Mycobacterium tuberculosis Modulates the Metabolism of Alternatively Activated Macrophages to Promote Foam Cell Formation and Intracellular Survival

The ability of Mycobacterium tuberculosis (Mtb) to persist inside host cells relies on metabolic adaptation, like the accumulation of lipid bodies (LBs) in the so-called foamy macrophages (FM). Indeed, FM are favorable to Mtb. The activation state of macrophages is tightly associated to different metabolic pathways, such as lipid metabolism, but whether differentiation towards FM differs between the macrophage activation profiles remains unclear. Here, we aimed to elucidate if distinct macrophage activation states exposed to a tuberculosis-associated microenvironment can accumulate LBs, and its impact on the control of infection. We showed that signal transducer and activator of transcription 6 (STAT6) activation in interleukin (IL)-4-activated human macrophages (M(IL-4)) prevents FM formation induced by pleural effusion from patients with tuberculosis. In these cells, LBs are disrupted by lipolysis, and the released fatty acids enter the β-oxidation (FAO) pathway fueling the generation of ATP in mitochondria. We demonstrated that inhibition of the lipolytic activity or of the FAO drives M(IL-4) macrophages into FM. Also, exhibiting a predominant FAO metabolism, mouse alveolar macrophages are less prone to become FM compared to bone marrow derived-macrophages. Upon Mtb infection, M(IL-4) macrophages are metabolically re-programmed towards the aerobic glycolytic pathway and evolve towards a foamy phenotype, which could be prevented by FAO activation or inhibition of the hypoxia-inducible factor 1-alpha (HIF-1α)-induced glycolytic pathway. In conclusion, our results demonstrate a role for STAT6-driven FAO in preventing FM differentiation, and reveal an extraordinary capacity by Mtb to rewire metabolic pathways in human macrophages and induce the favorable FM. IMPORTANCE Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis (Mtb). While its treatment was already standardized, TB remains one of the top 10 death causes worldwide. A major problem is the efficient adaptation of Mtb to the macrophage intracellular milieu, which includes deregulation of the lipid metabolism leading to the formation of foamy macrophages (FM) which are favorable to Mtb. A critical aspect of our work is the use of tuberculous pleural effusions (TB-PE) — human-derived biological fluid capable of mimicking the complex microenvironment of the lung cavity upon Mtb infection — to study the FM metabolic modulation. We revealed how the STAT6 transcription factor prevents FM formation induced by PE-TB, and how Mtb counteracts it by activating another transcription factor, HIF-1α, to re-establish FM. This study provides key insights in host lipid metabolism, macrophage biology and pathogen subversion strategies, to be exploited for prevention and therapeutic purposes in infectious diseases.

reveal an extraordinary capacity by Mtb to rewire metabolic pathways in human 1 macrophages and induce the favorable FM. 2 3 IMPORTANCE 4 Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis 5 (Mtb). While its treatment was already standardized, TB remains one of the top 10 6 death causes worldwide. A major problem is the efficient adaptation of Mtb to the 7 macrophage intracellular milieu, which includes deregulation of the lipid 8 metabolism leading to the formation of foamy macrophages (FM) which are 9 favorable to Mtb. A critical aspect of our work is the use of tuberculous pleural 10 effusions (TB-PE) -human-derived biological fluid capable of mimicking the 11 complex microenvironment of the lung cavity upon Mtb infection -to study the FM 12 metabolic modulation. We revealed how the STAT6 transcription factor prevents 13 FM formation induced by PE-TB, and how Mtb counteracts it by activating another 14 transcription factor, HIF-1 to re-establish FM. This study provides key insights in 15 host lipid metabolism, macrophage biology and pathogen subversion strategies, to 16 be exploited for prevention and therapeutic purposes in infectious diseases. 17

