How did we get here? Insights into mechanisms of immunity-related GTPase targeting to intracellular pathogens

Abstract

The cytokine gamma-interferon (IFNγ) activates cell-autonomous immunity against intracellular bacterial and protozoan pathogens by inducing a slew of antimicrobial proteins, some of which hinge upon immunity-related GTPases (IRGs) for their function. Three regulatory IRG clade M (Irgm) proteins chaperone about 20 effector IRGs (GKS IRGs) to localize to pathogen-containing vacuoles (PVs) within mouse cells, initiating a cascade that results in PV elimination and killing of PV-resident pathogens. However, the mechanisms that allow IRGs to identify and traffic specifically to “non-self” PVs have remained elusive. Integrating recent findings demonstrating direct interactions between GKS IRGs and lipids with previous work, we propose that three attributes mark PVs as GKS IRG targets: the absence of membrane-bound Irgm proteins, Atg8 lipidation, and the presence of specific lipid species. Combinatorial recognition of these three distinct signals may have evolved as a mechanism to ensure safe delivery of potent host antimicrobial effectors exclusively to PVs.

Introduction

Most non-immune cells are equipped with cell-intrinsic defense programs to protect against intracellular bacterial and protozoal pathogens, and these programs are frequently activated by cytokines called interferons [1]. While IFN signaling triggers a variety of different defense mechanisms in different cell types, many of these responses are conducted by families of IFN-inducible GTPases. Type I IFNs induce the Mx family of proteins, which interfere with viral replication to conduct antiviral immunity. Among the defense programs induced in response to both type I and type II interferon (IFNγ) are the p65 guanylate binding proteins (GBPs) and the p47 immunity-related GTPases (IRGs), which conduct defense against viral, bacterial, and protozoan pathogens [2,3].

Both the IRG and GBP families are structurally related to dynamins, GTPases that function in membrane biology and vesicle trafficking. The mouse IRG family consists of ~20 cytosolic effector IRGs that contain a canonical GxxxxGKS amino acid sequence in the P-loop of the G-domain, and three membrane-bound regulatory IRG clade M (Irgm) proteins that contain a non-canonical GxxxxGMS sequence (Figure 1,[4]). Effector GKS IRGs have been shown to target to and colocalize with intracellular pathogens, although the specific effector mechanism of each individual GKS IRG has not been shown. However, it has been observed that IRG-targeted PV membranes (PVMs) vesiculate and rupture, and it is thought that IRGs via their membrane binding and GTPase activity may directly mediate PVM disruption, exposing luminal pathogens for destruction by cytosolic defenses [5–7]. Deletion of individual effector IRGs results in partial defects in resistance to certain intracellular pathogens, but deletion of mouse Irgm1 and/or Irgm3, which reside on host organelle membranes, results in a failure of GKS IRGs to target to intracellular pathogens [2], leading to broad and severe defects in resistance to intracellular bacterial and protozoal pathogens including Chlamydia trachomatis and Toxoplasma gondii (Table 1). Thus, the murine IRG family comprises a robust IFNγ-inducible antimicrobial program, with Irgm proteins orchestrating the subcellular localization and activity of the cytosolic effector GKS IRGs.

Figure 1. Structural features of IRGs.

All IRGs contain a C-domain, G-domain, and N-domain. The G-domain contains the nucleotide binding pocket, with GKS IRGs such as Irga6, Irgb6, and Irgb10 containing a canonical GxxxxGKS amino acid sequence in the nucleotide-binding pocket and Irgm proteins containing a non-canonical GxxxxGMS sequence. Homo- and hetero-oligomerization among IRGs occurs via G-domain interactions. Membrane binding requires a conserved αK amphipathic helix in the C-domain. Additionally, Irga6 and Irgb6 possess an N-terminal myristoylation motif, and Irgm1 contains a C-terminal palmitoylation motif allowing for covalent lipid attachments that assist in membrane binding. In contrast, Irgb6 harbors a binding pocket for specific phospholipids such as PI5P. Figure created with BioRender.com

Table 1.

