Autophagy as an emerging dimension to adaptive and innate immunity.

Autophagy is an evolutionary conserved cellular process during which cytoplasmic material is engulfed in double membrane vacuoles that then fuse with lysosomes, ultimately degrading their cargo. Emerging evidence, however, now suggests that autophagy can form part of our innate and adaptive immune defense programs. Recent studies have identified pattern recognition molecules as mediators of this process and shown that intracellular pathogens can interact with and even manipulate autophagy. Recent translational evidence has also implicated autophagy in the pathogenesis of several immune-mediated diseases, including Crohn disease. In this review, we present autophagy in the context of its role as an immune system component and effector and speculate on imminent and future research directions in this field.

Introduction

Throughout evolution, facets of our cellular biology have been conserved and adapted. One process at the forefront of homeostasis and environmental interaction is autophagy. This complex process in eukaryotic cells involves the trafficking of cellular elements from the cytosol to the lysosome wherein they are degraded and processed. Ongoing developments in this field point to an inextricable link between autophagy and the innate and adaptive immune system. In this review, we summarize the pertinent research findings to date and suggest future research directions in this dynamic field, especially with respect to pattern recognition receptor interaction.

Three sub-types of autophagy have been described—chaperone-mediated autophagy, microautophagy and macroautophagy (hereafter called autophagy). The term ‘autophagy’ was first suggested by de Duve over 45 years ago [1]. Lamellar vesicles that encapsulated portions of the cytosol and organelle remnants had been described in early electron microscopy studies as vacuoles and lysosomes, and were speculated to arise from focal cytoplasmic degradation [2], [3], [4]. Such vesicles bore the hallmarks of what are now termed ‘autophagosomes’, the characteristic vacuoles synonymous with autophagy. Metabolic manipulation was shown to affect autophagy induction, demonstrating that autophagy was a malleable rather than a static process. The catabolic hormone glucagon and deprivation of amino acids and nutrients were shown to induce autophagy while insulin and certain exogenous amino acids impaired autophagy and proteolysis, defining a role for autophagy in adaptation to cellular stresses [2], [5], [6], [7], [8], [9], [10], [11]. The complexity of signaling molecules that influence autophagy is an ongoing focus of research and will be discussed later.

The stepwise process of autophagosome biogenesis is a cornerstone of autophagy. Over thirty governing autophagy genes (ATG) and their proteins (Atg) have been identified in elegant studies in yeast species [12], [13], [14]. While not all mammalian orthologs have been identified, some have numerous mammalian paralogs with striking similarities in structure and/or function to their yeast counterparts [15], [16]. Ultra-structural studies of autophagosome membranes have shown that they harbor relatively few transmembrane proteins and are thinner than other cellular membranes e.g. the plasma membrane [17]. The earliest identifiable structure in the sequence of autophagosome formation is the disk-shaped, isolation membrane or phagophore (Fig. 1 ). Once formed, this membrane progressively elongates, encircling its cytosolic target, e.g. bacterium, within a portion of the cytosol, eventually sealing to complete the autophagosome. Speculation continues whether the foundation template for the isolation membrane originates from the endoplasmic reticulum, golgi, mitochondria, a pre-formed organelle membrane or even de novo [18], [19], [20], [21], [22], [23], [24], [25]. The molecular mechanisms that lead to isolation membrane appearance continue to be elucidated in both yeast and mammalian cell systems.

Fig. 1

Autophagosome biogenesis. The earliest identifiable structure in the initiation (nucleation) sequence of autophagosome formation is the crescent-shaped isolation membrane or phagophore. Key elements include Atg9, the ULK1–FIP200–Atg13 complex, LC3-II, the Atg12–Atg5–Atg16L complex. Once formed, this membrane progressively elongates (elongation), encircling its cytosolic target, e.g. bacterium, within a portion of the cytosol. The membrane tips fuse and eventually seal, forming the autophagosome (completion). The autophagosome may fuse with the endosomal compartment, forming an amphisome, prior to its ultimate maturation step, whereby its outer membrane fuses with the lysosome to form an autolysosome (also termed autophagolysosome). This facilitates degradation, processing and recycling of the contents of the autophagosome.

