Membrane atg8ylation and autophagy in protection against Mycobacterium tuberculosis

Abstract

Membrane atg8ylation is a broad homeostatic process of immunological import. It encompasses membrane repair and remodeling pathways, including canonical autophagy, in cells subjected to stress, damage, infection, and immune or metabolic signaling under microbe-induced or sterile inflammatory conditions. The initial reports on autophagy, which is one of membrane atg8ylation outputs, as a defense against Mycobacterium tuberculosis and other intracellular pathogens have ushered a new direction for immunological research but proved to be controversial once the studies have moved from in cellulo to in vivo studies in murine models. Recent research is beginning to resolve these controversies by revealing that membrane atg8ylation in general is key to host protection against M. tuberculosis. These developments inform us of how membrane atg8ylation and autophagy shape the innate and adaptive immunity against pathogens and invite further studies to identify downstream immunological effector mechanisms.

Introduction

The process of membrane atg8ylation (1) is akin to protein ubiquitylation and represents reversible covalent modification of cellular membranes by mammalian ATG8 proteins (mATG8s) in response to membrane stress, damage, and signals encountered during conditions of sterile inflammation or infection (Fig. 1A) (2, 3). One of its presentations, canonical autophagy, is a well-established immune mechanism with primary roles in inflammation (4–7), cell-autonomous defense against microbes (8–15), and other aspects of innate and adaptive immunity (5, 16–18). The immunological manifestations of membrane atg8ylation processes other than canonical autophagy represent an area of emerging growth (3, 19–26) (Fig. 1B). In this brief review, we use the example of Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis (TB), which is according to the World Health Organization “the world’s top infectious killer” (27), to discuss the journey (Table 1) from the initial view that autophagy is a key host defense mechanism against it to the realization that the anti-TB defense functions may rest with other manifestations of membrane atg8ylation.

Figure 1.

Membrane atg8ylation and its different manifestations with immunological outputs of consequence for M. tuberculosis infection and tuberculosis disease control. A. Membrane atg8ylation and deatg8ylation cycle. PE/PS, phosphatidylethanolamine or phosphatidylserine. ATG8s: LC3A, LC3B, LC3B2, LC3C, GABARAP, GABARAPL1 and GABARAPL2. Note that ATG7 is an E2 enzyme and that ATG16L1 is a component of E3 ligases that specifies the location for membrane atg8ylation. B. Different intracellular presentations of membrane atg8ylation: VAIL (V-ATPase-ATG16L1-induced LC3 lipidation; V-ATPase is an adaptor for ATG16L1-containing E3 ligase), CASM (conjugation of ATG8 to single membranes), canonical autophagy (WIPI2b is an adaptor for ATG16L1, replacing V-ATPase), membrane repair, AMLaS (atg8ylation of multi-lamellar membranous structures), STIL (sphingomyelin-TECPR1-induced LC3 lipidation; TECPR1 is an alternative component for an E3 ligase replacing ATG16L1 and recognizing ectopic sphingomyelin), microautophagy, LAP (LC3-associated phagocytosis), LAM (LC3-associated micropinocytosis), LANDO (LC3-associated endocytosis), secretory autophagy (of multiple sub-pathways only one is depicted) and EVAC (V-ATPase-ATG16L1-LC3C). Squares labeled “Mtb”, membrane atg8ylation processes linked to immunological control of Mtb or pathogenesis (as in the case of PMNs, polymorphonuclear leukocytes/neutrophils).

Table 1.

Progression toward the current understanding of the roles of membrane atg8ylation, autophagy, and unique functions of certain Atg genes in control of Mtb

