Abstract
Guanylate binding proteins (GBPs), a family of interferon (IFN)-inducible GTPases, can promote cell-intrinsic defense by removal of intracellular microbial replicative niches through host cell death. GBPs target pathogen-containing vacuoles or the pathogen itself, and assist in membrane-disruption and release of microbial molecules that trigger cell death by activating the inflammasomes. We previously showed that GBP1 mediates atypical apoptosis or pyroptosis of human macrophages infected with Toxoplasma gondii (Tg) or Salmonella enterica Typhimurium (STm), respectively. In mice, the p47 Immunity-related GTPases (IRGs) control the recruitment of GBPs to microbe-containing vacuoles and subsequent cell death. However, humans are devoid of functional IRGs, and the pathogen-proximal immune detection mechanisms by GBP1 are poorly understood. Here, we describe two novel single-cell assays which show that GBP1 promotes the lysis of Tg-containing vacuoles and Tg plasma membrane, resulting in the cytosolic detection of Tg-DNA. In contrast, we show GBP1 only targets cytosolic STm and does not contribute to bacterial escape into the cytosol of human macrophages. GBP1 interacts with caspase-4 and recruits it directly to the bacterial surface, where caspase-4 can be activated by LPS. During STm infection, caspase-1 cleaves and inactivates GBP1 at Asp192, a site conserved in related mammalian GBP1 proteins but not in murine Gbps. STm-infected human macrophages expressing a cleavage-deficient GBP1 mutant exhibit higher pyroptosis due to the absence of caspase-1-mediated feedback inhibition of the GBP1-caspase-4 pathway. Our comparative studies elucidate microbe-specific spatiotemporal roles of GBP1 in detecting infection and the assembly and regulation of divergent caspase signaling platforms.
Introduction
Most nucleated cells can defend themselves against infection by viruses, bacteria and eukaryotic parasites in a process called cell-intrinsic immunity. These defense programs are set in motion in response to the detection of pathogens by membrane-bound or cytosolic pattern recognition receptors (PRRs) (MacMicking, 2012; Randow et al, 2013; Mostowy & Shenoy, 2015; Jorgensen et al, 2017). In addition to antimicrobial molecules that restrict or kill pathogens, host cell-death is a destructive yet effective mechanism of defense because it removes replicative niches and traps intracellular pathogens within the cell remnants (Jorgensen et al, 2016). Antimicrobial immunity and cell death are enhanced by the type II interferon (IFNγ) which induces the expression of up to 2000 IFN-stimulated genes (ISGs) (MacMicking, 2012; Schoggins, 2019). The guanylate binding protein (GBP) family of GTPases, which are highly abundant in cells exposed to IFNγ, consists of seven members in the human and eleven members in mice (Olszewski et al, 2006; Shenoy et al, 2007; Kresse et al, 2008; Shenoy et al, 2012). GBPs target intracellular pathogens and mediate host-defense through multiple mechanisms, including the regulation of autophagy, oxidative responses, inflammasomes and cell death (Jorgensen et al, 2016; Olszewski et al, 2006; Tripal et al, 2007; Kresse et al, 2008; Kim et al, 2011, 2012b; MacMicking, 2012; Shenoy et al, 2012; Haldar et al, 2013; Randow et al, 2013; Haldar et al, 2014; Meunier et al, 2014; Haldar et al, 2015; Man et al, 2015; Meunier et al, 2015; Mostowy & Shenoy, 2015; Shenoy et al, 2007; Feeley et al, 2017; Foltz et al, 2017; Jorgensen et al, 2017; Li et al, 2017; Lindenberg et al, 2017; Man et al, 2017; Piro et al, 2017; Wallet et al, 2017; Wandel et al, 2017; Zwack et al, 2017; Costa Franco et al, 2018; Liu et al, 2018; Santos et al, 2018; Gomes et al, 2019; Schoggins, 2019).
Once GBPs target to a pathogen vacuole or the pathogen itself, they are thought to disrupt these membranes by an as yet uncharacterized mechanism (Yamamoto et al, 2012; Selleck et al, 2013; Meunier et al, 2014; Kravets et al, 2016). Disruption of barrier membranes leads to pathogen growth-control and release of pathogen-associated molecular patterns (PAMPs) which are sensed by PRRs that can trigger host cell death. Whether GBPs directly recognize pathogen vacuolar membranes or PAMPs is an important question that has not yet been answered (Pilla et al, 2014; Meunier et al, 2014; Lagrange et al, 2018; Santos et al, 2018; Fisch et al, 2019a).
A large body of work on GBPs has been carried out in murine cells, wherein these proteins closely collaborate with members of a second family of IFN-induced GTPases, comprising 23 members in the mouse, the p47 immunity-related GTPases (IRGs) (MacMicking et al, 2003; Bernstein-Hanley et al, 2006; Singh et al, 2006; Henry et al, 2007; Miyairi et al, 2007; Shenoy et al, 2007; Coers et al, 2008; Hunn et al, 2008; Al-Zeer et al, 2009; Tiwari et al, 2009; Khaminets et al, 2010; Lapaquette et al, 2010; Singh et al, 2010; Brest et al, 2011; Haldar et al, 2014). For instance, mouse IrgB10 targets bacteria following mGbp recruitment and contributes to the release of bacterial LPS and DNA, and mouse IrgM1 and −3 are essential regulators of GBP-targeting of some pathogen-containing vacuoles (Singh et al, 2010; Meunier et al, 2014; Haldar et al, 2015; Man et al, 2016; Balakrishnan et al, 2018). However, only one IFN-insensible, truncated IRG, IRGM, can be found in the human genome (Bekpen et al, 2005, 2010). Therefore, how human GBPs target intracellular pathogens remains unknown. In addition, some PRRs are unique to humans, for example, LPS-sensing by both caspase-4 and caspase-5 in humans but only caspase-11 in the mouse (Kayagaki et al, 2011, 2013; Hagar et al, 2013; Shi et al, 2014; Casson et al, 2015; Ding & Shao, 2017). Moreover, unlike mouse cells, human cells can respond to tetra-acylated LPS (Lagrange et al, 2018) and possess additional DNA sensors, such as the DNA-dependent protein kinase (Burleigh et al, 2020). The mechanisms underlying GBP-mediated detection of pathogens and stimulation of human macrophage death therefore need to be investigated further.
All seven human GBPs have a conserved structure with an N-terminal globular GTPase domain and a C-terminal helical domain. GBP1, GBP2 and GBP5 are isoprenylated at their C-terminal CaaX-box, which can anchor them to membranes (Nantais et al, 1996; Olszewski et al, 2006; Tripal et al, 2007; Britzen-Laurent et al, 2010). The ability of human GBPs to target pathogen-containing vacuoles remains poorly characterized. Differences have also been noted depending on the pathogen and cell type. We previously showed that human GBP1 fails to target the apicomplexan parasite Toxoplasma gondii (Tg) and two intracellular bacterial pathogens, Chlamydia and Salmonella enterica subsp. enterica serovar Typhimurium (STm), in human A549 epithelial cells; however, GBP1 is required for the restriction of parasite growth, but not the bacterial pathogens (Johnston et al, 2016). On the other hand, in human macrophages GBP1 localizes to Tg, Chlamydia and STm, but whether it can disrupt membranes that enclose these pathogens is not known (Al-zeer et al, 2013; Fisch et al, 2019a).
Human GBP1 targets Tg and STm and promotes distinct forms of macrophage cell death. In the case of both pathogens, GBP1 targeting to pathogens is necessary, even though downstream mechanisms of cell death are distinct. Since Tg induces the loss of inflammasome proteins, including NLRP3 and caspase-1, human macrophages undergo atypical apoptosis through the assembly of AIM2-ASC-caspase-8 complexes. In contrast, GBP1 promotes activation of caspase-4 following its recruitment to STm resulting in enhanced pyroptosis (Fisch et al, 2019a). Although our previous work suggested that GBP1 is involved in PAMP release for detection by these PRRs during natural infection (Fisch et al, 2019a), the underlying mechanisms involved in liberating microbial ligands was not investigated.
In this study we show that GBP1 contributes to lyses of the parasite-containing vacuole and the plasma membrane of Tg by employing two newly developed assays. We also show that GBP1 only targets STm that are already cytosolic and does not contribute to their ability to reach the cytosol of human macrophages. In contrast, during STm infection, caspase-1 cleaves and inactivates GBP1 and thereby reduces its ability to recruit caspase-4. These studies reveal the feedback inhibition of GBP1/caspase-4-driven pyroptosis during STm infection and its dual membrane-disruptive actions during Tg infection.
