GBPs Inhibit Motility of Shigella flexneri but Are Targeted for Degradation by the Bacterial Ubiquitin Ligase IpaH9.8

1Pathogens inhabit specific niches in their host organism to which they are exquisitely adapted. The host cytosol appears a particularly hostile environment, considering the small number of bacteria able to replicate in this compartment despite its high nutrient content. In resting cells, anti-bacterial autophagy, inflammasome activation, and the induction of cell death are all potent effector mechanisms against cytosol-invading bacteria. Additional resistance to infection is caused by exposure to interferons, which induce a large number of interferon-stimulated genes (ISGs) to enhance cytosolic immunity. ISGs antagonize both viruses and bacteria, although their contribution to the interferon-induced “anti-viral” state is much better understood than their anti-bacterial action.

2Among the interferon-induced effector proteins with anti-bacterial function are guanylate-binding proteins (GBPs), which belong to a large IFN-induced GTPase family. GBPs have been suggested to compromise the structural integrity of bacteria, to release ligands that stimulate inflammasomes, and to activate anti-bacterial effector mechanisms such as xenophagy and the oxidative burst. The target of GBP action remains hotly disputed since evidence has emerged for GBPs attacking host membranes as well as bacterial surfaces. It also remains unknown whether individual GBPs perform specific functions or whether their action is largely redundant as suggested by mouse knockout experiments, in which phenotypes emerged only upon large chromosomal deletions encompassing several GBPs.

3A fundamental question in innate immunity is how professionally cytosol-dwelling bacteria avoid clearance by cell-autonomous immunity. Shigella flexneri appears to have evolved sophisticated countermeasures against cytosolic effector mechanisms, such as the IpaH family of E3 ubiquitin ligases, characterized by an N-terminal leucine-rich repeat domain and a uniquely folded C-terminal catalytic domain distinct from cellular E3 ligases. Secreted via the type 3 secretion system (T3SS), IpaH proteins target host processes critical for anti-bacterial defense. IpaH1.4, for example, degrades the LUBAC subunit HOIP to prevent the activation of NF-κB and the deposition of M1-linked ubiquitin chains on the bacterial surface, even on co-infecting Salmonella enterica serovar Typhimurium (S. Typhimurium), while IpaH9.8 degrades the IKK subunit NEMO and the splicing factor U2AF35, resulting in reduced inflammation and higher bacterial counts. On the other hand, cytosol-dwelling bacteria take advantage of the unique opportunities their cytosolic lifestyle offers and infect neighboring cells via actin-dependent motility, thus avoiding exposure to the hostile extracellular space, while host cells carefully monitor their cytosolic homeostasis and deploy septins to antagonize such actin-driven bacterial motility. As demonstrated in primate models of infection, cell-to-cell spread is indeed an essential virulence determinant for S. flexneri and inactivating icsA, the gene that mediates actin-based motility, may contribute toward development of a live-attenuated vaccine strain.

4By studying the fate of S. flexneri in human cells activated with IFNγ, we discovered that, in a hierarchical manner and dependent on GBP1, bacteria become coated with multiple GBPs. The GBP coat antagonizes actin-dependent bacterial motility and invasion of neighboring cells. However, efficient cell-to-cell spread of S. flexneri is restored by IpaH9.8, which, by ubiquitylating GBPs for proteasome-dependent degradation, removes existing GBP coats and induces a transient ubiquitin coat. The IpaH9.8-derived ubiquitin coat appears morphologically indistinguishable but functionally opposite to the classical host-generated ubiquitin coat on cytosol-invading bacteria in that it promotes the pathogen rather than immunity.

