Mycobacterium tuberculosis: Rewiring host cell signaling to promote infection

Abstract The ability of Mycobacterium tuberculosis to cause disease hinges upon successfully thwarting the innate defenses of the macrophage host cell. The pathogen's trump card is its armory of virulence factors that throw normal host cell signaling into disarray. This process of subverting the macrophage begins upon entry into the cell, when M. tuberculosis actively inhibits the fusion of the bacilli‐laden phagosomes with lysosomes. The pathogen then modulates an array of host signal transduction pathways, which dampens the macrophage's host‐protective cytokine response, while simultaneously adapting host cell metabolism to stimulate lipid body accumulation. Mycobacterium tuberculosis also renovates the surface of its innate host cells by altering the expression of key molecules required for full activation of the adaptive immune response. Finally, the pathogen coordinates its exit from the host cell by shifting the balance from the host‐protective apoptotic cell death program toward a lytic form of host cell death. Thus, M. tuberculosis exploits its extensive repertoire of virulence factors in order to orchestrate the infection process to facilitate its growth, dissemination, and entry into latency. This review offers critical insights into the most recent advances in our knowledge of how M. tuberculosis manipulates host cell signaling. An appreciation of such interactions between the pathogen and host is critical for guiding novel therapies and understanding the factors that lead to the development of active disease in only a subset of exposed individuals.


INTRODUCTION
Mycobacterium tuberculosis has coexisted with mankind for tens of thousands of years, claiming more lives than any other infectious agent. This long history of coevolution with humans has given rise to a pathogen uniquely capable of persisting even in the face of a plethora of host antimicrobial effector mechanisms. Mycobacterium tuberculosis thus continues to cause devastating morbidity and mortality, killing 1.8 million people in 2015 alone, and latently infecting an estimated one quarter of the world's population. 1 Mycobacterium tuberculosis is transmitted between hosts by aerosols, which are capable of traveling to distal regions of the host antimicrobial defenses. Mycobacterium tuberculosis lacks many of the classical bacterial virulence factors such as toxins and flagella, which are advantageous to pathogens that must compete with the mucosal microflora in order to colonize the host, but the need for which is obviated for pathogens that target sterile sites deep in the lung. 3 Instead, M. tuberculosis has evolved a cornucopia of refined adaptations to escape immunity and persist within the host. Perhaps most interesting and sophisticated are the mechanisms by which the pathogen systematically disables, stimulates, or reroutes normal host cell signaling pathways to promote its own survival.

Hijacking phagosomes as a replicative niche
Over 45 years ago, Armstrong and Hart reported the archetypical virulence mechanism of M. tuberculosis-the inhibition of phagosome-lysosome fusion in macrophages. 4 Later work revealed that macrophages additionally fail to acidify mycobacteria-laden phagosomes, which was attributed to their inability to recruit host vacuolar-H + -ATPase (V-ATPase). 5 It was proposed that M. tuberculosis resists the maturation of phagosomes in order to exploit the organelle as an intracellular replicative niche. Phagosomes normally interact with the endosomal compartment to recruit V-ATPase, which actively transports protons into the phagosome, generating a potently acidic lumen that is required for subsequent fusion with lysosomes and the activity of the antimicrobial molecules they deliver. 6 Recent work showed that the secreted mycobacterial protein tyrosine phosphatase PtpA permeates through the phagosome membrane into the cytosol and binds to subunit H of host V-ATPase. 7,8 This binding disrupts the tethering of V-ATPase to the phagosome membrane, and also localizes PtpA in close proximity to its catalytic substrate, vacuolar protein sorting (VPS)33B, which is involved in regulating endocytic membrane fusion. Interference with V-ATPase recruitment and dephosphorylation of VPS33B by PtpA are both required for the inhibition of phagosome acidification and phagosome-lysosome fusion. 7 Interestingly, although deletion of PtpA restricts bacterial growth within human THP-1 cells, 8 a deletion mutant was not attenuated for growth or virulence during in vivo infection of mice. 9 This discrepancy was attributed to potential species differences affecting the activity of PtpA, such that reduced activity in a particular species would obscure any defect caused by its deletion. However, equally likely is the existence of some degree of functional redundancy between particular mycobacterial virulence factors that only becomes apparent during in vivo infection. In fact, numerous virulence factors have been reported to interfere with phagosome-lysosome fusion in vitro, including SecA2, PknG, SapM, components of the ESX-1 secretion system, and various glycolipids (Fig. 1). [10][11][12][13][14] The deletion of some of these virulence factors was found to attenuate bacterial growth and disease in animal models. 11,15 While this could indicate that certain virulence factors play nonredundant roles in inhibiting phagolysosome formation in vivo, several other roles have been reported for these factors, some of which are discussed later in this review. Collectively, although the literature supports the inhibition of phagosome maturation as a potent virulence mechanism of M. tuberculosis, the reported multifunctional roles of many of the reported effectors of this inhibition makes it difficult to determine the extent to which their phagolysosome-inhibitory activity alone impacts on overall disease pathogenesis.

