Shiga toxin suppresses noncanonical inflammasome responses to cytosolic LPS

Abstract

Inflammatory caspase-dependent cytosolic LPS sensing is a critical arm of host defense against bacteria. How pathogens overcome this pathway to establish infections is largely unknown. Enterohemorrhagic Escherichia coli (EHEC) is a clinically significant human pathogen causing hemorrhagic colitis and hemolytic uremic syndrome. We found that a bacteriophage-encoded virulence factor of EHEC, Shiga toxin (Stx), suppresses caspase-11-mediated activation of the cytosolic LPS sensing pathway. Stx was essential and sufficient to inhibit pyroptosis and IL-1 responses elicited specifically by cytosolic LPS. The catalytic activity of Stx was necessary for suppression of inflammasome responses. Stx impairment of inflammasome responses to cytosolic LPS occurs at the level of gasdermin D activation. Notably, Stx suppresses inflammasome responses during LPS challenge as well as bacterial infection in vivo. Overall, this study assigns a previously undescribed inflammasome-subversive function to a well-known bacterial toxin, Stx, and reveals a new phage protein-based pathogen blockade of cytosolic immune surveillance.

One Sentence Summary:

Shiga toxin, a phage-encoded bacterial virulence factor, inhibits caspase-11-mediated inflammasome responses.

Introduction

The interaction of infectious microbes with the host represents a complex battle to gain selective advantage over each other. The host deploys well-coordinated immune defense strategies to detect and eliminate pathogens. One such strategy is to detect bacterial LPS that enters the cytosol via the caspase-11-dependent noncanonical inflammasome and terminate the infection by killing the infected cells via pyroptosis and secreting IL-1β, IL-1α, and IL-18 (1–4). On the other hand, driven by this selective pressure, successful pathogens are likely to have evolved mechanisms to subvert host defense in order to colonize and replicate within the host (5). Enterohemorrhagic Escherichia coli (EHEC) is a successful enteric pathogen that causes hemorrhagic colitis (HC) and hemolytic uremic syndrome (HUS) (6, 7). EHEC infections result in frequent hospitalizations and can be particularly fatal in children as they develop HUS at a higher rate. Treatment of EHEC infection is challenging as antibiotics enhance the severity of the disease and therefore, are contraindicated highlighting an acute need for new therapies (7–9). Our poor understanding of EHEC-host interactions remains a barrier to develop specific treatments. We have previously shown that caspase-11 senses EHEC infection via outer membrane vesicle (OMV)-associated LPS (2). Given that the cytosolic LPS sensing pathway plays a crucial role in the detection and clearance of Gram-negative bacterial infections, a bacterium with a greater capacity to evolve such as EHEC is likely to have developed mechanisms to counter such anti-bacterial efforts. However, the mechanisms by which EHEC overcomes host innate defense to establish infection is poorly understood.

EHEC has remarkably evolved from nonpathogenic ancestral E. coli by acquiring mobile genetic elements including bacteriophages (10). EHEC virulence determinants encoded on these acquired genetic elements manipulate host signaling networks to facilitate bacterial colonization (11, 12). Direct comparison of whole genome sequence of EHEC with that of nonpathogenic E. coli K12 has identified EHEC-specific genomic islands that are absent in E. coli K12, which were designated as O islands (10). O islands harbor two major virulence factors of EHEC, the Shiga toxin (Stx) and the type III secretion system (T3SS) (13). While EHEC remains attached to the intestinal epithelium during infections, Stx enters the systemic circulation and mediates more harmful aspects of EHEC pathogenesis including the development of hemorrhagic colitis resulting in bloody diarrhea and HUS leading to renal failure (14).

EHEC strains express two subtypes of Stx, Stx2 and Stx1, which are encoded by two lambdoid bacteriophages that were horizontally acquired by EHEC in a sequential manner during its evolution (15). While these Stx-encoding phages typically remain in a lysogenic state in EHEC, a bacterial DNA damage response can induce the lytic cycle of the phage leading to increased production of the toxin and thus, pathogenesis (16). Stx belongs to the type II ribosome inactivating protein family (17). Stx has a typical AB₅ toxin structure composed of an enzymatically active A subunit and a B-pentamer subunit that binds to its receptor, Gb3 (CD77) (18, 19). The A subunit is an N-glycosidase enzyme that removes a specific adenine residue from the highly conserved α-sarcin/ricin loop of the 28S rRNA of the 60S subunit of ribosome (18). This results in a translational block, inhibition of protein synthesis, and eventually apoptosis (18–20). Notably, these Stx effects are highly cell type specific; cells expressing high levels of lipid raft-associated Gb3 receptor such as vascular endothelial cells and renal epithelial cells are highly sensitive to Stx toxicity (20). Stx follows the retrograde intracellular trafficking in these cells, enter the cytosol from endoplasmic reticulum, and inhibits protein synthesis. In contrast, Gb3 receptors on innate immune cells including macrophages and monocytes are not associated with lipid rafts. While the exact intracellular trafficking route of Stx in these cells is not clear, it believed that the toxin follows the endosomal/lysosomal pathway in these cells (20, 21). Consequently, innate immune cells particularly primary macrophages are not susceptible to Stx-mediated inhibition of protein synthesis and subsequent toxic effects (21–23). Interestingly, previous studies have indicated that Stx interacts with innate immune cells during its systemic spread through circulation (14, 20, 21). In certain cell types such as THP1 monocytes, Stx induces the mitogen-activated protein kinase pathway resulting in the production of TNF and IL-6 (22, 24). However, the consequences of Stx-primary immune cell interaction, particularly its effects on host defense responses during EHEC infection is not known. Similarly, it is not clear if Stx has any additional biological activity other than toxicity during EHEC infection. Increasing evidence in the recent past hint to an important role for bacteriophage-encoded proteins in regulating host defense against pathogenic bacteria (25). Given that Stx is a phage-encoded protein, it is possible that Stx may exert immunoregulatory functions during EHEC infection.