5When entering its replicative niche in the host cytosol, S. flexneri avoids becoming coated with polyubiquitin (Figures 1A and 1B), suggesting that lack of ubiquitin coating enables S. flexneri to escape xenophagy and other effector mechanisms of cell-autonomous immunity. By contrast, S. Typhimurium, which is restricted by autophagy, becomes decorated with polyubiquitin (Figure 1C). To investigate whether exposure of cells to interferons or pro-inflammatory cytokines induces an “anti-bacterial” state, we employed HeLa cells, a well-established model system, to study the entry and subsequent intracellular trafficking of S. flexneri in epithelial cells. We found that polyubiquitin accumulated in the vicinity of S. flexneri in cells treated with IFNγ, but not IFNβ, TNFα, IL-1β, or IL-22 (Figures 1A and 1B). The ubiquitin coat occurred transiently with a peak at 2 hr post-infection (p.i.) (Figure 1A), consisted mainly of K48-linked ubiquitin chains (Figures 1D and 1E), and was associated with the bacterial surface as revealed by structured illumination microscopy, a super-resolution technique (Figure 1F). Ubiquitin accumulation on S. flexneri in IFNγ-treated cells was not caused by enhanced bacterial entry into the host cell cytosol, as indicated by the indistinguishable recruitment of galectin-8 to bacteria (Figure 1G). We conclude that IFNγ modulates a post-entry event, such as, for example, the ability of an E3 ubiquitin ligase to synthesize polyubiquitin in the vicinity of S. flexneri. Given the well-established role of the bacterial ubiquitin coat as an “eat-me” signal for anti-bacterial autophagy, we compared bacterial replication in mock and IFNγ-treated cells. IFNγ did not affect bacterial proliferation (Figure 1H). However, it significantly antagonized the development of actin tails (Figures 1I and 1J), a key feature of S. flexneri, induced by IcsA, and strictly required for bacterial motility in the host cytoplasm. However, IFNγ did not affect the typical polar localization of IcsA on bacteria (Figure 1K) or the fraction of bacteria that had attracted N-WASP, WIP, ARP2, or ARP3, all involved in actin tail formation (Figure 1L), suggesting IFNγ acts downstream of their recruitment. We next considered the formal possibility of actin-dependent motility antagonizing the development of the ubiquitin coat. However, S. flexneri ΔicsA, which completely lacks actin tails (Figure 1I), also became ubiquitin coated only in IFNγ-treated cells, similar to wild-type (WT) bacteria (Figure 1M), indicating that reduced motility of S. flexneri does not automatically lead to ubiquitin coating.

6To test whether the induction of the bacterial ubiquitin coat by IFNγ requires de novo gene expression, we knocked down STAT1, a transcription factor essential for IFNγ effects. In cells lacking STAT1, S. flexneri did not become coated by polyubiquitin (Figures 2A and S1A). To identify the gene(s) responsible for the phenotype, we compared gene expression in cells treated with IFNγ and IFNβ. However, microarray analysis did not reveal any E3 ubiquitin ligase prominently induced by IFNγ (Table S1), suggesting that the IFNγ-dependent coating of S. flexneri with polyubiquitin may rather require the induction of gene(s) regulating a pre-existing E3 ligase. GBPs comprise a family of seven GTPases in humans that control cell-autonomous immunity, of which GBP1, GBP2, GBP3, GBP4, and GBP5 were strongly and selectively upregulated by IFNγ (Figure 2B). When investigating the localization of GFP-tagged GBPs in cells stimulated with IFNγ, we observed that GBPs co-localized directly with bacteria and did not stain galectin-8-positive damaged endomembranes in the vicinity of bacteria that had originated from the entry of S. flexneri into the cytosol (Figure S2). Antibodies raised against GBP1 or GBP2, or pan-reactive for GBP1–5 revealed that, in both HeLa and THP1 cells, upon stimulation with IFNγ, endogenous GBPs were recruited to S. flexneri (Figures 2C and 2D). At 1 hr p.i. between 30% and 60% of S. flexneri co-localized with GFP-tagged GBP1, GBP2, and GBP3; 10% with GBP4; and 5% with GBP7, while very few or none were positive for GBP5 and GBP6 (Figure 2E). Overexpressed GBP1 associated with S. flexneri even in the absence of IFNγ (Figure 2E). GBP1 recruitment required a catalytically active GTPase domain (Figure 2F), and importantly, GBP1 was essential for the recruitment of GBP2, GBP3, or GBP4 (Figures 2G and S1B). In contrast, depletion of GBP2, GBP3, or GBP4 had no effect on the recruitment of other GBPs. We therefore conclude that in a GTPase-dependent manner, GBP1 performs an essential upstream role in initiating the hierarchical recruitment of GBPs to cytosol-invading bacteria.