INTRODUCTION 1
Tuberculosis (TB) is a highly contagious disease caused by Mycobacterium 2 tuberculosis (Mtb). Even though the treatment of the disease has been 3 standardized for a while, TB still remains one of the top 10 causes of death 4 worldwide (1). Chronic host-pathogen interaction in TB leads to extensive 5 metabolic remodeling in both the host and the pathogen (2). The success of Mtb as 6 a pathogen is in a large part due to its efficient adaptation to the intracellular milieu 7 of human macrophages, leading to metabolic changes in infected host cells. One 8 of these changes is the dysregulation of the lipid metabolism, which induces the 9 formation of foamy macrophages (FM). FM are cells filled of lipid bodies (LBs) that 10 are abundant in granulomatous structures in both experimentally infected animals 11 and patients (3, 4) and fail to control the infection (5). Recently, we used the pleural 12 effusions (PE) from TB patients (TB-PE) as a tool to recapitulate human lung TB-13 associated microenvironment and demonstrated that uninfected-macrophages 14 exposed to TB-PE form FM displaying an immunosuppressive profile though the 15 activation of the interleukin (IL)-10/STAT3 axis (6). 16 It is widely accepted that macrophages undergo different activation programs by 17 which they carry out unique physiological and defensive functions. Essentially, 18 macrophages can modify their metabolic functions from a healing/repairing/growth 19 promoting setting (M2 macrophages) towards a killing/inhibitory/microbicidal profile 20 (M1 macrophages) (7,8), representing opposing ends of the full activation 21 spectrum. The M1 macrophages, generally induced by interferon (IFN)- and/or 22 lipopolysaccharide (LPS) stimulation, are endowed with microbicidal properties; the 23 M2 macrophages, usually differentiated upon IL-4 or IL-13 stimulation, are reported 1 to be immunomodulatory and poorly microbicidal, resulting in impaired anti-2 mycobacterial properties (9)(10)(11). In addition, while M1 macrophages rely on 3 glycolysis dependent on the hypoxia-inducible factor 1-alpha (HIF-1α) activation 4 (12), M2 macrophages require the induction of fatty acid oxidation (FAO), at least 5 in the murine model, (13-16) through signal transducer and activator of 6 transcription 6 (STAT6) activation (17). FAO is the mitochondrial process of 7 breaking down a fatty acid into acetyl-CoA units, and requires the carnitine 8 palmitoyltransferase (CPT) system, which consists of one transporter and two 9 mitochondrial membrane enzymes: CPT1 and CPT2 (18). This facilitates the 10 transport of long-chain fatty acids into the mitochondrial matrix, where they can be 11 metabolized by the oxidative phosphorylation (OXPHOS) pathway and produce 12 ATP. Moreover, in the context of Mtb infection, macrophages have been recently 13 shown to be pre-programmed towards different metabolic pathways, according to 14 their ontogeny. Indeed, recruited lung interstitial macrophages are glycolytically 15 active and capable of controlling Mtb infection, while resident alveolar 16 macrophages are committed to FAO and OXPHOS bearing higher bacillary loads 17 (19). 18 Considering that mitochondria are the main site of lipid degradation, and that 19 mitochondrial metabolic functions are known to differ between macrophages 20 profiles, we investigated if the different activation programs of human macrophages 21 differ in their ability to form FM, and whether this impacts the control of Mtb growth. 