Results
GBP1 contributes to Toxoplasma parasite and vacuole disruption and infection control
As GBP1 elicits divergent host cell death programs in response to Tg and STm, we sought to investigate the upstream mechanistic details of GBP1 during infection by these two unrelated human pathogens. We previously correlated GBP1 recruitment to Tg parasitophorous vacuoles (PV) to activation of AIM2 and caspase-8 with the recognition of parasite DNA (Fisch et al, 2019a). Like some murine Gbps (Yamamoto et al, 2012; Selleck et al, 2013; Degrandi et al, 2013; Kravets et al, 2016), we therefore hypothesize that human GBP1 promotes PV opening and cytosolic access to intravacuolar pathogens.
Extending our previous finding of GBP1 recruiting to the PV, we also localized GBP1 directly to the surface of Tg using AiryScan super-resolution microscopy (Figure 1A). To test whether GBP1 can disrupt Tg PVs and potentially the parasites, we used cytosolic dye CellMask, which is excluded from PVs but enters once the PV membrane (PVM) is disrupted (Figure 1B). As positive control for this new assay, vacuoles were chemically disrupted by detergent-mediated permeabilization resulting in higher fluorescence within the vacuoles as compared to untreated cells (Figure 1B). Increased CellMask dye intensity within naturally disrupted Tg vacuoles could be reliably quantified using our artificial intelligencebased high-throughput image analysis workflow HRMAn (Fisch et al, 2019b), which enabled us to enumerate dye access within thousands of PVs formed upon infection of type I and type II strains of Tg. Analyses of CellMask fluorescence within PVs in IFNγ-primed THP-1 wildtype (WT) cells revealed increased intensities, indicating their disruption (Figure S1). IFNγ-primed THP-1 ΔGBP1 cells showed that Tg vacuoles were not disrupted, as seen by the exclusion of CellMask dye (Figure S1). Doxycycline induced reexpression of GBP1 (THP-1 ΔGBP1+Tet-GBP1 cells) rescued vacuole breakage; as control, empty vector transduced cells (THP-1 ΔGBP1+Tet-EV) behaved like ΔGBP1 cells (Figure S1). We next used Doxycycline-induced expression of mCherry-GBP1 (THP-1 ΔGBP1+Tet-mCH-GBP1 cells) to allow quantification of GBP1-recruitment to Tg and stratify data on whether PVs that were decorated with mCH-GBP1 lost their integrity. Indeed, a population of GBP1+ PVs were unable to exclude CellMask dye clearly indicating loss of membrane integrity (Figure 1C). Taken together, we conclude that GBP1 is contributing to opening of PVs and GBP1-targeted vacuoles preferentially undergo loss of membrane integrity.
Elegant microscopy previously localized murine Gbps directly onto the surface of the Tg plasma membrane (Kravets et al, 2016) similar to our finding of GBP1 recruiting directly to the parasite surface in presumably broken vacuoles. Whether direct recruitment of a GBP to a Tg parasite leads to disruption of the parasite plasma membrane has not been studied. We developed a second novel assay that measures parasite membrane integrity (Figure 1D). In a split-GFP complementation approach, Tg parasites only fluoresce upon access of a GFP11 fragment expressed in host cell cytosol (Figure S2A) to the GFP1-10 fragment expressed in the Tg cytosol (Figure S2B); neither fragment is fluorescent on its own (Romei & Boxer, 2019). If the PV and the Tg membranes are both disrupted, the fragments should assemble to form fluorescent GFP holo-protein (Figure 1D). Indeed, we could observe GFP-fluorescing parasites in IFNγ-primed THP-1 cells (Figure 1E) and quantify the proportion of parasites with GFP fluorescence using high-throughput imaging (Figure 1F). This revealed that Tg only become disrupted in the presence of GBP1 (Figure 1F). Moreover, all parasites within the same vacuole were disrupted suggesting that once PV integrity is lost the Tg within them are susceptible to membrane disruption (Figure 1F). The disruption of parasites was further investigated using flow cytometry of Tg from infected THP-1 cells which showed that parasites disrupted at 6 hours post infection (p.i.) or later (Figure S2C). Plaque assay of sorted parasites confirmed that green-fluorescing Tg were not viable (Figure S2D).
We validated our PV disruption assays by examining the ultrastructure of the vacuole membranes using correlative light and electron microscopy, which revealed ruffled and broken vacuole membranes in cells expressing GBP1 (Figure S3). In THP-1 ΔGBP1, most PVs analyzed by electron microscopy did not show structural defects or loss of membrane integrity (Figure S3). Together, our novel assays indicated that GBP1 contributes to disruption of both the membrane of the PV and the plasma membrane of Tg parasites.
GBP1 does not participate in Salmonella vacuolar escape but targets cytosolic bacteria
Having established an indispensable role for GBP1 in opening of Tg PVs and parasites, we wanted to test if GBP1 also contributed to the escape of STm from Salmonella-containing vacuoles (SCVs). In murine cells, Gbps have been found to both disrupt STm vacuoles, as well as directly recognize bacterial LPS in the cytoplasm (Meunier et al, 2014; Pilla et al, 2014). We used differential permeabilization (Meunier et al, 2014) to determine whether the escape of STm from its vacuole into the cytosol required GBP1. Similar number of cytosolic STm were detected in WT and ΔGBP1 cells, suggesting that GBP1 is dispensable for cytosolic escape of STm (Figure 1G). This indicated that GBP1 has a microbe-specific role in disruption of microbial compartments. Importantly, differential permeabilization revealed that GBP1 was exclusively recruited to cytosolic STm at all time points (Figure 1H). Although these results in human macrophages contrast findings of mGbp involvement in STm infection of mouse macrophages (Meunier et al, 2014), they are in agreement with the lack of a role for GBPs in the opening of bacterial pathogen-containing vacuoles, for example during infection by Legionella (Creasey & Isberg, 2012; Pilla et al, 2014; Feeley et al, 2017; Liu et al, 2018) and Yersinia (Feeley et al, 2017), or directly targeting cytosolic Francisella novicida (Meunier et al, 2015; Man et al, 2016). In a second assay to analyze the capacity of GBP1 to open STm vacuoles, we used the lectin galectin-8 (Gal-8) as a marker for cytosolic bacteria. Studies in human epithelial cells have shown that Gal-8 is recruited to disrupted SCVs, which promotes bacterial xenophagy and growth-restriction(Thurston et al, 2012). Consistent with the previously observed lack of a role for GBP1 in cytosolic escape of STm, similar proportions of STm were decorated with Gal-8 in WT and ΔGBP1 cells (Figure 1I). Temporal studies showed that SCVs were rapidly disrupted (become Gal-8+), but lose this marker over time (Figure 1I), as has been shown before in epithelial cells (Thurston et al, 2012). At later time points as the proportion of Gal-8+ vacuoles decreased, cytosolic STm still retained GBP1 coating (Figure 1J). These single-cell assays revealed that unlike during Tg infection, GBP1 does not contribute to cytosolic escape of STm, but recruits directly to cytosolic STm.
GBP1 promotes access to PAMPs for cytosolic host defense and interacts with caspase-4 on the surface of STm
As Tg infection activates the DNA sensor AIM2 and we demonstrated that GBP1 promotes PV and Tg plasma membrane disruption, we wanted to visualize release of Tg-DNA into the cytoplasm of infected macrophages and subsequent recognition by AIM2. To this end, we labelled Tg-DNA with EdU by growing them in human foreskin fibroblasts (HFFs), whose DNA remains unlabeled as they do not replicate due to contactdependent growth inhibition. Following infection of macrophages with EdU-labelled Tg, we visualized Tg-DNA with Alexa Fluor 647 dye using click-chemistry and quantified macrophages containing cytosolic Tg-DNA (Figure 2A). Approximately 35% of infected macrophages that had at least one PV targeted by GBP1 (GBP1+) contained Tg-DNA in their cytosol at 6h p.i. while uninfected macrophages or infected macrophages without targeted PVs did not show this phenotype (Figure 2A). Infection of myc-AIM2 expressing THP-1 macrophages (Figure S4C) furthermore showed association of the cytosolic DNA sensor with EdU-labelled Tg-DNA in the cytosol (Figure 2A). Taken together, these results corroborate the model that human GBP1 actively ruptures the Tg PV and parasites and releases Tg-DNA into the cytosol for downstream detection by AIM2.