Machinery of autophagosome formation

Two ubiquitin-like conjugation systems are pivotal to autophagosome formation and completion. The first system modifies a core autophagy protein–microtubule-associated protein 1 light chain 3 (LC3). Multiple paralogs of Atg8 exist in mammals – LC3A, LC3B, GATE16, GABARAP – hereafter referred to collectively as LC3 [26], [27]. LC3 has a diffuse cytosolic distribution. It is cleaved at its c-terminus by the cysteine protease Atg4 and in turn undergoes sequential ubiquitin-like modifications by the E1-like enzyme, Atg7, and the E2-like enzyme, Atg3, to form LC3-1. The c-terminal carboxyl group of LC3-I is ultimately conjugated to the amine of phosphatidylethanolamine, forming LC3-II. This lipidation of LC3-I to form LC3-II is notable in that LC3-II is exclusively found on autophagosome membranes. The conjugated yeast ortholog of LC3, Atg8, is known to have membrane tethering properties, which may explain one of its roles in autophagosome formation [28]. Atg4 also deconjugates LC3-II on the autophagosome membrane, releasing LC3, highlighting the plasticity of this process. The multifunctional protein p62 interacts with both ubiquitinated proteins and LC3, whereby it is incorporated into autophagosomes. p62 accumulates during autophagy inhibition and has been implicated in targeting proteins to autophagosomes, although it is not itself essential for autophagosome formation [29].

In the second conjugation system, the ubiquitin-like autophagy protein Atg12 is covalently conjugated to Atg5 via its c-terminal glycine, forming the dimeric Atg12–Atg5 complex, following ubiquitin-like reactions involving Atg7 and Atg10. The autophagy scaffold protein Atg16L1 is then conjugated to Atg5 via its N-terminus, forming the Atg12–Atg5–Atg16L1 complex. The Atg16L1 complex self-multimerizes, forming large 800 kDa complexes. These are found in the cytosol and on the evolving isolation membrane, and are likely necessary for the ultimate conjugation of LC3-I. Fujita et al. showed that the Atg16L1 complex behaves as an E3-like enzyme and targets LC3-I to its membrane site of lipid conjugation [30]. The Atg16L1 complexes dissociate from autophagosome membranes as they near completion.

Other essential groups of autophagy proteins participate in isolation membrane formation. The mammalian autophagy proteins ULK1 (Unc-51-like kinase), FIP200 (focal adhesion kinase family interacting protein) and Atg13 were recently identified in a complex which subsequently co-localized at the nascent isolation membrane on autophagy induction, similar to the Atg1–Atg13–Atg17 complex in yeast [31]. The c-terminus of ULK-1 binds to FIP200 and Atg13, and ULK-1 also interacts with LC3 [32]. Mammalian studies of the trans-membrane protein Atg9 have similarly underscored its essential role early in autophagosome formation. It associates with the trans-golgi network, late endosomes, LC3, the Rab-GTPase proteins (Rab7 and Rab9) and re-distributes following autophagy induction, localizing to the nascent autophagosome [33]. Yeast Atg9 was recently shown to self-multimerise via its c-terminus, facilitating its intra-cellular trafficking, independent of other autophagy proteins. This novel finding suggests a potential further role for such Atg9 complexes in contributing to early isolation membrane formation [25].

Controlling autophagy

The discovery of the target of rapamycin in yeast (TOR) and mammalian cells (mTOR) led to significant advances in understanding autophagy regulation, through the family of phosphatidylinositol kinase-related kinases [34], [35], [36]. These signaling networks are involved in broad cellular functions from metabolic responses to growth and proliferation.

The key serine/threonine kinase, Akt, links the mTOR and phosphatidylinositol-3 kinase (PI3K) pathways which are activated by a diverse array of stimuli, including cytokine receptors and toll-like receptors (TLR) [37]. Following receptor activation, class-I PI3Ks are recruited by receptor adaptor molecules to phosphorylate phosphatidylinositol-4,5-bisphosphate, which in turn phosphorylates and activates Akt [38], [39]. The mTOR complexes, down-stream positive effectors of Akt, integrate multiple cellular signals, including those from growth factors, amino acids and intracellular ATP. mTOR activation increases cellular anabolic activity and protein translation [40], [41], [42]. Autophagy is under negative regulation by activated Akt and mTOR [43]. Recently, mTOR was shown to phosphorylate and therefore inhibit the ULK kinase-complex activity, disrupting autophagosome formation [44], [45]. Rapamycin inhibition of mTOR and amino acid deprivation reversed these effects. mTOR may further affect autophagy through its control of autophagy gene transcription [40], [46].