In this study, we have identified an inflammasome-inhibitory role for Stx during EHEC infection. Stx specifically suppressed the activation of caspase-11-mediated cytosolic LPS sensing pathway in macrophages and mice. This inflammasome-suppressive function of Stx was mediated by its catalytic activity independent of its inhibitory effect on translation. Collectively, these findings reveal a new mechanism of pathogen blockade of intracellular surveillance pathways. In broader terms, this study reinforces the paradigm that successful bacterial pathogens employ a single protein to execute multiple functions in the host to facilitate pathogenesis.

Results

Shiga toxin (Stx) suppresses noncanonical inflammasome activation by EHEC

EHEC encodes multiple known and putative virulence factors on its genomic O-islands. In an attempt to identify O-island-encoded proteins with a potential to regulate inflammasome activation during EHEC infection, we screened a library of 57 EHEC mutants lacking major O-islands or known virulence factors for their ability to induce cell death and IL-1β secretion in mouse bone marrow-derived macrophages (BMDMs). The inflammasome responses induced by the majority of the EHEC mutants were comparable to that of the wild-type EHEC (Fig. 1A–B). Remarkably, a mutant that lacks Shiga toxin (ΔStx) induced significantly elevated levels of cell death, IL-1β, and IL-1α compared to wild-type EHEC indicating a potential inflammasome inhibitory role for Stx (Fig. 1C–E). Inflammasome activation is a multistep process where TLR activation induces pro-IL-1β, which is cleaved by caspase-1 into mature IL-1β and therefore, a change in IL-1β levels could also indicate modulation of TLR signaling (26, 27). However, unlike IL-1β, levels of TLR-dependent but inflammasome-independent cytokines, IL-6 and TNF, were comparable between wild-type EHEC- and ΔStx-infected cells indicating that Stx specifically inhibits inflammasome activation and not the TLR signaling-driven inflammasome priming (Fig. 1F–G). Furthermore, a similar elevation in inflammasome responses in the absence of Stx was observed in cells primed with Pam3CSK4 to promote pro-IL-1β levels prior to infection with wild-type and ΔStx EHEC (Fig. 1H–J). Importantly, infection with ΔStx also resulted in a marked increase in cleavage of pro-caspase-1 and pro IL-1β into their active caspase-1 p20 and IL-1β p17 forms (Fig. 1K and Fig. S1A–B). In contrast, levels of caspase-11, pro IL-1β, and procaspase-1 remained similar between cells infected with wild-type and ΔStx EHEC (Fig. 1K). As expected, the inflammasome responses elicited by both wild-type and ΔStx were caspase-11 dependent (Fig. S1C–D). These results suggested a role for Stx in suppressing caspase-11-mediated inflammasome responses during EHEC infection.

Fig. 1. Stx deficiency augments inflammasome responses.

(A-J) Cell death and secretion of indicated cytokines by BMDMs primed with Pam3CSK4 (A, B, H-J) or left unprimed (C-G) prior to infection with EHEC or the indicated isogenic mutant strains at MOI 50 for 16 h.

(K) Cleaved caspase-1 p20 and IL-1β p17 in the supernatants and procaspase-1, proIL-1β, caspase-11 (casp11), and β-actin in the lysates of Pam3CSK4-primed BMDMs stimulated with EHEC or ΔStx at MOI 50 for 16 h.

Pooled data from two (A-B) or three (C-J) independent experiments are presented as mean±SD. One experiment representative of three experiments is shown in K. Each experiment had three technical replicates **p<0.01; ****p < 0.0001; one-way ANOVA followed by Tukey’s post-test.

See also fig. S1.