GBP1 thus promotes the sensing of PAMPs and formation of cytosolic signaling platforms also known as supramolecular organizing centers (SMOCs) (Kagan et al, 2014). We investigated the structure of caspase activation SMOCs promoted by GBP1 actions using structured illumination microscopy (SIM). Upon Tg infection, AIM2 detects released Tg-DNA and nucleates the formation of an inflammasome containing ASC and caspase-8. Using SIM, we found that these atypical inflammasome complexes appear similar to previously described inflammasomes containing ASC and caspase-1 or caspase-8 (Man et al, 2013, 2014). We found a “donut”-like ASC ring enclosing caspase-8 in the center (Figure S4A-B). As we could not detect endogenous AIM2 by immunofluorescence microscopy, we resorted to using THP-1 cells expressing myc-AIM2 (Figure S4C), which revealed AIM2 recruitment to ASC specks in Tg-infected macrophages (Figure S4D-E). Altogether, these studies confirm that Tg-DNA is present within the macrophage cytosol as a result of GBP1-mediated disruption of the PVM and Tg membrane resulting in AIM2 activation.
We next decided to contrast GBP1 actions during STm infection, where we previously showed that caspase-4 is directly targeted to GBP1+ STm (Fisch et al, 2019a). The question was whether there is an interaction between GBP1 and caspase-4, which leads to recruitment of the LPS-sensor, and whether caspase-4 is recruited directly on the surface of STm. Indeed, 3D-rendered SIM imaging demonstrated GBP1 recruited caspase-4 directly to the surface of STm (Figure S4F). Bacteria were completely covered in GBP1, with a high degree of colocalisation with YFP-CASP4C258S (Figure S4F). Interestingly, immunofluorescence staining of Salmonella-LPS using a monoclonal antibody revealed that GBP1-CASP4+ bacteria stained not at all or poorly, suggesting access to the epitope to be blocked (Figure 2B+C). STm not decorated with caspase-4 but positive for GBP1 however were stained with anti-LPS antibody (Figure 2B). As caspase-4 can directly bind LPS with its CARD (Shi et al, 2014), this finding is consistent with the possibility that caspase-4 on the bacterial surface precludes antibody-mediated staining of LPS (Figure 2C).
Caspase-4 recruitment to bacteria mirrored that of GBP1. The majority of cytosolic (Gal-8+) GBP1 + STm were also positive for caspase-4. Notably, GBP1-caspase-4 were retained on STm over time even though Gal-8 staining had reduced (Figure 2D), which suggested that GBP1-caspase-4 are present on cytosolic STm longer during infection.
Our previous work showed that the translocation of GBP1 to STm and enhanced pyroptosis requires its GTPase function and isoprenylation (Fisch et al, 2019a) but did not determine whether these effects led to deficient caspase-4 targeting to STm. THP-1 ΔGBP1 cells reconstituted with Dox-inducible variants of GBP1 that lacked GTPase activity (GBP1K51A) or isoprenylation sites (GBP1C589A or GBP1Δ589-592; Figure S5A) revealed that none of these variants supported the recruitment of caspase-4 (Figure S5B+C). Taken together, through single-cell comparative analyses we have established that GBP1-targeting to Tg promotes the release of parasite DNA into the cytosol, whereas GBP1-targeting to STm enables caspase-4 recruitment to cytosolic bacteria. The reduced LPS staining on bacteria further suggest that GBP1 facilitates access to bacterial LPS ligand to caspase-4.
We additionally decided to use an unbiased proteomics approach to identify GBP1 binding-partners and other proteins recruited to GBP1-caspase-4 SMOCs on cytosolic STm. For this, we immunoprecipitated Dox-inducible Flag-GBP1 from STm-infected THP-1 ΔGBP1 cells following protein cross-linking (Figure 2E and Figure S6). Comparing infected to uninfected cells and correcting for nonspecific binding of proteins to the Flag-beads, identified several proteins that were enriched in infected samples above the significance cut-off (P ≤ 0.01, Figure 2E). Some of these proteins are known GBP1 interacting proteins such as γ-Actin (ACTG1), Myosin light polypeptide 6 (MYL6) and Myosin regulatory light chain 12a (MYL12A) (Ostler et al, 2014; Forster et al, 2014). The most prominent infection-specific GBP1 interaction partner we detected was caspase-4, which supported results from microscopy. We did not identify any other proteins that may be interacting with GBP1 and caspase-4 on the STm surface. To establish that the detected interaction is physiologically relevant during infection, we repeated immunoprecipitation experiments using antibodies against endogenous GBP1 from THP-1 WT cells (this time without prior cross-linking). In agreement with our proteomics results, endogenous GBP1 interacted with caspase-4 only during STm infection pointing towards its specific and crucial role in enabling LPS-sensing by caspase-4 (Figure 2F).
Taken together these results indicate that GBP1 has two modes of assembling caspasecontaining complexes depending on the infecting pathogen: (1) by proxy through Tg vacuole and parasite membrane disruption and release of Tg-DNA into the cytosol to trigger activation of the AIM2 inflammasome and (2) by direct recruitment and interaction with caspase-4 on the surface of STm.
Caspase-1, but not caspase-4, can cleave GBP1 during infection
The noncanonical inflammasome in mouse macrophages involves sequential activation of capsase-4/11 and caspase-1, wherein caspase-4/11 activation precedes caspase-1. As both caspases are independently activated in IFNγ-stimulated human macrophages infected with STm (Fisch et al, 2019a) we wanted to investigate whether a crosstalk existed between the two pathways. This was also pertinent given a previous report of caspase-1-mediated cleavage of GBP1 in human umbilical vein endothelial cells (HUVECs) (Naschberger et al, 2017); however, the functional consequences of GBP1 proteolysis during infection were not investigated in that study. Noncanonical inflammasome activation during LPS-transfection is a cell-intrinsic process that involves K+ efflux-mediated activation of caspase-1 (Kayagaki et al, 2011; Rühl & Broz, 2015), and release of inflammasome specks can activate inflammasomes in neighboring cells (Venegas et al, 2017). We therefore first wanted to verify that caspase-1 and caspase-4 are activated within the same STm-infected macrophage. Our results showed a perfect correlation between bacterial targeting by GBP1-CASP4 and pyroptosis and we indirectly quantified caspase-1 activation by measuring ASC speck formation. Indeed, single-cell microscopy confirmed that 80 % of cells with GBP1+cASp4+ STm (indicating active caspase-4), also had ASC specks (active caspase-1) (Figure 3A). Notably, caspase-4 was not recruited to ASC specks, which is consistent with previous work (Thurston et al, 2016) (Figure 3A+B). As these results suggested dual activation of caspase-1 and caspase-4 in the same cell, we investigated whether and how GBP1 proteolysis might affect caspase-4 recruitment to STm.
We therefore examined the impact of caspase-1-mediated cleavage of GBP1 at the surface exposed Asp192 residue that generates a stable p47 GBP1 C-terminal fragment (Figure 3C). Of note, phylogenetic analysis of representative GBPs (Shenoy et al, 2012), revealed that the Asp residue required for caspase-1 cleavage-site was present in all primates and absent in most rodents, including mice (Figure 3C). Infection of THP-1 with STm indeed confirmed that GBP1 is cleaved into a ~47 kDa fragment that is detected in supernatants. GBP1 proteolysis could be prevented by silencing caspase-1, but not caspase-4, confirming the dominant role of caspase-1 in the process (Figure 3D); LPS+Nigericin treatment served as a positive control and also led to p47 GBP1 production. As expected with the lack of caspase-1 activation during Tg infection (Fisch et al, 2019a), GBP1 proteolysis could not be detected in Tg-infected THP-1 cells (Figure 3D).
To confirm proteolysis of GBP1 at the Asp192 residue, we used a non-cleavable (D192E) variant. We created THP-1 ΔGBP1 cells expressing the non-cleavable GBP1D192E mutant without or with an mCherry tag (THP-1 ΔGBP1+Tet-GBP1D192E and THP-1 ΔGBP1+Tet-mCH-GBP1D192E cells; Figure S7A). Immunoblotting of GBP1 from STm-infected IFNγ-primed macrophages revealed caspase-1 activation and formation of p47 GBP1 from cells expressing wildtype GBP1 but not GBP1D192E (Figure 3E). Together, these results point towards the specificity of caspase-1 in cleaving GBP1 and that neither caspase-4 nor caspase-8 (active during Tg infection) can replace its role.