The class III PI3K enzyme, Vps34 (vacuolar protein sorting 34), solely phosphorylates phosphatidylinositol and is involved in regulating vesicular trafficking, nutrient sensing and autophagy [47], [48]. The pharmacological agent 3-methyadenine (3-MA) inhibits its function in vitro. Together with Vps15 (another kinase), Beclin-1, UVRAG (ultraviolet radiation resistance associated gene) and Ambra-1, Vps34 forms a multiprotein complex that is necessary for early stages of autophagosome biogenesis and can up-regulate autophagy overall [49], [50], [51]. However, its seemingly paradoxical role in signal transduction to the mTOR complex following amino acid sensing suggests that its signaling function may depend on the nature of its interacting protein complexes [52], [53].

Beclin-1, a tumor suppressor protein, is itself also involved in modulating autophagy through its interaction with Bcl-2, an anti-apoptotic protein that inhibits both autophagy and apoptosis. The Beclin-1/Bcl-2 interaction is an evolutionary conserved phenomenon, the balance of which determines either up- or down-regulation of autophagy. Silencing or over-expression of Bcl-2 was shown to enhance or suppress starvation-induced autophagy respectively [54]. These effects were specifically dependent on Beclin-1/Bcl-2 interaction, suggesting that nutrient sensing affects the equilibrium of the Beclin-1/Bcl-2 interaction. Bcl-2 dominant interactions with Beclin-1 likely disrupt Beclin-1/Vps34 complex formation, leading to autophagy suppression, although the mechanism has not been fully elucidated. Recently, the toll-like receptor (TLR) signaling molecules MyD88 and TRIF were shown to modulate the Beclin-1/Bcl-2 interaction, enhancing their interaction with Beclin-1 to induce autophagy [55].

A myriad of other signal transduction and effector molecules influence autophagy regulation, directly or indirectly. The Akt and JNK pathways have been shown to enhance or reduce expression of LC3 and Beclin-1 in response to tumor necrosis factor-α (TNF-α) and insulin-like growth factor-1 respectively [56]. The mammalian transcription factor, NFκB, is a key regulator of gene expression, modulating physiological processes including inflammation, apoptosis and also autophagy. TNFα-induced NFκB activation suppresses autophagy, while NFκB suppression enhances starvation-induced autophagy [57], [58]. NFκB may signal through mTOR activation or by affecting enhanced Bcl-2 expression to modulate autophagy. Autophagy itself may in turn influence NFκB activity since it is involved in degradation of IκB kinase, the upstream activator of NFκB, through association with the heat-shock protein, Hsp90 [59], [60]. Reactive oxygen species (ROS) are highly reactive molecules generated from mitochondrial respiratory activity and the products of oxidase enzymes, including NADPH oxidase, and are capable of modulating autophagy [61], [62]. Atg4 is redox-regulated via a conserved cysteine residue and, furthermore, starvation-induced autophagy depends on H2O2 signaling [62]. Starvation lead to local H2O2 formation, partly dependent on class III PI3K activity, and anti-oxidant treatment in vitro attenuated autophagy induction. Recently, a transgenic mouse model harboring a mutant form of super-oxide dismutase, a key anti-oxidant enzyme, also showed increased autophagic activity due to ROS accumulation [63]. Evidence also suggests that autophagic (type II) cell death may stem from ROS accumulation, as seen following in vitro treatments with TNFα and LPS [57], [64].

Pathogen recognition of autophagy

Microbial invasion of the cytosol presents a serious challenge to our innate defenses, including autophagy. While several agents succumb to autophagic destruction (xenophagy), others have evolved mechanisms of autophagy evasion and manipulation. Various Gram+ and Gram− bacteria, viruses and protazoa are known autophagy targets (Table 1 ). The mechanisms by which microbes are selectively sequestered in autophagosomes remain elusive and a combination of host and microbial factors are likely to be necessary. Microbial molecular motifs themselves may solicit autophagosome formation. Alternatively, the up-regulation of autophagy through activating multiple pattern recognition receptors could culminate in xenophagy or perhaps organelle or compartmental damage may lead to targeting by autophagic machinery. Microbial factors may be of equal importance for autophagy activation. For example, Group A streptococcus (GAS) is sequestered in autophagosomes following escape from its early endosomal compartment into the cytosol [65]. Lysosomal degradation of bacteria-containing autophagosomes ensues, effects not observed in autophagy deficient cells. Strains of GAS lacking the streptolysin O toxin remain within endosomes and avoid autophagic destruction indicating a role for streptolysin O in induction of autophagy.