7We next investigated a potential functional link between GBP recruitment to, and ubiquitin coating of, S. flexneri. We found that the majority of GBP1- and GBP3-positive S. flexneri were also coated with ubiquitin but that, importantly, only GBP1 occurred on all ubiquitin-coated bacteria (Figures 2H and S3). Structured illumination microscopy revealed an association of both GBP1 and K48-linked ubiquitin chains with the bacterial surface (Figure 2I). Depletion of GBP1 prevented the occurrence of polyubiquitin on S. flexneri in cells stimulated with IFNγ (Figure 2J), while overexpression of GBP1 in resting cells was sufficient to induce ubiquitin coating (Figures 2K–2M). Depletion or overexpression of GBP2, GBP3, or GBP4 had no effect. Taken together, we conclude that the ubiquitin coat of S. flexneri in cells stimulated with IFNγ is due to the induction of GBP1, which localizes to S. flexneri, initiates the hierarchical recruitment of other GBPs, and may recruit an E3 ubiquitin ligase.

8We considered the possibility that an E3 ubiquitin ligase encoded by S. flexneri rather than a cellular enzyme may generate the bacterial ubiquitin coat. To test our hypothesis, we used S. flexneri ΔmxiE, a strain deficient in the upregulation of many effector genes upon contact with host cells, including up to twelve bacterially encoded E3 ubiquitin ligases. S. flexneri ΔmxiE did not become coated with K48-linked ubiquitin chains (Figures 3A and 3B). We therefore investigated whether members of the Shigella-encoded and MxiE-controlled IpaH family of E3 ubiquitin ligases are recruited to and mediate ubiquitin coating of S. flexneri in IFNγ-stimulated cells. Among a panel of catalytically inactive IpaHs, only IpaH9.8 was recruited to S. flexneri (Figure 3C). Recruitment of exogenously expressed, catalytically inactive IpaH9.8 occurred specifically to bacteria that were also coated with K48-linked ubiquitin chains (Figures 3D and 3E) and only in cells stimulated with IFNγ (Figure 3F). To test whether IpaH9.8 is required for ubiquitin coating of S. flexneri in IFNγ-stimulated cells, we deployed strains lacking specific ipaH genes. S. flexneri ΔipaH9.8 did not become coated with K48-linked ubiquitin chains, in contrast to S. flexneri ΔipaH1.4 or ΔipaH7.8 (Figures 3G and 3H). An antibody against K27-linked ubiquitin also lost reactivity against S. flexneri ΔipaH9.8 (Figures 3I and 3J). Complementation with ipaH9.8, but not with catalytically inactive ipaH9.8C337A, restored ubiquitin deposition on S. flexneri ΔIpaH9.8 (Figure 3K). We therefore conclude that the GBP1-dependent ubiquitin coat of S. flexneri in IFNγ-stimulated cells is synthesized by IpaH9.8. The bacterial origin of the ubiquitin coat on S. flexneri in cells stimulated with IFNγ suggests that this particular ubiquitin coat may not have anti-bacterial functions but rather that IpaH9.8 counteracts a GBP-dependent cellular defense pathway, potentially through directly antagonizing GBP recruitment.