Stx expression in trans attenuates inflammasome activation by E. coli BL21

While the above observations indicate that Stx is inhibiting the activation of the cytosolic LPS sensing pathway during EHEC infection, it is important to note that these experiments were conducted using EHEC grown to stationary phase in Luria Bertani (LB) broth in which Stx expression remains minimal due to repression by a bacteriophage-encoded cI protein (28). In contrast, EHEC greatly up-regulates Stx expression in the intestinal milieu during infection (16, 28). Therefore, an experimental system that expresses substantially higher levels of Stx as in intestinal infections would be more relevant to address the effect of Stx on inflammasomes. To this end, we constructed an IPTG-inducible promoter system-based expression of Stx in E. coli BL21. An expression vector carrying full length EHEC stx2 was introduced into E. coli BL21(DE3)/pLysS strain generating E. coli BL21/pStx2 (BL21/pStx2) that expresses high levels of Stx upon IPTG treatment. IPTG treatment did not interfere with the growth of BL21/pStx2 or an isogenic BL21strain carrying an empty vector (BL21/pEmpty) (Fig. S2A). The levels of biologically active Stx2 in BL21/pStx2 supernatants was quantified by Vero cell cytotoxicity assay. Vero cells express high levels of Gb3 receptor and are routinely used in bioassays to measure Stx quantity (29, 30). In the absence of IPTG treatment, BL21/pStx2 produced minimal basal levels of Stx, whereas IPTG treatment resulted in a marked increase in Stx production (Fig. 2A). Infection of macrophages with H₂O- or IPTG-treated BL21/pEmpty and H₂O-treated BL21/pStx2 induced comparable levels of inflammasome responses. On the other hand, infection of macrophages with IPTG-treated BL21/pStx2 resulted in a marked reduction in cell death and IL-1 cytokine production, but not in TNF or IL-6 levels, indicating that Stx specifically and robustly inhibits inflammasome activation (Fig. 2B–F). A similar reduction in cell death and IL-1 secretion was observed with Pam3CSK4 primed cells infected with IPTG-treated BL21/pStx2 as well (Fig. S2B–C). Interestingly, infection with H₂O-treated BL21/pStx2 also resulted in a decrease in IL-1β in certain instances indicating that even at basal expression levels Stx2 is capable of suppressing IL-1β responses (Fig. 2C, S2C, and S2J–K). Several recent studies have discovered that caspase-11 cleaves a pore forming protein, gasdermin D, and the cleaved N-terminal fragment of gasdermin D oligomerizes to form pores in the plasma membrane, eventually leading to pyroptotic cell death (31–34). While the full length gasdermin D (p55) levels remained unchanged, a pronounced decrease in the level of cleaved gasdermin D (p30) was observed upon infection with IPTG-treated BL21/pStx (Fig. 2G and S2D). Immunoblot analysis also confirmed that IPTG-induced high Stx expression results in a dramatic decrease in cleavage of caspase-1 and IL-1β (Fig. 2G and S2E–F) whereas pro IL-1β levels remained unchanged in response to high Stx expression (Fig. 2G). All the inflammasome responses induced by both BL21/pEmpty and BL21/pStx2 were dependent on caspase-11 and gasdermin D further confirming that Stx is in fact suppressing the noncanonical cytosolic LPS sensing pathway (Fig. 2H–J and Fig. S2G–K). Notably, IPTG-induced high Stx expression did not lead to a change in caspase-11 protein levels in the cells (Fig. 2J) indicating that Stx-mediated inflammasome inhibition is not due to the suppression of caspase-11 expression.

Fig. 2. Induction of Stx expression impairs inflammasome responses elicited by E. coli BL21.

(A) Stx2 levels as measured by Vero cell cytotoxicity assay in the culture supernatants of E. coli BL21/pET21a (BL21/pEmpty) or E. coli BL21 harboring pET21a-Stx2 (BL21/pStx2) treated with H₂O or 0.5 mM IPTG for 5 h.

(B-F) Cell death and secretion of indicated cytokines at 16 h post infection by BMDMs infected with H₂O- or IPTG-treated BL21/pEmpty or BL21/pStx2 at an MOI of 50.

(G) Cleaved caspase-1 p20 and IL-1β p17 in the supernatants and gasdermin D (full length p55 and cleaved p30), pro IL-1β, and β-actin in the lysates of Pam3CSK4-primed BMDMs infected with H₂O- or IPTG-treated BL21/pEmpty or BL21/pStx2 or F. novicida at an MOI of 50 for 16 h.

(H-J) LDH assay for cell death (H), ELISA for IL-1β (I), or immunoblots for the indicated proteins (J) in the supernatants or lysates of Pam3CSK4-primed C57BL/6 or Caspase-11^−/− BMDMs infected with H₂O- or IPTG-treated BL21/pEmpty or BL21/pStx2 or F. novicida or treated with poly(dA:dT) for 16 h.

Pooled data from three independent experiments (A-F and H-I; mean±SD) or one experiment representative of two experiments (G and J) are presented. Each experiment had two or three technical replicates. ***p<0.001; ****p < 0.0001; two-way ANOVA followed by Tukey’s or Sidak’s post-test. See also fig. S2.

