Autophagy: An Emerging Immunological Paradigm

In this Brief Review, we cover the immunological roles of macroautophagy (1), a specific autophagic process that will be referred to herein as sensu stricto autophagy or simply autophagy. Autophagy is unique in its capacity to sequester, remove, or process bulk cytosol, cytoplasmic organelles (1), invading microbes, and immunological mediators (2), as depicted in Fig. 1. Another special property illustrated in Fig. 1 is that autophagy acts as a topological inverter—bringing molecules and objects from the cytosolic side to the luminal side for degradation or processing, interaction with luminal receptors, or secretion from cells. In this review, we discuss the four principal manifestations of immunological autophagy (Fig. 1): 1) direct pathogen elimination assisted by sequestosome 1-like receptors (SLRs); 2) regulation and effector functions of pattern recognition receptors (PRRs); 3) regulation of inflammasome activation and alarmin secretion; and 4) cytoplasmic Ag processing for MHC II presentation and T cell homeostasis. We relate these processes to conventional immunological functions, defense against infectious agents, chronic inflammatory diseases, and other immunological disorders.

The key morphological features of autophagy are endomembranous organelles, called autophagosomes (Fig. 1), whose formation is controlled by the autophagy-related gene (Atg) and additional factors comprehensively reviewed elsewhere (1). Briefly, the Atg system includes Ser/Thr kinases Ulk1 and Ulk2 (Atg1), Beclin 1 (Atg6; a subunit of the class III PI3K human vacuolar protein sorting 34 [hVPS34 complexes]), Atg5–Atg12/Atg16L1 complex, and microtubule-associated protein L chain 3s (LC3s) (multiple Atg8 orthologs), with LC3B being a commonly used marker for identification of autophagosomes (1). Ulk1/2 and Beclin 1-hVPS34 integrate upstream signals and direct the downstream Atg conjugation cascade, which involves Atg5–Atg12/Atg16L1 assembly as an “E3 enzyme” for LC3 lipidation. Lipidated LC3s in conjunction with other factors assemble, elongate, and close nascent autophagic organelles. Autophagosomes interact with endosomal and lysosomal organelles to mature into autolysosomes (1) or promote unconventional secretion of cytoplasmic constituents, as first demonstrated in yeast (3) and recently shown to include immune mediators (4, 5). In addition to its immunological functions (2), autophagy plays a general cellular homeostatic role by supplying nutrients (e.g., amino acids) through cytosol autodigestion at times of starvation or growth factor withdrawal, and serves as a quality and quantity control mechanism for intracellular organelles (1).

At the transcriptional level, regulation of autophagy is coupled to the lysosomal system via TFEB (6) and other proteolytic systems via FoxO3A (7). However, autophagy is primarily a rapid-response remodeling of membranes that occurs in the cytoplasm under the control of signaling systems faster than transcriptional changes. The classical nutritional/energy regulation of autophagy is via mTOR and AMP-activated protein kinase (AMPK) inhibiting and activating, respectively, Ulk1/2 (1). This pathway merges with signaling via the inhibitor of NF-κB kinases (IKK), frequently involved in immune signaling. IKKα and IKKβ transduce the classical signal for autophagy induction—starvation (8)—but this signaling is not based on nuclear NF-κB responses. Instead, IKK and AMPK signaling merge via TGF-β–activated kinase 1 (TAK1) and its activators TAB2 and TAB3. Upon autophagy induction, TAB2 and TAB3 dissociate from and thus activate Beclin 1 and also bind to and activate TAK1 (8), whereas TAK1 in turn phosphorylates and activates AMPK.

In T cells, autophagy is activated upon TCR engagement and CD28 costimulation and supports their effector functions and proliferation (9). Recently, class III PI3K hVPS34 was found to be dispensable for autophagy induction in T cells, albeit required for T cell homeostasis via its regulation of receptor endocytosis (10), bringing up the possibility of alternative pathways in PI3P signaling, as suggested by the positive role of class I PI3K p110β (11).

Innate immune signaling can induce autophagy. TRAF6 downstream of TLR4 activates autophagy (12). Alarmins or damage-associated molecular patterns (DAMPs) induce autophagy (13, 14). High-mobility group protein B1 (HMGB1), an alarmin, undergoes translocation from the nucleus into the cytoplasm and then out of the cells by unconventional secretion (5, 15) or cell death-associated release, inducing autophagy at each stage: cytoplasmic through de-repression of Beclin 1 by displacing its negative regulator Bcl-2, or extracellularly via RAGE signaling (13, 14). In addition to HMGB1, DAMPs such as ATP, IL-1β, and DNA complexes are known to induce autophagy (reviewed in Ref. 16).

