The Lipid Virulence Factors of Mycobacterium tuberculosis Exert Multilayered Control over Autophagy-Related Pathways in Infected Human Macrophages

Autophagy is an important innate immune defense mechanism that controls Mycobacterium tuberculosis (Mtb) growth inside macrophages. Autophagy machinery targets Mtb-containing phagosomes via xenophagy after damage to the phagosomal membrane due to the Type VII secretion system Esx-1 or via LC3-associated phagocytosis without phagosomal damage. Conversely, Mtb restricts autophagy-related pathways via the production of various bacterial protein factors. Although bacterial lipids are known to play strategic functions in Mtb pathogenesis, their role in autophagy manipulation remains largely unexplored. Here, we report that the lipid virulence factors sulfoglycolipids (SLs) and phthiocerol dimycocerosates (DIMs) control autophagy-related pathways through distinct mechanisms in human macrophages. Using knock-out and knock-in mutants of Mtb and Mycobacterium bovis BCG (Bacille Calmette Guerin) and purified lipids, we found that (i) Mtb mutants with DIM and SL deficiencies promoted functional autophagy via an MyD88-dependent and phagosomal damage-independent pathway in human macrophages; (ii) SLs limited this pathway by acting as TLR2 antagonists; (iii) DIMs prevented phagosomal damage-independent autophagy while promoting Esx-1-dependent xenophagy; (iv) and DIMs, but not SLs, limited the acidification of LC3-positive Mtb compartments. In total, our study reveals an unexpected and intricate role for Mtb lipid virulence factors in controlling autophagy-related pathways in human macrophages, thus providing further insight into the autophagy manipulation tactics deployed by intracellular bacterial pathogens.


Introduction
Each year, over 1.5 million people die from tuberculosis. This pulmonary infectious disease, which is caused by Mycobacterium tuberculosis (Mtb), remains the leading cause of death worldwide of bacterial origin. Mtb pathogenicity is closely related to its ability to survive in macrophages through the deployment of strategies to escape or neutralize host defenses [1]. One of these defenses is autophagy, a cellular process that allows for the capture of intracellular bacteria and their killing by lysosomes [2][3][4][5].
After phagocytosis by macrophages, Mtb resides in a single-membrane-bound compartment called the phagosome. The autophagy machinery targets Mtb-containing phagosomes via two main degradative pathways, xenophagy and LC3-associated phagocytosis (LAP) [6][7][8][9][10]. The LAP pathway is triggered upon bacterial phagocytosis (after the engagement of pattern recognition receptors (PRRs) on the cell surface) to promote LC3 conjugation to the phagosomal membrane: this phagosome is called a LAPosome [11]. Interestingly, Mtb RNA delivered by extracellular vesicles from infected macrophages in combination with IFN-β can also promote LAP via an RIG-I/MAVS-dependent pathway [12]. On the other hand, damage to the phagosomal membrane, which is triggered by the Type VII secretion system Esx-1, promotes xenophagy via the exposure of pathogen-associated molecular patterns (PAMPs) to the cytosolic sensor cGAS (cyclic GMP-AMP synthase), followed by STING signaling and the ubiquitination of damaged phagosome or via galectin recruitment onto the damaged phagosome [7,[13][14][15]. Alternatively, the xenophagy of Mtb-containing phagosomes can be achieved exogenously using pharmaceutical drugs or immunomodulators, such as cytokines [5,16]. Xenophagy requires the formation of an LC3-decorated double-membrane-bound compartment, which is called an autophagosome and engulfs the bacteria. Importantly, both the LAP and xenophagy pathways result in acidification, fusion with lysosomes of the Mtb compartment, and eventually bacterial death [7,10].
Mtb evades various host defense mechanisms (autophagy is one of them) [1,5]. Over the past decade, several studies have identified the mycobacterial proteins implicated in autophagy inhibition. For instance, Eis, an N-acetyl transferase, and PE_PGRS47 prevent autophagy initiation, whereas Esx-1 limits autophagic flux, i.e., fusion with lysosomes [17][18][19]. Recently, CpsA has been shown to restrict the LAP pathway by impairing the assembly of the NADPH oxidase NOX2 [10]. While quite a few mycobacterial proteins have been investigated, very little is known about the role of Mtb lipids in macrophage autophagy. In particular, little information is available on the noncovalently associated lipids located in the outermost part of the cell envelope, despite the fact that several of them contribute efficiently to pathogenesis [20,21].
These complex lipids are present on the surface of the bacteria and can be trafficked inside infected cells, as well as extracellularly. Some of them are potent immunomodulators. They can act as ligands of toll-like receptor 2 (TLR2) and can trigger autophagy upon phagocytosis [22,23]. Others, such as mannose-capped lipoarabinomannan, limit LC3 recruitment onto latex beads containing phagosomes via an unknown mechanism [23,24]. A few envelope lipids also have a role in Mtb virulence, among which are major lipid virulence factors phthiocerol dimycocerosates (DIMs, also called PDIMs), and Mtb-specific lipids sulfoglycolipids (SLs, also called SGLs). Recently, DIMs have been proposed to induce autophagy. Indeed, THP-1 macrophages infected with an Mtb mutant with DIM transport deficiencies have lower levels of GFP-LC3 associated with their membranes than do those with wild-type (WT) Mtb, suggesting either a decrease in autophagy or an activation of LC3 degradation [25,26]. Whether DIMs promote the targeting of Mtb through autophagy machinery also remains to be investigated [26]. Another work has reported that purified Mtb SLs promote GFP-LC3 puncta accumulation in macrophages [27]. Again, it is unclear whether SLs activate autophagy or prevent LC3 turnover. In addition, the importance of SLs in the bacterial context has not been addressed. Collectively, these published data suggest that DIMs and SLs may modulate macrophage autophagy, but how these lipids precisely control autophagy-related pathways that target Mtb-containing phagosomes needs to be explored.
Here, we investigated the contributions of Mtb lipid virulence factors (DIMs and SLs) to the modulation of autophagy during the course of human macrophage infection. Our work unveiled a multifaceted role for Mtb lipid virulence factors in controlling autophagy-related pathways. Overall, this study deepens our understanding of autophagy manipulation by Mtb and the functions of lipid virulence factors in infection.

