Partners in anti-crime: how interferon-inducible GTPases and autophagy proteins team up in cell-intrinsic host defense

Abstract

Once pathogens have breached the mechanical barriers to infection, survived extracellular immunity and successfully invaded host cells, cell-intrinsic immunity becomes the last line of defense to protect the mammalian host against viruses, bacteria, fungi and protozoa. Many cell-intrinsic defense programs act as high-precision weapons that specifically target intracellular microbes or cytoplasmic sites of microbial replication while leaving endogenous organelles unharmed. Critical executioners of cell-autonomous immunity include interferon-inducible dynamin-like GTPases and autophagy proteins, which often act cooperatively in locating and antagonizing intracellular pathogens. Here, we discuss possible mechanistic models to account for the functional interactions that occur between these two distinct classes of host defense proteins.

Introduction

Macroautophagy (hereafter simply referred to as autophagy) is a highly conserved homeostatic process by which eukaryotic cells recycle portions of their own cytoplasm through sequestration into double-membraned autophagosomes and subsequent delivery into degradative lysosomes. This process not only removes obsolete or damaged organelles and promotes cell survival during nutrient starvation [1] but can also capture and destroy intracellular pathogens through a process known as xenophagy [2]. Multiple steps of autophagy and xenophagy, including autophagosome formation and fusion with lysosomes, require an ancient ubiquitin-like conjugation system conserved from yeast to mammals. This system is composed of several autophagy proteins (ATGs) that covalently conjugate members of the ubiquitin-like protein ATG8 protein family, such as mammalian microtubule associated protein 1 light chain 3 (LC3), to the headgroup of the membrane lipid phosphatidylethanolamine (Table 1). Similar to protein ubiquitination, ATG8 conjugation to phosphatidylethanolamine requires the function of E1-, E2- and E3-like enzymes. Following cleavage of its C-terminal arginine by the cysteine protease ATG4, the exposed penultimate glycine of ATG8 is activated by the E1-like ATG7 enzyme, then transferred to the E2-like ATG3 enzyme, before ATG8 is conjugated to phosphatidylethanolamine by the E3-like enzymatic complex consisting of ATG12–ATG5-ATG16 [1]. In addition to its essential role in degradative autophagy, the ATG8 conjugation system also regulates LC3-associated phagocytosis of pathogens and dead cells [3], the unconventional secretion of host proteins [4], exocytosis of viruses [5], and the localization of interferon (IFN)-inducible GTPases to subcellular sites of microbial colonization [6,7].

Table 1.

Glossary of host proteins discussed in the text and their functions

Host Factors	Functions
ATG3	An E2-like conjugating enzyme required for the lipidation of ATG8-like proteins.
ATG5	As a complex with ATG12 and ATG16, it serves as an E3-like ligase for the lipidation of ATG8-like proteins.
ATG7	An E1-like activating enzyme that activates ATG12 for its conjugation to ATG5 as well as ATG8 proteins for their conjugation to the lipid phosphatidylethanolamine.
ATG8	A family of ubiquitin-like proteins consisting of the LC3 and GABARAP subfamilies. ATG8 proteins can be conjugated to phosphatidylethanolamine.
ATG12	An ubiquitin-like protein that forms a conjugate with ATG5.
GBPs	A family of IFN-inducible GTPases involved in cell-intrinsic immunity and the regulation of inflammation.
GKS proteins	An ‘effector’ IRG subfamily that exists in mice but not in humans.
IRGs	A vertebrate family of IFN-inducible GTPases controlling cell-intrinsic host defense to bacterial and protozoan pathogens.
IRGM proteins	An IRG subfamily controlling autophagy-related processes. IRGM proteins exist both in mice and humans.
MX proteins	A family of IFN-inducible GTPases with antiviral functions.
NOD2	A cytoplasmic sensor, which detects the bacterial peptidoglycan sub-fragment muramyl dipeptide and activates the autophagy pathway.
Syntaxin-17	A membrane integrated protein that controls the fusion of autophagosomes with lysosomes.
ULK1	A serine/ threonine kinase controlling autophagy initiation.
VLIGs	A family of very large IFN-inducible GTPases with poorly defined functions.