22 To this end, we employed our already characterized model of TB-induced FM 23 differentiation, using TB-PE (6), and explored the molecular and metabolic 24 PE-treated M(IL-4) cells were even more efficient at efferocytosis than untreated-11 M(IL-4) cells. As IL-4 levels in TB-PE were undetectable by ELISA (less than 6 12 pg/ml, data not shown), the observed effect could not be due to IL-4 within TB-PE. 13 In line with this, M(IL-10) or M(IFN-γ) macrophages did not show detectable 14 expression of the phosphorylated form of STAT6 in the presence of TB-PE (Fig.  15 S5D). Hence, important features associated with the alternative activation driven 16 by IL-4 were not impaired by TB-PE treatment. Moreover, mitochondrial respiration 17 is associated with M(IL-4) macrophages (22). In this regard, we found a higher 18 mitochondrial respiration in TB-PE-treated M(IL-4) macrophages in comparison to 19 untreated cells (Fig. 5A-B). We also measured the release of lactate as evidence 20 of the activation of the glycolytic pathway, and found that lactate release was lower 21 in M(IL-4) cells regardless its treatment with TB-PE compared to M(IFN-γ) cells, 22 which are known to be glycolytic cells ( Fig. 5C and (27)). The glycolytic pathway is 23 highly regulated by HIF-1α activity (12,28). In fact, the stabilization of HIF-1α 24 expression resulted in both the upregulation of glycolysis and the suppression of 1 FAO (29). For this reason, we next treated M0 and M(IL-4) with 2 dimethyloxaloylglycine (DMOG), which leads to HIF-1α stabilization (30), and 3 evaluated its impact on FM formation upon TB-PE treatment. As shown in figure  4 5D and S6A-B, an increase in LBs accumulation was observed after HIF-1α 5 stabilization. Importantly, when we looked at murine alveolar macrophages (AM), 6 which are known to be committed to an FAO and OXPHOS pathway (19), we 7 observed that they did not accumulate LBs in response to mycobacterial lipids in 8 comparison to bone marrow derived macrophages (BMDM) (Fig. 5E). By contrast, 9 when exposed to either a HIF-1α potentiator (DMOG) or a FAO inhibitor 10 (Etomoxir), AM did become foamy ( Fig. 5F and S6C). Of note, murine IL-4-11 activated BMDM were less prone to become foamy upon mycobacterial lipids 12 exposure compared to M0 macrophages ( Fig. 5G and S6D). These results 13 demonstrate that an oxidative metabolic state in alternatively activated 14 macrophages is associated with less LBs accumulation and FM formation. infection. We confirmed that the infection with Mtb promoted LB accumulation at 24 comparable levels in the different activation programs, including the M(IL-4) 1 macrophages (Fig. 6A). Since we observed that STAT6 activation mediates the 2 inhibition of FM formation in M(IL-4) macrophages, we next assessed whether Mtb 3 infection resulted in a reduction in STAT6 phosphorylation in these cells. However, 4 pSTAT6 was not reduced after Mtb infection of M(IL-4) cells (Fig. 6B), suggesting 5 that Mtb targets another molecule downstream to STAT6 making M(IL-4) cells 6 capable of accumulating LBs. Given the importance of the metabolic state in our 7 model, we inferred that the M2-like metabolism was probably rewired due to 8 infection. In the case of the glycolytic activity, we found that Mtb infection slightly 9 increased glucose uptake and lactate release in both M0 and M(IL-4) cells ( Fig.  10 6C-D). In addition, the expression levels of HIF-1α was higher in the Mtb-infected 11 M(IL-4) cells compared to uninfected cells (Fig. 6E). As a proof-of-concept that might be very low. Additionally, while CD206 is also known to be induced by IL-4 18 and IL-13 (36), we did not find such an association between this marker and the 19 numbers of FM. This can be due to the fact that its expression is also induced upon 20 other activation programs such as M(IL-10), which are demonstrated to express 21 high levels of CD206 (21, 37) and prone to synthetize and accumulate LBs (6). In 22 fact, we have previously demonstrated that FM were formed after TB-PE treatment 23 by increasing the biogenesis of LBs through ACAT up-regulation induced by the IL-24 10/STAT3 axis, generating cells accumulating LBs and displaying 1 immunosuppressive properties (6). In this study, we observed that the STAT3 2 pathway is also induced in TB-PE-treated M(IL-4) macrophages (data not shown) 3 resulting in ACAT induction, but as the newly formed