9We therefore investigated whether IpaH9.8 negatively controls the accumulation of GBPs on S. flexneri. Bacteria lacking IpaH9.8 were much more frequently associated with GBP1, GBP2, GBP3, and GBP4; complementation with IpaH9.8 reversed the effect (Figure 4A). To test whether IpaH9.8 controls GBP accumulation on bacteria by degrading GBPs, we analyzed GBP levels by flow cytometry in cells infected with S. flexneri. At 10 min p.i., GBP levels were indistinguishable between infected and uninfected cells, while at 180 min p.i., levels of GBP1, GBP2, and, to a lesser extent, GBP4 were reduced specifically in cells carrying a high burden of S. flexneri, i.e., in cells with proliferating bacteria (Figures 4B and S4). GBP3 levels were unaffected. Infection with S. flexneri ΔipaH9.8 did not reduce GBP levels, revealing that IpaH9.8 is essential for reducing GBP levels in infected cells. Endogenous GBP1 in cells stimulated with IFNγ was similarly reduced upon overexpression of IpaH9.8, but not other IpaH family members (Figure 4C), thus revealing exquisite specificity of IpaH proteins in the absence of other bacterial proteins. To further investigate the specificity of IpaH9.8 and to reveal how GBPs are degraded, we co-expressed IpaH9.8 and GBPs in 293ET cells. Similar to infection with S. flexneri, expression of WT, but not catalytically inactive IpaH9.8, drastically reduced levels of GBP1, GBP2, and GBP4, but not GBP3 (Figure 4D). Such specificity of GBP degradation correlates with and is likely explained by the ability of IpaH9.8 to bind GBP1, GBP2, and GBP4, but not GBP3 (Figures 4E and S5). To investigate whether GBPs are direct substrates for IpaH9.8, we performed in vitro ubiquitylation assays using purified proteins. IpaH9.8 catalyzed almost quantitative ubiquitylation of GBP1, resulting in a high molecular weight ubiquitin smear, but was inactive toward GBP3 (Figure 5A), consistent with the binding specificity of IpaH9.8 for GBPs (Figure 4E) and the autoinhibited state of IpaH E3 ligases in the absence of bound substrate. Co-expression of GBP1 and IpaH9.8 resulted in ubiquitylation of GBP1 and its degradation (Figure 5B). Treatment with the proteasome inhibitor Carfilzomib antagonized the effects of IpaH9.8 on GBPs — Carfilzomib caused accumulation of polyubiquitin in immunoprecipitated GBP1 samples (Figure 5B), it rescued the degradation of endogenous GBP1 in cells infected with S. flexneri (Figure 5C), and it restored GBP1 coats on S. flexneri (Figure 5D). Taken together, we conclude that S. flexneri secretes IpaH9.8 to ubiquitylate specific GBPs and cause their proteasome-mediated degradation, resulting in reduced GBP levels in infected cells and escape of bacteria from labeling by GBPs.

10To investigate whether the inhibition of actin tail formation on S. flexneri observed upon IFNγ treatment is mediated by GBPs and whether GBP degradation by IpaH9.8 precedes actin-driven motility of S. flexneri, we performed live microscopy in cells expressing GFP::GBP1 and labeled with an F-actin binding peptide known as Lifeact (Movies S1 and S2; Figure 6A). Upon infection, both WT bacteria and S. flexneri ΔipaH9.8 became coated with GBP1, a status that, once established, was sustained throughout division by bacteria of either strain. However, while S. flexneri ΔipaH9.8 maintained their GBP1 coat over the time course of the movie, i.e., for at least 3 hr, WT bacteria lost their coat between 1 and 1.5 hr p.i., followed by the acquisition of actin-dependent motility and invasion of neighboring cells, where sometimes de novo GBP coating occurred. In contrast, S. flexneri ΔipaH9.8, due to lack of actin-driven motility, formed tightly clustered micro-colonies and failed to invade neighboring cells. Quantification confirmed the absence of GBP1-positive bacteria displaying actin tails in WT and ΔipaH9.8 strains, as well as the lower and higher percentage of actin tails and GBP1 coats, respectively, in S. flexneri ΔipaH9.8 compared to WT bacteria (Figure 6B). Complementation with ipaH9.8 reversed the actin phenotype (Figure 6C) and treatment with the proteasome inhibitor Carfilzomib antagonized the IpaH9.8 effect on actin tail formation (Figure S5). To test whether GBPs mediate the IFNγ-induced suppression of actin tail formation, we depleted cells of GBPs using small interfering RNAs (siRNAs) (Figure 6D) and discovered that GBP1 was essential for IFNγ to antagonize actin tail formation on S. flexneri. Finally, to investigate the functional consequences of IpaH9.8-mediated degradation of IFNγ-induced GBPs and its effect on actin tail formation, we measured proliferation of S. flexneri and bacterial load in host cells. Bacterial proliferation in either resting or IFNγ-treated cells was unaffected by deletion of IpaH9.8 (Figure 6E). However, bacteria lacking IpaH9.8 did not spread efficiently between cells, resulting in a much smaller number of cells becoming infected (Figures 6F, 6G, and S7) and those cells carrying a much higher bacterial load (Figures 6F, 6H, and S7). Taken together, we conclude that GBPs restrict actin-driven motility of S. flexneri, thus impairing the spread of bacteria into neighboring, uninfected cells and containing the bacterial burden in a smaller number of infected cells. IpaH9.8 antagonizes GBP-mediated cell-autonomous immunity by targeting GBPs for proteasomal degradation, resulting in unrestricted spread of bacteria.