Study id	Authors	Year	Finding	Comments	Ref. #
a.	Sturgill-Koszycki et al.	1994	Mtb phagosomes excluded V-ATPase	A landmark in cellulo study that links with the recent V-ATPase studies on membrane atg8ylation	(143)
b.	Gutierrez et al.	2004	Autophagy controlled virulent Mtb in macrophages	In cellulo demonstration of autophagy as a cell-autonomous antimicrobial defense	(8)
c.	Watson et al.	2012	Atg5 protected mice against Mtb	In vivo demonstration that Atg5 protects against Mtb	(110)
d.	Castillo et al.	2012	Atg5 protected mice against Mtb	In vivo Atg5 protects against Mtb + plays a role in neutrophilic inflammation and IL-17	(109)
e.	Kimmey et al.	2015	Atg5 protected against Mtb but not the other autophagy genes tested	In vivo + in depth neutrophil role; casts doubts re autophagy as a pathway protecting against Mtb	(111)
f.	Koster et al.	2017	LAP (a form of membrane atg8ylation) protected against mutant Mtb	In vivo role for LAP protecting against CpsA mutant but not against fully virulent Mtb	(21)
g.	Jia et al.	2022	Membrane atg8ylation recruited stress granule proteins upon Mtb-induced damage	In cellulo, endolysosomal damage induces membrane atg8ylation and recruits stress granule proteins to organelles	(24)
h.	Bussi et al.	2023	Stress granules plugged holes on damaged endomembranes	In cellulo, stress granules protect from Mtb-caused damage	(142)
i.	Golovkine et al.	2023	Mice with a more complete inactivation of Atg7 and Atg16l1 were sensitized to Mtb	In vivo, clearly establishes a role for the atg8ylation genes Atg5, Atg16l1, and Atg7 in protection against Mtb	(113)
j.	Aylan et al.	2023	In human induced pluripotent stem cell-derived macrophages, ATG7 controlled Mtb	In cellulo, ATG14 controlls Mtb in phagosomes through a proposed non-autophagic role	(119)
k.	Kinsella et al.	2023	In vivo & in vitro, canonical autophagy in lung macrophages and DCs controlled inflammation but not the Mtb burden	Atg5 exceeds effects of Becn1 and Atg16l1 in neutrophil recruitment; autophagy-independent role of Atg5 proposed	(114)
l.	Wang et al.	2023	Atg5 in vivo & in cellulo showed a specific and unique role in neutrophils not extending to other Atgs	In vivo and in cellulo studies show that Atg5 loss triggers an unconventional secretory pathway leading to neutrophil activation and degranulation	(25)
m.	Mittal et al.	2024	None of the genes tested (Atg5, Atg7, Atg14) mattered for survival of mice infected with virulent Mtb	In vivo, Atg5, Atg7 (atg8ylation), and Atg14 (canonical autophagy) protect mice against mutant Mtb missing the lipid PDIM	(117)
n.	Feng et al.	2024	Canonical autophagy Fip200, Becn1, Atg14 and membrane atg8ylation genes Atg16l1 and Atg7 protected against high dose Mtb infection	In vivo, canonical autophagy protects against high dose Mtb infection. Amends the conclusions of the previous study by Kimmey et al. 2015	(115)
o.	Paddar et al.	2025	ATG5 interacted with retromer and controlled plasma membrane glucose transporter GLUT1	In cellulo, absence of ATG5 traps GLUT1 in the cytoplasm. In vivo, loss of Atg5 does not affect Mtb reactivation in a latency model of TB	(120)
p.	Javed et al.	2025	Myeloid inactivation of Atg9a did not additionally sensitize mice to Mtb, low or high dose infection	In vivo, Atg9a, a key canonical autophagy gene not involved in atg8ylation, has no effect on Mtb infection. However, its loss in myeloid cells increases spontaneous mortality	(59)
q.	Chen et al.	2025	A novel presentation of membrane atg8ylation reported with tubulovesicular structures (TVS) in the proximity of Mtb phagosomes.	In cellulo, membrane atg8ylation of TVS depends on V-ATPase and ATG16L1 and suppresses Mtb replication	(53)
r.	Duque et al.	2025	Mice with a truncated version of Atg16l1E230 that fully supports canonical autophagy but not V-ATPase dependent atg8ylation succumbed to Mtb	In vivo demonstration that membrane atg8ylation but not canonical autophagy is key to protection against Mtb, low or high infectious dose; additionally, ATG16L1 has a unique role in control of V-ATPase assembly and activity	(26)

Membrane atg8ylation processes include canonical autophagy

The mATG8s and ubiquitin are closely related and activated via dedicated conjugation cascades involving ATP, E1, E2 and E3 ligases resulting respectively in protein ubiquitylation or membrane atg8ylation (1). The specific factors leading to membrane atg8ylation include two enzymatic cascades with ATG12-ATG5 and mATG8-phosphatidylethanolamine (PE) or less frequently mATG8-phosphatidylserine (PS) conjugates as their end products (28, 29). The ATG12-ATG5 protein-protein conjugate combines with additional proteins to form E3 ligases (23, 28, 30–32) to guide mATG8-lipid conjugation resulting in atg8ylation of specific membrane domains.