IFN-inducible GTPases are grouped into four families (Table 1): Immunity Related GTPases (IRGs), Guanylate Binding Proteins (GBPs), Myxovirus-resistance (Mx) proteins and Very Large Inducible GTPases (VLIGs). While the physiological functions of VLIG proteins are poorly described, the pivotal roles for IRGs, GBPs and Mx proteins in cell-intrinsic host defense have been widely reported [7]. Current evidence suggests that Mx proteins provide resistance exclusively to viruses [8]. IRGs and GBPs on the other hand execute host defenses against a diverse group of pathogens that includes viruses, bacteria, microsporidia and protozoa [7]. In order to perform many of their antimicrobial functions, IRGs and GBPs specifically associate with intracellular microbes residing in the host cell cytosol or at pathogen-occupied supramolecular structures [9], which include viral replication compartments [10] and pathogen-containing vacuoles [7]. While these pathogen-occupied supramolecular structures bear distinct molecular signatures, some common principles that underlie their recognition by IRGs and GBPs have emerged. We will discuss the role of ATG proteins in coordinating the targeting of IRGs and GBPs to pathogen-containing vacuoles and viral replication compartments as well as the regulatory functions of IRGs and GBPs in antimicrobial autophagy.

Autophagy proteins control the translocation of IFN-inducible GTPases to microbial replication compartments

The IFN-inducible IRG resistance system of mice is highly polymorphic and is typically encoded by 10 to 20 paralogous genes in the genome of any given mouse strain [11]. Mouse IRG proteins are grouped into GKS and IRGM protein subfamilies [7,12,13]. The GKS ‘effector’ proteins are defined by a canonical GKS sequence in the nucleotide-binding pocket, which is changed to a non-canonical GMS sequence in IRGM proteins [12]. GKS proteins translocate to pathogen-containing vacuoles formed by pathogens such as Toxoplasma gondii or Chlamydia trachomatis [13-16] and promote the lytic destruction of pathogen-containing vacuoles and their occupants [17-19]. IRGM proteins regulate this translocation process, possibly through multiple mechanisms. Endomembrane-bound IRGM proteins undergo transient interactions with cytoplasmic GKS proteins to inhibit the exchange of GDP for GTP by GKS proteins [13]. Through these interactions IRGM proteins not only maintain a cytoplasmic pool of GDP-bound deployable GKS proteins [13] but also guard self-structures against off-target attacks by activated GTP-bound GKS proteins [16,20]. Whereas most endomembranes are IRGM-decorated, the paucity of IRGM on pathogen-containing vacuoles marks these microbe-occupied structures with a ‘missing-self’ pattern that permits the transition of GKS proteins into the GTP-bound, ‘membranophilic’ state and their consequential binding to pathogen-containing vacuoles [9,16] (Figure 1). However, the absence of IRGM proteins per se is insufficient to drive GKS translocation to IRGM-devoid membranes [13,20], indicating that additional ‘second’ or ‘third’ signals are required to accurately deliver GKS proteins to pathogen-containing vacuoles.

Figure 1. Multilayered regulation of GKS protein localization to pathogen-containing vacuoles.

In interferon-activated cells, the majority of endomembranes are shielded from GKS proteins by the presence of IRGM proteins. The sparsity of IRGMs at pathogen-containing vacuoles formed by T. gondii or C. trachomatis provides a ‘missing-self’ signal that permits the recruitment of GKS to these vacuoles; however, additional signals in the form of lipidated ATG8 are required to efficiently recruit GKS proteins to pathogen-containing vacuoles. In support of this ‘two-signal model’ forced localization of the ATG8 conjugation machinery (ATG12–ATG5-ATG16L1 complex) to the IRGM-free plasma membrane results in GKS recruitment at that site. In contrast, artificial placement of the ATG8 conjugation machinery at IRGM-decorated mitochondria is insufficient to recruit GKS proteins unless additional cues, like T. gondii infections, are provided. How ATG8 overrides the inhibitory function of IRGM at the mitochondria upon infection is unknown; potential mechanisms may include posttranslational modifications that block IRGM function or increase GKS affinity for ATG8-decoraetd membranes (denoted by “?”) or lead to the removal of IRGMs from mitochondrial membranes (denoted by “??”).