Table 1

Microbial agents interacting with autophagy.

Microbe	Host autophagy interaction	Biological factors and outcomes
Bacteria
Streptococcus pyogenes	Induction	Bacterial clearance
Staphlococcus aureus	Induction	Bacterial clearance
Francisella tularensis	Induction	Bacterial clearance
Salmonella Typhimurium	Induction	Bacterial clearance
Rickettsia conorii	Induction	Bacterial clearance
Escherichia coli	Induction	Bacterial clearance
Mycobacterium tuberculosis	Induction	IFNγ treatment enhances clearance
Vibrio chloerae (exotoxin)	Induction	Limits cytotoxicity, enhances survival
Legionella pneumophilia	Manipulation	Autophagosome maturation delayed
Brucella abortus	Manipulation	Autophagy harnessed for replication
Coxiella burnetti	Manipulation	Autophagosome maturation delayed
Listeria monocytogenes	Evasion	Dependent on ActA, PLC
Shigella flexneri	Evasion	Bacterial escape, dependent on IcsB
Burkholderia pseudomallei	Evasion	Bacterial escape. Dependent on BopA
Viruses
Parvovirus B19	Induction	Cell cycle arrest, virus sequestration
Herpes simplex virus	Evasion	Dependent on neurovirulence factor
Kaposi sarcoma-associated virus	Evasion	Viral Bcl-2 inhibits Beclin-1
Rotavirus	Manipulation	Impaired autophagosome maturation
Human poliovirus	Manipulation	Autophagy harnessed for replication
Hepatitis C virus	Manipulation	Autophagosome maturation delayed
Coxsackievirus	Manipulation	Autophagy harnessed for replication
Protozoa
Toxoplasma gondii	Induction	Parasite elimination

The Gram− human diarrheal agent Shigella flexneri is a highly adapted pathogen harboring a type III secretion system (TTSS) for delivery of its effector proteins to host cells. In epithelial cells, wild-type (WT) strains secreting the effector IcsB are capable of evading entrapment in autophagosomes, in comparison to mutant strains lacking IcsB [66]. Interestingly, IcsB did not appear to confer autophagy protection in a subsequent study in murine marrow derived macrophages, suggesting a cell-type specific phenomenon [67]. The intracellular bacterium Burkholderia pseudomallei, also avoids autophagic destruction in murine macrophages through secretion of its TTSS-delivered effector protein BopA, which shares some homology with IcsB [68].

Salmonella enterica serovar Typhimurium resides within salmonella-containing vacuoles following intracellular invasion. Salmonella employs its TTSS to disrupt these vacuoles, facilitating cytoplasmic entry. Autophagy promptly contributes to subsequent restriction of intracellular proliferation by targeting bacteria from damaged vacuoles—effects that were dependent on a functioning TTSS and reversed in autophagy deficient cells [69], [70]. Listeria monocytogenes, a Gram+ bacillus, replicates within the host cytoplasm following phagosome escape, evading autophagic destruction [71], [72]. The virulence factors listeriolysin O, ActA and phospholipase C were recently shown to be of importance in modulating Listeria-containing phagosomal compartments, blocking lysosomal degradation and facilitating replication and survival [73], [74]. Mycobacterium tuberculosis has adapted to survive within host macrophages by interfering with and blocking phagosome fusion with lysosomes [75]. Autophagy up-regulation with rapamycin or IFN-γ overcame this evasion, and lead to phagosome degradation [76], [77].

Secreted bacterial toxins are themselves capable of interacting with the autophagy pathway. The non-invasive pathogen, Vibrio cholerae, causes a potentially fatal secretory diarrhea. Its secreted exotoxin, VCC, induces vacuole formation consistent with autophagy induction [78]. Furthermore, cell viability was adversely affected following autophagy inhibition, suggesting that in this situation, autophagy may defend against cell toxicity. Our group has recently reported autophagy induction following infection with Helicobacter pylori, which was dependent on the vacuolating cytotoxin, VacA. Autophagy limited the stability of intracellular VacA, again suggesting a cytoprotective function of autophagy in response to secreted toxins [79].