There are seven mATG8s in humans: LC3A, LC3B, LC3B2, LC3C, GABARAP, GABARAPL1 and GABARAPL2 and their membrane conjugation was historically referred to as ‘LC3 lipidation’ and ‘LC3 puncta’ formation (28, 33). Initially, LC3B was widely used to monitor autophagosome formation (28, 33), however it has become evident that ‘LC3’ (i.e. mATG8s) can be present on other organelles (34). These ‘noncanonical autophagy’ processes (1) include phagosomes harboring microbes (19, 22, 23, 29, 35, 36) and pathogen associated molecular patterns (PAMPs) or danger associated molecular patterns (DAMPs) in endolysosomal compartments or in the cytosol (24, 31, 35, 37–42), immunologically or otherwise elicited extracellular vesicles and exosomes in a process collectively referred to as ‘secretory autophagy’ (20, 25, 43, 44), and stressed endoplasmic reticulum (45) or lipid droplets (46). They have become known by various acronyms (Fig. 1B): LAP (LC3-associated phagocytosis) (19), LANDO (LC3-associated endocytosis) (47), LAM (LC3-associated micropinocytosis) (37), CASM (conjugation of ATG8 to single membranes) (48, 49), VAIL (V-ATPase-ATG16L1-induced LC3 lipidation) (22, 36, 39, 50), and EVAC (V-ATPase-ATG16L1-LC3C) (45), with most recent additions being STIL (sphingomyelin-TECPR1-induced LC3 lipidation) (51), HALD (hemi-layer atg8ylation on lipid droplets) (46, 52), and AMLaS (atg8ylation of multi-lamellar membranous structures) (53). Many of the above processes are elicited by microbial-induced membrane damage or mediators of sterile inflammation and play a role in immunity (3).

Until recently, the most frequently studied form of membrane atg8ylation in immunity has been canonical autophagy (54, 55) – from cell-autonomous defense against intracellular microbes to other aspects of innate immunity, immunometabolism, and development and function of adaptive immunity systems (5). Canonical autophagy depends on the formation in the cytoplasm of the specialized organelles called autophagosomes (54), morphologically distinguished by their double membranes, which capture intracellular targets (protein aggregates, defunct organelles, invading microbes) leading to their elimination. The specific factors that differentiate canonical autophagy apparatus from other forms of membrane atg8ylation are the protein complexes ULK1- ATG13-FIP200 (56) and ATG2A/B-ATG9A (57), which together initiate phagophore formation (58) and its expansion via lipid transfer (57) to the nascent autophagosomal membranes (phagophores). ATG9A also plays a role in closing the phagophores to generate double membrane autophagosomes (59), which sequesters the cargo and delivers it to lysosomes for degradation. Whereas canonical autophagy involves membrane atg8ylation, autophagosomes can form in cells devoid of mATG8s or atg8ylation cascade (60–63), albeit the frequency and kinetics of their formation (64), their size (61), contents (63), quality (62) and maturation into autolysosomes (65, 66) are affected. Furthermore, selective autophagy implicated in microbial clearance of bacteria, viruses and protozoa via their capture into autophagosomes (9, 10, 15, 30, 67–83) depends on binding of the sequestosome-like receptors (SLRs) to the atg8ylated phagophores via SLRs’ LC3-interacting regions (LIR) (84). SLRs such as SQSTM1 (p62), CALCOCO2 (NDP52), CALCOCO3 (TAXBP1), OPTN (Optineurin) and possibly other selective autophagy receptors play a role in elimination of intracellular microbes by: (a) their association with ubiquitylated macromolecules on pathogens’ surface; (b) their binding to mATG8s on atg8ylated autophagic membranes (13–15); and (c) organizing the canonical-autophagy specific factors FIP200, ATG13, ATG1, and ATG9A (85). They may also have a role in noninfectious inflammatory diseases such as neurodegeneration (86, 87), cancer (88), and nephropaties (89). Of note, studies in the past have not differentiated canonical autophagy from other membrane atg8ylation processes, but regardless, some form of autophagy or membrane atg8ylation impacts general immune process (4, 16, 90) including shaping adaptive immune responses (16–18, 91–93) and immunometabolism (5, 94–96).