The first evidence for the existence of such additional signals came from studies demonstrating that ATG5 was required to deliver GKS proteins to T. gondii-containing vacuoles, resulting in gamma-interferon (IFNγ)-mediated control of T. gondii infection in tissue culture and in vivo mouse models [21]. Genetic dissection of the autophagy pathway demonstrated that GKS targeting to pathogen-containing vacuoles is dependent on the ATG8 conjugation system but independent of factors that promote autophagosome initiation and lysosomal degradation [22-26], thus indicating a noncanonical function of the ATG8 conjugation system in the delivery of GKS proteins to pathogen-containing vacuoles (Figure 1). This model was further substantiated by findings that several ATG8 homologs occupy T. gondii-containing vacuolar membranes and control GKS and GBP recruitment to pathogen-containing vacuoles [22,27,28], with a notable role for the gamma-aminobutyric acid receptor associated protein (GABARAP) subfamily of ATG8 proteins in this process [28]. Analogously, decoration of murine norovirus replication compartments with ATG8 proteins prompts GKS and GBP recruitment to replication compartments and blocks replication of murine norovirus [10], highlighting the central role for the ATG8 conjugation system in the recruitment of IFN-inducible GTPases to diverse types of microbe-occupied supramolecular structures inside infected cells. The presence of putative LC3-interacting regions in some members of the GKS and GBP protein families suggests that a subset of IFN-inducible GTPases directly interact with ATG8-decorated membranes [6] and subsequently recruit additional GKS and GBP family members to targeted membranes in a hierarchical manner. In support of this model the forced delivery of the ATG12–ATG5-ATG16L1 complex to the IRGM-devoid inner leaflet of the plasma membrane is sufficient to trigger GBP and GKS recruitment to the plasma membrane [27] (Figure 1). The engineered placement of the ATG8 conjugation machinery to IRGM-decorated mitochondria on the other hand is insufficient to mark these organelles as targets for GBPs or GKS effector GTPases, unless concomitant T. gondii infections override the inhibitory function of mitochondrial IRGM proteins [27,29]. Thus, infections may provide a third signal that strengthens the interaction of GBPs and GKS proteins with ATG8-decorated membranes. The nature and consequence of this signal is unknown but may involve posttranslational modifications that increase binding affinities between ATG8-positive membranes and its protein interaction partners [29].

Although ATG-independent mechanisms of pathogen-containing vacuolar targeting by IFN-inducible GTPases also exist [30], compelling evidence demonstrates that ATG8 lipidation of pathogen-containing vacuoles and viral replication compartments promotes the recruitment of GBPs and GKS proteins to these intracellular centers of microbial replication. However, the mechanism by which the ATG8 conjugation machinery is deployed to pathogen-containing vacuoles or replication compartments is unknown. In vitro reconstitution experiments demonstrated direct binding of recombinant ATG5 protein to liposomes in a negative charge-dependent manner [31]. Therefore, the transient association of ATG5 with T. gondii-containing vacuoles following shortly after infection [27] may similarly depend on interactions between ATG5 and negatively charged lipids present in nascent T. gondii-containing vacuoles. Alternatively, ATG5 or an ATG5 adaptor protein could directly interact with microbial molecules present on pathogen-containing vacuoles or viral replication compartments, paralleling the interactions described for ATG5 and the Shigella protein IcsA [32]. Lastly, microbe-mediated activities such as the insertion of protein secretion channels into pathogen-containing vacuolar membranes could induce aberrant-self signals that trigger the recruitment of ATG5 and IFN-inducible GTPases to pathogen-containing vacuoles [30,33]. The latter two models would account for the specificity with which the ATG8 conjugation machinery is delivered to vesicles containing pathogenic microbes.

IFN-inducible GTPases promote ubiquitination of pathogen-containing vacuoles and trigger autophagy-related defenses

GBPs are recruited to sites of intracellular microbial colonization, where they promote antimicrobial responses that eliminate the pathogen. Current evidence indicates that these antimicrobial responses are diverse, vary with the particular GBP, and may be host species-specific. They include: (i) lysis of pathogen-containing vacuoles to extricate vacuolar pathogens from their safe intracellular niches [34,35]; (ii) recruitment of antimicrobial effectors such as the NADPH oxidase to bacteria-containing phagosomes [36]; (iii) promotion of vacuole maturation and elimination through lysosomal fusion [36]; (iv) lysis of cytosolic bacteria [37-39] and (v) suppression of the actin-based motility used by some cytosolic bacteria for cell-to-cell spread [40,41]. Knowledge of GBP functioning at the molecular level is still insufficient to explain such a broad array of activities.

There is now considerable evidence that some of the antimicrobial functions of IFN-inducible GTPases are linked to their ability to regulate cellular ubiquitination events. Studies in mouse cells have shown that a sentinel event controlled by IRG proteins is the recruitment of ubiquitin E3 ligases such as TRAF6 and Trim21 to the C. trachomatis- and T. gondii-containing vacuoles [17,42,43]. The ubiquitin moieties then serve as binding sites for the ubiquitin-binding protein and autophagy receptor p62/SQSTM1, which in turn attracts mouse GBPs to pathogen-containing vacuoles [17,44]. In human cells, IFNγ cell priming similarly promotes ubiquitination of C. trachomatis- and T. gondii-containing vacuoles but in a seemingly IRG- and GBP-independent manner [45-48]. There is also evidence that GBPs may interact with autophagy proteins to drive downstream responses that occur subsequent to GBP loading on the vacuole. For instance, mouse GBP7 facilitates ATG4 recruitment to mycobacterial phagosomes [36], and GBP1 and GBP2 mediate xenophagic destruction of C. trachomatis in human macrophages [49]. However, such studies are few at this point, underscoring the lack of broad knowledge concerning the interphase between GBPs and autophagy proteins, and the manner in which this impacts their antimicrobial functions.