Viruses also interact with autophagy. The herpes virus HSV-1 evades autophagy in part through Beclin-1 inhibition by its neurovirulence protein ICP34.5 [80]. Poliovirus manipulates autophagic machinery following cellular infection, as evidenced by a marked reduction in viral release following pharmacological inhibition of autophagy and siRNA silencing of key autophagy proteins [81]. Rotavirus has been suggested to harness the autophagic apparatus to facilitate replication. Its enterotoxin, NSP4, was found to co-localize with LC3+ structures on immunofluorescence microscopy, and the study authors speculate that NSP4 may interfere with autophagosome–lysosome fusion, enabling viral recruitment of autophagosomes as replication niches [82]. The antiviral protein kinase, PKR, participates in viral induced autophagy, functioning upstream of Beclin-1. It is possible that other viruses which inhibit PKR function, including Influenza and Ebstein–Barr virus, may in turn inhibit autophagy to enhance their own survival [83], [84]. The possibility of a cell-type dependent autophagy response to viral infection was suggested by coronavirus studies, wherein mouse hepatitis virus replication was impaired in ATG5−/− stem cells, but not in ATG5−/− embryonic fibroblasts or marrow derived macrophages [85], [86], [87].

These diverse examples of autophagy–microbial interactions underpin the conserved primary innate role of autophagy as an anti-microbial, protective mechanism and how certain pathogenic organisms have evolved to recognize and commandeer this process for their own advantage.

Autophagy and innate immunity

The innate immune system is responsible for the early detection and destruction of pathogens. This first line of defense relies mostly on a set of receptors called pattern recognition molecules (PRM) that sense molecular motifs that are common to a wide range of pathogens, triggering different signaling cascades that culminate with the elimination of pathogens and the initiation of an adaptive response [87], [88]. The findings that autophagy can specifically target cytosolic pathogens immediately prompted the investigation of the role of PRM in the autophagic detection and elimination of intracellular microbes.

The TLRs are transmembrane proteins, mostly located at the cell surface, with a Toll-IL-1 receptor (TIR) domain facing the cytosol. This domain is able to recruit four different adapter molecules: the myeloid differentiation primary response protein 88 (MyD88), the TIR domain-containing adaptor protein (TIRAP, also called MyD88 adaptor-like—MAL), the TIR domain-containing adaptor-inducing IFN-β-Trif, also called TIR-domain-containing adaptor molecule 1—TICAM-1) and the Trif-related adaptor molecule (TRAM or TICAM2) [88], [89]. As we will see in this section, recent data suggest that induction of autophagy after TLR engagement requires the recruitment of specific adaptors (Fig. 2 ).

Fig. 2

TLR activation triggers autophagy. LPS triggers autophagy after recruitment of Trif (also RIP1 and p38, not shown) and MyD88. The latter seems to interact with Beclin-1, reducing its binding to the anti-autophagic molecule BCL-2. TLR2 engagement induces the incorporation of LC3 to phagosomes (unkown mechanism). Viruses are able to induce autophagy through TLR3, RIG-I (dsRNA) or TLR7/8-MyD88 (ssRNA). Conventional DCs sense viral ligads through the RIG-I/MAVS axis to secrete type I interferon, while the conjugate ATG5/12 seems to be a down regulator of such response. Plasmocytoid DCs deliver TLR7 ligands from the cytosol to the compartments containing TLR7 using basal autophagy. IPAF inhibits autophagy through an unclear mechanism.

Eissa and colleagues provided the first evidence that TLRs are able to trigger an autophagic response by showing the formation of numerous autophagosomes in response to LPS stimulation in the murine macrophage RAW264.7 cell line [90]. Furthermore, silencing TLR4 using RNA interference resulted in significant reduction in autophagosomes. The TLR4-induced autophagic response was dependent on p38, RIP1 and Trif-, but not MyD88. As TLR4 can use both MyD88 and Trif adapter molecules for downstream signaling, the authors proposed that by recruiting both signaling cascades, TLR4 could promote both a fast phagocytic response (through MyD88) and a slower autophagic response (via Trif).