In cellulo studies of membrane atg8ylation in microbial pathogenesis

The initial report (Table 1b) on the control of intracellular Mtb in infected macrophages by nutritionally- or IFN-γ−induced autophagy (8, 12) has played a pivotal role leading to the currently prevailing notion that autophagy is a cell-autonomous antimicrobial process (11, 13, 97). This initial work and many of the follow-up studies have been carried out in cultured cells infected with viruses, bacteria or other microbes (9, 10, 53, 69, 71, 72, 78, 83, 98–101). These studies have demonstrated that SLRs target intracellular bacteria such as Salmonella, Shigella, Listeria, and others, leading to their capture and delivery to autolysosomes for degradation (13, 97). This form of selective autophagy resembles elimination of mitochondria via a specialized canonical autophagy process termed mitophagy. Mitochondria have evolved from a Rickettsia-like endosymbiont, thus parallels between capture of intracellular bacteria and mitochondria may have common evolutionary roots (14, 102, 103). In conclusion, selective canonical autophagy of microbes is a process that manifests itself clearly in infected cells studied ex vivo/in vitro.

In vivo studies of autophagy and atg8ylation: infection vs. sterile inflammation

Initially, the in vivo studies of microbial infections have not discriminated canonical autophagy vs. other forms of membrane atg8ylation. The data indicated that Atg5 (80), and then Atg7, Atg3, as well as the whole membrane atg8ylation E3 ligase Atg12-Atg5-Atg16L1 (82) control Toxoplasma gondii in mice. The Atg16l1 alleles mimicking human polymorphisms associated with Crohn disease in humans (104) affect bacterial clearance and inflammation in mice (81). Atg5 also protects mice against Sindbis virus (68). A hyperactive ATG16L1 (rs6861 polymorphism) protects again HIV-1 in human populations (74). Deletion of Atg5 in T cells is associated with increase in IL-17A + IFN-γ + Th17 T cells which drive liver fibrosis in mice (105). Of note, the above genes are all specific for membrane atg8ylation in general and their inactivation does not distinguish canonical autophagy form other outputs of membrane atg8ylation,

Further examples from in vivo modeling also show that loss of membrane atg8ylation or specifically of canonical autophagy can paradoxically confer nonspecific protection against certain microbes due to heightened background inflammation. Such studies contradict the notion that autophagy is a protective antimicrobial mechanism. There are examples whereby loss of canonical autophagy genes but not membrane atg8ylation genes creates hyperimmune state protective against Listeria monocytogenes (106). Other compensatory mechanisms in myeloid cells lacking Atg16l1 enhance elimination of Shigella flexneri in mice (76). Loss of either membrane atg8ylation genes or canonical autophagy-specific genes paradoxically inhibits herpes reactivation in mice infected with murine herpesvirus by artificially creating a hyperinflammatory state (107). These examples can be best interpreted by general anti-inflammatory nature of both membrane atg8ylation and canonical autophagy (5, 108). However, this is primarily manifested when probed with infectious agents. In contrast, when it comes to chronic sterile inflammation, canonical autophagy can be pro-inflammatory (6). In conclusion, membrane atg8ylation in general and canonical autophagy as one of its downstream pathways modulate inflammatory processes.