Mouse and human IRGM proteins regulate autophagy induction and autophagosome maturation in response to infections

As alluded to above, IRGM proteins are biochemically and functionally distinct from the GKS proteins in possessing a methionine substitution for the canonical lysine residue in the G₁ GTP-binding motif [12]. Humans possess one IRGM gene that is expressed as 5 differentially spliced mRNAs (IRGMa-e) [12,50], while mice possess three distinct genes, named Irgml (previously known as LRG-47), Irgm2 (GTPI) and Irgm3 (IGTP) [12]. Interest in the IRGMs has been piqued by a large body of literature associating human IRGM gene variants with inflammatory and infectious diseases, including Crohn’s disease [51-53], non-alcoholic fatty liver disease [54], ankylosing spondylitis [55], sepsis [56], and mycobacterial infection [57,58]. Increased risks for the development of inflammatory diseases is linked to carriage of the IRGM minor allele, which encompasses several polymorphisms including a 20-kb deletion within the IRGM promoter region [52]. The various polymorphisms associated with the IRGM minor allele are in nearly complete linkage disequilibrium [59,60], making it difficult to assess the functional consequences of individual polymorphisms and their relevance to disease in the absence of more detailed studies. Because the genetic changes common to the IRGM minor allele are predominantly found outside the protein-coding region, we can nonetheless presume that these genetic alterations affect gene expression rather than protein function. This is supported by studies showing that mRNA expression of the minor allele is diminished relative to that of the major allele [56,59]; though conversely, one study proffers that the minor allele mRNA is impervious to negative regulation by mir-196, a microRNA expressed in colonic epithelial cells during inflammation [61]. Collectively, these studies suggest then that either decreases or increases in IRGM levels may promote disease dependent on the cellular context.

The degree to which human and mouse IRGMs share cellular functions has been debated [60,62,63]. Mouse and human IRGMs differ from each other in some notable ways: mouse IRGMs do possess the aforementioned GKS IRG regulatory function (described above) that human IRGM lacks. Further, expression of mouse IRGMs but not human IRGM is induced by IFN signaling [13,16,64]. In spite of these differences, a growing body of evidence suggests shared functions, particularly in modulating antimicrobial autophagy (Figure 2). Absence of either mouse IRGM1 or human IRGM leads to impaired xenophagic clearance of pathogens, especially in macrophages [20,44,50,65-67]. For human IRGM, the underlying mechanism involves the ability of the protein to associate with autophagy proteins including the Unc-51-like autophagy activating kinase 1 (ULK1) of the autophagy initiation complex and to mediate activation of this complex via phosphorylation [68-70]. Human IRGM also interacts with the bacterial peptidoglycan sensor nucleotide-binding oligomerization domain-containing protein 2 (NOD2) and ATG16L1 [68], known mediators of infection-induced autophagy [71]. Whereas these observations suggest that human IRGM plays a role in the nucleation or elongation of autophagosomes, a recent study demonstrates a role for human IRGM in the recruitment of the membrane fusion protein Syntaxin-17 to the autophagosome, with Syntaxin-17 then stimulating autophagosome-lysosome fusion [72]. In sum, these studies indicate that human IRGM controls several stages of the autophagic process in response to infectious triggers such as cell invasion-mediated NOD2 activation. The mechanism through which mouse IRGM1 or its paralogs regulate autophagy is less well studied.

Figure 2. Functional consequences of IRGM dysregulation.

Defects in IRGM function perturb autophagy and related processes in mouse and human cells, thereby nurturing a pro-inflammatory milieu linked to Crohn’s and other diseases. The absence of IRGM results in defective processing of secretory granules in Paneth cells, defects in antimicrobial autophagy (xenophagy), disrupted T cell homeostasis and changes in mitochondrial dynamics and metabolism that lead to exacerbated cytokine production during inflammation. In many cases, defects in IRGM closely mirror defects in another autophagy gene, ATG16L1, which like IRGM is an established susceptibility factor for the development of Crohn’s disease. WT = wild type; KO = knockout; STX17 = Syntaxin-17; IFNG = gamma-interferon; ROS = radical oxygen species.