LBs are rapidly disrupted in 4 these macrophages, the foamy appearance was reduced drastically. In this regard, 5 we showed that, while LBs are formed, they are quickly disrupted by an enhanced 6 lipolytic activity induced in M(IL-4) macrophages, and those fatty acids produced 7 through lipolysis are then incorporated into the mitochondria and used for FAO. 8 Moreover, we demonstrated that cell population noted to rely on oxidative 9 metabolism, such as AMs and M(IL-4) macrophages, are less prone to become 10 FM. Hence, our findings are strengthened by previous results on the inhibition of 11 the LBs accumulation in macrophages treated with resveratrol (38), since this 12 phytoalexin is known to promote mitochondrial biogenesis and to increase the 13 cellular respiratory capacity (39). Moreover, in agreement with our previous report and others, the differentiation of 3 macrophages into FM (prior to infection) renders the cells more susceptible to the 4 intracellular replication of Mtb, (6,44,45); in this study, we extended this notion to 5 the M(IL-4) macrophage profile exposed to lipases inhibitors. We have previously 6 shown that FM induced by TB-PE had immunosuppressive properties such as: i) 7 high production of IL-10, ii) low production of TNF-α, iii) poor induction of IFN-γ 8 producing T clones in response to mycobacterial antigens, and iv) more 9 permissiveness to intracellular mycobacterial growth (6). Along with other data that 10 associated FM with the persistence of infection (2, 32), these results support the 11 idea that the generation of FM would have a negative impact for mycobacterial 12 control; conversely, the refractoriness of macrophages to become foamy mediated 13 by the IL-4/IL-13-STAT6 axis could have positive consequences for the host. 14 Surprisingly, however, the infection with Mtb per se promoted the LB accumulation 15 in M(IL-4) macrophages despite the activation of the IL-4/STAT6 axis. In fact, Mtb 16 may induce the foamy phenotype by hijacking a metabolic pathway downstream to 17 STAT6 activation. In addition, Mtb infection can decrease the lipolytic activity of 18 macrophages at early time points (5, 34). Therefore, we propose that modulation of 19 lipolysis is key for determining lipid accumulation in the context of Mtb infection. Of 20 note, although we were able to inhibit FM differentiation upon Mtb infection by 21 fostering the FAO pathway with L-carnitine, the cells were still susceptible to Mtb 22 intracellular growth. Since macrophages having a more active FAO, and acquiring 23 higher amounts of fatty acid, were reported as a preferred site for bacterial growth 24 and survival (19), the effect of reducing the LB formation by L-carnitine may be 1 counteracted by the exacerbation of the FAO metabolism, resulting in a net 2 persistence of Mtb. 3 In this study, we demonstrated that M(IL-4) macrophages are reprogrammed 4 metabolically by Mtb infection, leading to the formation of foam cells through the 5 positive regulation of HIF-1α and/or the decrease in the FAO, without modulation in 6 STAT6 phosphorylation. We consider that this mechanism constitutes a strategy of 7 mycobacterial persistence. Indeed, a high lipid content guarantees the survival of 8 the pathogen, and the activation of the IL-4 / STAT6 axis is associated with the 9 establishment of a poorly microbicidal profile (9). According to our results, HIF-1α 10 can be proposed as a molecular target to be modulated by Mtb infection impacting 11

the LB accumulation. Previous reports have demonstrated that the infection with 12