The journey from canonical autophagy to membrane atg8ylation in control of Mtb

In the case of virulent Mtb, it has been reported in 2012 using Atg5^Fl/Fl LysM-Cre mice (Table 1c,d), with inactivated Atg5 in the myeloid lineage, that this gene, which does not discriminate between canonical autophagy and other manifestations of membrane atg8ylation, affords protection against TB in mouse models (109, 110). A heightened Th17, IL-17 and neutrophilic inflammation have been reported in Atg5^Fl/Fl LysM-Cre mice infected with virulent Mtb (109). The role of Atg5 and neutrophils was affirmed in subsequent studies (111) (Table 1e). Although the initial interpretation was that autophagy is responsible for protection (109, 110), it has been cautioned in 2014 (112) that a related but fundamentally different membrane atg8ylation process may contribute to the control of TB. An influential study was subsequently published in 2015 (111) suggesting that it is not the pathway of canonical autophagy that protects against TB while at the same time confirming the initial reports (109, 110) that Atg5 protects against Mtb.

These studies (Table 1b–e) raised several questions: (i) whether canonical autophagy plays at all a role in control of TB; (ii) is it only ATG5 that is important and what are the processes that it uniquely controls; (iii) do membrane atg8ylation processes other than canonical autophagy afford protection against Mtb. At first, Stallings and colleagues appropriately argued that autophagy does not protect against Mtb in mice (111) albeit this position was revised with further experimentation. In their tour de force study, Cox and colleagues have performed a more complete inactivation of Atg5, Atg7 and Atg16l1 genes in mice and clearly demonstrated that these genes are protective against Mtb in a chronic, low infection dose model (113). In their subsequent elegant studies, Stallings and colleagues observed effects of Atg16l1 and Becn1 in a standard low dose infection model (114), and confirmed this in an acute infection murine model using high dose inoculum (1,000 cfu) (115). Thus, even though not a complete canonical autophagy pathway was required (111), some form of membrane atg8ylation played a role in protection against Mtb in vivo (113, 114) (Table 1i,k).

The studies by Philips and colleagues (Table 1f,m) using disabled strains of Mtb with inactivated genes cpsA (21, 116) or ppsD (117) are vulnerable to LAP (one of the membrane atg8ylation manifestations; Fig. 1B) or to a form of autophagy, depending upon the mutant Mtb used (117). In another study, Stallings and colleagues tested the key LAP-defining gene, Rubicon (Rubcn^−/− mice), and found no effects against fully virulent Mtb potentially ruling out LAP as a dominant mechanism (115) (Table 1n). This however does not exclude contributions of LAP, which manifests its potency against attenuated Mtb strains (21, 117, 118) (Table 1f,m). In their studies, Mittal et al., used Atg5, Atg7 and Atg14 to define the process as canonical autophagy (117) but Atg14 was recently shown to play functions other than canonical autophagy (119) (Table 1j), and thus Atg14 cannot be used to qualify a process as canonical autophagy. A series of recently published findings (25, 26, 59, 120) (Table 1o,p,r,l) with fully virulent Mtb in murine models backed up by mechanistic ex vivo studies concur with the notion that the exact protective membrane atg8ylation process remains to be fully characterized at the cell biological and immunological levels.

Immune manifestations of protection against Mtb by membrane atg8ylation genes

The precise role in Mtb control by membrane atg8ylation, canonical autophagy, or individual membrane atg8ylation or canonical autophagy genes is not completely understood. A part of the issues is uncertainty regarding the exact processes controlled by the ATG genes.