IRG table

IRG	Susceptibility	Citation	Other phenotype	Citation
Irgm1 (LRG-47)	Toxoplasma	(Collazo et al., 2001; Mahmoud et al., 2015)	Immune cell homeostasis	(Alwarawrah et al., 2022; Coers et al., 2011; Feng et al., 2004; Feng et al., 2008; King et al., 2011; Santiago et al., 2005; Taylor et al., 2020; Xu et al., 2010)
	L. monocytogenes	(Collazo et al., 2001)	Autophagy	(Dong et al., 2015; Feng et al., 2009; Gutierrez et al., 2004; He et al., 2012; Singh et al., 2006; Tian et al., 2020; Traver et al., 2011; Zhou et al., 2020)
	M. tuberculosis	(MacMicking et al., 2003)	Macrophage adhesion/motility	(Henry et al., 2007; Henry et al., 2010)
	M. bovis BCG	(Singh et al., 2006; Tiwari et al., 2009)	TLR4-mediated cytokine production	(Bafica et al., 2007)
	M. avium	(Feng et al., 2004)	Phagosome-lysosome fusion	(Tiwari et al., 2009)
	T. cruzi	(Santiago et al., 2005)	Mitophagy/mitochondrial fission	(Henry et al., 2014; Singh et al., 2010)
	L. major	(Taylor et al., 2007)	EAE	(Wang et al., 2013; Xu et al., 2010; Xu et al., 2017)
	C. trachomatis	(Coers et al., 2008)	Autophagic flux	(Traver et al., 2011)
	S. typhimurium	(Henry et al., 2007)	Cholesterol/lipid metabolism	(Schmidt et al., 2017; Xia et al., 2013)
	L. donovani	(Murray et al., 2015)	Intestinal inflammation	(Liu et al., 2013; Mehto et al., 2019; Rogala et al., 2018; Taylor et al., 2020)
			Epithelial-mesenchymal transition	(Tian et al., 2015)
			Lysosome stability	(Maric-Biresev et al., 2016)
			Atherosclerosis	(Fang et al., 2021; Fang et al., 2016; Xia et al., 2013)
			NLRP3 inflammasome activity	(Mehto et al., 2019)
			Sjogren’s-like syndrome	(Azzam et al., 2017)
			Type I interferonopathy	(Rai et al., 2021)
Irgm2 (GTPI, IIGP2)	Toxoplasma	(Dockterman et al., 2021; Pradipta et al., 2021)	Noncanonical inflammasome activity	(Eren et al., 2020; Finethy et al., 2020)
	C. psittaci	(Miyairi et al., 2007)
Irgm3 (IGTP)	Toxoplasma	(Halonen et al., 2001; Ling et al., 2006; Melzer et al., 2008; Taylor et al., 2000; Zhao et al., 2009)	Cross presentation, lipid droplet homeostasis	(Bougneres et al., 2009)
	C. trachomatis	(Bernstein-Hanley et al., 2006)	ER stress	(Liu et al., 2012)
	L. donovani	(Murray et al., 2015)	CD8+ T cell activation	(Guo et al., 2015)
	L. major	(Taylor et al., 2004)
Irga6 (IIGP)	C. trachomatis	(Al-Zeer et al., 2009; Nelson et al., 2005)
	Toxoplasma	(Liesenfeld et al., 2011; Martens et al., 2005)
	Rabies virus	(Tian et al., 2020)
Irgb6 (TGTP)	Toxoplasma	(Lee et al., 2020)
	Vesicular stomatitis virus	(Carlow et al., 1998)
Irgb10	C. trachomatis	(Bernstein-Hanley et al., 2006; Coers et al., 2008)	Inflammasome activation in infection	(Christgen et al., 2022; Man et al., 2016)
	F. novocida	(Man et al., 2016)
Irgd (IRG-47)	Toxoplasma	(Collazo et al., 2001)

