and glycolytic response
Data acquisition
Oxygen consumption (OCR) and extracellular acidification rates (ECAR) were measured using the Seahorse XF96 extracellular flux analyzer (Agilent, Santa Clara, CA). Two different assays were performed using the XF96: mitochondrial respiration assay, and glycolytic stress assay. Acquired real-time data were into the XF Report Generators using the Wave Desktop 2.6 software for calculation of the parameters from the specific assays.
Results
1. Respiratory profile and respiratory parameters of BMDMs treated with FX11 or DMSO (vehicle control)
1.1. Box plots showing respiratory response (OCR value) stratified by experiment replicate (related to Fig. 1A and B)
1.2. Linear regression models for each of the six parameters (see above box plots in 1.1)
For each parameter (readout), the influence of FX11 concentration on the parameter readout was tested using log-linear regression. To this end, the FX11 concentrations were logarithmized (with the control, DMSO, assumed to have a concentration below 0.0143 mM) and a linear model (lm) was fit on the resulting data with the lm() function in R.
1.3. Linear regression modeling results (related to respiratory parameters presented in Fig. 1A and B)
2. Glycolytic stress profile and glycolytic parameters of BMDMs-treated with FX11 or DMSO (vehicle control)
2.1. Box plots showing glycolytic response (ECAR value) stratified by experiment replicate (related to Fig. 1C and D)
2.2. Linear regression models for each of the four outputs collected (see above box plots in 2.1.)
For each parameter (readout), the influence of FX11 concentration on the parameter readout was tested using log-linear regression. To this end, the FX11 concentrations were logarithmized (with the control, DMSO, assumed to have a concentration below 0.0143 mM) and a linear model (lm) was fit on the resulting data with the lm() function in R.
2.3 Linear regression modeling results (related to glycolytic function parameters presented in Fig. 1C and D)
ACKNOWLEDGEMENTS
Animal protocols were approved by Landesamt für Gesundheit und Soziales, Berlin, Germany. Experiments were conducted in accordance with the European directive 2010/63/EU on Care, Welfare and Treatment of Animals.
We thank Manuela Primke, Ines Neumann, Jens Otto, Uwe Klemm, and Gesa Rausch for their help in mouse breeding and maintenance; Marion Klemm, Manuela Stäber and Dagmar Oberbeck-Mueller for technical assistance. We gratefully acknowledge the partial financial support (to S.H.E.K) from “PreDiCT-TB” and the intramural funding of Max Planck Society to S.H.E.K.