Stx is sufficient to inhibit the cytosolic LPS sensing pathway

To directly assess the role of Stx on inflammasome activation by EHEC, we used commercially available purified Stx2. Lack of LPS contamination in purified Stx2 was confirmed by LAL assay. The biological activity of Stx2 was verified by the Vero cell cytotoxicity assay (Fig. 3A). Previous studies have indicated that primary macrophages are not susceptible to Stx-mediated toxicity (21, 23). Corroborating this, and unlike Vero cells, BMDMs did not undergo cell death upon treatment with purified Stx2 (Fig. 3B and S3A). Further, BMDMs were infected with EHEC and increasing doses of Stx2 were added to the cells at the time of infection. Stx2 is likely to be internalized by macrophages via Gb3 receptor-mediated endocytosis (20, 21). Stx2-treatment led to a significant dose-dependent decrease in pyroptosis and IL-1β production in EHEC infected cells (Fig. 3C–D). This reduction was specific to inflammasome responses as treatment with purified Stx did not affect EHEC-induced TNF or IL-6 production (Fig. 3E–F). Consistent with previous reports on its ability to activate MAP kinases in THP1 cells (23, 24), treatment of uninfected BMDMs with Stx induced secretion of low levels of TNF and IL-6 (Fig. 3E–F). Complementation of ΔStx infection with exogenous Stx2 protein at the time of infection brought ΔStx-induced elevated responses back to wild-type EHEC levels (Fig. 3G–H). Purified Stx2 also significantly reduced inflammasome activation, but not TNF or IL-6 production, by E. coli K12 (Fig. 3I–J and S3B–C).

Fig. 3. Treatment with purified Stx specifically inhibits the noncanonical inflammasome.

(A-B) LDH assay for cell death in Vero cells (A) or BMDMs (B) left untreated or treated with 4 or 8 μg/ml of purified Stx2 for 24 h.

(C-F) Cell death and secretion of indicated cytokines at 16 h post infection by BMDMs treated with PBS or indicated μg/ml (C-D) or 4 μg/ml (E-F) of Stx2 at the time of infection with EHEC at MOI 50.

(G-L) Cell death and IL-1β secretion by Pam3CSK4-primed BMDMs treated with PBS or 4 μg/ml of purified Stx2 at the time of infection with indicated bacterial strains or transfection (trxn) with 1 μg/10⁶ cells of LPS or poly(dA:dT) for 16 h.

(M) Immunoblot of gasdermin D in the lysates of BMDMs treated with 4 μg/ml of purified Stx2 at the time of transfection with 1 μg/10⁶ cells of LPS or poly(dA:dT) for 16 h.

(N-S) Cell death and IL-1β secretion by Pam3CSK4-primed BMDMs treated with PBS or 4 μg/ml of purified Stx2 at the time of infection with F. novicida for 16 h (MOI=50), or treatment with nigericin (10 μM) for 1 h, or infection with S. Typhimurium (MOI=1) for 4 h.

Pooled data from two or three independent experiments (K-L; Mean±SD) or one experiment representative of three experiments (A-J and M-S; mean±SEM) are presented. Each experiment had two or three technical replicates. *p<0.05; **p<0.01; ***p<0.001; ****p < 0.0001; one-way followed by Tukey’s or two-way ANOVA followed by Sidak’s post-test. See also fig. S3.

To further determine if Stx2 is capable of suppressing the cytosolic LPS sensing pathway in the absence of other bacterial factors, Stx2 was added to the cell supernatant just prior to transfection of purified LPS directly into the cytosol. Remarkably, Stx2 potently inhibited the inflammasome responses induced by LPS transfection including cell death, secretion of IL-1β, and cleavage of gasdermin D into its active form (3K-M and S3F). As predicted, Stx2 did not reduce the TLR-mediated responses, TNF and IL-6 (S3D-E). Interestingly, the suppressive effect of Stx2 was highly specific to the noncanonical cytosolic LPS sensing pathway as treatment with purified Stx2 had no effect on poly(dA:dT)- or Francisella novicida-induced activation of the AIM2 inflammasome (Fig. 3K–O), or nigericin-induced activation of the canonical NLRP3 inflammasome (Fig. 3P–Q), or Salmonella Typhimurium-induced activation of the NLRC4 inflammasome responses (Fig. 3R–S). Collectively, these data demonstrate that Stx is sufficient to suppress the noncanonical inflammasome responses induced by caspase-11. Importantly, these observations also demonstrate that Stx-mediated inflammasome suppression is not mediated by Stx-induced translational block and/or protein synthesis inhibition as the production of proteins such as TNF, IL-6, proIL-1β, caspase-11, and full length gasdermin D occurred normally in the presence of Stx (Fig. 1–3). Furthermore, Stx did not affect IL-1β secretion, gasdermin D cleavage, or pyroptosis downstream of other inflammasome pathways such as AIM2, canonical NLRP3, and NLRC4 (Fig. 3K–S).