The phagosome-a Trojan horse?
Until recently, M. tuberculosis was thought to reside within the phagosomal compartment of macrophages for the entire duration of the infection cycle, with release occurring upon lytic cell death. This paradigm shifted, however, with reports describing rupture of the phagosomal membrane and translocation of M. tuberculosis to the cytosol. [16][17][18] This escape from phagosomes appears to be dependent on both ESX-1 and particular bacterial outer membrane lipids (phthiocerol dimycocerosates). 19 In 2015, 3 independent studies simultaneously reported the activation of cytosolic DNA sensing pathways resulting from this egress of M. tuberculosis from phagosomes. [20][21][22] The DNA sensor cyclic GMP-AMP synthase (cGAS) was shown to associate with M. tuberculosis DNA in the cytosol to stimulate cyclic GAMP synthesis. This secondary messenger activates stimulator of interferon genes (STING), which drives the STING-TBK1-IRF3 signaling pathway, leading to the transcriptional induction of type I IFNs, particularly IFN- (Fig. 1). Collectively, the studies showed that deletion of the cGAS-STING pathway abrogated IFN-secretion by macrophages infected in vitro with M. tuberculosis. However, this in vitro phenotype was only partially replicated in mice deficient in either cGAS or STING, which had slightly reduced serum IFN-levels but otherwise unaltered inflammatory markers. 20,21 Furthermore, this reduction did not translate into any defect in resistance to M. tuberculosis infection, with Cgas −/− or Sting −/− mice harboring similar bacterial numbers in all organs at all time points. One study did, however, report earlier mortality of Cgas −/− but not Sting −/− mice compared with wild-type mice. 21 This raises two points that require further investigation; first, additional pathways besides cGAS-STING contribute to M. tuberculosis induced IFN-production in vivo, and second, perhaps cGAS has other functions conferring resistance to chronic infection.

Modulation of type I IFN signaling
Type I IFNs are potent mediators of antiviral immunity, but are often associated with disease progression in bacterial infections. C57BL/6 mice deficient in the type I IFN receptor (IFNAR) are indistinguishable from wild-type mice following M. tuberculosis infection, except for a reduction in splenic bacterial burden. 23 However, IFNAR deletion in more susceptible mouse strains substantially improves survival, with mice having much lower bacterial burdens. 24