11GBPs provide important but poorly understood immunity against invasive bacteria in interferon-stimulated cells. Here we provide evidence for the GBP1-dependent hierarchical recruitment of multiple GBPs to cytosol-exposed S. flexneri, where they form a dense GBP coat surrounding the bacterium that inhibits actin-dependent bacterial motility and, consequently, cell-to-cell spread. S. flexneri antagonizes GBP-mediated cellular defenses by secreting the E3 ubiquitin ligase IpaH9.8, which ubiquitylates and degrades GBPs in a proteasome-dependent manner, causing existing GBP coats to dissolve and actin-dependent motility to become re-established. The existence of an E3 ubiquitin ligase in the genome of S. flexneri that efficiently degrades GBPs provides compelling evolutionary evidence for the importance of GBPs in anti-bacterial defense.

12GBPs clearly play an important role in defending the cell interior against parasites and bacteria, although the target of GBP attack and their mode of action remain poorly characterized. While both host membranes and bacterial surfaces have been found to be targeted by GBPs, our SIM super-resolution data provide unequivocal evidence for the accumulation of GBPs on the bacterial surface with no observable enrichment on galectin-8-positive remnants of Shigella-containing vacuoles. Importantly, the accumulation of GBPs on S. flexneri required catalytically active GBP1, suggesting a clear hierarchy in GBP recruitment and the possibility that GBP1 either serves as a receptor for S. flexneri and possibly other pathogens or that other cellular proteins are required to load GBPs in a hierarchical manner onto the bacterial surface. Similar to the ubiquitin coat formed by cellular E3 ubiquitin ligases when they encounter bacteria in the host cytosol, the GBP coat also represents a highly polyvalent display of host proteins on the bacterial surface. Therefore, just like the ubiquitin coat, the GBP coat may also transform the bacterial surface into a signaling platform. Considering the time required to upregulate IFNγ-dependent GBP expression and the model character of the cells used in this study, follow-up investigations will be required to test the importance of the GBP coat for the defense against S. flexneri in vivo.

13While several effector molecules of the ubiquitin coat have been identified, including pro-inflammatory signaling molecules such as the Nemo-recruited IKK complex and autophagy cargo receptors such as NDP52, Optineurin, and p62, GBP effectors remain to be identified, although IRGB10, recruited to Francisella novicida via GBPs and contributing to the liberation of bacterial ligands for the inflammasome, may represent an important effector molecule of the GBP coat.

14Since the bacterial ubiquitin coat serves anti-bacterial purposes, S. flexneri, as a professional cytosol-dwelling pathogen, has evolved sophisticated countermeasures against ubiquitin coating, such as, for example, degrading anti-bacterial E3 ligases. However, in cells stimulated with IFNγ, S. flexneri does become ubiquitin coated, although not due to the action of cellular E3 ubiquitin ligases, but as a result of IpaH9.8 ubiquitylating the GBP coat. Therefore, rather than promoting anti-bacterial defense mechanisms by providing “eat-me” signals for cargo receptors of xenophagy, the coating of S. flexneri with polyubiquitin by IpaH9.8 results from the bacterial attempt to counteract a cellular defense pathway. Care must therefore be taken when attempting to assign anti-bacterial function to the ubiquitin coat on cytosolic bacteria. Our results on the ubiquitin coating of S. flexneri by a bacterially encoded E3 ubiquitin ligase are congruent with a growing body of literature suggesting that S. flexneri possesses multiple mechanisms to evade anti-bacterial autophagy. Previous studies have shown that Toca1 recruitment to intracellular S. flexneri antagonizes xenophagy and that IcsB and VirA are bacterially encoded factors that help in evading anti-bacterial autophagy.