Cell Culture and Infection
Human blood from fully anonymous nontuberculous donors was purchased from the Etablissement Français du Sang of Toulouse. Human macrophages derived from monocytes (hMDMs) were prepared as described in Reference [32]. In brief, monocytes plated into 24-well plates were isolated by adhesion from peripheral blood mononuclear cells (PBMCs) and allowed to differentiate for 7 days in RPMI (Roswell Park Memorial Institute) 1640 medium (Gibco, USA) supplemented with 2 mM glutamine (Gibco) and 7% (v/v) heat-inactivated human AB serum at 37 • C in 5% CO 2 . The culture medium was renewed on the third day. Before use, hMDMs were washed twice with fresh RPMI medium containing glutamine.
A mycobacterial infection was performed as described in Reference [32]. Briefly, exponentially growing mycobacteria were pelleted by centrifugation and subsequently dispersed in serum-free RPMI 1640 medium using sterile glass beads. The number of bacteria per ml in the suspension was estimated through a measurement of the optical density at 600 nm. Human macrophages were infected for 1 h at an MOI (multiplicity of infection) between 5 and 10 in RPMI medium at 37 • C in 5% CO 2 . Extracellular bacteria were then removed by three successive washes with fresh medium, and infected cells were further incubated in RMPI medium supplemented with heat-inactivated serum at 37 • C in 5% CO 2 .