Mtb leads to glycolysis in bone-marrow derived macrophages (46, 47), in lungs of 13 infected mice (48), and in lung granulomas from patients with active TB (49). Here, 14 we showed that Mtb infection leads to HIF-1α activation and lactate release in 15 M(IL-4) macrophages. Of note, Mtb infection induces the increase of HIF-1α 16 expression in IFN--activated macrophages (M1), which is essential for IFN--17 dependent control of infection (50). Despite the fact that HIF-1α activation engages 18 the microbicidal program in macrophages, we consider that Mtb can also take 19 advantage of HIF-1α stabilization in M(IL-4) macrophages through the 20 enhancement of LB accumulation, which is known to be detrimental for the 21 inflammatory phenotype of macrophages (6, 51). It is also important to highlight 22 that Mtb can reprogram the metabolic state of M(IL-4) macrophages without 23 affecting STAT-6 activation. This means that the pro-inflammatory program, which 1 could have been established upon HIF-1α activation, may also be tuned by the 2 inhibitory signals driven by the IL-4/STAT-6 pathway. Hence, we speculate that 3 foamy M(IL-4) macrophages can bear high bacillary loads despite HIF-1α 4 activation, especially considering that LBs were described as a secure niche for 5 mycobacteria conferring protection even in the presence of bactericidal 6 mechanisms, such as respiratory burst (5). Moreover, our findings agree with a 7 recent article by Zhang et al, in which they demonstrated that the deficiency of the 8 E3 ligase von Hippel-Lindau protein (VHL), an enzyme that keeps HIF-1α at a low 9 level via ubiquitination followed by proteasomal degradation, uplifted glycolytic 10 metabolism in AM. Although not stressed by the authors, it is notable that these 11 AM displayed a foamy phenotype unlike AM from WT mice (52), arguing that the 12 uncontrolled activation of HIF-1α may contribute to FM formation. In line with this, it 13 has recently been described that the activation of HIF-1α during late stages of from Santa Cruz, Biotechnology (Palo Alto, CA, USA). 7

Preparation of human monocyte-derived macrophages 15
Buffy coats from healthy donors were prepared at Centro Regional de Hemoterapia 16 Garrahan (Buenos Aires, Argentina) according to institutional guidelines (resolution 17 number CEIANM-664/07). Informed consent was obtained from each donor before 18 blood collection. Monocytes were purified by centrifugation on a discontinuous 19 Percoll gradient (Amersham, Little Chalfont, United Kingdom) as previously 20 described (6). After that, monocytes were allowed to adhere to 24-well plates at 21 the presence of concurrent infectious diseases or non-infectious conditions 20 (cancer, diabetes, or steroid therapy). None of the patients had multidrug-resistant 21 TB. PE samples derived from patients with pleural transudates secondary to heart 22 failure (HF-PE, n=5) were employed to prepare a second pool of PE, used as 23 control of non-infectious inflammatory PE. The PE were collected in heparin tubes 24 and centrifuged at 300 g for 10 min at room temperature without brake. The cell-1 free supernatant was transferred into new plastic tubes, further centrifuged at 2 12000 g for 10 min and aliquots were stored at -80°C. The pools were 3 decomplemented at 56°C for 30 min and filtered by 0.22 µm in order to remove any 4 remaining debris or residual bacteria. 5

Foamy macrophage induction 6
Activated macrophages were treated with or without 20% v/v of PE, 10 μg/ml of 7 Mtb lipids (BEI resources) or infected with Mtb (MOI 2:1) for 24 h. When indicated, 8 cells were pre-incubated with either STAT6 inhibitor AS1517499 (100 nM, Sigma-9 Aldrich) for 30 min prior to IL-4 and TB-PE addition, orlistat (100 µM) and lalistat 10 antimycin (AA). Basal respiration was calculated as the last measurement before 24 addition of OM minus the non-mitochondrial respiration (minimum rate 1 measurement after ROT/AA). Estimated ATP production designates the last 2 measurement before addition of OM minus the minimum rate after OM. Maximal 3 respiration rate (max) was defined as the OCR after addition of OM and FCCP. 4 Spare respiration capacity (SRC) was defined as the difference between max and 5 basal respiration. 6 Quantitative RT-PCR 7 Total RNA was extracted with Trizol reagent (Sigma-Aldrich) and cDNA was 8 reverse transcribed using the Moloney murine leukemia virus reverse transcriptase 9 and random hexamer oligonucleotides for priming (Life Technologies). The 10 expression of CPT1 was determined using PCR SYBR Green sequence detection 11 system (Eurogentec, Seraing, Belgium) and the CFX Connect™ Real-Time PCR 12 Detection System (Bio-Rad, CA, United States). Gene transcript numbers were 13 standardized and adjusted relative to eukaryotic translation elongation factor 1 14 alpha 1 (EeF1A1) transcripts. Gene expression was quantified using the ∆∆Ct 15 method. 16