From an evolutionary perspective, the most primal manifestation of immunological autophagy is direct capture and degradation of invading intracellular microbes by autophagy (Fig. 1, panel 1, left). This cell-autonomous defense function of autophagy is often countered by microbial adaptation mechanisms, and a number of highly adapted pathogens can convert autophagic organelles into growth-supporting compartments (17). Autophagic capture of intracellular microbes is facilitated by autophagic adaptors, referred to as sequestosome 1/p62-like receptors (SLRs) (18). SLRs have LC3 interacting regions and cargo-tag (e.g., ubiquitin) recognition domains and are modulated by protein kinases. Salmonella requires multiple SLRs [p62, nuclear domain 10 protein/Ag nuclear dot 52 kDa protein (NDP52), optineurin] (19, 20), phosphorylation of at least one of the SLRs (optineurin) with an IKK-related kinase, TBK-1 (20), and an intracellular DAMP receptor (galectin 8) (21). The SLRs p62 and NDP52 are also engaged in clearance of Shigella and Listeria (22–24), whereas streptococci are affected by NDP52 (19). Sindbis virus interacts with p62 (25). Candidate E3 ligases contributing to target ubiquitination have been identified in some instances: SMURF1 for sindbis virus (26) and LRSAM1 as a candidate for Salmonella (27).

The most recent player in these processes is galectin 8, a cytosolic lectin binding to β-galactoside glycans. The membrane glycans are normally present only on the luminal side of parasitophorous vacuoles. However, upon membrane damage the glycans come in contact with the cytosol and thus become recognized by cytosolic galectins (21) (Fig. 1, panel 1, hatched square). Galectin 8 is important in restricting Salmonella proliferation, and plays an early role until supplanted by a phase dominated with ubiquitin and ubiquitin-recognizing SLR—NDP52. It appears that the phases and the sequence of recognizing membrane damage could be ushered in by the appearance of diacylglycerol (28), followed by galectin-β–galactoside recognition, followed by NDP52–ubiquitin recognition. Because galectin 8 and NDP52 interact, a sequential action is doubly ensured. Galectin 8 is important for the recruitment of NDP52, as the requirement for galectin 8 to restrict Salmonella proliferation could be bypassed by expressing a fusion hybrid between galectin 3 and NDP52. Galectin 3 per se is not required for restriction, although it is found on Salmonella vacuoles, primarily because it, unlike galectin 8, cannot interact with NDP52. Galectin 8 recognizes host membrane glycans and, indirectly, Salmonella carbohydrates, albeit it can directly recognize blood group B-positive E. coli O86. Galectin 8 is also important for Shigella and Listeria, and can even detect sterile damage to endosomes and lysosomes.

SLRs can also act in a completely different manner to promote autophagic killing of intracellular microbes (Fig. 1, panel 1, right). They gather cytoplasmic proteins (e.g., ubiquitin and ribosomal proteins) to be converted in autolysosomes into antimicrobial products that, upon delivery to cytoplasmic compartments harboring microbes, transform them into autophagolysosomes, organelles with enhanced antimicrobial capacities relative to conventional phagolysosomes (29–31).

Autophagy interacts with classical PRRs, including TLRs, Nod-like receptors (NLRs), and RIG-I–like receptors (RLRs). TLRs and autophagy intersect in two ways, illustrated in Fig. 1, panel 2. First, autophagy is an effector mechanism (e.g., elimination of microbes illustrated in Fig. 1, panel 1) downstream of TLR activation. TLR4 triggers autophagy via TRAF6 E3 ligase, ubiquitination of Beclin 1, and Bcl-2 dissociation from the BH3 domain of Beclin 1 (12). Second, autophagy as a topological inverter device can bring cytosolic pathogen-associated molecular pattern (PAMP) molecules into the lumen, where they can bind the ligand recognition side of the TLR. This has been demonstrated for TLR7 (32), TLR4 ligands (33), and TLR9 in the context of BCR signaling (34).

NLR and autophagy interactions are evolutionarily conserved from Drosophila (35) to humans (16). Nod1 and Nod2 interact with Atg16L1 (36, 37), of significance for Crohn’s disease (CD) because Nod2 and Atg16L1 are risk loci for CD (38). NLRC4 (Ipaf) and NLRP4 inhibit autophagy (39) and are found in macromolecular complexes with Beclin 1. RLRs activate autophagy with biologically important effects (40), but thus far, more attention has been given to negative regulation of RLR signaling by autophagy factors Atg5–Atg12 (41) and Atg9 (42). Atg9 negatively regulates trafficking and activation of TBK-1 in the type I IFN response to dsDNA (42).