Western Immunoblotting and a Proteome Profiler Human Phospho-Kinase Array Kit
THP-1 cells were seeded and differentiated in 6-well plates (5 × 10 6 cells). Cells were infected with Mtb strains at MOI = 10 for 1 h, and then extracellular bacteria were washed away. Infected cells were incubated for 48 h in the presence or absence of 100 nM Bafilomycin A1 (Santa Cruz, CA, USA) (for the last 2 h) (0.1% DMSO (dimethyl sulfoxide) was used as a negative control, Sigma, USA). For the Pam3CSK4 experiment, cells were treated with the indicated concentration for 2 h, and then 100 nM Bafilomycin A1 or the DMSO control was added and cells were incubated for 2 h to block autophagic flux and accumulate LC3-II. After incubation, the cells were lysed with a Laemmli buffer containing beta-mercaptoethanol and boiled for 10-15 min at 95 • C. Denaturated proteins were subjected to SDS (sodium dodecyl sulfate-polyacrylamide gel electrophoresis (4-16% gradient) using a Tris/glycine buffer system (BioRad, USA). After electrophoresis, proteins were transferred to a nitrocellulose transfer membrane. Blots were blocked with 5% dried milk or bovine serum albumin (BSA) in PBS (phosphate buffered saline) and then incubated with primary antibodies and the corresponding horseradish peroxidase-conjugated secondary antibody (Thermo Scientific, USA). Immunostaining was detected with SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific). A beta-actin band was used as a loading control. A proteome profiler array kit (ARY003B) was purchased from R&D Systems (USA). Each array contained two membranes spotted in duplicate with antibodies against 43 different phosphorylated proteins and 2 related total proteins. The assay was performed according to the manufacturer's instructions. Membranes were visualized with a ChemiDoc TM Touch Imaging System (BioRad), and integrated densities of Western blot bands and phospho-kinase array dots were measured using Image J software (National Institutes of Health).

Immunofluorescence, Lysotracker Staining, and Confocal Microscopy
Macrophages cultured on sterile glass coverslips were infected with GFP (green fluorescent protein)-expressing mycobacteria and fixed with 4% (v/v) paraformaldehyde (Electron Microscopy Sciences, USA) at different times after infection. Cells were then permeabilized for 5 min through a treatment in PBS containing 0.1% Triton X100. After blocking with 4% (w/v) BSA (bovine serum albumin) and 2% (v/v) goat serum in PBS, permeabilized cells were incubated with primary antibody followed by secondary Alexa 568-conjugated antibody or Alexa 647-conjugated antibody (Invitrogen, France). For the LysoTracker Red (LTR) experiment, cells were incubated during the last two 2 h of postinfection with LTR (Invitrogen) at 1 µM, followed by several washes and fixation. The coverslips were mounted onto glass slides with fluorescent mounting medium (Dako) and were analyzed on a Zeiss LSM510 or Olympus FV1000 confocal microscope. Around 15 to 30 random images were taken per condition per independent experiment. Images were analyzed using LSM510 or Image J software.

Statistical Analyses
The data were from at least three independent experiments, and they are shown as the mean ± SEM (standard error of the mean). Statistical analyses were performed with Graphpad Prism version 5 software using Student's two-tailed t-test. Differences were considered significant when the p-value was inferior to 0.05.

The Loss of SLs and DIMs Promoted LC3 Recruitment to M. tuberculosis Compartments
In our previously published work, we reported that the Mtb mutant PMM127, which is lacking in both DIMs and SLs, is trafficked more efficiently to lysosomes than is WT Mtb (in human macrophages) [33]. Importantly, the Mtb mutant PMM127 does not represent a significant change to the bacterial cell envelope structure [33]. However, whether this Mtb mutant interacts differently with autophagy machinery than does WT Mtb was not investigated. First, we examined endogenous LC3's association with the Mtb intracellular compartment using immunofluorescence and confocal microscopy. Human monocyte-derived macrophages (hMDMs) were pulsed with GFP-expressing Mtb strains for 1 h, chased for different time points, and stained with a specific antibody against LC3. At both early (2 h) and late time points (72 h) postinfection (p.i.), the percentage of Mtb positive for LC3 was two times higher with PMM127 than with WT ( Figure 1A,B). These results were also observed in PMA-differentiated THP-1 macrophages up to 96 h p.i ( Figure 1C). To determine the contribution of each type of lipid to the inhibition of LC3 recruitment, macrophages were infected with PMM127 that had been complemented genetically for either DIM or SL production [31]. At 2 h p.i., the complemented strains behaved like PMM127. However, at a late time point p.i. (72 h), DIM and SL production significantly reduced LC3 colocalization with PMM127-containing compartments by 50% and 90%, respectively ( Figure 1A,B). These results reveal that Mtb producing SLs, and to a lesser extent DIMs, are able to prevent LC3 recruitment to Mtb compartments in human macrophages.

