Autophagy is a membrane trafficking pathway that results in the formation of autophagosomes which deliver portions of the cytosol to lysosomes for degradation. When autophagosomes engulf intracellular pathogens, the pathway is called ‘xenophagy’ because it leads to the removal of foreign material. Autophagy is activated during infection by Toll-like receptors that recognize pathogen-associated molecular patterns. This allows autophagy to kill micro-organisms and present pathogen components to the innate and acquired immune systems. The targeting of pathogens by autophagy is selective and involves a growing family of autophagy receptors that bind to the autophagosome membrane protein LC3 (light-chain 3)/Atg8 (autography-related protein 8). Ubiquitination of microbes identifies them as substrates for autophagy and they are delivered to autophagosomes by autophagy receptors that bind both ubiquitin and LC3/Atg8. Bacteria can also be detected before they enter the cytosol by autophagy receptors that scan the surface of membrane compartments for evidence of damage. The observation that some pathogens survive in cells suggests they can evade complete destruction by autophagy. For some bacteria this involves proteins that shield the surface of the bacteria from recognition by autophagy receptors. Other viruses and bacteria are resistant to degradation in lysosomes and use autophagosomes and/or lysosomes as sites for replication. Most of our current understanding of the role played by autophagy during microbial infection has come from studies of bacteria and viruses in tissue culture cell lines. Future work will focus on understanding how autophagy determines the outcome of infection ‘in vivo’, and how autophagy pathways can be exploited therapeutically.
Early free-living eukaryotic cells needed to adapt to rapidly changing environments and uncertain food supply. Lack of nutrients imposed serious stress on these cells and this is thought to have driven the evolution of a membrane trafficking pathway called autophagy. Autophagy, which literally means ‘self-eating’ allows cells to deliver cytosolic organelles and proteins to lysosomes for degradation to provide a short-term supply of amino acids and this allowed early eukaryotes to generate the amino acids they needed to move and search for food. The capacity to degrade large quantities of cytoplasm also provided cells with a powerful mechanism to degrade intracellular pathogens. When autophagy engulfs pathogens, the pathway is called ‘xenophagy’ because it leads to the removal of foreign organisms . Xenophagy therefore represents a very early stage in the evolution of innate immunity.
The process of autophagy
Several autophagy pathways have been described that deliver proteins to lysosomes. Microautophagy and chaperone-mediated autophagy deliver proteins directly from the cytoplasm into the lumen of the lysosome. Macroautophagy generates new membranes in cells called autophagosomes which engulf portions of the cytosol within double-membraned vesicles that fuse with lysosomes. Macroautophagy is important for the removal of pathogens and will be referred to as ‘autophagy’ in the rest of the chapter. At least 36 proteins are required for autophagy , and these are conserved from yeast through to fully differentiated mammalian cells. Autophagy is regulated by the TOR (target of rapamycin) kinase which senses amino acid levels in cells. When amino acids are abundant the TOR kinase inhibits autophagy and at the same time increases protein translation to increase cell mass. When food is scarce, amino acid levels fall and the TOR kinase is inactive. This slows translation and activates autophagy to generate amino acids through protein degradation. Autophagy is initiated by a class III PI3K (phosphoinositide 3-kinase) called vps34 that forms a complex with Beclin1/Atg6 to phosphorylate lipids at sites of autophagosome formation. Many membrane compartments can generate autophagosomes, but the best-characterized pathway involves the ER (endoplasmic reticulum) and/or mitochondria (Figure 1).