Mechanisms of cell-autonomous immunity can differ substantially between mice and humans. There are several mouse and human GBPs that share significant structural and functional homology, but the IRG system is far more divergent [2,8]. While mice feature an expansive set of IRG genes, humans lack antimicrobial effector IRGs. Only one functional IRG is broadly expressed in human tissues, the human clade M paralog IRGM [4]. Human IRGM and the murine Irgm paralogs do share roles in orchestrating inflammation but execute divergent functions in cell-autonomous immunity. Specifically, murine Irgm1 and Irgm3 are essential in regulating GKS IRG localization and cell-autonomous immunity, but also play additional roles in cellular homeostasis (Table 1). While Irgm2 is less critical for GKS IRG localization [9,10], it was recently found to regulate noncanonical inflammasome activation in response to LPS or gram-negative bacterial infection [11,12]. While the phenotypes in GKS-deficient mice appear to be limited to defects in cell-autonomous immunity, Irgm-deficient animals display pleiotropic phenotypes that include altered host defense, exacerbated inflammation, and disruption of cellular homeostasis [2].

While the importance of the IRG family in murine cell-autonomous immunity has been readily demonstrated, many questions remain regarding their function and regulation. GKS IRGs specifically localize to cytosolic pathogens or the membranes of PVs rather than host membranes. A variety of features on cytosolic bacteria, such as lipopolysaccharide (LPS) and flagella, is evident and accessible to the host for identification and differentiation from host membranes. However, the PVM of pathogens such as Chlamydia and Toxoplasma are originally derived from the host cell plasma membrane and are not known to display such obvious identifying features. How, then, are GKS IRGs able to identify and specifically target to the PVM? In this review, we integrate different mechanistic concepts for controlling the subcellular localization of IRGs, incorporating recent structural insights in IRG-lipid interactions into a unified model for IRG targeting to PVs.

The ”missing self” hypothesis

Initial models of IRG targeting to PVs derived from early observations regarding IRG subcellular localization. In IFNγ-primed cells, it was initially observed that GKS IRGs Irga6 and Irgb6 accumulate at the Toxoplasma PVM, but in uninfected IFNγ-primed cells, these proteins adopt a diffuse cytosolic staining pattern with a slight enrichment on the endoplasmic reticulum [5,13]. Ectopically expressed Irga6 in unprimed cells formed cytoplasmic aggregates, implying that other IFNγ-induced proteins are required to regulate the localization of Irga6 and other GKS proteins [13].

A possible explanation for the formation of GKS aggregates was provided by biochemical studies revealing that Irga6 and Irgb6 form GTP-dependent oligomers in vitro [14]. In the absence of IFNγ, ectopic expression of Irga6 or Irgb6 could therefore result in GKS protein oligomerization in the host cell cytosol, thereby sequestering proteins otherwise destined for delivery to Toxoplasma PVMs. However, ectopic expression of Irga6 or Irgb6 along with the three Irgm proteins abolished the cytoplasmic aggregation of GKS IRGs and restored their targeting to the Toxoplasma PV. Yeast two-hybrid assays demonstrated direct nucleotide-dependent interactions between GKS IRGs and Irgm proteins, indicating that Irgm proteins regulate GKS localization and function by acting as guanine dissociation inhibitors (GDIs) [13]. These observations led to the “missing self” hypothesis: Irgm proteins reside on host organellar membranes and inhibit the GTP-dependent oligomerization and binding of GKS IRGs. In contrast, the PVM, which is devoid of Irgm proteins, allows for disinhibited GKS oligomerization and subsequent antimicrobial function. Although GKS proteins can oligomerize in the absence of lipids [14], it is likely that oligomerization occurs with accelerated kinetics in the presence of an appropriate lipid template, as it has been shown for many other dynamin-related proteins including GBP1 [15–17]. Therefore, it is the absence of Irgm proteins that acts as a signal for oligomerization-dependent GKS localization and activation on a target membrane such as the PVM (Figure 2).

Figure 2. Three-signal model for IRG recruitment.