The Mtb Mutant Deficient in DIMs and SLs Promoted Autophagy Activation
Several studies have revealed that LC3 can be involved in autophagy-independent processes [34]. To confirm that PMM127 is more targeted by autophagy machinery than is WT Mtb, we examined the intracellular localization of endogenous Beclin-1 and Atg16L1, two autophagy-related gene (Atg) products acting upstream of LC3. An immunofluorescence confocal microscopy analysis and quantification showed that around 65% to 50% of PMM127 was colocalized with Beclin-1 and Atg16L1 independently of the time postinfection, whereas only 30% to 20% of WT Mtb was associated with these proteins (Figure 2A-D). To determine whether the increase in Atg recruitment was associated with an upregulation of these proteins, we evaluated the protein levels of Beclin-1 and Atg16L1 through immunoblotting. Infection with PMM127 induced an approximately two-fold increase in the Beclin-1 protein amount compared to WT Mtb and noninfected macrophages ( Figure S1A,B). Although not significant, a similar trend was observed for Atg16L1 ( Figure S1C,D). Using quantitative RT-PCR, we found that both WT Mtb and PMM127 induced an increase in BECN1 gene expression, but not in ATG16L1 expression ( Figure S1E,F). However, no difference was detected between the two strains for either gene, indicating that the difference observed upon infection with WT or PMM127 was not due to differential gene expression. Collectively, these data indicate that the lipid-deficient Mtb mutant promoted autophagy machinery recruitment to bacterial compartments. To further explore autophagy activation, LC3-II lysosomal turnover was determined using Bafilomycin A1 (BafA1), an inhibitor of lysosomal degradative activities, and immunoblotting. WT Mtb and PMM127 infections resulted in an increase in LC3-II turnover compared to noninfected macrophages ( Figure 2E). However, the turnover was almost two-fold higher with PMM127 (LC3-II/actin with BafA1 minus LC3-II/actin without BafA1 = 5.85) than with WT (LC3-II/actin with BafA1 minus LC3-II/actin without BafA1 = 3.25) ( Figure 2E). Furthermore, in the presence of BafA1, the total level of LC3-II was much greater with PMM127 than with the WT, which supported the idea of the activation of autophagy initiation. Altogether, these results indicate that the lipid-deficient Mtb mutant unleashed functional autophagy in macrophages.

The Mtb Mutant Deficient in DIMs and SLs Promoted Autophagy via MyD88
In order to get some insight into the signaling pathway involved in autophagy activation induced by PMM127, we used an unbiased approach based on a phosphoprotein-specific antibody array [35]. We found that the two major signaling pathways implicated in autophagy regulation, i.e., mTOR (mammalian target of rapamycin) (p70S6K phosphorylation) and AMPK (5 AMP-activated protein kinase), were activated in the WT Mtband PMM127-infected macrophages ( Figure S2A,B) [36]. However, the levels of activation were similar between WT and PMM127, indicating that these two pathways were not responsible for the autophagy upregulation observed with Mtb mutants deficient in DIMs and SLs.
The TLR/MyD88 signaling pathway is known to promote autophagy [3], in particular in mycobacteria-infected macrophages [22,37]. Therefore, we investigated its role in PMM127-induced autophagy. Differentiated THP-1 macrophages (WT or MyD88 knockouts (KOs)) were infected with WT Mtb or PMM127 and treated with BafA1 to determine the total amount of LC3-II formed upon infection. Immunoblot analyses showed that the LC3-II level in infected cells was reduced in MyD88-deficient macrophages compared to WT macrophages ( Figure 3A). Importantly, while in WT macrophages PMM127 induced more LC3-II than did WT Mtb, this difference was drastically reduced in MyD88-deficient macrophages. These results indicated that Mtb mutants deficient in DIMs and SLs promoted autophagy via TLR/MyD88. This is consistent with published data showing that SLs limit the TLR/MyD88 pathway in macrophages by acting as a competitive TLR2 antagonist [28]: it is also consistent with a recent report indicating that DIMs evade TLR signaling in macrophages [38], although the mechanism of action remains unclear [39].