Generation of the autophagosomes
Anchoring of Atg14 to the ER and/or ER–mitochondrial contact sites [3–5], plays a key role in recruiting the vps34–Beclin1 complex from the cytosol to sites of autophagosome formation in response to starvation. Localized lipid phosphorylation generates small cup-shaped membranes called phagophores or isolation membranes which recruit effector proteins such as WIPI (WD-repeat protein interacting with phosphoinositides) 1/2 (Atg18) to prime phagophore expansion. The major structural protein of the autophagosome is called Atg8/LC3 and this is recruited from the cytosol to the phagophore following conjugation to PE (phosphatidylethanolamine). Generation of the LC3–PE conjugate (also known as LC3II) requires the Atg12–Atg5–Atg16 complex and results in phagophore expansion and eventual release of autophagosomes into the cytosol. During this stage, most of the proteins involved in autophagosome formation are recycled back into the cytosol. LC3II is the major protein of the autophagosome and remains with the autophagosome until fusion with the lysosomes. GFP tagging of LC3 (GFP–LC3) has become indispensable for visualizing autophagosomes in mammalian cells .
Microbes that evade lysosomes are captured by autophagy
Most bacteria enter cells by a combination of phagocytosis and endocytosis and are taken to lysosomes for degradation (Figure 2, i–ii). Lysosomes contain a wide range of hydrolytic enzymes maintained at acidic pH levels that are able to kill bacteria and increase presentation of microbial antigens to the immune system. The observation that several medically important pathogens survive in cells suggests they have evolved ways of evading direct delivery to lysosomes after endocytosis. Interestingly, recent work shows that these lysosome-evasion strategies trigger autophagy, allowing cells a second chance to deliver pathogens to lysosomes (reviewed in [7–10]). Mycobacteria, for example, remain in endocytic vesicles, but modify endosome/phagosome membranes so they cannot fuse with lysosomes (Figure 2, iii). These modified endosomes are taken up into autophagosomes and delivered to lysosomes. Listeria, Shigella and Streptococci secrete lysins to facilitate early release from endosomes and replicate in the cytosol (Figure 2, iv), but these cytosolic forms can be captured by autophagosomes. Some microbes, for example Coxiella burnetii, are resistant to degradation in lysosomes (Figure 2, v) and activate autophagy to gain access to the lysosome which they convert into a parasitophorus vacuole for replication. Intracellular microbes can also use parts of the autophagy pathway to promote replication through a process called ‘non-canonical’ autophagy . For Mycobacterium marinum this involves generation of double-membraned vacuoles containing bacteria by pathways independently of Atg5 and LC3 . Brucella abortus avoids delivery to lysosomes by generating a vacuole from the smooth ER. In common with autophagy, this requires localized lipid phosphorylation by PI3K, Beclin1/Atg6 and Atg14, but does not require later stages of autophagosome expansion powered by Atg5, Atg16L1 and LC3 . FMDV (Foot and Mouth Disease virus) activates autophagy during cell entry by a non-canonical pathway which does not require PI3K , but is dependent on Atg5 and LC3.
Autophagy pathways activated by intracellular microbes
Viruses infect cells by delivering genomes or nucleoprotein core particles into the cytoplasm, either directly through the plasma membrane, or following endocytosis. In contrast with studies of bacteria, relatively few studies have focused on the role played by autophagy in removing viruses immediately after they enter cells. Viruses are obligate intracellular pathogens and many viruses activate autophagy at the onset of viral genome replication. This may result from the recognition of viral genomes by cytosolic helicases such as RIG-I that bind viral RNA and trigger innate immunity to viral infection. Autophagy can also be activated by cell stress pathways, such as ER stress and UPRs (unfolded protein responses) that are activated when viruses use cellular membrane compartments as platforms for virus replication and synthesis of envelope proteins. Some viruses actually benefit from autophagy because the autophagosomes can become incorporated into sites of replication. This has been demonstrated for RNA viruses such as the poliovirus . Autophagy can also inhibit virus replication and this can be important for controlling replication ‘in vivo’. VSV (vesicular stomatitis virus) can infect insects and studies using fruitflies lacking genes essential for autophagy show that autophagy protects against infection . SINV (Sindbis virus) is transmitted by mosquitoes and certain strains cause acute encephalitis in mice. Autophagy is activated during SINV infection leading to degradation of viral capsids. Studies using mice with neuron-specific loss of autophagy show a marked delay in the removal of viral antigens and increased susceptibility to SINV infection . Similarly, neurovirulence of HSV1 (herpes simplex 1) is linked to expression of viral genes that bind Beclin1 and inhibit autophagy .