Multiple variants of Stx are capable of suppressing the cytosolic LPS sensing pathway in murine and human cells

EHEC strains express two subtypes of Stx; Stx1and Stx2. Stx2 is more potent and is associated more frequently with hemorrhagic colitis and HUS (35). Stx1 and Stx2 are further divided into variants such as Stx1a, Stx1c, Stx1d and Stx2a, Stx2b, Stx2c, Stx2d, Stx2e, Stx2f, and Stx2g (35). We used purified Stx2a variant of Stx2 subtype in all the experiments described thus far. Considering the strong association between Stx subtypes and the disease pathogenesis, we sought to examine if the inflammasome-suppressive effect is common to additional Stx variants. Toward this end, we treated BMDMs with purified Stx variants, Stx2d and Stx1a, at the time of LPS transfection and analyzed inflammasome responses. Interestingly, both Stx2d and Stx1 significantly inhibited cell death and IL-1β secretion elicited by cytosolic LPS without affecting IL-6 or TNF production (Fig. 4A–D). Similar to Stx2a, Stx2d and Stx1 also inhibited the cleavage of gasdermin D into its active form (Fig. 4E–F and S4A–C). Caspase-11 expression was not affected by treatment with Stx2a, Stx2d, or Stx1 (Fig. 4E). The suppressive activity of these toxins were specific to cytosolic LPS-mediated responses as poly(dA:dT)-induced AIM2 inflammasome responses were not affected by any of the Stx variants (Fig. 4A–B and 4F). This data show that multiple Stx subtypes have the potency for inflammasome inhibition.

Fig. 4. Multiple Stx variants sufficiently inhibit inflammasome responses in murine and human cells.

(A-D) Cell death and secretion of indicated cytokines by Pam3CSK4-primed BMDMs treated with PBS or 4 μg/ml of indicated Stx variant at the time of transfection (trxn) with 1 μg/10⁶ cells of LPS or poly(dA:dT) for 16 h.

(E-F) Gasdermin D cleavage or caspase-11 expression assessed by immunoblot in the lysates of Pam3CSK4-primed BMDMs treated with PBS or 4 μg/ml of indicated Stx variant at the time of transfection with LPS or poly(dA:dT) for 16 h.

(G-I) Cell death in Caco-2 cells primed with 10 ng/ml of human IFNγ followed by treatment with PBS or 4 μg/ml indicated Stx variant at the time of transfection with 1 μg/10⁶ cells of LPS (G) or infection with indicated strains of bacteria (MOI=100) (H-I) for 16 h.

(J-K) Cell death (J) or immunoblots for human gasdermin D (hGSDMD) C-terminal p23 fragment in the supernatant and full length hGSDMD p53 or β-actin in the lysates (K) of A431 cells upon infection with EHEC or ΔStx at MOI=100 for 16 h.

Pooled data from two independent experiments (G-J; Mean±SD) or one experiment representative of three experiments (A-D; mean±SEM) are presented. Each experiment had three technical replicates. *p<0.05; **p<0.01; ***p<0.001; ****p < 0.0001; one-way or two-way ANOVA followed by Tukey’s post-test. See also fig. S4.

We further tested if Stx inhibits the caspase-4/5-mediated cytosolic LPS sensing pathway in human cells. A recent study (36) has shown that IFNγ priming followed by LPS transfection activates caspase-4 in Caco2 (human epithelial colorectal carcinoma) cells resulting in pyroptosis. Caco-2 cells did not undergo cell death upon treatment with Stx alone indicating that similar to primary macrophages they are not sensitive to Stx-toxicity (Fig. 4G). We tested if Stx inhibits caspase-4 activation by LPS transfection and bacterial infection in IFNγ-primed Caco-2 cells. Treatment with Stx2a, Stx2d, or Stx1 led to a significant decrease in cell death induced by LPS transfection in Caco-2 cells (Fig. 4G). Similarly, infection of Caco-2 cells with IPTG-treated E. coli BL21/pStx2, which expresses high levels of Stx, resulted in a significantly lower level of cell death compared to infection with H₂O- or IPTG-treated E. coli BL21/pEmpty and H₂O-treated E. coli BL21/pStx2 (Fig. 4H). In contrast, infection with ΔStx EHEC induced a higher level of cell death compared to infection with wild-type EHEC (Fig. 4I). We also used A431 (human epidermoid carcinoma) cells, another human epithelial cell line that has a functional caspase-4 pathway (37), to test Stx-mediated cytosolic LPS sensing pathway inhibition. We treated A431 cells with multiple doses of Stx2 to confirm that they are not susceptible to Stx-toxicity (Fig. S4D–F). Importantly, similar to Caco-2 cells, infection of A431 cells with ΔStx induced significantly higher cell death and gasdermin D cleavage compared to wild-type EHEC (Fig. 4J–K and S4G). Together, these data from Caco-2 and A431 cells indicate that Stx has an inhibitory effect on caspase-4 activation in human cells as well.