Caspase-1-cleaved GBP1 fragments cannot traffic to microbial vacuoles or mediate cell death
As GBP1 can be cleaved by caspase-1, we wanted to investigate how this affects the pathogen-proximal activities of GBP1 in enabling PAMP access and triggering cell death. We infected mCH-GBP1D192E expressing cells with STm and quantified GBP1 recruitment to bacteria. The proportion of GBP1+ STm was similar in cells expressing GBP1 WT and GBP1D192E (Figure 4A). However, the mean fluorescence intensity of mCH-GBP1 around decorated STm was markedly higher in cells expressing the GBP1D192E variant (Figure 4A), even though the expression and fluorescence of GBP1 WT and GBP1D192E was comparable in uninfected cells (Figure S7B). In agreement with increased GBP1 amounts covering cytosolic bacteria, STm-infected GBP1D192E cells underwent higher pyroptosis than wildtype cells but released similar levels of IL-1β (Figure 4B). This finding is consistent with a major role for GBP1 in promoting caspase-4-driven pyroptosis, but not canonical caspase-1 activation, which is responsible for IL-1β production (Kortmann et al, 2015; Reyes Ruiz et al, 2017). These results led us to speculate that cleavage of GBP1 reduces the cellular pool of functional full-length GBP1, and its cleaved fragments do not support cell death-related roles. Indeed, ΔGBP1 cells reconstituted with GBP11-192 or GBP1193-592 with or without mCherry-tag using our Dox-inducible system (Figure S7C) revealed that neither fragment was recruited to STm (Figure 4C) nor supported enhanced pyroptosis (Figure 4D). As caspase-1 is not activated during Tg infection, we anticipated that Tg targeting and apoptosis would be similar in cells expressing GBP1 WT or GBP1D192E. Indeed, the proportion of Tg-PVs decorated with GBP1 WT and GBP1D192E was similar (Figure 4E) and apoptosis remained unaffected (Figure 4F).
In summary, these results suggested that active caspase-1 cleaves a portion of cellular GBP1 and generates protein fragments that cannot (1) target cytosolic STm, (2) subsequently recruit caspase-4 and (3) enhance pyroptosis induction. Because IL-1β maturation was not affected by GBP1D192E mutation, we speculate that this caspase-1-driven feedback mechanism balances cell death and IL-1β secretion during STm infection. Moreover, as caspase-1 does not contribute to cell death during Tg-infection, this feedback regulatory mechanism is pathogen-specific.
Discussion
IFNγ-inducible GBPs have emerged as important proteins in host defense against a range of pathogens in the murine system (Meunier & Broz, 2016; Pilla-Moffett et al, 2016; Man et al, 2017; Saeij & Frickel, 2017; Tretina et al, 2019). In this study we have established that human GBP1 is essential for the breakdown of PVMs and Tg parasites through the use of two new single-cell assays combined with the artificial intelligence-driven image analysis pipeline HRMAn that are adaptable for other pathogens (Fisch et al, 2019b). In contrast to Tg, GBP1 only decorates cytosolic STm and forms a complex with caspase-4, which it recruits onto the surface of bacteria. Caspase-1, but not caspase-4, also cleaves GBP1 at Asp192 to limit pyroptosis. These findings provide important new insights on this key GTPase in human macrophages.
As the forerunner of the human GBP family, GBP1 has been extensively studied structurally and biochemically. For instance, high-resolution structural and biophysical studies point towards an exceptionally fast GTP hydrolysis capacity to form GDP and GMP (Cheng et al, 1991; Schwemmlel & Staeheli, 1994; Praefcke et al, 1999; Ghosh et al, 2006), ability to homo- and hetero-oligomerize (Wehner et al, 2012; Ince et al, 2017; Barz et al, 2019; Lorenz et al, 2019), and undergo isoprenylation (Nantais et al, 1996;
Olszewski et al, 2006; Tripal et al, 2007; Britzen-Laurent et al, 2010). Further, GBP1 is a dynamin-like GTPase and may actively alter biological membranes (Huang et al, 2019); indeed, recombinant farnesylated GBP1 can bend giant unilamellar vesicle membranes in vitro (Shydlovskyi et al, 2017). We add the role of recruiting caspase-4 to STm dependent on functional GTPase activity and isoprenylation, which is in line with previous findings on mouse and human GBPs in vitro and in cellulo (Nantais et al, 1996; Stickney & Buss, 2000; Prakash et al, 2000; Britzen-Laurent et al, 2010; Fres et al, 2010; Piro et al, 2017; Shydlovskyi et al, 2017; Kohler et al, 2019).
Targeting of Tg vacuoles by murine GBPs and their interplay with the IRG proteins has been extensively studied (MacMicking et al, 2003; Bernstein-Hanley et al, 2006; Singh et al, 2006; Degrandi et al, 2007; Henry et al, 2007; Miyairi et al, 2007; Shenoy et al, 2007; Coers et al, 2008; Hunn et al, 2008; Al-Zeer et al, 2009; Tiwari et al, 2009; Khaminets et al, 2010; Lapaquette et al, 2010; Singh et al, 2010; Brest et al, 2011; Traver et al, 2011; Virreira Winter et al, 2011; Kim et al, 2012a; Haldar et al, 2013, 2014). Uniquely in mice, two chromosomal loci each encode members of the GBP (~11 genes on Chr 3 and Chr 5) and IRG (~23 genes on Chr 11 and Chr 18) families. Deletion of all mGbps on Chr3 (ΔGbpChr3) abrogates Tg vacuole rupture in macrophages and these mice are highly susceptible to Tg infection (Yamamoto et al, 2012). Single deletion of mGbp1 (Selleck et al, 2013) or mGbp2 (Degrandi et al, 2013) also results in enhanced susceptibility to Tg in vivo and in vitro. mGbp2 can homodimerize or form heterodimers with mGbp1 or mGbp5 before recruitment and attack of Tg vacuoles (Kravets et al, 2016). However, in the mouse, the hierarchical recruitment of IRG family GTPases to Tg vacuoles precedes the recruitment of GBP family members. No GBPs are recruited to Tg in Irgm1/Irgm3-/- murine cells pointing to their pivotal role in this process (Haldar et al, 2013) In addition to the absence of IRGs in humans, a direct role for individual GBPs in Tg vacuole disruption has not been demonstrated for GBPs before, even though mouse Gbp2 has been found to localize inside Tg (Kravets et al, 2016). Indeed, while targeting of murine Gbps to Tg vacuoles with subsequent vacuolar lysis has been demonstrated for mGbp1, 2 and the collective Gbps located on chromosome 3, parasite plasma membrane lysis has not been observed before (Yamamoto et al, 2012; Degrandi et al, 2013; Selleck et al, 2013).
During STm infection GBP1 only targeted cytosolic bacteria even though a proportion of bacteria remained vacuolar. Our finding that GBP1 only targets bacteria already in the cytosol are consistent with bacterial staining with Gal-8, which binds to glycans on endogenous damaged membranes and recruits other proteins, including the autophagy machinery (Thurston et al, 2012). Furthermore, mouse Gbp-recruitment is reduced in murine macrophages lacking Gal-3, which normally labels Legionella (Creasey & Isberg, 2012; Pilla et al, 2014; Feeley et al, 2017; Liu et al, 2018) or Yersinia (Feeley et al, 2017) expressing secretion systems that trigger damage of bacterial-containing vacuoles. Work with bacterial mutants that readily access the cytosol, such as Legionella pneumophila ΔsdhA and STm ΔsifA, also revealed no differences in cytosolic bacteria in mouse ΔGbpChr3 macrophages (Pilla et al, 2014). Similarly, release of Francisella novicida into the cytosol was shown to be independent of mouse Gbps (Man et al, 2015; Meunier et al, 2015). It is plausible that in human macrophages GBP1 is dispensable for release of STm into the cytosol even though Gbps encoded at mouse Chr3 and mGbp2 have previously been implicated in this process in murine cells (Meunier et al, 2014). It is tempting to speculate that human GBP1 recruitment to vacuolar STm is prevented by a bacterial virulence factor. Indeed, anti-GBP1 bacterial effectors have been identified in Shigella flexneri (Li et al, 2017; Piro et al, 2017; Wandel et al, 2017). Further work should investigate whether other human GBPs also assemble alongside or assist GBP1 during STm infection.