Other TLR family members have also been implicated in the control of autophagy (Fig. 2). Deretic and colleagues have recently demonstrated that when RAW264.7 macrophages were stimulated with a panel of TLR ligands such as Pam3CSK4 (TLR2), flagellin (TLR5), CpG DNA (TLR9), poly (I:C) (TLR3), LPS (TLR4) and ssRNA (TLR7), the latter three were able to up regulate autophagy [91]. In contrast to TLR3 (that recruits only Trif) and TLR4 (that recruits both MyD88 and Trif), TLR7 recruits only MyD88, suggesting that MyD88 may trigger autophagy after TLR7 activation. However, TLR9 activation by CpG DNA also activates MyD88 but did not induce autophagy. Therefore, a simple analysis of which downstream adaptor protein is recruited by TLRs does not fully explain the induction of autophagy by some pathogen associated molecular patterns (PAMPs) and not others. The mechanistic explanation is still elusive and seemingly conflicting evidence remains difficult to reconcile. For example, TLR7 recruitment of MyD88 also leads to the activation NFκB, which is thought to inhibit autophagy [57]. Trif-dependent signaling leads to the induction of type I interferon, which was previously shown not to affect autophagy [76], [88]. Two recent studies have proposed a mechanism by which TLRs might regulate autophagy. Kehrl and Shi demonstrated that not only Trif, but also MyD88 targets Beclin1 and reduces its binding to Bcl-2, upon stimulation with an array of TLR ligands [54], [55]. Alternatively, Wagner and colleagues observed that TLR activation leads to the activation of mTOR, which in turn interacts with the adaptor proteins MyD88 and interferon-regulatory factors (IRFs) 5 and 7, thus controlling the transcription of cytokines such as TNF-α, IL-10, IL12, type I interferons but, surprisingly, not IL-1β [92]. These lines of evidence suggest a more elaborate TLR control of autophagy whereby TLR-adapter molecules interact with proteins from the autophagic pathway rather than by simply activating the classic hierarchical signaling cascades described heretofore.

In contrast with the general notion that TLR ligands up regulate autophagy, Green and colleagues suggested a model in which some TLRs, when engaged by their cognate ligands, usurp the autophagic pathway, recruiting LC3 to the phagosome membrane instead of forming classic autophagosomes [93]. However, as pointed out by the authors, it is not possible to exclude the possibility that the LC3 recruited to phagosomes has its origin in rapidly forming autophagosomes. If confirmed, these data would have a deep impact on the understanding on the role of autophagy in the enhancement of antigen presentation for example.

PAMP recognition as an autophagy trigger seems to be an evolutionary conserved feature. In Drosophila, peptidoglycan-recognition protein (PGRP) family members sense peptidoglycan (PG) from gram-negative bacteria [94]. One of the PGRP family members, PGRP-LE was recently implicated in PG sensing and induction of autophagy upon infection with L. monocytogenes thereby leading to clearance of bacteria [95]. Cytosolic PRMs have also been implicated in regulation of autophagy. Suzuki and colleagues demonstrated that Ipaf, a Nod-like protein previously shown to sense flagellin, down regulates autophagy during infection with the non-flagellated bacterium S. flexneri [67]. The down regulation of autophagy did not involve the ASC adapter protein, normally required for the induction of IL-1β after Ipaf activation. One can speculate that Nod proteins, which sense PG in mammalian cells, may play a similar role in the regulation of autophagy. Up-coming studies addressing this question are eagerly awaited.

In summary, the data above suggest a dynamic interaction between receptors from the innate immune system and regulation of autophagy. Additional studies with knockout mice are now needed in order to demonstrate a definitive role for TLR- or NLR in autophagy during infection with pathogens known to activate specific PRMs.

Autophagy and cytokine responses

In the last few years, autophagy induction has been frequently reported as a consequence of innate immune system activation. However, there is compelling evidence that the relationship between autophagy and the immune system is reciprocal. Cytokines from the innate and adaptive systems regulate autophagy by different mechanisms. Two of the prototypical Th1 cytokines, IFN-γ and TNF-α, were shown to up-regulate autophagy. Gutierrez and colleagues first demonstrated that mouse macrophages harboring Mycobacterium within phagosomes were able to clear bacteria after stimulation with IFN-γ in an autophagy-dependent manner [76]. Follow up studies implicated GTPases in this process. The mouse genome contains 23 different immunity-related GTPases, most of which respond to IFN-γ stimulation and play a role in the defense against intracellular pathogens via a mechanism which is as yet unclear [96]. The studies from Deretic's group showed that both mouse immunity-related GTPase (Irgm1) and its human ortholog IRGM are the key molecules driving the induction of autophagy upon IFN-γ stimulation, leading to the clearance of Mycobacterium from infected macrophages [76], [97].