15IpaH9.8 is an important pathogenicity factor for S. flexneri since, in a murine pneumonia model, S. flexneri ΔipaH9.8 proliferates less and induces higher levels of pro-inflammatory cytokines. Degradation of both U2AF35, a splicing factor, and Nemo, an essential subunit of the NF-κB-inducing IKK complex, has been suggested to cause the observed phenotype. Our work revealed that IpaH9.8, among a panel of twelve IpaH proteins, specifically targets GBP1, GBP2, and GBP4, but not GBP3. As demonstrated for GBP1 and GBP3, binding of GBPs to IpaH9.8 parallels the ability of IpaH9.8 to ubiquitylate GBPs in vitro, consistent with the autoinhibited state of IpaH E3 ligases in the absence of bound substrate, and the ability of IpaH9.8 to induce proteasome-dependent degradation in cells. GBP degradation does not require GBP oligomerization on the bacterial surface, since expression of IpaH9.8 in uninfected cells was sufficient to deplete GBP from cells, suggesting sufficiently high affinity of IpaH9.8 for soluble GBPs consistent with binding of IpaH9.8 to GBP1, GBP2, and GBP4 in LUMIER interaction assays. The ability of IpaH9.8 and potentially other IpaH proteins to selectively degrade key host defense molecules suggests that IpaH proteins hold significant potential for the experimental manipulation of host-pathogen interactions, although it will be important to first gain a better understanding of the structural basis of IpaH specificity.

16Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Felix Randow (randow@mrc-lmb.cam.ac.uk).

17HeLa, 293ET and THP1 cells, as well as all stable cell lines, were grown in IMDM supplemented with 10% FCS at 37c°C in 5% CO2. HeLa and 293ET are of female, THP1 of male origin. Cell lines have not been authenticated. All cell lines were tested to be Mycoplasma free.

18S. flexneri (strain M90T) was grown in tryptic soy broth (TSB) or on tryptic soy agar containing 0.003% Congo red.

19S. Typhimurium (strain 12023) was grown in Luria broth (LB) or on LB agar plates.

20E. coli strains MC1061, BL21 and Rosetta2 were grown on tryptic yeast extract (TYE) agar plates or in LB.

21Cytokines were added for 10-20 hours before experiments at the following concentrations: IFNγ 1 ng/ml, IFNβ 100 U/ml, TNFα 10 ng/ml, IL1-β 10 ng/ml and IL-22 10 ng/ml.

22M5P or closely related plasmids were used to produce recombinant MLV for the expression of proteins in mammalian cells. Open reading frames encoding IpaH proteins, N-WASP, WIP, ARP2, ARP3 and GBP1-7 were amplified by PCR. Mutations were generated by PCR and verified by sequencing. For complementation IpaH9.8 and FLAG::IcsA were amplified by PCR and cloned into pFPV25.1.

23S. flexneri was grown overnight in tryptic soy broth (TSB) and sub-cultured (1:100) in fresh TSB for 2.5 h before infection. Such cultures were consecutively washed in PBS and re-suspended in antibiotic-free IMDM plus 10% FCS immediately before 100 μL was used to infect HeLa cells in 24-well plates. Samples were centrifuged for 10 min at 670 g followed by incubation at 37°C for 20 or 30 min. Following two washes with warm PBS, cells were cultured in 100 μg/ml gentamycin for 2 h and 20 μg/ml gentamycin thereafter. For FACS-based infection assays, bacteria were diluted 1:3 in warm antibiotic-free IMDM before infection and cells were washed 10 min after infection.

24S. Typhimurium was grown overnight in Luria broth (LB) and sub-cultured (1:33) in fresh LB for 3.5 h before infection. Such cultures were further diluted (1:5) in antibiotic-free IMDM plus 10% FCS immediately before 20 μL was used to infect HeLa cells in 24-well plates for 15 min at 37°C. Following two washes with warm PBS, cells were cultured in 100 μg/ml gentamycin.

25To enumerate intracellular bacteria, cells from triplicate wells were lysed in 1 ml cold PBS containing 0.1% Triton X-100. Serial dilutions were plated in duplicate on TYE agar.