Irgm proteins reside on the membranes of host organelles such as mitochondria, lipid droplet (LD), endoplasmic reticulum, and Golgi apparatus. As Irgm proteins inhibit GTPase activity of GKS IRGs, their absence (i.e. “missing self”) on the membranes of vacuoles containing intracellular pathogens such as C. trachomatis (green circles) renders those membranes permissive to GKS IRG recruitment. Conjugation of Atg8 proteins to phosphotidylethanolamine by Atg5/12/16L1 at the PVM comprises a second signal for GKS IRG recruitment. Finally, some IRGs such as Irgb6 specifically bind PV-enriched lipid species such as PI5P and may go on to recruit other IRGs such as Irga6 and Irgb10. Figure created with BioRender.com.

The “missing self” hypothesis was corroborated and expanded in subsequent work. Irgb10 was found to localize to inclusions (vacuoles containing the bacterial pathogen C. trachomatis) in a manner dependent on Irgm proteins. Similar to Irga6 and Irgb6, Irgb10 GTP-dependent oligomerization was required for stable recruitment to non-self target membranes. Forced tetramerization or oligomerization of an Irgb10 mutant lacking its G-domain was sufficient to target Irgb10 to C. trachomatis inclusions independent of Irgm proteins [18]. Collectively, these observations indicated that Irgm proteins regulate proper IRG localization by regulating nucleotide-dependent protein oligomerization. Further supporting this model, lipid droplets were found to be decorated with Irgm3 but largely devoid of GKS proteins in IFNγ-primed wildtype cells. In contrast to wildtype cells in which lipid droplets showed limited enrichment for GKS IRGs, lipid droplets in Irgm3-deficient cells were heavily decorated with Irga6, Irgb6, and Irgb10 [18]. These findings demonstrating vastly amplified GKS IRG localization to host organelles in the absence of inhibitory Irgms further supported the “missing self” hypothesis.

“Targeting by AutophaGy (TAG)”

The “missing self” hypothesis was eventually found to be insufficient to explain the complex process of GKS protein delivery to PVs, as components of the autophagy pathway were implicated in regulating GKS targeting. Briefly, a complex of autophagy-related proteins (E1-like Atg7, E2-like Atg3, and E3-like Atg5/12/16L1) conjugates the Atg8 family of small proteins to phosphotidylethanolamine (PE) for incorporation into autophagosomes at various stages of development. Murine Atg8 homologs include the two microtubule-associated protein light chain 3 family members LC3a and LC3b as well as three γ-butyric acid receptor-associated proteins Gabarap, GabarapL1, and GabarapL2 [19]. Adaptor proteins such as p62 bind mammalian Atg8 homologs as well as ubiquitin, targeting ubiquitinated substrates for autophagic consumption [19]. It was initially found that Atg5-deficient cells were defective for IFNγ-mediated destruction of Toxoplasma PVs and resistance to Listeria monocytogenes, and subsequent findings hinted at a potential link between IRGs and autophagy proteins [20]. GKS IRG aggregates in Irgm-deficient cells were found to colocalize with Atg8 and p62 [21], and GKS IRGs also aggregated and failed to target C. trachomatis or Toxoplasma PVs and execute cell-autonomous immunity in Atg3- or Atg5-deficient cells [22,23]. These findings demonstrated that Atg8 conjugation to PE is required for proper GKS IRG localization and PV targeting.

Subsequent studies revealed that proteins involved upstream of Atg8 lipidation in canonical autophagy were not required for cell-autonomous immunity or GKS IRG localization, implying that the Atg8 lipidation process specifically is coopted by the IRG system rather than other components of canonical autophagy (Figure 2). Indeed, Atg5 was found to transiently associate with the Toxoplasma PV early in infection, and Atg8 proteins were found to decorate the Toxoplasma PV alongside GKS IRGs [24]. Additionally, conjugation-dependent localization of Atg8 proteins on viral replication centers promoted GKS IRG recruitment independently of canonical autophagy [25]. Deletion of individual Atg8 proteins – especially Gate-16/GabarapL2 – disrupted targeting of GKS IRGs to Toxoplasma PVs and induced their cytoplasmic aggregation [26]. In a landmark study, Hwang and colleagues showed that ectopic expression of the Atg8 lipidation complex, Atg5/12/16L1, was sufficient to drive GKS IRG recruitment to certain host membranes. Atg16L1 mutants that localized specifically to mitochondria or plasma membrane were expressed in cells deficient for wildtype Atg16L1, resulting in Atg8 conjugation at those target membranes and Irga6 recruitment to those sites in IFNγ-treated cells under most conditions [27]. Given that some IRGs contain putative LC3-interacting regions, these findings situate Atg8 conjugation to PE upstream of IRG targeting to intracellular pathogens and suggest that Atg8 conjugation and incorporation into target membranes may act as a second signal for GKS IRG recruitment via direct Atg8-IRG interactions [26–28].