There is evidence that altered autophagy regulation as a consequence of human IRGM or mouse IRGM1 deficiencies promotes the onset of inflammatory diseases such as colitis through multiple mechanisms (Figure 2). (i) Xenophagic removal of bacterial pathogens is defective in the absence of human IRGM or mouse IRGM1 [7,65-67,73], which may lead to persistent infections with pathogens or pathobionts, thereby causing chronic intestinal inflammation. (ii) Absence of IRGM1 leads to alterations in autophagic processing of the secretory granules in intestinal Paneth cells [74], the components of which are important for homeostasis of bacterial populations in the ileum. This connection between autophagic regulation of Paneth cells and Crohn’s disease was first established by the identification of another gene, ATG16L1, as a Crohn’s disease susceptibility gene [51,75], and by the phenotype of Atg16l1 hypomorphic mice [76,77] that parallels that of Irgm1-deficient mice with respect to altered Paneth cell function and increased susceptibility to dextran sodium sulfate-induced colitis [74]. (iii) Decreased levels of human IRGM expression or disruption of Irgm1 gene function in mice also lead to altered mitochondrial dynamics [50,74,78], suggesting a possible role for IRGM proteins in the clearance of mitochondria via mitophagy. Components of damaged mitochondria that are not removed through mitophagy are known to activate the Nucleotide-binding oligomerization domain-like receptor pyrin domain-containing-3 (NLRP3) inflammasome and to trigger inflammation [79,80]. Further, diminished mitophagy has also been connected to altered Paneth cell function [81]. (iv) Irgm1 deletion prompts striking metabolic changes in mouse cells, particularly increases in glycolysis, that lead to escalated production of proinflammatory cytokines [78]. Again, a parallel exists with ATG16L1, as impaired autophagy through its deficiency also leads to increases in glycolysis and cytokine production [77,82,83]. (v) IRGM1 regulates autophagic survival of proliferating T cells during immune activation and accordingly Irgm1-deficient mice become lymphopenic in response to infections [84-87]. Crohn’s disease is well documented to be a T cell-driven disease, and further, autophagy deficiencies disrupt the functional integrity of the regulatory T cell compartment leading to increased intestinal inflammation in mouse models [82,83]. Clearly, much work is needed to clarify which of these mechanisms are relevant to inflammatory diseases influenced by IRGM proteins, as well as to determine whether phenotypes associated with IRGM1 deficiency in mice are also present in humans carrying IRGM disease variants, and vice versa.

Conclusions

The innate immune system employs pattern recognition receptors to detect the presence of invasive microbes. While the field of immunology has made tremendous progress in our understanding of how this antimicrobial alarm system is activated [88], we know far less about how the innate immune system subsequently ‘handcuffs’ microbial intruders, many of which hide within infected host cells. The latter task requires for the host to detect the microbe’s precise location inside a host cell. IFN-inducible dynamin-like GTPases and autophagy proteins have emerged as cooperative partners in host defense that are able to localize and destroy intracellular pathogens; IRGs and GBPs can detect ‘non-self,’ ‘aberrant-self’ and ‘missing-self’ patterns associated with intracellular microbes [9,16,30,40]; autophagy proteins are also able to detect the location of pathogen-occupied supramolecular structures independent of IFN-inducible GTPases [10,27,32] and subsequently recruit GBPs and GKS proteins via ATG8-lipidated membranes. Nevertheless, a molecular and biochemical appreciation of these processes is still lacking and the underlying mechanisms need to be defined.

Once bound to their microbial targets, GBPs and mouse GKS proteins solicit a diverse repertoire of defense programs that include the ubiquitination and autophagic destruction of the captured microbes [17,36,42,49]. Human IRGM and its mouse orthologs appear to intersect with the autophagy machinery in a different manner. While dispensable for canonical autophagy [70,89], IRGM proteins promote the formation and maturation of autophagosomes in response to microbial stimuli [50,66-69,72]. Recent evidence reveals that human IRGM promotes autolysosome formation by assisting in the recruitment of the vesicle fusion protein Syntaxin-17 to immature autophagosomes [72]. Whether other human IRGM isoforms or their murine orthologs also act as adaptors for Syntaxin-17 or instead function as adaptors for other membrane remodeling proteins remains to be tested. Nonetheless, the physical and functional interactions between IRGM and several autophagy proteins set a framework for the future exploration of the role of IRGM isoforms in autophagy-related host defense programs.