26Total RNA from HeLa cells was extracted using the RNeasy Plus Mini Kit. RNA samples were prepared using the Ambion WT Expression Kit and Affymetrix GeneChip WT Terminal Labeling and Hybridization Kit. The generated cocktails were hybridized to Human Affymetrix Gene ST 1.0 cartridge arrays. The arrays were then scanned on the Affymetrix GCS3000 by the Addenbrooke’s Hospital Genomics Core Laboratory.

275 × 104 cells per well were seeded in 24-well plates. The following day, cells were transfected with 40 pmol of siRNA using Lipofectamine RNAiMAX. Experiments were performed after 3 days.

28The non-targeting negative control was used as control.

29HeLa cells were grown on glass coverslips before infection. After infection, cells were washed twice with warm PBS and fixed in 4% paraformaldehyde for 20 min. Cells were washed twice in PBS and then simultaneously permeabilized and blocked in PBSB (PBS, 0.01% saponin, 2% BSA). Coverslips were incubated with primary followed by secondary antibodies for 1 h in PBSB. Samples were mounted either in mounting medium with DAPI or Prolong Antifade mounting medium for confocal imaging and super resolution microscopy, respectively. Marker positive bacteria were scored by eye among at least 200 bacteria per coverslip. Confocal images were taken with a × 63, 1.4 numerical aperture objective on either a Zeiss 710 or a Zeiss 780 microscope. Live imaging was performed on a Nikon Eclipse Ti equipped with an Andor Revolution XD system and a Yokogawa CSU-X1 spinning disk unit. Super resolution images were acquired using an Elyra S1 structured illumination microscope (Carl Zeiss Microscopy Ltd, Cambridge, UK). The system has four laser excitation sources (405nm, 488nm, 561nm and 640nm) with fluorescence emission filter sets matched to these wavelengths. SIM Images were obtained using a 63X 1.4 NA oil immersion lens with grating projections at 3 rotations and 5 phases in accordance with the manufacturers instructions. The number of Z planes varied with sample thickness. Super resolution images were calculated from the raw data using Zeiss ZEN software.

30293ET cells were grown on 6-well plates and transfected with 2 μg plasmid DNA (1.5 μg GFP:GBP1 and 0.5 μg IpaH9.8) using PEI. 24h after transfection the proteasome was inhibited for 18h using 100 nM Carfilzomib.

31Post-nuclear supernatants from 2 × 106 293ET cells expressing tagged proteins were obtained following lysis (150 mM NaCl, 0.1% Triton X-100, 20 mM Tris-HCl pH 7.4, 5 mM EDTA and proteinase inhibitors). Proteins were precipitated for 2 h with anti-GFP or Flag agarose before washing. Samples were eluted with Flag peptide or Laemmli buffer and separated on 4%–12% denaturing Bis-Tris gels (Thermo Fisher Scientific). Visualization following immunoblotting was performed using ECL detection reagents.

32Proteins were expressed from pETM30 in E.coli BL21. Bacteria were grown to an OD600 of 0.7 at 37°C before overnight induction at 16°C in the presence of 100 μM IPTG. Cells were mechanically lysed in lysis buffer (20mM Tris (pH 8.0), 150mM NaCl, 1mM DTT, protease inhibitors) and cleared by centrifugation. Lysates were snap frozen and stored at −80°C.

33LUMIER binding assays with pairs of putative interactors, one fused to luciferase and the other fused to GST, were performed in LUMIER lysis buffer (150 mM NaCl, 0.1% Triton X-100, 20 mM Tris-HCl (pH 7.4), 5% glycerol, 5 mM EDTA and proteinase inhibitors) as previously described. GST-fusion proteins were immobilized on beads before incubation with the luciferase tagged binding partner for 2 h. Luciferase-tagged proteins were expressed in 293ET cells. After washing in lysis buffer, proteins were eluted with glutathione in Renilla lysis buffer (Promega). Relative luciferase activity represents the ratio of activity eluted from beads and present in lysates.