The other Th1 cytokine shown to stimulate autophagy is TNF-α. Codogno and colleagues observed that cells stimulated with TNF-α are committed to die when NFκB is blocked [57], [59], [60]. These findings are of great interest as the activation of autophagy may represent a way to overcome the resistance of cancer cells to anti-cancer drugs targeting NFκB.

In contrast to the autophagy enhancing effect of some Th1 cytokines, Th2 cytokines such as IL-4 and IL-13, seem to counteract starvation and IFN-γ-induced autophagy by different pathways [97]. While IL-4 and IL-13 block starvation-induced autophagy by activating the Akt-mTOR axis, these cytokines inhibit IFN-γ-induced autophagy in an Akt-independent but STAT6-dependent manner.

The regulation of cytokine secretion by autophagy, has also been reported. Jounai and colleagues demonstrated that in response to infection with RNA viruses or immunostimulatory RNA, IFN-β levels were increased in ATG5 knockout embryonic fibroblasts [98]. The authors demonstrated that the Atg5–Atg12 conjugate negatively regulates the antiviral immune response by interacting with the RIG-I-like receptor (s protein retinoic acid-inducible gene I (RIG-I) and IFN-β promoter stimulator 1 (IPS-1) thus, implying autophagy contributes to viral replication. Iwasaki and colleagues showed that in autophagy-impaired cells the increased cytokine secretion in response to immune-stimulatory RNA is due to the accumulation of defective mitochondria and consequent IPS-1 and ROS accumulation, further strengthening the importance of autophagy in the maintenance of cellular homeostasis [99].

Autophagy has also been proposed to regulate cytokine secretion in Crohn disease. Crohn disease (CD) is a chronic inflammatory intestinal disease with a complex and multifactorial etiology. Several recent independent genome wide association studies have implicated a number of heretofore unappreciated biological pathways in CD pathogenesis, including autophagy [100], [101], [102], [103]. Since the identification of a non-synonymous single nucleotide polymorphism in the ATG16L1 gene as a causal risk variant for CD, several groups have sought to elucidate its functional impact on development of CD. Akira and colleagues generated mice lacking the coiled–coil domain of ATG16L1 and observed aberrant IL-1β secretion upon LPS stimulation of fetal derived liver macrophages. In contrast to previous studies, LPS did not induce autophagy in control macrophages indicating the enhanced IL-1β was not due to disruption of LPS-mediated autophagy [104]. Chimeric mice with ATG16L1-deficient hematopoietic cells had an unremarkable baseline intestinal phenotype, but displayed increased susceptibility to DSS-induced colitis compared with controls. Even though this study used mice expressing a truncated form of ATG16L1, rather than the ATG16L1 risk allele, the results point to the importance of functional autophagy machinery for normal intestinal function. Using an alternative mouse model hypomorphic for ATG16L1 protein expression, Cadwell and colleagues noted paneth cell-specific abnormalities including degenerating mitochondria, loss of lysozyme granule integrity and absence of apical microvilli [105]. Parallel findings were observed when intestinal ATG5 expression was suppressed. Transcriptional profiling analysis revealed that, among other differences, transcripts for the adipocytokines leptin and adiponectin were highly enriched. Similar increased expression profiles were observed previously in patients with CD [106], [107]. The above findings, while not specific to ATG16L1 suppression, underscore the importance of autophagy pathway integrity to normal paneth cell function. Interestingly, ATG7 knockout of pancreatic islet cells resulted in abnormal cellular morphology on EM, including mitochondrial swelling, distension of the endoplasmic reticulum and a paucity of insulin granules when compared with controls [108]. It remains unclear why paneth cells, above others, are susceptible to autophagy interference and how autophagy is involved in maintaining integrity of its lysozyme exocytosis pathway. However, the interaction between autophagy and multivesicular body biogenesis may provide a potential explanation for abnormal granule formation and exocytosis. Once again, the paneth cell is placed at the convergence of several innate immune pathway aberrations and CD pathogenesis. Translational clinical data are keenly awaited.