Catalytic activity of Stx is essential for suppressing the cytosolic LPS sensing pathway

While the variants of Stx differ in their toxic potency, they all share the enzymatic N-glycosidase activity of the A subunit (18). To determine if this catalytic activity of Stx A subunit is required for inflammasome inhibition, we used catalytically inactive mutants of Stx2a and Stx1 that contain the following replacements in the amino acids located within the enzymatically active cleft of the Stx A subunit; Stx2A-E167Q, R170L and Stx1A-E167Q, R170L (38). These mutations in the A subunit do not affect the B-subunit, thereby leaving the B subunit pentameric structure and its Gb3 receptor binding ability intact (39–41). The catalytic inactivity of purified mutant Stx preparations were confirmed by the Vero cell cytotoxicity assay (Fig. 5A). While Stx2a and Stx1 suppressed cell death, IL-1β secretion, cleavage of IL-1β and gasdermin D in response to EHEC or LPS transfection, the catalytically inactive mutants of these toxins failed to do so (Fig. 5B–J and S5A–F). Similarly, purified Stx B subunit alone did not suppress inflammasome responses to EHEC infection (Fig. 5B–C). As expected, neither the toxins nor the inactive mutants had an effect on poly(dA:dT)-induced AIM2 inflammasome responses (Fig. 5D–E and 5H–I). All these data show that Stx suppresses the noncanonical inflammasome responses through its catalytic activity.

Fig. 5. Catalytic activity of Stx is essential for inflammasome inhibition.

(A) LDH assay for cell death in Vero cells treated with 4 μg/ml of indicated versions of Stx for 16 h.

(B-I) Cell death and secretion of IL-1β by Pam3CSK4-primed BMDMs treated with PBS or 4 μg/ml of Stx2 or Stx1, or their corresponding enzymatically inactive versions (Stx2a-Inactive and Stx1-Inactive), or purified B subunit at the time of infection with EHEC or transfection with 1 μg/10⁶ cells of LPS or poly(dA:dT) for 16 h.

(J) Cleavage of gasdermin D and IL-1β assessed by immunoblot in the lysates or supernatants of Pam3CSK4-primed BMDMs treated with Stx2 or Stx1, or their corresponding enzymatically inactive versions (Stx2a-Inactive and Stx1-Inactive) at the time of transfection with LPS or poly(dA:dT) for 16 h.

Pooled data from two independent experiments (B-I; Mean±SD) or one experiment representative of three experiments (A; mean±SEM) are presented. Each experiment had three technical replicates. *p<0.05; **p<0.01; ***p<0.001; ****p < 0.0001; two-way ANOVA followed by Tukey’s post-test. See also fig. S5.

Stx inhibits cytosolic LPS sensing pathway at the level of gasdermin D activation

To define the target of Stx in the cytosolic LPS sensing pathway, we examined various steps of the pathway in the presence or absence of Stx. The entry of LPS into the cytosol is a critical proximal step in the activation of the cytosolic LPS sensing pathway. We recently demonstrated that during infection with bacteria such as EHEC, bacterial OMVs mediate cytosolic LPS delivery (2). Both wild-type EHEC and ΔStx produced similar levels of OMVs indicating that Stx does not affect OMV production (Fig. 6A). Additionally, we assessed the intracellular survival of EHEC and E. coli BL21 in the presence or absence of Stx. There was no significant difference in the intracellular bacterial count between cells infected with wild-type EHEC or ΔStx (Fig. S6A). Furthermore, treating the cells with purified Stx did not change the bacterial survival inside the cells (Fig. S6A). Similarly, IPTG-induction of Stx expression did not change the intracellular load of E. coli BL21 (Fig. S6B). These data indicate that Stx does not alter the intracellular survival of EHEC or E. coli BL21.

Fig. 6. Stx does not interfere with the upstream events in the cytosolic LPS sensing pathway.

(A) BCA assay to assess the quantity of OMVs purified from EHEC and isogenic ΔStx cultures at the same growth phase.

(B-C) LAL assay to assess LPS quantity in the cytosolic and residual fractions of BMDMs treated with H₂O or Stx2 (Stx2a variant) at the time of infection with EHEC or ΔStx (B) or infected with the indicated BL21 strains (C) at MOI 10 at 5 h (B) or 16 h (C) post infection.

(D) Immunoblot for caspase-11 or β-actin in the lysates (Input) or elute from streptavidin-pull down from the lysates of BMDMs transfected with unlabeled (LPS) or biotinylated-LPS (Biotin-LPS) in the presence or absence of indicated variants of Stx for 6 h.

(E) Release of AMC fluorescence in the lysates of gasdermin D-deficient BMDMs left untreated or transfected with LPS in the presence or absence of indicated variants of Stx at 5 h post treatment.