Nevertheless, at the immunological level, progress has been made in understanding how ATG genes affect host responses to Mtb infection. A prolonged Th17 and persistent neutrophilic infiltration have been reported for Atg5^Fl/Fl LysM-Cre infected with virulent Mtb (109, 111) (Table 1d,e). The Th17 and neutrophilic inflammation in conditional Atg5 mutant mice was further dissected using Cre drivers specific for different myeloid cell types (111, 114). Comparing Atg5^fl/fl LysM-Cre (loss of Atg5 in myeloid lineage), Atg5^fl/fl Mrp8-Cre (loss of Atg5 in neutrophils) and Atg5^fl/fl Cd11c-Cre (loss of Atg5 in dendritic cells (DCs) and in lung but not all macrophages) Kinsella et al., have shown that Atg5^fl/flLysM-Cre and Atg5^fl/fl Cd11c-Cre but not Atg5^fl/fl Mrp8-Cre mice accumulated neutrophils at 14 days post-infection (dpi) with no changes in bacterial burden (114). Nevertheless, one week later at 21 dpi, Atg5^fl/fl LysM-Cre mice showed increased Mtb bacillary burden in the lungs (114). In contrast, at that time point the Atg5^fl/fl Cd11c-Cre mice did not show increase in Mtb bacillary burden in the lungs whereas a subset of Atg5^fl/fl Mrp8-Cre mice termed ‘susceptible’ did (114). The authors have concluded that Atg5 in lung macrophages and DCs plays a role in regulating initial neutrophilic infiltration whereas Atg5 in neutrophils controls their accumulation and overall bacterial burden (114). Of note, myeloid/neutrophilic granulomas are a hallmark of progression to active TB disease in animal models (121) and a strong neutrophilic signature is observed in TB patients (122). Thus, Atg5 is important for both neutrophil recruitment and neutrophilic function as it pertains to control of active TB. In contrast to active disease, loss of Atg5 in myeloid lineage did not affect reactivation from experimentally induced TB latency in mice (120) (Table 1o).

Further studies have indicated that ATG genes affect the dissemination of Mtb from alveolar macrophages and the airways, which contributes to the generation of adaptive immune response (116) and B-cell accumulation in lymphoid follicles in mice (117) (Table 1m). Such B cells may play an important role. Antigen-specific B cells contribute to Mtb control in mice and non-human primates (123, 124). They do so by directing formation of better organized granulomas associated with lymphoid follicles within tertiary lymphoid organs in macaques and outbred mice (albeit not in inbred strains of mice used in transgenic work) thus contributing to Mtb control (123, 125). This contrasts with neutrophilic granulomas that are less capable of controlling infection and lead to active disease (121). At least some of these aspects such as levels of B-cells in lymphoid follicles depend in vivo on LAP or other forms of membrane atg8ylation when using attenuated, CpsA-defective or phthiocerol dimycocerosate (ppsD)-defective Mtb (21, 117). These effects are at least partially masked when fully virulent Mtb is tested (115).

In conclusion, membrane atg8ylation genes affect various stages of active TB (109, 110, 126)(21, 111, 113–115, 117) (Table 1c–f,i,k,m,n). However, at least in the case of Atg5, this may be less linked to TB latency (120) (Table 1o) with this aspect remaining to be fully explored.

Cellular mechanisms of protection against Mtb

Of note, in all mouse Mtb infection studies, loss of Atg5 had the most pronounced effect (25, 113) (Table 1i,l) relative to other membrane atg8ylation genes even as others, such as Atg7 and Atg16l1, have been demonstrated to play a role (113). Thus, ATG5 must have additional functions. At the cell biological level, most recent studies have linked ATG5 to a unique hyperactivation and hypersecretory phenotype in neutrophils (25) as well as in regulating glucose transporter GLUT1 (120). The intrinsic neutrophil hyperactivation phenotype in the absence of Atg5 manifests itself as increased presence of CD11b on bone marrow-derived neutrophils (BMN), a key component of the Mac-1/CR3 integrin which promotes neutrophil migration to sites of inflammation, and other markers of excessive BMN degranulation such as elevated extracellular elastase released by BMNs from Atg5^Fl/Fl LysM-Cre mice upon in vitro stimulation (25). Mechanistically, in the absence of ATG5 an anomalous conjugation between ATG12—ATG3 occurs leading to increased exocytosis and degranulation of neutrophils (25) (Table 1l). ATG12—ATG3 is an alternative conjugation product formed when ATG12 cannot be conjugated to its usual partner ATG5 (127, 128). ATG12—ATG3 drives secretory autophagy (Fig. 1B) via its interactions with ALIX, a component of the ESCRT system known to promote secretion of extracellular vesicles (43). In addition, ALIX is recruited away by this complex from lysosomal organelles, where ALIX contributes to the maintenance of lysosomal membranes, which in turn leads to excessive lysosomal exocytosis and overexuberant neutrophil degranulation (Fig. 1B) (25). These cell-intrinsic phenotypes that are specific to Atg5 are not manifested by other canonical autophagy or membrane atg8ylation genes (25) and likely contribute to excessive neutrophilic inflammation in mice lacking Atg5 in the myeloid lineage.