Autophagy and antigen presentation

The products of the two main cellular degradation systems – the proteasome and the lysosome – are not merely unwanted material but are, instead, key molecules utilized to instruct the immune system. This instruction step is achieved by the presentation of these products to cells from both innate and adaptive immune systems. CD8+ T cells monitor mainly cytosolic and nuclear antigens degraded by the proteasome (a large cytosolic enzyme complex) and loaded into MHC class I. In contrast, CD4+ T cells respond to extracellular or membrane peptides generated by lysosomal degradation and presented in the context of MHC class II at the cell surface [109], [110]. However, this paradigm has been challenged by the demonstration that dendritic cells (DCs) are also capable of presenting extracellular antigens on MHC class I, and not just on MHC class II as initially thought, through a mechanism called cross-presentation. Cross-presentation allows DCs to instruct also CD8+ T cells, generating a more efficient T-cell response [111].

Functional evidence for the presentation of endogenous antigens on MHC class II was first provided by Long and colleagues, who demonstrated that measles and influenza antigens could be presented in the context of MHC class II [112], [113]. Indeed, the affinity purification of MHC class II from Epstein–Barr virus (EBV)-transformed B lymphoblastoid cells, murine B cell lymphoma and myeloid cells showed that more than 20% of natural MHC class II ligands had their origin in intracellular proteins [109]. Together these studies suggested that an alternative and unknown route could deliver antigens from the cytosolic compartment for presentation on MHC class II. Knecht and colleagues were the first to suggest a role for autophagy in this process by showing that glyceraldehyde-3-phosphate dehydrogenase, an important source of human MHC class II ligands, is degraded via chaperone-mediated autophagy [114]. Additionally, peptides from two Atg8 homologues, LC3 and GABARAP, have been isolated from human and mouse MHC class II molecules, respectively, providing further support to the notion of autophagy as an alternative route for delivery of cytosolic antigens for MHC class II.

More direct evidence came from studies using pharmacological inhibition of macroautophagy with PI3K inhibitors (such as 3-MA and wortmannin), which are thought to block the sequestration step of autophagy. Stockinger and colleagues demonstrated that over-expressed C5 was processed and loaded onto MHC class II in an autophagy-dependent manner, as the loading was decreased in the presence of 3-MA [114]. A similar approach was used to show that an endogenously expressed bacterial peptide, NeoR (neomycin-phosphotransferase II), was sequestered in autophagosomes and processed in endosomal/lysosomal compartments for loading onto MHC class II. Brossart and colleagues also used pharmacological inhibition to demonstrate that DCs electroporated with RNA coding for the tumor-associated antigen Muc-1 requires not only lysosomal antigen degradation and processing, but also autophagy in order to prime CD4+ T cells [115].

Further evidence that autophagy contributes to MHC class II presentation came from Munz and colleagues, in which they analyzed the endogenous MHC class II processing of the nuclear antigen 1 (EBNA 1) from EBV, the dominant EBV-latent antigen for CD4+ T cell. Inhibition of autophagy by Atg12siRNA in EBV-transformed B cells reduced recognition by EBNA1-specific CD4+ T cells [116]. In another study, the same group demonstrated that the fusion of influenza matrix protein 1 (MP1) with Atg8/LC3 drives this molecule to autophagosomes in different cell types and enhances recognition by antigen specific CD4+ T cells [117]. Knockdown of Atg12 confirmed that the localization of the fusion proteins with MHC class II molecules was dependent of autophagy. Importantly, these results represent great potential for vaccine design, since targeting antigens to autophagosomes induces a more robust T cell response. In support of this contention, Jagannath et al. demonstrated that induction of autophagy enhances BCG vaccine efficacy in a murine model [118].

In contrast to model or viral antigens, very little is known about bacterial antigens requiring autophagy for proper presentation on MHC class II. So far, only the 85B antigen from M. tuberculosis was shown to be presented more efficiently on MHC class II upon induction of autophagy [119]. Accordingly, Atg6 silencing dampened this process, while rapamycin treatment enhanced priming of 85B-specific CD4+ T cells, strongly suggesting a role for autophagy in MHC class II presentation of antigens of bacterial origin. Even though studies with other bacterial models are lacking, it is possible to speculate a role for autophagy in MHC class II presentation during infections with bacteria that escape from phagosomes, such as L. monocytogenes

Conclusion

Autophagy is steadily emerging from its historic ‘house-keeping’ role as a new dimension in our host defense program. Taken together, the data highlighted above suggest that autophagy impacts on the development of both innate and adaptive immune responses to diverse pathogens and that, conversely, components of the immune system themselves also regulate autophagy. This biological process is now a major target for researchers who want to enhance understanding of and develop strategies to modulate immune responses in a variety of inflammatory and infectious conditions.