Our work also shows unique GBP1 action during infection by these diverse pathogens whose distinct PAMPs are recognized by downstream innate immune pathways. Click-chemistry revealed that Tg-DNA is present in the cytoplasm of GBP1-expressing macrophages that subsequently induces the assembly of the atypical AIM2-ASC-caspase-8 SMOC and apoptosis. Super-resolution imaging structure of a caspase-8 containing AIM2 inflammasome closely resembles previously published structures of caspase-8 in NLRP3/NLRC4 inflammasomes (Man et al, 2013, 2014), revealing donut-like ASC rings enclosing AIM2 and caspase-8. Super-resolution microscopy during STm infection showed that GBP1 and caspase-4 formed a dense coat on STm, which reduced bacterial staining with anti-LPS antibody. Whether this was due to reduced antibody access due to the GBP1/caspase-4 coat or blocking of the LPS epitope by caspase-4 cannot be definitively distinguished. As caspase-4 by itself could not recruit to the bacteria, we speculate that GBP1 is involved in exposing parts of the LPS that are buried deeper within the membrane potentially through direct interaction with LPS as has been suggested for mGbp5 (Santos et al, 2018). We therefore hypothesize that GBP1 ‘opens’ the bacterial outer membrane for caspase-4 to gain access to otherwise hidden PAMPs.
Our results also uncovered a physiological role for GBP1 proteolysis by caspase-1 that was previous reported in vitro using HUVEC cells and in vivo from cerebrospinal fluid of meningitis patients (Naschberger et al, 2017), which we confirmed during natural infection of macrophages with STm. Not surprisingly, GBP1 is not proteolyzed during Tg infection due to the absence of active caspase-1 in this setting (Fisch et al, 2019a). Notably, despite the 40-98 % sequence similarity between human and mouse GBPs (Shenoy et al, 2007; Kim et al, 2011), and the conservation of Asp192 in other primate GBP1 sequences, Asp192 is absent in the closest murine homologue, mGbp2 (Olszewski et al, 2006), which is therefore unlikely to be regulated in this manner. Intriguingly, this finding mirrors our recent identification of the proteolysis of human, but not mouse p62, by caspase-8 at a conserved residue found in other mammalian p62 sequences (Sanchez-Garrido et al, 2018). During STm infection, caspase-1 plays a dominant role in IL-1β maturation whereas IFNγ-induced GBP1 enhances caspase-4-driven pyroptosis. As a result, caspase-1-dependent proteolysis of GBP1 affected pyroptosis but not IL-1β maturation. Furthermore, GBP1 fragments produced by caspase-1 failed to target STm, recruit caspase-4 and support pyroptosis. Thus, besides directly aiding the release or access to PAMPs for detection by caspases, GBP1 itself is a target of caspase-1 and a key regulatory hub that modulates host cell death. This contrasts our discovery of the ubiquitin conjugating enzyme UBE2L3 as an indirect target of caspase-1 that specifically controls IL-1β production but not pyroptosis (Eldridge et al, 2017). At the whole organism level, these mechanisms potentially enable differential responses based on the strength of the activating stimulus. Enhanced IL-1β or IL-18 production for adaptive immunity may be balanced by cell death that could enable pathogen uptake by other cell types such as neutrophils. Studies on cellular targets of caspases may therefore provide new insights on homeostasis and disease.
Common themes also emerge from work on human and mouse GBPs. For instance, human GBP1 and mouse Gbp2 accumulate on vesicles generated through sterile damage, which suggests they could detect endogenous luminal ligands in the cytosol, for example endogenous sulfated lipids (Bradfield, 2016). The presence of Gal-3/Gal-8 and GBPs at sites of damaged membranes suggests GBPs may be assisted in sensing damage by other proteins, including IFN-induced genes. Undoubtedly, future work in the area will focus on finding how human GBPs are targeted to diverse microbes, the ligands they sense and how they are regulated.
Author Contributions
DF, ARS and EMF conceived the idea for the study, DF, BC, MCD and VE performed experiments. HB, MY provided essential reagents. LMC and APS provided essential expertise and equipment. DF, ARS and EMF analyzed and interpreted the data and wrote the manuscript. All authors revised the manuscript.
Conflict Of Interest
The authors declare that they have no conflict of interest.
Funding
This work was supported by the Francis Crick Institute, which receives its core funding from Cancer Research UK (FC001076 to EMF, FC001999 to LMC and APS), the UK Medical Research Council (FC001076 to EMF, FC001999 to LMC and APS), and the Wellcome Trust (FC001076 to EMF, FC001999 to LMC and APS). EMF was supported by a Wellcome Trust Career Development Fellowship (091664/B/10/Z). DF was supported by a Boehringer Ingelheim Fonds PhD fellowship. ARS would like to acknowledge support from the MRC (MR/P022138/1) and Wellcome Trust (108246/Z/15/Z). MY was supported by the Research Program on Emerging and Re-emerging Infectious Diseases (JP18fk0108047) and Japanese Initiative for Progress of Research on Infectious Diseases for global Epidemic (JP18fk0108046) from Agency for Medical Research and Development (AMED). HB was supported by Grant-in-Aid for Scientific Research on Innovative Areas (17K15677) from Ministry of Education, Culture, Sports, Science and Technology.
Materials And Methods
Cells, parasites and treatments
THP-1 (TIB-202, ATCC) were maintained in RPMI with GlutaMAX (Gibco) supplemented with 10% heat-inactivated FBS (Sigma), at 37°C in 5% CO2 atmosphere. THP-1 cells were differentiated with 50 ng/mL phorbol 12-myristate 13-acetate (PMA, P1585, Sigma) for 3 days followed by a rested for 2 days in complete medium without PMA. Cells were not used beyond passage 20. HEK293T and human foreskin fibroblasts (HFF) were maintained in DMEM with GlutaMAX (Gibco) supplemented with 10% FBS at 37°C in 5% CO2 atmosphere. Tg expressing luciferase/eGFP (RH type I and Prugniaud (Pru) type II) were maintained by serial passage on monolayers of HFF cells. All cell culture was performed without addition of antibiotics unless otherwise indicated. Cell lines were routinely tested for mycoplasma contamination by PCR and agar test. An overview of all cell lines made/ used in this study is provided in Table S1.
Cells were stimulated for 16 h prior to infection in complete medium at 37°C with addition of 50 IU/mL human IFNγ (285-IF, R&D Systems). Induction of Doxycycline-inducible cells was performed with 200 ng/mL Doxycycline overnight (D9891, Sigma). To chemically activate caspase-1, cells were treated with 10 μM Nigericin (N1495, Invitrogen) and 100 μg/mL LPS-Sm (IAX-100-011, Adipogen).
Toxoplasma gondii infection
Parasite were passaged the day before infection. Tg tachyzoites were harvested from HFFs by scraping and syringe lysing the cells through a 25 G needle. The Tg suspension was cleared by centrifugation at 50 x g for 5 min and then the parasites were pelleted by centrifugation at 550 x g for 7 min from the supernatant, washed with complete medium, and finally resuspended in fresh medium. Viable parasites were counted with trypan blue and used for infection at a multiplicity of infection (MOI) of 3 for most experiments or 1 for immunofluorescence imaging. Infection was synchronized by centrifugation at 500 x g for 5 min. Two hours after infection, extracellular parasites were removed with three PBS washes.
Flow cytometry and sorting
For flow cytometry analysis of GFP-fluorescence, Tg ΔHpt+GFP1-10 were harvested from host cells by syringe lysis, washed twice with warm PBS and then re-suspended in PBS + 1% BSA. Parasites were analyzed on a LSR Fortessa (BD Biosciences) and data were processed using FlowJo version 10.3 (FlowJo, LLC). For viability determination of GFP-fluorescing versus non-fluorescing Tg the parasites were harvested and prepared identically, sorted on a FACSAria™ III (BD Biosciences) based on their GFP-fluorescence and then plated onto HFFs grown confluent in wells of a 24-well plate. 5 days post infection of the HFFs, cells were fixed with ice-cold methanol and stained with crystal violet. Following 5 washes with PBS, plaques were imaged on a GelCount™ Colony Counter (Oxford Optronix) and cell covered area determined using FIJI ImageJ.