M. tb virulence factors
F I G U R E 1 Mycobacterium tuberculosis manipulates essential components of the macrophage antimicrobial response. Mycobacterial virulence factors such as ESX-1, PtpA, and SecA2 interfere with the ability of lysosomes to kill phagocytosed M. tuberculosis, thereby establishing a secure, intracellular replicative niche. The pathogen then disables NF-B and MAPK signal transduction pathways, blunting the macrophage's hostprotective cytokine response. However, bacilli may eventually burst from phagosomes to activate host cytosolic DNA sensors, either directly by shedding their own DNA, or indirectly by inducing mitochondrial stress and DNA release. The consequence of this is the transcriptional induction of type I IFNs, which are detrimental to the host support the notion that type I IFNs are associated with TB disease progression. The mechanism for this remains unclear, but recent work suggests that the inhibition of host-protective cytokines (TNF, IL-12, IL-1 ), blunting of IFN-responsiveness, and induction of immunosuppressive IL-10 may be involved. 25,27 In fact, Wassermann and colleagues 22 showed that secretion of IL-10 was reduced in infected cGAS-and STING-deficient macrophages. They further demonstrated that macrophages infected with the hypervirulent HN878 strain, which induces a stronger type I IFN response, benefited from cGAS or STING deletion in terms of reduced IL-10 production and enhanced survival.
It would be interesting to examine the outcome of infection of Cgas −/− mice with HN878, and indeed, cGAS deficiency in a more susceptible mouse strain.
IFN-production following M. tuberculosis infection is dependent on a functional ESX-1 secretion system, 20,22,23 which indicates that the mycobacterial DNA that activates cGAS originates from living rather than dead/degrading bacteria. Although it remains unclear exactly how M. tuberculosis releases its DNA into the cytosol, it is tempting to speculate that this occurs deliberately in order to coopt host signaling in favor of IFN-production. However, subsequent work showed that strain-dependent differences in IFN-production were at least partially related to the level of mitochondrial stress and mitochondrial DNA released into the cytosol due to M. tuberculosis infection, and not due to variability in bacterial access to the cytosol or shedding of bacterial DNA. 28 The detection of host mitochondrial DNA by cGAS may thus contribute to type I IFN production. It remains unclear exactly why strains vary in their ability to induce mitochondrial stress/DNA release, but is most likely due to variability in the expression of particular virulence factors. It is almost certain, however, that additional factors or pathways contribute to differences in IFNproduction between strains, as mitochondrial involvement does not completely account for these differences. 28

Short-circuiting signal transduction pathways
Macrophages infected with M. tuberculosis produce a number of cytokines in order to orchestrate an effective immune response to the pathogen. The synthesis of many host-protective cytokines is regulated by the NF-B and MAPK signaling pathways. Given that several such cytokines, such as TNF, IL-1 , and IL-6, are potent, nonredundant mediators of anti-TB immunity, it is perhaps not surprising that their regulatory pathways represent attractive targets for dampening the host immune response to infection. For instance, the M. tuberculosis virulence protein PtpA inhibits the JNK, p38, and NF-B pathways in macrophages (Fig. 1). 29 The inhibition of the JNK and p38 MAPK pathways is dependent on the phosphatase activity of PtpA, which dephosphorylates phospho-JNK and phospho-p38. Interestingly, this activity is itself stimulated by the binding of ubiquitin-interacting motif-like region of PtpA to host ubiquitin, which may serve to restrain its phosphatase activity until the protein is secreted into the cytosol.
PtpA also competitively binds to TAB3, blocking its ability to bind to K63 ubiquitin chains and thereby partially interfering with NF-B activation. A subsequent study identified the RING domain of host TRIM27 as an additional target of PtpA. 30 TRIM27 has E3 ubiquitin ligase activity and promotes JNK/p38 pathway activation to restrict M. tuberculosis. By deleting PtpA from Mycobacterium bovis bacillus Calmette-Guérin (BCG), these studies showed that the protein suppresses the production of TNF, IL-1 , and IL-12 by macrophages both in vitro as well as in mice, which also harbored fewer bacteria in the lungs compared to mice infected with wild-type BCG. 29,30 This observation contrasts with the deletion of PtpA in M. tuberculosis that, as discussed earlier, does not attenuate the pathogen during infection in vivo. 9 The discrepancy is most likely explained by the fact that M. tuberculosis possesses many additional virulence factors (such as ESX-1) that BCG lacks, and that may render the function of PtpA redundant in vivo.
The ability of PtpA to inhibit cytokine production appears to be mostly due to its impairment of NF-B signaling, with contribution from its MAPK-inhibitory activity, particularly JNK and p38, but not ERK1/2. 29 Other virulence factors of M. tuberculosis have also been reported to interfere with NF-B activation. For example, treatment of cells with purified early secreted antigenic target 6 (ESAT-6)the major ESX-1 substrate-inhibits NF-B activation downstream of TLRs, thereby attenuating TNF and IL-6 release. 31 Interestingly, ESAT-6 is capable of preventing TLR-mediated NF-B activation downstream of all TLRs, despite only binding directly to TLR2. This binding activates cytosolic Akt kinase that appears to prevent the formation of MyD88 signaling complexes, and therefore NF-B activation, following ligation of other TLRs. The mammalian cell entry protein 3E (Mce3E), which is encoded by the mce3 operon, is secreted by phagocytosed M. tuberculosis and is expressed during infection in humans. 32,33 MceE3 was recently shown to interfere with the ERK1/2 MAPK signaling pathway by entering the cytosol and localizing to the endoplasmic reticulum (ER). 33 Here, it interacts with ERK1/2, tethering it to ER. Collectively, these studies reveal novel mechanisms by which M. tuberculosis may interfere with several major host signal transduction pathways, and thereby modulate cytokine production to the detriment of the host. Further work will be required to elucidate the in vivo contributions of the various virulence factors that have been implicated.