Transmission Electron Microscopy (TEM) 17
M0 and M(IL-4) cells exposed to TB-PE were prepared for TEM analysis. For this 18 purpose, cells were fixed in 2.5 % glutaraldehyde / 2 % paraformaldehyde (PFA, 19 EMS, Delta-Microscopies) dissolved in 0.1 M Sorensen buffer (pH 7.2) during 2 h 20 at room temperature and then they were preserved in 1 % PFA dissolved in 21 Sorensen buffer. Adherent cells were treated for 1 h with 1% aqueous uranyl 22 acetate then dehydrated in a graded ethanol series and embedded in Epon. 23 Sections were cut on a Leica Ultracut microtome and ultrathin sections were 24 mounted on 200 mesh onto Formvar carbon-coated copper grids. Finally, thin 1 sections were stained with 1% uranyl acetate and lead citrate and examined with a 2 transmission electron microscope (Jeol JEM-1400) at 80 kV. Images were acquired 3 using a digital camera (Gatan Orius). For the determination of LBs size and 4 number, TEM images were quantified with the ImageJ "analyze particles" plugins in 5 thresholded images, with size (µm 2 ) settings from 0.01 to 1 and circularity from 0.3 6 to 1. For quantification, 8-10 cells of random fields (1000x magnification) per 7 condition were analyzed. 8

Infection of human macrophages with Mtb 9
Infections were performed in the biosafety level 3 (BSL-3) laboratory at the Unidad 10

Operativa Centro de Contención Biológica (UOCCB), ANLIS-MALBRAN (Buenos 11
Aires), according to the biosafety institutional guidelines. Macrophages seeded on 12 glass coverslips within a 24-well tissue culture plate (Costar) at a density of 5x10 5 13 cells/ml were infected with Mtb H37Rv strain at a MOI of 2:1 during 1 h at 37ºC. 14 When indicated, PX-478 (100 µM) was added or not 1 h prior Mtb infection and 15 renewed in fresh complete media after cell-washing. Then, extracellular bacteria 16 were removed gently by washing with pre-warmed PBS, and cells were cultured in 17 RPMI-1640 medium supplemented with 10 % FBS and gentamicin (50 µg/ml) for 18 24 h. In some experiments, L -carnitine (10 mM) was added or not after removing 19 the extracellular bacteria. The glass coverslips were fixed with PFA 4% and stained 20 with ORO, as was previously described. 21 Measurement of bacterial intracellular growth in macrophages by CFU assay 22 Macrophages exposed (or not) to TB-PE, were infected with H37Rv Mtb strain at a 23 MOI of 0.2 bacteria/cell in triplicates. After 4 h, extracellular bacteria were removed 24 by gently washing four times with pre-warmed PBS. At different time points cells 1 were lysed in 1% Triton X-100 in Middlebrook 7H9 broth. Serial dilutions of the 2 lysates were plated in triplicate, onto 7H11-Oleic Albumin Dextrose Catalase 3 (OADC, Becton Dickinson) agar medium for CFU scoring 21 days later. 4

Visualization and quantification of Mtb infection 5
Macrophages seeded on glass coverslips within a 24-well tissue culture plate 6 (Costar) at a density of 5 × 10 5 cells/ml were infected with the red fluorescent 7 protein (RFP) expressing Mtb CDC 1551 strain at a MOI of 5:1 during 2 h at 37°C. 8 Then, extracellular bacteria were removed gently by washing with pre-warmed 9 PBS, and cells were cultured in RPMI-1640 medium supplemented with 10% FBS 10 for 48 h. The glass coverslips were fixed with PFA 4% and stained with BODIPY 11 493/503 (Life Technologies). Finally, slides were mounted and visualized with a 12 FluoView FV1000 confocal microscope (Olympus, Tokyo, Japan) equipped with a 13 Plapon 60X/NA1.42 objective, and then analyzed with the software ImageJ-Fiji. We