Salmonella Typhimurium infection
STm SL1344-GFP (with pFPV25.1 plasmid) was maintained under Ampicillin selection (11593027, Gibco). STm SL1344 wildtype strain was maintained in the presence of streptomycin (11860038, Gibco) selection. One day before infection bacteria from a single colony were inoculated into 9 mL LB and grown overnight at 37°C. The overnight culture was diluted 1:50 into LB + 300 mM NaCl (746398, Sigma) and grown shaking in a closed container until an OD600 of 0.9. Bacteria were harvested by centrifugation at 1000 x g for 5 min, washed with serum-free cell culture medium twice and re-suspended in 1 mL medium. Cells were infected with STm at an MOI of 30 and infections were synchronized by centrifugation at 750 x g for 10 min. Infected cells were washed 30 min post-infection three times with warm PBS (806552, Sigma) to remove extracellular bacteria and fresh medium containing 100 μg/mL Gentamicin (15750060, Gibco) was added for 1 h. Medium was then replaced with medium containing 10 μg/mL gentamicin and the infection continued for indicated times. Bacterial MOI used for infections were confirmed by plating on LB agar plates.
Creation of transgenic Toxoplasma gondii
To create new Tg lines that constitutively express non-fluorescent GFP1-10 fragment, the GFP1-10 ORF was amplified from pEGFP-C1 (Clontech) and Gibson-assembled into NsiI and PacI digested pGRA-HA-HPT (a gift from Moritz Treeck) (Coppens et al, 2006), to have expression of the ORF under control of the TgGRA1 promoter.
Next the plasmid was transfected into type II (Pru) Tg ΔHpt (a gift from Moritz Treeck) using nucleofection as established by Young et al (Young et al, 2019): The plasmid was linearized using PsiI-V2 and purified using phenol-chloroform precipitation and resuspended in P3 solution (Lonza). Successful linearization was confirmed using agarose-gel electrophoresis. Next, type II (Pru) Tg ΔHpt were harvested from HFFs by syringe lysis and washed with PBS twice and then 5×106 parasites resuspended in P3 solution. Prior to nucleofection, 25 μg linearized DNA were added to the parasites and then nucleofected using 4D-NucleofectorTM (Lonza) with setting EO-115. Transfected parasites were then incubated for 12 minutes at room temperature, followed by platting onto fresh HFF cells into a T25 tissue culture flask. The next day, medium was replaced with complete DMEM containing 50 μg/mL xanthine and mycophenolic acid (MPA) each for selection. The selection medium was replaced every two days and the parasites passaged normally for two weeks when all Tg in the untransfected control had died. Successful integration of the plasmid and expression of GFP1-10 was confirmed by immunofluorescence and immunoblotting.
Creation of new cell lines
Creation of the Dox-inducible GBP1 and caspase-8 cell lines
THP-1 ΔGBP1+Tet-GBP1 and THP-1 ΔGBP1+Tet-mCH-GBP1 have been published before and the THP-1 WT+Tet-CASP8-Flag cells were created identically using Lentiviral transductions (Fisch et al, 2019a).
To create the caspase-8-3xFlag expressing Dox-inducible plasmid (pLenti-Tet-CASP8-3xFlag), the empty vector backbone was digested with BamHI, CASP8 ORF was amplified from pcDNA3-CASP8 by PCR (Addgene #11817, a gift from Guy Salvesen) (Stennicke & Salvesen, 1997), 3xFlag was amplified from lentiCRISPRv2 (Addgene #52961, a gift from Feng Zhang) (Sanjana et al, 2014) and all fragments assembled with a Gibson assembly. Similarly, to create the 3xFlag-GBP1 expressing Dox-inducible plasmid (pLenti-Tet-3xFlag-GBP1), the empty vector backbone was digested with BamHI, GBP1 ORF was amplified from pGene-GBP1 by PCR (Frickel lab), 3xFlag was amplified from lentiCRISPRv2 and all fragments assembled with a Gibson assembly. In the same way, GBP1 fragments 1-192 and 193-592 were amplified from pGene-GBP1 (Frickel lab) and Gibson assembled into BamHI digested pLenti-Tet (Fisch et al, 2019a) with and without addition of a mCherry tag. The obtained plasmids were then transduced into THP-1 ΔGBP1+Tet cells (Fisch et al, 2019a) using lentiviral particles.
To make the cells expressing GBP1D192E, the pLenti-Tet-mCH-GBP1 and pLenti-Tet-GBP1 plasmids (Fisch et al, 2019a) were mutated using site-directed mutagenesis and transduced into the THP-1 ΔGBP1+Tet target cells using Lentiviral transduction as described before (Fisch et al, 2019a). To make cells expressing YFP-CASP4C258S and mutated GBP1 versions, THP-1 ΔGBP1+Tet-mCH-GBP1K51A, +Tet-mCH-GBP1C589A or +Tet-mCH-GBP1Δ589-592 (Fisch et al, 2019a) were transduced with pMX-CMV-YFP-CASP4C258S (Fisch et al, 2019a) using retroviral particles. All primers used for cloning PCRs can be found in Table S3.
Creation of myc-AIM2 expressing cell line
To create a lentiviral vector for constitutive expression of myc-AIM2, the ORF was amplified from pcDNA3-myc-AIM2 (Addgene #73958, a gift from Christian Stehlik) (Khare et al, 2014) and Gibson assembled into BstBI and BsrGI digested pLEX-MCS-ASC-GFP (Addgene #73957, a gift from Christian Stehlik) (de Almeida et al, 2015) to create pLEX-MCS-myc-AlM2. The newly made vector was then transduced into THP-1 +Tet-CASP8-Flag cells to create THP-1+Tet-CASP8 + myc-AIM2 cells using Lentiviral transduction as described above.
Creation of GFP11 expressing cell lines
To create an lentiviral vector for constitutive expression of GFP11, the sgRNA cassette from lentiCRISPRv2 was removed by digestion with KpnI and EcoRI and the plasmid re-ligated using annealed repair oligo pair 1 (see Table S3) and Quick Ligation™ Kit (M2200L, NEB). Next, the Cas9-ORF was removed by digestion with XbaI and BamHI and again the vector re-ligated using annealed repair oligo pair 2 (see Table S3), also adding a multiple cloning site, which created the vector pLenti-P2A-Puro. Next, the GFP11 ORF was amplified from pEGFP-C1 (Clontech) and ligated into BamHI and XbaI digested pLenti-P2A-Puro, to have the GFP11-ORF in frame with the P2A-Puro cassette, for Puromycin-selectable, constitutive expression of GFP11. The newly made vector was then transduced into THP-1 WT and THP-1 ΔGBP1+Tet-GBP1 cells using Lentiviral transduction as described above.
Real-time cell death assays and IL-1β ELISA
To measure live kinetics of cell death, 60,000 cells were seeded per well of a black-wall, clear-bottom 96-well plate (Corning) for differentiation with PMA, treated and infected as described above. Medium was replaced with phenol-red-free RPMI supplemented with 5 μg/mL propidium iodide (P3566, Invitrogen). The plate was sealed with a clear, adhesive optical plate seal (Applied Biosystems) and placed in a plate reader (Fluostar Omega, BMG Labtech) pre-heated to 37°C. PI fluorescence was recorded with top optics every 15 min for times as indicated.
Apoptosis kinetics were analyzed using the RealTime-Glo™ Annexin V Apoptosis Assay (JA1001, Promega) according to the manufacturer’s instruction. Simultaneously with infection, detection reagent was added. Luminescence was measured using a Fluostar Omega plate reader (BMG Labtech). No-cell, medium-only controls were used for background correction.
For IL-1β ELISA, the cell culture supernatant was harvested, cleared by centrifugation at 2000 x g for 5 minutes and diluted in the buffer provided with the ELISA kit. ELISA was performed according to the manufacturer’s instruction. IL-1β ELISA kit was from Invitrogen (#88-7261, detection range 2 - 150 pg/mL).