Synthetic SL, but not Purified DIM, Molecules Limited TLR2-Dependent Autophagy
TLR2, one of the main PRRs (pattern recognition receptors) that recognize Mtb [40], has been implicated in autophagy activation in the context of mycobacterial infection [22,41,42]. Next, we investigated whether SLs and DIMs could per se limit TLR2-dependent autophagy by testing purified SL and DIM molecules. TLR2-dependent autophagy was induced by incubating macrophages with a synthetic lipopeptide TLR2 ligand, Pam3CSK4. First, we determined the concentration of Pam3CSK4 required for autophagy activation in THP-1 macrophages and found that 2.5 µg/mL was necessary to obtain about a two-fold increase in the LC3-II total in the presence of BafA1 ( Figure 3B). Cells were then pretreated with a synthetic analog of SLs [28] or with purified DIMs, followed by Pam3CSK4 treatment in the presence of BafA1. We found that synthetic SLs at 10 µg/mL inhibited TLR2-mediated autophagy by around 20% but had a negligible effect on basal autophagy ( Figure 3C). Similar results were obtained with SLs at 20 µg/mL ( Figure S3A,C) and another TLR2 ligand, zymosan ( Figure S3B). In contrast, purified DIMs at 100 µg/mL did not inhibit Pam3CSK4-induced autophagy, but rather seemed to activate basal autophagy ( Figure 3D, Figure S3D). These results show that SL, but not DIM, molecules could directly prevent TLR2-mediated autophagy.

Autophagy Mediated by a Lipid-Deficient Mtb Mutant Was Not Associated with Phagosomal Damage
TLR/MyD88 is involved in autophagy pathways that target either damaged or intact phagosomes [37,43,44]. Since Mtb can reside in both types of phagosome, we examined whether enhanced LC3 recruitment was associated with enhanced phagosomal membrane damage using ubiquitin and galectin-3 as markers [45]. Immunofluorescence and confocal microscopy analyses showed that, at 3 days p.i., a small percentage of LC3-positive compartments containing WT Mtb, around 20%, were positive for ubiquitin and galectin-3 ( Figure 4A-D). This result highlights the coexistence of two autophagy-related pathways targeting Mtb, one linked to phagosomal damage and a second that targets intact phagosomes. Importantly, the percentage of LC3 compartments containing Mtb that were positive for ubiquitin and galectin-3 dropped significantly to approximately 5% with PMM127. These findings indicate that the loss of DIMs and SLs did not enhance LC3 recruitment to the damaged phagosome, but rather promoted LC3 recruitment to the damage-free phagosome.

DIMs Limited Phagosomal Damage-Independent Autophagy while Promoting Esx-1-Dependent Autophagy
As shown in Figure 4, we noted a higher proportion of damage-associated autophagy with WT Mtb than with a lipid-deficient mutant. To determine which lipids, DIMs or SLs, were involved in this process, macrophages were infected with PMM127 complemented genetically for DIM or SL production, and the percentage of LC3 compartments containing Mtb that were positive for ubiquitin or galectin-3 was counted. Mtb complementation for DIM production (but not SL) significantly increased the percentage of LC3-positive compartments containing Mtb that were positive for ubiquitin and for galectin-3 ( Figure 5A). These results were consistent with our published data, which showed an increase in phagosomal damage when Mtb was complemented for DIM (but not for SL) production [31]. Thus, DIMs seemed to favor LC3 associated with phagosomal damage ( Figure 5A). However, overall, DIMs inhibited LC3 recruitment to Mtb compartments ( Figure 1A,B), suggesting that these lipids may inhibit LC3 recruitment to damage-free phagosomes. To validate this hypothesis, we investigated the respective contribution of these two pathways using Mycobacterium bovis Bacillus Calmette-Guérin (BCG), which is naturally deficient in SLs; Esx-1, a secretion system required for phagosomal damage; and recombinant BCG strains expressing different combinations of DIMs and Esx-1 [31]. Macrophages were infected with different GFP-expressing BCG strains and stained for endogenous LC3 and ubiquitin. As expected from a previous work [7], damage-dependent autophagy was induced by Esx-1, as seen in the occurrence of ubiquitin in LC3-positive compartments containing BCG::Esx-1 but not BCG ( Figure 5B). Furthermore, we observed that DIM-deficient BCG::Esx-1 was less able to promote ubiquitin-associated autophagy than was DIM-proficient BCG::Esx-1. This confirmed that DIMs promoted damage-dependent autophagy in the presence of functional Esx-1. Nevertheless, the percentage of mycobacteria positive for LC3 was significantly higher in the absence of DIM, even in the absence of Esx-1, indicating that DIMs inhibited damage-independent autophagy ( Figure 5C) while promoting Esx-1-dependent autophagy ( Figure 5B).