Metabolic reprogramming of host cells
The metabolic versatility of M. tuberculosis enables it to grow on a variety of carbon sources, and the available evidence indicates that during intracellular growth, the pathogen relies primarily on cholesterol ester and fatty acid metabolism. [35][36][37] Mycobacterium tuberculosis adjusts macrophage metabolism under hypoxic conditions to promote the accumulation of lipid bodies, giving rise to the "foamy" macrophages characteristically found at the interface of central necrotic regions within granulomas. 38 Mycobacterium tuberculosis laden phagosomes have been shown to interact with, and release bacilli into, host lipid bodies, which serve as a critical source of nutrients in an otherwise nutritionally devoid phagosome. 38 This interaction with host lipid bodies also generates a secure niche within which the pathogen is protected from bactericidal mechanisms such as respiratory burst.
Additionally, mycobacteria contained within lipid bodies acquire a dormancy phenotype, which confers tolerance to several frontline antibiotics. 39 Although the advantages that lipid body accumulation confers to M. tuberculosis are well established, insights into the mechanisms by which the pathogen coopts the macrophage to induce the foamy phenotype have only recently come to light (Fig. 2). Macrophages are driven into an anabolic state by ESAT-6, which stimulates the translocation of GLUT-1 glucose transporters from the cytosol to the cell membrane, thereby drastically enhancing glucose uptake and metabolism. 40,41 ESAT-6 also appears to heighten the activity of several glycolytic enzymes, thus perturbing the normal flux between glycolysis and the tricarboxylic acid cycle. This leads to the accumulation of dihydroxyacetone phosphate, which is used as a substrate for the synthesis of triacylglycerol (TAG). 40 Concurrently, intracellular concentrations of acetyl CoA also rise, which promotes ketogenesis and results in shunting of acetyl CoA toward the synthesis of the ketone body D-3-hydroxybutyrate (3HB), 40,42 although some of this acetyl CoA may also be directed toward de novo lipid synthesis. 41 Secreted 3HB activates the antilipolytic G protein coupled receptor GPR109A, which inhibits the pathway leading to the phosphorylation of perilipin. 42 The absence of phosphorylated perilipin prohibits the translocation of hormone-sensitive lipase to lipid bodies, thereby preventing the mobilization of stored TAG. Thus, ESAT-6 perturbs host glucose metabolism to both promote the accumulation of lipid bodies and therefore the differentiation of "foamy" macrophages, and simultaneously arrest lipid catabolism. However, this enhancement of glycolysis also induces IL-1 and suppresses IL-10, which is required for controlling intracellular replication. 43 Additionally, studies showed that lipid body formation does not require live bacilli and can be driven by mycobacterial cell wall components such as lipoarabinomannan and trehalose dimycolate, in a partially TLR2-dependent manner. [44][45][46] This would suggest that modulation of glycolysis by ESAT-6 may not represent the sole or major tuberculosis has devised strategies to both stockpile essential nutrients in the form of lipid bodies, as well as to tap into these reserves as required, perhaps during reactivation from latency. ing their migration to lymph nodes and delaying the onset of adaptive immunity. 63,64 It is also thought that the cross-presentation of M. tuberculosis antigens is prevented by the inhibition of apoptosis of infected phagocytes, 65 as discussed below.