These findings supporting Atg8 conjugation as a second signal are consistent with the “missing self” hypothesis. Mitochondria are decorated with Irgm1 and the plasma membrane is devoid of Irgm proteins. Accordingly, in cells expressing plasma membrane targeted Atg16L1, IFNγ alone was sufficient to drive GKS IRG recruitment to Atg8-decorated plasma membrane. However, in cells expressing mitochondria-targeting Atg16L1, infection with Toxoplasma was required in addition to IFNγ to drive Irga6 localization to mitochondria [27]. These observations suggest that Toxoplasma infection somehow interferes with the inhibitory activity of Irgm proteins on mitochondria. While the absence of Irgm proteins and the presence of PE-conjugated Atg8 on target membranes comprises a first and second signal, respectively, for GKS IRG localization, the detailed mechanisms underlying how these signals result in GKS IRG recruitment remain to be fully elucidated.

Direct lipid binding to deliver IRGs to vacuolar pathogens

While most investigations aimed to identify protein signals to trigger IRG recruitment, some studies have evaluated the capabilities of IRGs to bind to specific membrane compartments. Palmitoylation of Irgm1 supports localization with Golgi and mitochondrial membranes [29], and Irgm1 has been shown to colocalize with some endosomes and phagosomes, although conflicting reports exist [30–32]. Irgm-decorated host vesicles generally do not contain luminal pathogens, however Irgm1 appears to localize to phagosomes containing the avirulent Mycobacterium bovis Bacille Calmette-Guérin (BCG) and promote fusion with lysosomes [30,33]. Irgm1 recruitment to BCG-containing phagosomes was mediated via direct preferential binding of phosphatidylinositol-(3,4,5)P3 (PtdIns(3,4,5)P3) and PtdIns(3,4)P2, as well as diphosphatidylglycerol (cardiolipin). Modulation of phosphoinositide kinase (PI(3)K) activity disrupted Irgm1 localization, demonstrating that membrane lipid composition may serve as a signal for Irgm1 recruitment [30]. However, these observations did not address GKS IRG recruitment to bona fide PVs containing virulent pathogens and fell short of explaining how Irgm1 may regulate GKS IRG localization.

Some GKS IRGs display membrane-binding activity via conserved amphipathic helical domains and covalent lipid attachments (Figure 1). While Irgb10 localizes to C. trachomatis and Toxoplasma PVs [18,22,34], it also has been shown to localize to the inner membrane of cytosolic bacteria such as Francisella novocida, albeit in a GBP-dependent manner [35]. The precise binding partner for Irgb10 on bacterial membranes is unknown, but oligomerized Irgb10 requires a C-terminal amphipathic helix for recruitment to C. trachomatis inclusions [18]. This amphipathic helix comprises a putative membrane-binding domain that is found among many GKS IRGs including Irga6, suggesting a possible conserved membrane docking mechanism. In addition to these amphipathic helices, Irgb10 and Irga6 are also myristoylated at their N-terminal domains [18,36,37], and recent structural analysis of Irgb10 has generated a putative model for GTP-dependent insertion of this myristoyl moiety into target membranes [38]. Therefore, spatial control over GTPase activity and the resulting myristoyl switch could promote Irgb10 binding specifically to PVMs. However, forced tetramerization of Irgb10, which eliminates any spatial regulation of Irgb10 oligomerization, still resulted in preferential localization of Irgb10 to C. trachomatis inclusions, in a manner dependent upon the N-terminal myristoylation and the C-terminal amphipathic helix motifs [18]. These findings imply that myristoylation provides non-specific membrane anchoring properties but that amphipathic helices, commonly found in GKS IRGs, delineate target membrane specificity.