DIMs, but not SLs, Limited the Acidification of LC3-Positive Compartments Containing Mtb
The autophagy machinery of Beclin-1, Atg16L1, and LC3 targeted Mtb mutants deficient in DIMs and SLs (Figures 1 and 2). To determine whether the Mtb-containing compartments positive for LC3 could be acidified, infected macrophages were stained with LysoTracker Red (LTR) and analyzed using immunofluorescence confocal microscopy [22]. At 72 h p.i., while only 30% of WT Mtb within the LC3-positive compartments was colocalized with LTR, we observed that more than 50% of the LC3-positive compartments containing PMM127 were acidified ( Figure 6A,B), indicating that PMM127 not only enhanced LC3 recruitment compared to the WT (Figure 1) but also increased the acidification of LC3-positive bacterial compartments. To determine the contribution of each type of lipid to preventing the acidification of LC3-positive compartments containing Mtb, macrophages were infected with PMM127 complemented genetically for either DIM or SL production. Only 30% of PMM127/DIMs within the LC3-positive compartment were acidified compared to 50% with PMM127/SLs, indicating that DIM, but not SL, synthesis limited the acidification of mycobacterial compartments and restored the Mtb WT phenotype ( Figure 6B).

Discussion
In recent years, an increasing number of studies have reported that Mtb can circumvent autophagy pathways in macrophages by producing different bacterial proteins [10,17,18]. The present study reports, for the first time, that Mtb lipid virulence factors, namely DIMs and SLs, control autophagy at multiple levels in infected human macrophages. First, we found that Mtb mutants deficient in DIMs and SLs upregulated autophagy pathways that rely on MyD88 but not on phagosomal damage. SLs prevented this pathway by acting as an TLR2 antagonist. DIMs also limited autophagy independently of phagosomal damage, but in addition promoted an autophagy pathway that relied on Esx-1-mediated phagosomal damage. Finally, the presence of DIMs, unlike SLs, prevented acidification of the LC3-positive compartments containing Mtb (Figure 7).
Our previous work indicated that Mtb deficient in DIMs and SLs was attenuated in macrophages and in a mice model of infection (compared to WT Mtb) [33]. Furthermore, using an autophagy inhibitor, Bafilomycin A1, we were able to restore the WT Mtb growth phenotype [31], indicating that autophagy could contribute to growth attenuation. Here, we show that Mtb mutants deficient in DIMs and SLs promoted an autophagy-related pathway that required MyD88 but did not involve phagosomal damage. The LAP pathway is triggered upon phagocytosis after the engagement of PRRs on the cell surface (which includes TLR2) to promote LC3 conjugation in undamaged phagosomal membranes [44]. MyD88 is an important TLR adaptor protein that is involved in autophagy activation during the course of mycobacterial infection [37]. Recently, the LAP has been shown to target Mtb phagosomes [11], suggesting that Mtb mutants deficient in DIMs and SLs may stimulate a LAP process that involves TLR/MyD88 signaling. However, we cannot rule out the possibility that intact Mtb phagosomes might be captured in autophagosomes by a yet undefined molecular mechanism [46].
A growing amount of literature has demonstrated the key role of TLR/MyD88 signaling in bacterial infections [2,3,37,47]. Mtb produces numerous TLR ligands involved in triggering immune responses, but it has also developed several strategies to avoid recognition or to alter downstream signaling cascades [40]. Here, we found that SLs were able to restrict LC3 recruitment to an intact phagosome containing Mtb. As anticipated from our previous published data, which showed that SLs act as an antagonist of TLR2 [28], synthetic SLs prevented TLR2-induced autophagy. In contrast, purified DIMs did not prevent TLR2-mediated autophagy. This could be explained by either a requirement for the entire bacteria or the inhibition of MyD88-induced autophagy (which is independent of TLR2).
In human macrophages, DIMs are transferred to host cell membranes and disorganize the lipid bilayer in a way that modulates phagocytic receptor functions [30,32]. We thus suspect that DIMs' effects on membranes can affect membrane receptors such as TLR. Recently, the Mtb lipoprotein LprE has been shown to suppress TLR2-dependent autophagy [42]. Since Mtb produces several factors that antagonize TLR2, such as Esat-6 and phenolic glycolipids [48,49], it is tempting to speculate that these factors might also participate in limiting TLR2-induced autophagy during the course of Mtb infection.