Directing the macrophage death modality
Failure to eliminate intracellular M. tuberculosis inevitably leads to death of the infected phagocyte. However, the manner in which M. tuberculosis infected macrophages die remains enormously controversial. A substantial body of research supports the prevailing opinion that virulent M. tuberculosis limits apoptosis of the cells they infect, instead promoting a lytic or necrotic form of death (Fig. 3). [66][67][68][69][70][71][72][73] Apoptosis is thought to result in the killing of intracellular mycobacteria and the priming of adaptive immunity when apoptotic bodies of infected cells are engulfed and degraded by uninfected macrophages in a process termed efferocytosis. 65,[74][75][76][77] The finding that avirulent strains of M. tuberculosis primarily induce apoptosis, while virulent strains induce mostly a lytic death modality is in keeping with the notion that apoptosis is protective for the host, and is thus a target for virulent mycobacteria. Multiple proteins secreted by M. tuberculosis into the macrophage cytoplasm have been reported to inhibit apoptosis by diminishing the availability of caspase 8, 71 interfering with apoptotic envelope formation, 68 dephosphorylating GSK3 , 78 inactivating TNF by inducing release of TNF receptor 2 67 and inducing antiapoptotic Mcl-1, 69 as well as other proteins for which the mechanisms have not been elucidated. 79,80 However, some groups maintain that virulent M. tuberculosis exclusively induces apoptosis. [81][82][83] This discrepancy is difficult to reconcile, but is likely due to differences in the experimental systems used in these studies, especially in relation to the species of origin and mortality of the cells being used.
While it is clear that a small proportion of macrophages infected even with virulent M. tuberculosis die by apoptosis, 76 the majority succumb to pathogen-induced lytic death. This death modality is believed to allow M. tuberculosis to evade host immunity, disseminate to neighboring cells, breach the airways, and ultimately transmit to new hosts in order to complete its pathogenic life cycle. 73,84 The detailed mechanism by which M. tuberculosis achieves this is unclear, but may relate to cell membrane microdisruptions caused by the pathogen, and the simultaneous induction of lipoxin A 4 , which blocks prostaglandin E 2 -mediated membrane repair. 65,85 Recent work also showed that M. tuberculosis secretes tuberculosis necrotizing toxin (TNT) into the macrophage cytosol, which hydrolyzes the coenzyme NAD + . 86 Depletion of NAD + results in macrophage necrosis, although the downstream events leading to this remain undefined. Intriguingly, M. tuberculosis produces an immunity factor for TNT (IFT), which protects mycobacterial cells from TNT-mediated toxicity by inhibiting its activity. 86 It is tempting to speculate that TNT secretion may enable the pathogen to fine-tune the timing of macrophage death.
Collectively, the literature supports a model whereby M. tuberculosis dictates the mode, and most likely the timing, of death of their host cells, in order to prevent mycobacterial killing, blunt the immune response, and disseminate.

CONCLUSIONS
Given the enormous body of literature concerning the manipulation of host signaling by M. tuberculosis, we have attempted to summarize the most recent gains in our understanding of host-pathogen interactions in TB. The pathogen has at its disposal an assortment of virulence factors that display incredible breadth in terms of host cellular targets. It is pertinent to bear in mind, however, that the human host is by no means a defenseless counterpart in the pathogenic process. On the contrary, we possess a potent repertoire of countermeasures that, in the majority of individuals, are reasonably successful in at least containing the infection. It is interesting to note that the individual deletion of many of the virulence factors of M. tuberculosis causes only a mild or no effect on disease in vivo. This provides qualified support for the notion that many of these virulence factors play redundant roles during infection, perhaps because they target the same host signaling pathways. Speculatively, this redundancy may have evolved in response to the multifaceted defense systems of the host. Hence, deciphering how M. tuberculosis thwarts host signaling to cause disease is imperative, not only in paving the way for novel therapeutic targets, but perhaps in beginning to understand why some individuals are more susceptible to active TB disease than others.

DISCLOSURES
The authors declare no conflict of interest.