(F) Immunoblots for Stx2 A subunit or caspase-11 in the elute from Strep-Tactin-mediated immunoprecipitation (Streptactin IP) and for Stx2 A subunit, caspase-11, gasdermin D, and β-actin in the lysates (Input) from HEK293T cells stably expressing Twin-strep-tagged caspase-11 and HA-2x-FLAG-tagged murine gasdermin D at 2 h following electroporation with PBS or LPS in the presence or absence of Stx2 (Stx2a variant).

(G) Immunoblots for Stx2 A subunit and gasdermin D in the elute from anti-FLAG-mediated immunoprecipitation (FLAG IP) or lysates (Input) from HEK293T cells stably expressing untagged caspase-11 and HA-2x-FLAG-tagged murine gasdermin D at 2 h following electroporation with PBS or LPS in the presence or absence of Stx2 (Stx2a variant).

(H) SDS-PAGE gel stained with Coomassie blue showing recombinant mouse gasdermin D (mGsdmd) protein, which was subjected to cleavage by recombinant active caspase-11 (p22/p10) in the presence or absence of indicated quantities of Stx2 (Stx2a variant). Red ‘+’ indicate these proteins were pre-incubated for 30 min prior to adding the third (black ‘+’) component in the assay. Arrows indicate full length Gsdmd and cleaved Gsdmd D-N terminus p30 fragment. The band right below Gsdmd p30 is Stx2.

Pooled data from three independent experiments (B-C; Mean±SD) or one experiment representative of three experiments (A and E; mean±SEM, D-H) are presented. Each experiment had three technical replicates. Data analyzed by one-way or two-way ANOVA followed by Tukey’s post-test. See also fig. S6.

Following cellular uptake, OMVs traffic through the endocytic pathway and LPS accesses the cytosol most likely from the early endosomal compartments (2). To test if Stx inhibits entry of LPS into the cytosol, we used a well-established cell fractionation technique (2) to purify cytosolic fractions from cells infected with EHEC or ΔStx in the presence or absence of Stx2. Analysis of LPS quantity in the cytosolic and residual fractions (fraction containing organelles such as endosomes and lysosomes) demonstrated that LPS entered the cytosol at similar levels upon infection with EHEC and ΔStx regardless of the presence of Stx (Fig. 6B). Likewise, we observed similar levels of LPS in the cytosol of macrophages infected with H₂O- or IPTG-treated BL21/pEmpty or BL21/pStx2 showing that infection with high Stx-expressing IPTG-treated BL21/pStx2 did not reduce cytosolic LPS levels (Fig. 6C). Consistent with this, Stx was able to reduce cell death and IL-1β production even when LPS was directly introduced into the cytosol via electroporation bypassing the endocytic pathway (Fig. S6C–D). Together, these observations indicate that Stx-mediated inflammasome inhibition occurs at a level downstream of LPS trafficking to the cytosol. Importantly, this observation also implies that the interferon-inducible host factors such as guanylate binding proteins (GBPs) that are necessary for cytosolic LPS entry are not affected by Stx, further ruling out a role for Stx in the priming of noncanonical inflammasome.

We next assessed if Stx interferes with caspase-11 binding of LPS by utilizing a biotin-LPS-based pull down assay. BMDMs were transfected with biotin-LPS and subjected to pulldown with streptavidin beads followed by immunoblotting for caspase-11. Caspase-11 was detected in precipitates from cells transfected with biotin-LPS, but not with unlabeled-LPS, confirming the specificity of the pull-down assay (Fig. 6D). Interestingly, biotin-LPS pulled down similar levels of caspase-11 from BMDMs not treated with Stx or treated with any of the variants of Stx, Stx2a, Stx2d, or Stx1 (Fig. 6D). This suggested that Stx does not interfere with the binding of cytosolic LPS with caspase-11, but potentially regulates a downstream step in the pathway. LPS binding activates caspase-11 into a catalytically active protease, and this active caspase-11 subsequently cleaves gasdermin D at D276 (42). We tested if Stx inhibits the activation or catalytic activity of caspase-11 following LPS-binding using Ac-YEVD-AMC, a fluorogenic substrate that can be cleaved by active caspase-11 and caspase-1 (43). We used gasdermin D-deficient (Gsdmd^−/−) BMDMs in this assay to exclude the possibility of Ac-YEVD-AMC cleavage by caspase-1, as gasdermin D-deficient cells fail to activate caspase-1 upon cytosolic LPS sensing (Fig S2K) (32, 42). Gsdmd^−/− BMDMs were left untreated or transfected with LPS in the presence or absence of Stx. The lysates were mixed with Ac-YEVD-AMC, and the release of AMC fluorescence was used as an indicator of the catalytic activity of caspase-11. LPS transfection induced markedly higher AMC release compared to untreated cells indicating a strong activation of caspase-11. Interestingly, treatment with Stx variants did not affect the AMC fluorescence release in response to LPS transfection suggesting that Stx does not inhibit the activation of caspase-11 or its catalytic activity per se (Fig. 6E).