Autophagy is activated following recognition of pathogen-associated molecular patterns and damage signals
A striking feature emerging from recent studies is that recognition of intracellular pathogens by autophagy is linked to the exposure of DAMPs (damage-associated molecular patterns) and/or PAMPs (pathogen-associated molecular patterns). DAMPs are cell-derived molecules that generate ‘danger’ signals when they are displaced following cell stress or damage. Damage signals can be generated when cellular components are released from cells, or more subtly, when specific cellular compartments are ruptured during infection. PAMPs are pathogen-derived molecules that are only encountered during infection and examples include peptidoglycan and lipopolysaccharides generated by bacteria, the zymosan of yeast cell walls and dsRNA generated during viral infection. DAMPs and PAMPs are recognized by pattern recognition receptors such as the TLRs (Toll-like receptors), and this activates pro-inflammatory responses and/or production of interferon. Many of these signalling pathways also activate autophagy, allowing autophagy to play a crucial role in innate immunity (reviewed in ).
Many pathogens enter cells by endocytosis and/or phagocytosis allowing microbial PAMPs to be recognized by TLRs in endocytic compartments. Examples include detection of viral RNA in endosomes by TLR7 and recognition of bacterial lipopolysaccharide and lipopeptides by TLR1, TLR2 and TLR4. The precise mechanisms leading to activation of autophagy by TLRs remain to be understood, but one pathway appears to involve the TLR adaptor protein MyD88 (myeloid differentiation factor 88), which can activate Beclin1 by releasing it from an inhibitory complex with Bcl-2. The cytosol contains NOD (nucleotide-binding and oligomerization domain)-like receptors that bind PAMPs exposed by pathogens once they are released from endosomes. NOD1 and NOD2 proteins activate autophagy when they bind peptidoglycan within bacterial cell walls. NOD proteins bind autophagy protein Atg16L1 and may seed autophagosome formation at sites of bacterial entry resulting in efficient delivery of bacteria to lysosomes for degradation [20,21]. Genome-wide association studies have linked mutations in NOD2 and Atg16L1 to Crohn's disease making it possible that the inflammation associated with Crohn's disease is linked to defects in autophagy and microbial handling in intestinal epithelial cells. Recent studies show that mice that are deficient in autophagy in intestinal epithelial cells show greater sensitivity to Salmonella infection .
Selective autophagy involves autophagy receptors with LC3-interacting regions
Activation of autophagy in response to starvation is generally thought to lead to non-specific degradation of proteins and organelles. In contrast, activation of autophagy following recognition of pathogens leads to the selective degradation intracellular microbes. This has resulted in the discovery of a new class of innate immunity receptors called SLR (sequestasome-like receptor; or p62) or LIR (LC3-interacting region) proteins. The founding member of this family, p62/SQSTM1, was discovered during studies of inherited diseases where mutations lead to protein misfolding and aggregation. It had been known for many years that ubiquitin can target misfolded proteins for degradation by proteasomes, but many protein aggregates are too big for the proteasome and accumulate in the cytosol where they are degraded by autophagy. The targeting of aggregates for autophagy involves a family of linker proteins that can bind ubiquitin through UBDs (ubiquitin-binding domains), and at the same time bind LC3 in the autophagosome membrane using an LIR. Thus far, p62/SQSTM1 and NBR1 are the best studied and have clear roles in the removal of protein aggregates and damaged mitochondria .