Additionally, ATG5 interacts with the intracellular trafficking protein complex called retromer (120), which has important function in glucose transport via GLUT1 (129). The glucose transporter GLUT1 trafficking is perturbed in ATG5 knockout cells (120), and is likely to affect multiple functions due to the heavy immunometabolic dependence of neutrophils on glucose and glycolysis (5, 130). Although ATG5 knockout had the strongest effect on GLUT1 trafficking (Table 1o), inactivation of other membrane atg8ylation (but not canonical autophagy) genes affected it too (120), indicating that membrane atg8ylation integrates with immunometabolism via glucose transport. Of note, global diabetes and obesity epidemic with associated glucose uptake dysregulation has been a factor in worldwide prevalence of active TB (131, 132). In conclusion, in addition to its essential role in membrane atg8ylation, ATG5 possesses unique functions affecting neutrophil activation and glucose transport.

Membrane atg8ylation in protection against Mtb-induced damage

Gaining further insight, recent in cellulo studies indicate that canonical autophagy is not engaged with Mtb phagosomes (133). Another report showed that membrane atg8ylation is important in control of Mtb in infected macrophages studied in vitro and that neither canonical autophagy nor LAP were involved (53). Instead, V-ATPase atg8ylation of multi-membranous structures/AMLS in the vicinity of Mtb phagosomes termed tubulovesicular structures plays a role in membrane repair that may affect Mtb replication (Fig. 1B) (53). Endolysosomal membrane damage is induced by various agents (134–136) including Mtb (24, 137–140). Specific repair mechanisms of endolysosomal organelles have been linked to membrane atg8ylation (Fig. 1B) (24, 141). Membrane atg8ylation recruits lipid transfer protein ATG2 to damaged lysosomes (141), a key effector of membrane repair acting independently of its role in canonical autophagy (134). Membrane atg8ylation also recruits components of stress granules (24), which, among other effects, plug holes on endolysosomal membranes and may be protective against Mtb shown ex vivo in infected macrophages (142). Thus, membrane atg8ylation and membrane repair play an important role in protection against Mtb-induced damage.

Membrane atg8ylation and V-ATPase

The defect in acidification of Mtb phagosomes has been linked to active exclusion of V-ATPase (proton pump) from vacuoles harboring tubercle bacilli as a key to intracellular survival of Mtb (143) (Table 1a). V-ATPase is a physiologically critical enzyme responsible for acidification of intracellular compartments in the endolysosomal system as well as proton secretion by specialized cells (144). V-ATPase consists of two domains, V₀ and V₁. V₀ proteolipid is integral to membranes whereas V₁ domain can be free in the cytosol or associated with V₀ (144). A principal mechanism for regulating V-ATPase activity is the assembly and disassembly of the V₀ and V₁ domains, which when assembled into the V₁V₀ holoenzyme pumps protons across membrane while hydrolyzing ATP (144). When the primary function of V-ATPase is to acidify the lumen of endolysosomal organelles, including phagosomes, it also has another function whereby it serves as an adaptor for atg8ylation of stressed, damaged or ion-leaking membranes by ATG12—ATG5-ATG16L1 E3 ligase. In its role of an adaptor, the V1 domain of V-ATPase directly binds ATG16L1 thus bringing the membrane atg8ylation apparatus to target organelles: phagosomes or endosomes for VAIL (22, 38, 49, 145) or ER exit sites (45), and Golgi membranes for EVAC (146) or Mtb phagosome-associated tubulovesicular structures for AMLaS (53). The most recent studies (26, 147) (Table 1r) have shown that association of ATG16L1 and the V1 domain of V-ATPase has reciprocal functions. Whereas V1 recruits ATG16L1 to damaged membranes (22, 38, 49, 145) to carry out atg8ylation leading to membrane repair (24, 141), ATG16L1’s binding to V-ATPase inhibits its function including the pump’s ATP hydrolysis activity and promotes the disassembly of the V₁ and V₀ domains (26). In the absence of ATG16L1, fully assembled V-ATPase on damaged membranes performs futile ATP hydrolysis, unnecessarily pumping protons since leaky membranes cannot hold H⁺ gradient, and exhausts energy charge in the cells (26). Importantly, mice homozygous (whole body) for a mutant form of ATG16L1 (Atg16l1^E230) that can fully support canonical autophagy but cannot bind and inhibit V-ATPase are highly susceptible to Mtb infection at any (low or high) dose of aerosol inoculum (26). Mechanistically, several processes, i.e., low energy charge (26) (Table 1r) and inability to carry out AMLaS (53) (Table 1q) in the vicinity of Mtb phagosomes, may contribute to the sensitivity of Atg16l1^E230 mice to Mtb observed in the murine aerosol infection model of TB (26). A further, potentially critical function, is that ATG16L1-directed membrane atg8ylation recruits factors such as DMXL1 via its LIR domain which in turn bring together the V₁ to V₀ domains (147) and assemble a functional V-ATPase thus resuming normal luminal acidification and restoring functional phagosomes and lysosomal organelles.