Autophagy and inflammasomes interact in two ways (Fig. 1, panel 3). All reports thus far (5, 43–46) agree on the observation that autophagy plays a negative role in inflammasome activation. Autophagy lowers the basal level of inflammasome activation by continually removing endogenous irritants (43, 44). For example, autophagy prevents spurious inflammasome activation by eliminating defunct mitochondria that otherwise represent endogenous sources of inflammasome agonists, such as reactive oxygen species (ROS) and mitochondrial DNA (43, 44) (Fig. 1, panel 3). In the absence of basal autophagy, endogenous factors lead to inflammasome activation and increased IL-1β processing and represent sources of sterile inflammation. This explains how loss of Atg16L1 elevates IL-1β levels in a murine model of CD (47).

On the flip side, autophagy plays a positive (but only acute, short-term) role in delivering outside the cell the effector products of inflammasome activation, such as IL-1β and potentially other alarmins, in a process referred to as the unconventional secretion of IL-1β (5). Although IL-1β and IL-18 do not have signal peptides to deliver them into the lumen of the organelles of the conventional secretory pathway (ER–Golgi–plasma membrane), they are released extracellularly upon inflammasome activation. This is, at least in part, supported by the topological inversion properties of autophagy, ferrying molecules from the cytosolic side into the lumen of putative secretory vesicles. However, this effect wanes quickly with time, and the downregulation of inflammasomes by autophagy becomes dominant once again (46). Thus, autophagy negatively controls inflammasome activation (5, 43–46) and positively controls IL-1β secretion per se (5). The topological inversion action and the positive role of autophagy in secretion of alarmins are not limited to IL-1β and extend to HMGB1 (5).

The role of autophagy as a topological inverter (transport from cytosol to lumen) and its other functions contribute to MHC II presentation of endogenous cytosolic Ags (33, 48, 49) (Fig. 1, panel 4). The physiological role of this is manifested in immune surveillance of viral infections (48) and inhibition of this process by HIV-1 (49). Autophagy-dependent presentation of endogenous Ags plays a part in positive and negative selection of naive T cell repertoires in the thymus (50). It has been hypothesized that peripheral tissue autophagic activities may have to be matched by central tolerance mechanisms dependent on autophagy in the thymus to prevent autoimmunity (50). Autophagy functions in mature T cell homeostasis, and is essential for T cell survival following exit from the thymus, in part based on the requirement for autophagy to physiologically reduce the mitochondrial and ER content in maturing T cells (51–53).

Genetic links between autophagy and chronic inflammatory disorders and autoimmune diseases continue to be uncovered by genome-wide association studies (GWAS). Genetic variations in the PRDM1–ATG5 intergenic region have been associated with rheumatoid arthritis (54). Autophagy specifically favors presentation of citrullinated proteins, which may contribute to autoimmune disorders such as rheumatoid arthritis (55). The initial GWAS linking of ATG16L1 and IRGM [immunity-related GTPase M; a modulator of autophagy (56, 57)] with CD (38) has been replicated in nearly 50 independent population studies. Polymorphisms in another autophagy gene, ULK1, are also associated with CD (58). Genetic associations of CD with IRGM have been extended to IRGM copy number variants in human populations (59). IRGM has further been linked to systemic lupus erythematosus in a recent meta-analysis of autoimmune diseases (60). GWAS in different populations link ATG5 variants to systemic lupus erythematosus (61, 62). This genetic evidence and other studies implicate autophagy in chronic inflammatory diseases and autoimmune disorders.

The initial sporadic observations that autophagy can play a role in cell-autonomous defense against intracellular bacteria such as Mycobacterium tuberculosis (63) and streptococci (64) have been extended in the past several years to various facets of immunity. The connections of autophagy with the normal function of innate and adaptive immunity at almost every level, the genetic and functional associations with immunological disorders, and the unique, specialized mechanisms of autophagy as stand-alone immune processes presented in this article and elsewhere (2) are consistent with the thesis of this review that autophagy represents a new and growing immunological paradigm.

I apologize to colleagues for omissions imposed by space limitations, including microbial defenses against autophagy, nonautophagic functions of the ATG factors, and roles of autophagy processes other than macroautophagy. I thank Carolyn Mold for comments on the text and Dara Elerath for graphic design.

The author has no financial conflicts of interest.