Irgb6 has been identified as a “pioneer” GKS IRG that loads onto Toxoplasma PVs prior to other IRGs such as Irga6 [22]. Irga6 and Irgb10 recruitment to Toxoplasma PVs was decreased in Irgb6-deficient cells, tempting the speculation that Irgb6 may identify PV-specific lipid species and trigger recruitment of other IRGs. Indeed, Irgb6 was found to specifically bind phosphoinositol monophosphates in vitro, especially phosphatidylinositol-5-phosphate (PI5P) and phosphotidylserine (PS), lipid species that are enriched on the Toxoplasma PVM [39]. Structural analysis of Irgb6 and in silico phospholipid-binding simulations characterized a putative mechanism for direct binding between Irgb6 and PI5P which was corroborated in cell culture [40]. These findings indicate that some IRGs such as Irgb6 may have developed a binding pocket with affinities for specific lipid moieties, conferring a preference for certain membranes enriched for these lipid species, and that these IRGs may act as “pioneers” to promote recruitment of other IRGs such as Irga6 (Figure 2).

Conclusions and open questions

Since the discovery of the IRG family, the working model for the immune detection of intracellular pathogens by IRGs has progressed substantially. GDP-bound GKS IRGs were shown to exist in a cytosolic pool and oligomerize and localize to the PVM in a manner dependent on GTP binding and hydrolysis. Host membrane-bound Irgm proteins were shown to be required for proper localization of GKS IRGs, and to act as guanine dissociation inhibitors of GKS IRGs. These observations implicated the absence of Irgm proteins as a necessary but not sufficient signal for GKS IRG membrane targeting. The lipid conjugation of Atg8 proteins was shown to drive GKS IRG recruitment, suggesting a second signal, yet these two signals were also insufficient in some circumstances. Therefore, additional factors must exist that enable GKS IRGs to differentiate the PVM from host membranes.

The preference of Irgb6 for PI5P as a preferred lipid binding partner represents a viable third signal (Figure 2). Upon IFNγ stimulation, cytosolic Irgb6 may identify potential target membranes by the presence of lipidated Atg8 proteins, and in the absence of inhibitory Irgm proteins, may oligomerize, bind tightly to PI5P, and execute its antimicrobial function. It has been shown that in defense against Toxoplasma, Irgb6 acts as a “pioneer” IRG, initially localizing to PVs and subsequently promoting recruitment of other GKS IRGs such as Irga6 and Irgb10 [22,39]. While Irgb6 contains a PI5P-specific binding pocket, it is currently unknown whether amphipathic helices found in other GKS IRGs such as Irga6 and Irgb10 exert any specificity for certain lipid species. In sum, PVM-specific lipid moieties as well as conjugated Atg8 together drive recruitment of GKS IRGs, which, in the absence of Irgm proteins, undergo GTP-dependent oligomerization and subsequent antimicrobial function.

Deficiencies in one or more IRGs are associated with susceptibilities to an extremely broad range of intracellular bacterial pathogens (Table 1), and based on their mechanisms of cell entry, nutrient acquisition, and host cell modulation, the PVMs of different pathogens are surely composed of different lipid species, although these differences have not been robustly explored. Different “pioneer” IRGs may therefore identify the PVMs of different pathogens based on their membrane compositions, subsequently recruiting other GKS IRGs that lack membrane specificity thereby initiating a specialized antimicrobial defense pathway.

Other interferon-inducible GTPases may also assist in identifying and exposing target membranes for IRG recruitment. Recent work has demonstrated that GBPs may directly bind LPS on cytosolic bacterial membranes and recruit other host effectors such as caspase-4 or ApoL3 [15,41–44]. Additionally, Irgb10 has been shown to be recruited to cytosolic bacterial pathogens in a GBP-dependent manner [35]. Work from our lab has shown that GBPs are able to localize to the LPS-coated outer membrane of cytosolic bacterial pathogens and act as detergent-like membrane permeabillizer to render bacteria susceptible to antimicrobials [15]. The surfactant activity of GBPs may thus explain their role in facilitating the delivery of Irgb10 to the inner bacterial membrane of cytosolic bacteria, where Irgb10 promotes bacterial killing and the liberation of bacterial ligands for sensing by the host [35]. Whether Irgb10 interacts directly with specific microbial ligands is currently unknown, but this mechanism highlights the possibility that, while some GKS proteins may function to disrupt the PVM, others may execute downstream antimicrobial functions at the level of the microbial membrane itself.