Autophagy receptors control selective autophagy of intracellular pathogens
The role played by autophagy receptors in the control of intracellular pathogens is best illustrated through studies of Salmonella typhimurium. These bacteria enter cells in large endosomes which shrink to enclose one or two bacteria within membrane vesicles called a SCV (Salmonella-containing vacuole) (Figure 3). The bacteria then acquire autophagy markers LC3/Atg8, p62/SQSTM1 and ubiquitin suggesting uptake into autophagosomes. One signal for activation of autophagy is generated by the production of DAG (diacylglycerol) in response to secretion of bacterial proteins from the vacuole via the type 3 secretion system . DAG acts as a lipid second messenger that promotes association of LC3/Atg8 with the SCV (Figure 3, i). A number of Salmonella escape from the vacuole, or reside in damaged vacuoles, exposing PAMPs to the cytosol, allowing access of autophagy receptors to the surface of the bacteria. Many pattern recognition proteins, for example TLRs and NOD-like receptors, contain LLR (leucine-rich) domains. This has stimulated a search for new LLR domain proteins that might play a role in linking pattern recognition with autophagy. A key goal has been to identify enzymes with E3 ubiquitin ligase activity that could transfer ubiquitin to the surface of bacteria. One such protein, LRSAM1, looks promising (Figure 3, ii). LRSAM1 is an LLR-domain protein with E3 ubiquitin ligase activity that binds an autophagy receptor called NDP52 (nuclear dot protein 52) that, in common with p62/SQSTM1, can bind ubiquitin and LC3 . The LLR domain of LRSAM recognizes Gram-positive and Gram-negative bacteria and the enzyme can transfer ubiquitin to bacteria in vitro. A functional role for LRSAM has been demonstrated by knockdown experiments which show reduced autophagy of Salmonella in HeLa cells. Further evidence is provided from lymphocytes from patients suffering from Charcot–Marie–Tooth disease. These lymphocytes do not express LRSAM1 and are defective in ubiquitination of Salmonella.
Recognition of S. typhimurium by autophagy receptors
Ubiquitin surrounding the bacteria may result in recognition by two (or more) autophagy receptors, p62/SQSTM1 and NDP52 (Figure 3, iii–iv) and both proteins contribute to elimination of Salmonella . NDP52 binds TBK [TANK (tumour-necrosis-factor-receptor-associated factor-associated nuclear factor-κB activator)-binding kinase] and optineurin (Figure 3, v). Optineurin also binds ubiquitin but has a low affinity for LC3; however, phosphorylation of optineurin by TBK generates a high-affinity-binding site for LC3, allowing TBK to play an important role in regulating autophagy of bacteria [26,27]. In addition to exposing PAMPs to the cytosol, disruption of the SCV by Salmonella exposes a newly discovered damage signal that is recognized by galectin-8 . Galectin-8 is a cytosolic lectin that binds sugars that would normally be hidden on the inside of endosomes and lysosomes. When membrane compartments are damaged by pathogens the sugars provide an ‘eat me’ signal that recruits galectin-8 to the vacuole, the lectin then recruits autophagy receptor NDP52 to recruit LC3 (Figure 3, vi). This recent work on S. typhimurium has shown that recognition for autophagy is driven by pattern recognition and damage signals. This makes it likely that similar mechanisms will operate for any microbes that expose PAMPs to the cytosol or damage membrane compartments. This is illustrated by studies on Shigella and Listeria which secrete lysins to damage endosomes and are detected by galectin-8, NDP52 and p62/SQSTM1 [28,29].
Recognition of viruses for autophagy also involves autophagy receptors (Figure 4). Genetically distinct viruses such as SINV and FMDV, which are RNA viruses, and HSV1, which is a DNA virus, are recognized by p62/SQSTM1 suggesting a common mechanism of detection [14,16,17]. Chikingunya virus capsids are toxic to cells and recognition of Chikingunya virus capsids by p62/SQSTM1 promotes cell survival by degrading capsids in autophagosomes . A recent genome-wide silencing screen has identified SMURF1 (SMAD-specific E3 ubiquitin ligase 1) as a protein that is required for capture of SINV capsids by autophagosomes  and subsequent clearance of SINV and HSV1 from cells. SMURF1 has ubiquitin E3 ligase activity, but surprisingly, this is not required for selective autophagy, instead SMURF1 uses a membrane-targeting domain that may bind directly to autophagosome lipids.