Recently, DIMs, but not SLs, have been reported to be required for phagosomal damage and rupture in human macrophages [26,31]. Furthermore, a flow cytometry analysis has shown that Mtb mutants deficient in DIMs accumulated less membrane-bound LC3-II than did WT Mtb, which suggested reduced autophagy initiation, although this could also be interpreted as an increase in autophagic flux [26]. Along the same lines, in lymphatic endothelial cells, DIMs promote phagosomal rupture and recruitment of the xenophagy receptors NDP52 and p62 to Mtb, hinting toward a pro-xenophagy role for DIMs [50]. As anticipated, our findings demonstrate that DIMs, but not SLs, triggered the xenophagy of Mtb in human macrophages. Importantly, we found that DIMs induced xenophagy only when the mycobacteria expressed functional Esx-1, which underscores the significance of the interplay between DIM and Esx-1 in modulating host defense mechanisms, including autophagy.
Another interesting finding is the role of DIMs in preventing the acidification of LC3-positive Mtb compartments (Figure 7). Mtb seems to limit autophagic flux by preventing Rab7 recruitment in autophagosomal membranes [51]. One major Mtb factor known to block the last stage of autophagy is Esx-1 and its effector Esat-6, albeit their molecular mechanisms of action remain to be elucidated [19,51]. We have previously shown that DIMs prevent the acidification of the Mtb phagosome inside human macrophages [32]. Here, we found that DIMs also interfered with the acidification of LC3-positive compartments containing Mtb. How DIMs block the acidification of phagosomes is under intensive investigation. Since DIMs and Esx-1 synergize to promote phagosomal damage and xenophagy, it will be interesting to determine whether they also act together in preventing the acidification of autophagic compartments. Importantly, DIM action seems to be specific, as we did not observe such inhibition with SLs, even though purified SLs can modulate phagosome acidification [52,53]. Other mycobacterial lipids are known to prevent phagosome acidification [23,[54][55][56][57]. Whether such lipids impair the acidification of autophagosomes or LAPosomes needs to be explored.
Compelling evidence has shown that DIMs and SLs exert multiple activities during the infection of macrophages, including the modulation of immune responses [20,26,28,31,38]. Here, we established a new role for DIMs and SLs in the control of autophagy. To survive in the adverse environment of the host, Mtb alters its gene expression. It should be noted that the genes involved in the biosynthesis of SLs and DIMs are upregulated upon infection [58,59]. In addition, the mass of these lipids increases as a result of their synthesis being coupled to cholesterol metabolism [60][61][62]. Given the multifaceted functions of SLs and DIMs, it is therefore reasonable to propose that the expansion of lipid pools in M. tuberculosis has important consequences in pathogenesis.
To conclude, our work uncovered novel Mtb virulence factors involved in the evasion of macrophage autophagy. To our knowledge, DIMs and SLs are the first bacterial lipids reported to modulate autophagy in an infection context. In addition to being attractive potential therapeutic targets in curtailing Mtb infection, DIMs and SLs, or synthetic functional analogs, might be useful tools for further deciphering the mechanisms of autophagy inhibition. Exploring these molecular mechanisms should help in the development of new compounds to either stimulate autophagy in the context of host-directed therapy against tuberculosis [5] or to limit detrimental autophagy in some other pathological contexts [63,64]. A deeper understanding of the multiple roles of autophagy in Mtb infection is warranted to discover important clues for the development of autophagy-based vaccines against and therapies for tuberculosis [65,66]. Mtb resides inside either a damaged or an intact phagosome [1]. The Type VII secretion system Esx-1 promotes phagosomal damage, which leads to xenophagy [7]. However, Esx-1 can also inhibit autophagic flux via an unknown mechanism [19,51,67]. Here, we found that (i) SLs limited TLR/MyD88-dependent and phagosomal damage-independent autophagy by acting as a TLR2 antagonist (1). (ii) DIMs prevented, to some extent, phagosomal damage-independent autophagy (dotted line) (1) while triggering xenophagy by favoring Esx-1-dependent phagosomal damage (2). (iii) Finally, DIMs, but not SLs, limited the acidification of LC3-positive compartments containing Mtb (3). Ubi: ubiquitin; Gal3: galectin-3. In terms of colors, autophagy modulation by SLs and DIMs is shown in red (inhibition) and green (activation).