Mechanisms of cell-autonomous immunity in mice and humans share some similarities but are significantly divergent. Although there are several GBPs that share functional homology between mouse and human systems, IRGM is the sole human IRG that is broadly expressed. Throughout human evolution, the IRGM gene family contracted to a single locus which was pseudogenized. However, the insertion of a transposon restored the open reading frame of human IRGM, which expresses a protein that is not IFN-inducible and is truncated compared to murine Irgms but that otherwise retains some structural homology [45,46]. Without GKS IRGs to regulate, human IRGM was initially considered to be a cryptic pseudogene until genetic association studies linked IRGM variants with Crohn’s disease, sepsis, resistance to mycobacterial disease, and other effects on immunity and inflammation [47–51]. Mechanistic studies investigating human IRGM function revealed that it plays a role in autophagic flux and autophagosome-lysosome fusion [33,52], paralleling roles of murine Irgm paralogs [53]. Humans may have evolved different mechanisms of cell-autonomous immunity to compensate for their lost IRGs: for example, in infection with C. trachomatis or Toxoplasma, IFNγ stimulation of human cells induces indoleamine 2,3-dioxygenase (IDO), which depletes the cell of tryptophan and deprives the auxotrophic microbes of a necessary metabolite [54,55]. Additional mechanisms of human cell-autonomous immunity to C. trachomatis or Toxoplasma are not well understood but appear to rely more heavily on xenophagy-related mechanisms compared to membrano-lytic defenses employed by the mouse systems [56,57]. However, both mouse and human cells rely on ubiquitination of pathogens and PVs, as well as noncanonical components of the autophagy pathway to resist intracellular pathogens [24,58–61].

One interesting observation was that, in IFNγ-primed cells, forced localization of the Atg8 lipidation machinery on the plasma membrane was sufficient to drive GKS IRG localization, while Atg8-loaded mitochondria required infection with Toxoplasma to display GKS IRG recruitment [27]. Because Irgm proteins are present on the mitochondria and not the plasma membrane, one possible interpretation of this finding is that Toxoplasma infection interferes with the inhibitory effects of mitochondrial Irgm proteins on GKS IRG recruitment. Irgm1 and human IRGM promote the fusion of early phagosomes and other vesicles with lysosomes [30,33,52], and it is possible that vacuolar pathogens such as Toxoplasma evolved effector mechanisms that block Irgm1/IRGM function to avoid direct delivery of PVs to degradative lysosomes. However, blocking Irgm recruitment to PVs would therefore render those PVs accessible to GKS IRGs, which may themselves represent a co-adaptation by the host to identify and destroy Irgm-deficient vesicles. Further study would be required to test this mechanism, but similar “guard models” of host resistance provide precedents, especially in plant immunity [62].

Many questions remain regarding mechanisms of IRG membrane targeting. While humans lack GKS IRGs, Atg8 lipidation appears to play important roles in both humans and mice, and human IRGM also promotes autophagy and lysosomal fusion. It therefore prompts the question as to whether humans also use the absence of IRGM as a marker for PVs, and if so, which host factors mediate this recognition. Other questions include whether lipidated Atg8 proteins also mark PVs for an antimicrobial attack in human cells, how the Atg8 lipidation machinery is delivered to PVs, whether the lipid composition of PVs is relevant for this type of immune recognition, and whether this process differs in mouse and human cells. Recent advances have highlighted IRG recruitment to PVs as a model system to understand fundamental principles of detection of vacuolar pathogens by the mammalian immune system. Future studies should aim to further identify the factors that differentiate host from foreign membranes and break down the differences between mouse and human systems of cell-autonomous immunity.