34His6-GBP1 and His6-GBP3 (both in the pETM-11 vector) and His6-GST-IpaH9.8 (in pETM-30 vector) were expressed in Rosetta2 (DE3) cells. Cells were grown at 30°C in 2xTY medium supplemented with 30 μg/ml kanamycin and 34 μg/ml chloramphenicol. Cultures were induced with IPTG (400 μM) at 18°C and harvested after 16 hours. Proteins were purified by immobilized metal-affinity chromatography and the His6-GST tag was removed from IpaH9.8 by incubation with TEV protease. His6-GBP1, His6-GBP3 and untagged IpaH9.8 were further purified by anion exchange chromatography using a Resource Q column (GE Healthcare) and size exclusion chromatography using a Superdex 200 16/60 column (GE Healthcare), the latter in buffer 20 mM Tris pH 8.5, 200 mM NaCl and 4 mM DTT.

35Reactions were set up in 40 mM Tris pH 8.5, 10 mM MgCl2, 0.6 mM DTT with combinations of the following reagents: E1 (0.5 μM), UBE2D1 (3 μM), IpaH9.8 (5 μM), His6-GBP1 (30 μM), His6-GBP3 (30 μM), ubiquitin (300 μM) and ATP (10 mM). The reaction was incubated at 37°C for 30 min after which, NiII resin was added for 15 minutes at r.t.. Following binding, resin was washed three times in 20 mM Tris pH 7.4, 300 mM NaCl, 50 mM Imidazole, 2 mM β-mercaptoethanol and 0.01% tween.

36After infection, cells were washed twice with warm PBS, detached with trypsin and fixed in 4% paraformaldehyde for 20 min. Cells were washed twice in PBS and quenched with PBSG (PBS, 1M Glycine). For intracellular staining samples were fixed and stained using Fix & Perm (Life Technologies). All samples were analyzed on a BD LSR ii flow cytometer, using the high throughput system (HTS) to score spread of bacteria. Data were analyzed in FlowJo.

37A Tryptic Soy Broth (TSB) overnight culture was diluted 1:100 in 10 mL of TSB and grown at 37°C to an OD600 of 0.6 to 0.8. Bacteria were cooled on ice for 10 min, centrifuged (4,300 g x 4 min, 4 C) and washed once in 10 mL then twice in 1 mL of ice cold electroporation buffer (1 mM MOPS, 20% glycerol, pH 7.2). Bacteria were pelleted a final time, resuspended in 100 μL of buffer and mixed with 150 ng of plasmid DNA. The mixture was electroporated in a chilled 2mm cuvette (Flowgen Bioscience) using 2,500V, 600 Ω and 10 μF. Electroporated bacteria were regenerated for 1 h at 37 C in 1 mL of Super Optimal Broth (SOB) medium and plated on TSB agar supplemented with Ampicillin (100 μg/mL) and Congo red (0.003%).

38All data were tested for statistical significance with Prism software (GraphPad Prism 7). The unpaired Student’s t test was used to test whether two samples originate from the same population. For more than two samples with only one variable an analysis of variance (one-way ANOVA) was performed. Either Dunnett’s multiple comparison test (to compare all samples against a control) or Tukey’s multiple comparison test (to compare all samples against each other) was applied. Performed tests are indicated in Figure Legends. No specific method was used to determine whether the data met assumptions of the statistical approach. Unless otherwise stated, all experiments were performed at least three times and the data were combined for presentation as mean ± SEM. All differences not specifically indicated as significant were not significant (ns, p > 0.05). Significant value are indicated as ∗, p < 0.05; ∗∗, p < 0.01. Statistical details, including sample size (n), are reported in the Figures and Figure Legends.

39For scoring marker positive bacteria three independent experiments with three replicates each were performed. Bacteria were scored by visual enumeration as n > 100 (for 1h p.i.), n > 200 (for 2h p.i.) or n > 300 (for 3h p.i.) bacteria per replicate. Graphs show mean ± SEM.

40To score bacterial burdens, cells from triplicate wells were lysed and bacteria were plated in duplicate on TYE agar. Each experiment was performed three times. Bacterial colonies were counted using the aCOLyte3 system (Synbiosis).

41Graphs show mean ± SD of a representative experiment or mean ± SEM for combined datasets.

42The accession number for the DNA microarray data reported in this paper is NCBI Gene Expression Omnibus, GEO: GSE103363.

43M.P.W., C.P., E.I.W., C.J.E., K.B.B., and A.v.d.M. performed experiments and analyzed data. J.R. provided knockout strains and reviewed the manuscript. M.P.W., C.P., and F.R. wrote the manuscript.