Immunoblotting and gel staining
For immunoblotting, 0.5×106 cells were seeded per well of a 48-well plate, differentiated with PMA, pre-treated and infected. Cells were washed with ice-cold PBS and lysed for 5 min on ice in 50 μL RIPA buffer supplemented with protease inhibitors (Protease Inhibitor Cocktail set III, EDTA free, Merck) and phosphatase inhibitors (PhosSTOP, Roche). Lysates were cleared by centrifugation at full speed for 15 min at 4°C. BCA assay (Pierce BCA protein assay kit, 23225, Thermo Scientific) was performed to determine protein concentration. 10 μg of total protein per sample were run on Bis-Tris gels (Novex, Invitrogen) in MOPS running buffer and transferred on Nitrocellulose membranes using iBlot transfer system (Invitrogen). Membranes were blocked with either 5% BSA (A2058, Sigma) or 5% dry-milk (M7409, Sigma) in TBS-T for at least 1 h at room temperature. Incubation with primary antibodies was performed at 4°C overnight (all antibodies used in this study can be found in Table S2). Blots were developed by washing the membranes with TBS-T, probed with 1:5000 diluted secondary antibodies in 5% BSA in TBS-T and washed again. Finally, the membranes were incubated for 2 minutes with ECL (Immobilon Western, WBKLS0500, Millipore) and luminescence was recorded on a ChemiDoc MP imaging system (Biorad). For silver staining of protein gels, following SDS-PAGE, the gels were washed in ddH2O and then silver stained following the manufacturers instruction (Silver Stain Plus Kit, 1610449, Biorad).
For immunoblots of culture supernatants, cells were treated in OptiMEM (1105802, Gibco) without serum. Proteins in the supernatants were precipitated with 4x volume cold acetone (V800023, Sigma) overnight at −20°C, and pelleted by centrifugation. Pellets were air dried and re-suspended in 40 μL 2x Laemmli loading dye.
Identification of GBP1 interacting proteins by mass spectrometry
Sample preparation
10×106 THP-1 ΔGBP1+Tet-Flag-GBP1 cells were seeded in 6-well plates and differentiated, pre-treated with IFNγ and Doxycycline and infected with STm as described before. 2 hours p.i. the interacting proteins were crosslinked with 1% formaldehyde (28906, Thermo Scientific) for 10 minutes at room temperature and the reaction quenched by addition of 125 mM glycine (Sigma). Cell were washed in ice-cold PBS and scraped from the plates. Cells were then pelleted by centrifugation and washed in PBS. Whole-cell lysates were prepared by adding 500 uL lysis buffer (1% Triton X-100, 20 mM Tris-HCl [pH 8], 130 mM NaCl, 1 mM dithiothreitol, 10 mM sodium fluoride, protease inhibitors (Protease Inhibitor Cocktail set III, EDTA free, Merck), phosphatase inhibitor cocktails (PhosSTOP, Roche)) and incubation for 15 minutes on ice. Lysates were cleared by centrifugation and then added to Flag(M2)-agarose beads (A2220, Sigma) washed three times with lysis buffer. Flag-GBP1 was captured by incubation on a rotator overnight at 4°C. Beads were then washed once with lysis buffer, three times with lysis buffer containing 260 mM NaCl and then again twice with lysis buffer. Proteins were eluted using 200 μg/mL 3xFlag peptide (F4799, Sigma) in lysis buffer by incubation on an orbital shaker (1400 rpm) for 2 hours at room temperature. Samples were then prepared by adding loading dye containing 5% β-Mercaptoethanol (Sigma) to reverse crosslinking and run on a 12% Bis-Tris polyacrylamide gel until the running front had entered the gel roughly 5 mm.
Trypsin digestion
Samples on the SDS-PAGE were excised as three vertical lanes each. The excised gel pieces were destained with 50% acetonitrile/50 mM ammonium bicarbonate, reduced with 10 mM DTT, and alkylated with 55 mM iodoacetamide. After alkylation, the proteins were digested with 250 ng of trypsin overnight at 37°C. The resulting peptides were extracted in 2% formic acid, 1% acetonitrile and speed vacuum dried. Prior to analysis the peptides were reconstituted in 50 μl of 0.1% TFA.
Mass spectrometry
The peptides were loaded on a 50 cm EASY-Spray™ column (75 μm inner diameter, 2 μm particle size, Thermo Fisher Scientific), equipped with an integrated electrospray emitter. Reverse phase chromatography was performed using the RSLC nano U3000 (Thermo Fisher Scientific) with a binary buffer system at a flow rate of 275 nL/min. Solvent A was 0.1% formic acid, 5% DMSO, and solvent B was 80% acetonitrile, 0.1% formic acid, 5% DMSO. The in-gel digested samples were run on a linear gradient of solvent B (2 - 30%) in 95.5 min, total run time of 120 min including column conditioning. The nano LC was coupled to an Orbitrap Fusion Lumos mass spectrometer using an EASY-Spray™ nano source (Thermo Fisher Scientific). The Orbitrap Fusion Lumos was operated in data-dependent acquisition mode acquiring MS1 scan (R=120,000) in the Orbitrap, followed by HCD MS2 scans in the Ion Trap. The number of selected precursor ions for fragmentation was determined by the “Top Speed” acquisition algorithm with a cycle time set at 3 seconds. The dynamic exclusion was set at 30s. For ion accumulation the MS1 target was set to 4×105 ions and the MS2 target to 2×103 ions. The maximum ion injection time utilized for MS1 scans was 50 ms and for MS2 scans was 300 ms. The HCD normalized collision energy was set at 28 and the ability to inject ions for all available parallelizable time was set to “true”.
Data processing and analysis
Orbitrap .RAW files were analyzed by MaxQuant (version 1.6.0.13), using Andromeda for peptide search. For identification, peptide length was set to 7 amino acids, match between runs was enabled and settings were kept as default. Parent ion and tandem mass spectra were searched against UniprotKB Homo sapiens and Salmonella Typhimurium databases. For the search the enzyme specificity was set to trypsin with maximum of two missed cleavages. The precursor mass tolerance was set to 20 ppm for the first search (used for mass re-calibration) and to 6 ppm for the main search. Product mass tolerance was set to 20 ppm. Carbamidomethylation of cysteines was specified as fixed modification, oxidized methionines and N-terminal protein acetylation were searched as variable modifications. The datasets were filtered on posterior error probability to achieve 1% false discovery rate on protein level. Quantification was performed with the LFQ algorithm in MaxQuant using three replicate measurements per experiment.
Quantitative RT-PCR (qRT-PCR)
RNA was extracted from 0.25×106 cells using Trizol reagent (15596026, Invitrogen). RNA (1 μg) was reverse transcribed using high-capacity cDNA synthesis kit (4368813, Applied Biosystems). qPCR used PowerUP SYBR green (A25742, Applied Biosystems) kit, 20 ng cDNA in a 20 μL reaction and primers (all primer used for qPCR can be seen in Table S3) at 1 μM final concentration on a QuantStudio 12K Flex Real-Time PCR System (Applied Biosystems). Recorded Ct values were normalized to Ct of human HPRT1 and data plotted as ΔCt (Relative expression).
siRNA transfection
Cells were transfected with siRNAs two days prior to infection, at the same time the THP-1 differentiation medium was replaced with medium without PMA. All siRNAs were used at a final concentration of 30 nM. For transfection, a 10x mix was prepared in OptiMEM containing siRNA(s) and TransIT-X2 transfection reagent (MIR 600x, Mirus) in a 1:2 stoichiometry. All siRNAs used in this study can be found in Table S3.
Fixed immunofluorescence microscopy
For confocal imaging 0.25×106 THP-1 cells were seeded on gelatin-coated (G1890, Sigma) coverslips in 24-well plates. Following differentiation, treatments and infection, cells were washed three times with warm PBS prior to fixation to remove any uninvaded pathogens and then fixed with 4% methanol-free formaldehyde (28906, Thermo Scientific) for 15 min at room temperature. Following fixation, cells were washed again with PBS and kept at 4°C overnight to quench any unreacted formaldehyde. Fixed specimens were permeabilized with PermQuench buffer (0.2% (w/v) BSA and 0.02% (w/v) saponin in PBS) for 30 minutes at room temperature and then stained with primary antibodies for one hour at room temperature. After three washes with PBS, cells were incubated with the appropriated secondary antibody and 1 μg/mL Hoechst 33342 (H3570, Invitrogen) diluted in PermQuench buffer for 1 hour at room temperature. Cells were washed with PBS five times and mounted using 5 μL Mowiol.
Specimens were imaged on a Leica SP5-inverted confocal microscope using 100x magnification and analyzed using LAS-AF software. For structured-illumination super-resolution imaging, specimens were imaged on a GE Healthcare Lifesciences DeltaVision OMX SR imaging system and images reconstructed using the DeltaVision software. All images were further formatted using FIJI software. 3D rendering of image stacks and distance measurements of AIM2-ASC-CASP8 inflammasome specks was performed using Imaris 8.3.1.