Recognition of viruses by autophagy receptors
Microbial evasion of autophagy
The observation that some pathogens survive in cells suggests they have found ways to avoid complete destruction by autophagy. Listeria monocytogenese and Shigella flexneri move rapidly within the cell using the actin cytoskeleton. S. flexneri secretes a protein called VirG to activate actin polymerization and VirG appears to activate autophagy because it binds the Atg5 protein involved in early stages of autophagosome formation. Virulent Shigella isolates may not activate autophagy because they generate the IcsB protein that blocks binding of VirG to Atg5 . L. monocytogenese triggers autophagy when lysins rupture the endosome membrane to release the bacteria into the cytosol. Once in the cytosol, L. monocytogenese expresses the ActA protein to harness actin-based motility through recruitment of cellular Arp2/3 and Ena/VASP proteins. Early studies showed that autophagy markers are not recruited to mobile bacteria, suggesting that L. monocytogenese uses actin-based motility to escape from autophagy. Interestingly, strains that recruit Arp2/3 and Ena/VASP to polymerize actin, but do not move in the cell, also escape autophagosomes. This implies that polymerization of actin on the surface of the bacteria, rather than motility, can shield the surface of the microbe from recognition by autophagy receptors .
Conclusions and future research
It is becoming clear that autophagy plays a key role in innate immunity against infection. Selective autophagy of intracellular pathogens is triggered by detection of PAMPs and damage signals that recruit autophagy receptors to deliver pathogens, or vacuoles containing them, to lysosomes for degradation. Delivery to lysosomes kills microbes and increases presentation of microbial antigens to the immune system. Recent experiments suggest that the recruitment of autophagy receptors may vary between pathogens . It will be important to determine if this reflects targeting of microbes to different autophagy pathways, for example, through binding to specific ATG8/LC3 family members , or the fine tuning of microbial-autophagosome evasion strategies. Ubiquitin plays a key role in recruiting autophagy receptors during selective autophagy. Studies on cargo capture need to be extended to identify the E3 ubiquitin ligases that mark microbes for ubiquitination and the microbial proteins that are recognized for ubiquitination. Galectin-8 and DAG provide examples of non-ubiquitin-based signals for targeting bacteria for autophagy, there may be other signals on the basis of recognition of membrane damage. Studies in cell culture need to be extended to animal models to determine the effects of autophagy on the outcome of infection. Conditional expression of autophagy proteins and/or autophagy receptor proteins can be used to determine the role played by autophagy in immunological surveillance and determining virulence and pathogenesis. Recent genome-wide screens have linked mutations in autophagy proteins to human diseases. Crohn's disease is linked to mutations in NOD sensors and autophagy protein Atg16 L, and mutations in autophagy receptor p62/SQSTM1 are linked to Paget's disease. It will be important to determine whether these mutations cause disease because they compromise microbial handling by autophagy.
Autophagy evolved as a response to starvation and results in the formation of autophagosomes that engulf portions of the cytosol and fuse with lysosomes.
When autophagy engulfs intracellular microbes the pathway is called xenophagy, indicating the removal of foreign material.
Autophagy is emerging as an important arm of anti-microbial immunity because lysosomes are able to kill microbes and increase presentation of microbial antigens to the innate and acquired immune systems.
Activation of autophagy during infection and recognition of intracellular pathogens is linked to the exposure of DAMPs and/or PAMPs.
Recognition of pathogens is selective and involves a family of autophagy receptors that bind the major membrane protein of the autophagosome LC3/Atg8.
Autophagy receptors bind microbes that have been tagged with ubiquitin, whereas others scan the surface of membrane compartments containing microbes for evidence of damage.
Many pathogens have evolved autophagy-evasion strategies that shield them from recognition by autophagy receptors. Viruses and bacteria that are resistant to degradation in lysosomes can use autophagosomes and/or lysosomes as sites for replication.