Conclusions

The concept of membrane atg8ylation in immunity can be viewed as a “version 2.0” of an increasingly complex field that is hard to fit under the catchall term “autophagy”. Whereas canonical autophagy, which is one of the many manifestations of membrane atg8ylation, can be clearly demonstrated in vitro in infected cells as a mechanism eliminating various intracellular microbes via selective autophagy, the immune defense role of canonical autophagy as a pathway is far more difficult to prove in vivo. This is well-illustrated in the studies focused on TB (148). To recap, the in vivo studies with Mtb in mouse models by independent groups of investigators now indicate that canonical autophagy plays only a minor role in immunological control of TB (59, 109, 111, 114, 115). Instead, other membrane atg8ylation processes are of significance (21, 25, 26, 113). The Mtb infection studies with Atg16l1^E230 mice (26) definitively demonstrate that canonical autophagy does not play a key role in control of Mtb in the murine model of TB. This conclusion is reinforced by the in vivo studies with Atg9a^fl/fl LysM-Cre mice (59) (Table 1p). ATG9A is a uniquely canonical autophagy factor – it is the only integral membrane core autophagy factor that controls all stages of canonical autophagosome formation and completion (149). In the infection model, Atg9a^fl/fl LysM-Cre mice are not sensitized to Mtb (59). However, Atg9a^fl/fl LysM-Cre mice show spontaneous mortality (59) due to spontaneous increase in type I interferon signaling and autoimmune disease (150), which is not further exacerbated by Mtb infection (59). Instead of canonical autophagy, membrane atg8ylation as an overarching pathway with multiple manifestations including membrane repair, combined with the special functions of ATG16L1 in the control of V-ATPase and the unique functions of ATG5 in neutrophilic inflammation are key determinants of host defense against TB infection. The murine modeling studies (Table 1) in TB have been essential in uncovering these processes. Similarly extensive in vivo studies remain to be applied to other infectious agents where autophagy has been implicated in vitro.

Acronyms

AMLaS

Atg8ylation of multi-lamellar membranous structures

CASM

Conjugation of ATG8 to single membranes

DAMPs

Danger associated molecular patterns

DC

Dendritic cells

EVAC

V-ATPase-ATG16L1-LC3C

HALD

Hemi-layer atg8ylation on lipid droplets

Mtb

Mycobacterium tuberculosis

LAM

LC3-associated micropinocytosis

LANDO

LC3-associated endocytosis

LAP

LC3-associated phagocytosis

PAMPs

Pathogen associated molecular patterns

PE

Phosphatidylethanolamone

STIL

Sphingomyelin-TECPR1-induced LC3 lipidation

TB

Tuberculosis

VAIL

V-ATPase-ATG16L1-induced LC3 lipidation

Data availability statement

All data reviewed in this article are published in original articles as referenced in the text and the authors of those studies are responsible for data availability.