Extended microscopy sample preparation and image analysis
Quantification of protein recruitment to pathogen vacuoles and ASC speck formation
Specimens were prepared as described above. Images were acquired using a Ti-E Nikon microscope equipped with LED-illumination and an Orca-Flash4 camera using a 60x magnification. All intracellular parasites/bacteria of 100 fields of view were automatically counted based on whether they showed recruitment of the protein of interest using HRMAn high-content image analysis (Fisch et al, 2019b). Further, the analysis pipeline was used to measure the fluorescence intensity of GBP1 on STm vacuoles using the radial intensity measurement implemented in HRMAn.
For quantification of ASC speck formation, 100 Tg-infected cells were manually counted per condition using a Ti-E Nikon microscope equipped with LED-illumination using 60x magnification based on whether they contain an ASC speck and whether STm was decorated with GBP1/CASP4. The experiment was repeated independently three times.
EdU labeling for visualization of Tg-DNA release
Type I (RH) Tg were grown in fully confluent and nonreplicating HFFs for 3 days in the presence of 20 μM EdU to incorporate the nucleotide into their DNA. Labelled parasites were then harvested and used for infection as described above. 6 hours p.i. cells were fixed and EdU incorporated into Tg-DNA visualized by staining the specimens using Click-iT™ EdU Cell Proliferation Kit for Imaging, Alexa Fluor™ 647 dye (C10340, Invitrogen) according to the manufacturers instruction. Coverslips were further stained with Hoechst and mounted before imaging on a Ti-E Nikon microscope equipped with LED-illumination and an Orca-Flash4 camera using a 100x magnification.
Correlative light and electron microscopy
1.25×106 THP-1 cells were seeded in a μ-Dish35 mm, high Glass Bottom Grid-500 (81168, ibidi) and differentiated with PMA as described before. Cells were then pre-stimulated with IFNγ overnight and infected with type II (Pru) Tg at an MOI =1 for 6 hours. One hour prior to fixation, 1 μg/mL CellMask Deep Red (H32721, Invitrogen) and 20 μM Hoechst 33342 (H3570, Invitrogen) was added to the culture medium to label the cells for detection in fluorescence microscopy. Cells were fixed by adding warm 8% (v/v) formaldehyde (Taab Laboratory Equipment Ltd) in 0.2 M phosphate buffer (PB) pH 7.4 directly to the cell culture medium (1:1) for 15min. The samples were then washed and imaged in PB using a Zeiss AiryScan LSM 880 confocal microscope. Samples were then processed using a Pelco BioWave Pro+ microwave (Ted Pella Inc) and following a protocol adapted from the National Centre for Microscopy and Imaging Research protocol (Deerinck et al, 2010) (See Table S4 for full BioWave program details). Each step was performed in the Biowave, except for the PB and water wash steps, which consisted of two washes followed by two washes in the Biowave without vacuum (at 250 W for 40 s). All chemical incubations were performed in the Biowave for 14 min under vacuum in 2 min cycles alternating with/without 100 W power. The SteadyTemp plate was set to 21°C unless otherwise stated. In brief, the samples were fixed again in 2.5% (v/v) glutaraldehyde (Taab) / 4% (v/v) formaldehyde in 0.1 M PB. The cells were then stained with 2% (v/v) osmium tetroxide (Taab) / 1.5% (v/v) potassium ferricyanide (Sigma), incubated in 1% (w/v) thiocarbohydrazide (Sigma) with SteadyTemp plate set to 40°C, and further stained with 2% osmium tetroxide in ddH2O (w/v). The cells were then incubated in 1% aqueous uranyl acetate (Agar Scientific), and then washed in dH2O with SteadyTemp set to 40°C for both steps. Samples were then stained with Walton’s lead aspartate with SteadyTemp set to 50°C and dehydrated in a graded ethanol series (70%, 90%, and 100%, twice each), at 250 W for 40 s without vacuum. Exchange into Durcupan ACM® resin (Sigma) was performed in 50% resin in ethanol, followed by 4 pure Durcupan steps, at 250 W for 3 min, with vacuum cycling (on/off at 30 s intervals), before embedding at 60°C for 48 h. Blocks were serial sectioned using a UC7 ultramicrotome (Leica Microsystems) and 70 nm sections were picked up on Formvar-coated slot copper grids (Gilder Grids Ltd). Consecutive sections were viewed using a 120 kV Tecnai G2 Spirit transmission electron microscope (Thermo Fischer Scientific) and images were captured using an Orius CCD camera (Gatan Inc). Individual TEM images of ~25-30 consecutive sections per Tg parasite were converted as Tiff in Digital Micrograph (Gatan Inc.) and aligned using TrakEM2, a plugin of the FIJI framework (Cardona et al, 2012). The stacks were used to check the integrity of the PV and for coarse alignment with the AiryScan data.
Vacuole breakage assay (HRMAn)
For quantification of Tg vacuole integrity, cells seeded in black-wall 96-well imaging plates were infected and treated as described before. One hour prior to fixation, 1 μg/mL CellMask Deep Red (H32721, Invitrogen) was added to the culture medium to load the cytosol of host cells with this fluorescent dye. Following fixation and staining with Hoechst (H3570, Invitrogen), plates were imaged at 40x magnification on a Cell Insight CX7 High-Content Screening (HCS) Platform (Thermo Scientific) and 25 fields of view per well were recorded. Fluorescence of the dye within detected Tg vacuoles was then analyzed using HRMAn (Fisch et al, 2019b). Additionally, HRMAn was used to classify Tg vacuoles based on recruitment of mCH-GBP1 using the implemented neural network and the dataset stratified into decorated and non-decorated vacuoles.
Differential stain for detection of cytosolic STm
To distinguish between STm contained in vacuoles and bacteria that had escaped into the cytosol of infected macrophages, cells were differentially permeabilized using 25 μg/mL digitonin for one minute at room temperature as has been described before (Meunier & Broz, 2015). Cytosolic STm were then stained using anti-Salmonella antibody (ab35156, Abcam) that has been pre-labelled using Alexa Fluor™ 647 Protein Labeling Kit (A20173, Invitrogen) for 15 minutes at 37°C, prior to immediate fixation with 4% paraformaldehyde. Following fixation cells were permeabilized as described above and all STm were stained using the same but unlabeled antibody and corresponding Alexa Fluor 488 labelled secondary antibody. Cells were further stained with Hoechst (H3570, Invitrogen) and imaged on a Leica SP5-inverted confocal microscope using 100x magnification. For quantification, 100 fields of view per coverslip (typically >1000 individual STm overall) were acquired using a Ti-E Nikon microscope equipped with LED-illumination and an Orca-Flash4 camera at 60x magnification and analyzed with HRMAn (Fisch et al, 2019b) for colocalization of fluorescent signal of all and cytosolic STm.
Data handling and statistics
Data analysis used nested t-test, one-way ANOVA or two-way ANOVA as groups that were compared are indicated in the figure legends. Benjamini, Krieger and Yekutieli false-discovery rate (Q = 5%) based correction for multiple comparisons as implemented in Prism was used when making multiple comparisons. Graphs were plotted using Prism 8.1.1 (GraphPad Inc.) and presented as means of n = 3 experiments (with usually 3 technical repeats within each experiment) with error bars representing SEM, unless stated otherwise. Structure image of GBP1 was created using MacPymol v.1.74.
Data availability
All datasets generated during and/or analyzed during the current study are available from the corresponding author on request.
Supplementary Information
Acknowledgements
We would like to thank Matt Renshaw from the Crick Advanced Light Microscopy (CALM) STP for help with super-resolution SIM imaging, Julia Sanchez-Garrido for help in optimizing immunoblots and advice on reagents, Michael Howell from the Crick High-throughput screening (HTS) STP for help in performing automated imaging experiments, the Crick Genomics and Equipment Park STP for performing Sanger sequencing and DNA minipreps for cloning, Debipriya Das from the Crick Flow Cytometry STP for sorting Tg parasites and Caia Dominicus, Jeanette Wagener and Joanna Young from Moritz Treeck’s lab for help with creation of new transgenic Tg lines. We thank all members of the Frickel and the Shenoy labs for productive discussion and Crick Core facilities for assistance in the project.
Footnotes
The manuscript was re-written for clarity. Additionally further experiments were added to strengthen the conclusions.