It is possible that Stx directly binds to gasdermin D blocking its interaction with caspase-11. To assess if Stx binds to gasdermin D or caspase-11 via immunoprecipitation assay, we generated HEK293T cell lines stably expressing (i) twin-strep-tagged caspase-11 and HA-2x-FLAG-tagged murine gasdermin D or (ii) untagged caspase-11 and HA-2x-FLAG-tagged murine gasdermin D. HEK293T cells were treated with purified Stx2 to confirm that they are not susceptible to Stx-mediated cell death (Fig. S6E–G). Electroporation of these stable cell lines with LPS resulted in cleavage of gasdermin D into p30 fragment and treatment with Stx2 following LPS electroporation led to a decrease in this cleavage (Fig. S6H, inputs in Fig. 6F–G, and S6I–J) illustrating that the Stx-mediated inhibition of the cytosolic LPS sensing pathway occurs in this HEK293T system as well. Subsequently, these cell lines were electroporated with purified LPS and treated with purified Stx2 or left untreated before immunoprecipitation with Strep-Tactin (to pull down caspase-11) or anti-FLAG (to pull down gasdermin D) magnetic beads followed by immunoblotting for Stx2. The Strep-Tactin and anti-FLAG beads efficiently pulled down caspase-11 and gasdermin D, respectively (Fig. 6F–G). While the Stx2 antibody detected the Stx A subunit in the lysate input of Stx2-treated cells, it did not detect Stx A subunit in the immunoprecipitates of either cell line (Fig. 6F–G). This result suggested that Stx may not be directly interacting with caspase-11 or gasdermin D.

To further verify this possibility, we employed a cell free gasdermin D cleavage assay. Here, recombinant murine gasdermin D was incubated with enzymatically active recombinant caspase-11 (p22/p10) in the presence or absence of Stx. While recombinant caspase-11 efficiently cleaved gasdermin D, we did not observe a decrease in this cleavage upon addition of Stx to the reaction suggesting that Stx does not directly interfere with caspase-11-gasdermin D interaction (Fig. 6H). The data from the immunoprecipitation and cell free gasdermin D cleavage assays suggest the possibility of Stx directly binding to caspase-11 or gasdermin D or directly blocking caspase-11-gasdermin D interaction is less likely and that Stx’s effect on gasdermin D cleavage possibly involves an intermediary protein or a cellular process.

Stx inhibits inflammasome responses to intracellular LPS in vivo

Having demonstrated a noncanonical inflammasome-inhibitory function of Stx in vitro, we sought to determine if Stx exerts such a function in vivo. Intraperitoneal LPS administration is well-documented to activate the caspase-11 pathway in mice in vivo (3, 32, 44, 45). To directly test if Stx modulates the cytosolic LPS sensing pathway in vivo, we adapted this LPS injection model with Stx co-administration. Mice were injected with PBS or a low dose of Stx2 30 min prior to injection with purified LPS. As expected, injection of LPS elicited robust IL-18 and IL-1β secretion to plasma. Importantly, these inflammasome responses were significantly reduced in mice pretreated with Stx2 (Fig. 7A–B). This Stx-mediated inhibition was specific to inflammasome-dependent cytokines as IL-6 levels remained unchanged in Stx-pretreated mice (Fig. 7C). A similar decrease in plasma IL-18 levels was observed when mice were injected with purified Stx1 prior to LPS injection (Fig. 7D). A reduction, albeit to a lesser extent, was observed in IL-1β also upon Stx1 injection (Fig. 7E), whereas IL-6 levels remained unchanged in the presence of Stx1 (Fig. 7F). These findings demonstrate that Stx is capable of suppressing LPS-induced inflammasome responses in vivo as well.

Fig. 7. Stx inhibits inflammasome responses elicited by cytosolic LPS sensing pathway in vivo.

(A-F) Indicated cytokines in the plasma of mice injected i.p. with PBS or 20 ng Stx2 (Stx2a variant) (A-C) or 50 ng Stx1 (D-F) 30 min prior to i.p. injection with 100 μg LPS assessed by ELISA at 6 h post LPS injection.

(G) Bacterial colonization determined by viable fecal bacterial counts in the colon content of mice infected with 1×10⁸ CFU of C. rodentium (C. rod) or C. rodentium-Stx (C. rod-Stx) at day 2 post infection.

(H-K) IL-18 or IL-6 in the colon or kidney homogenates of mice left uninfected or infected with 1×10⁸ CFU of C. rodentium or C. rodentium-Stx at day 2 post infection. n=5 for uninfected, n=10 for C. rodentium and C. rodentium-Stx.

(L) Immunoblot for gasdermin D, caspase-1, caspase-11, or β-actin in the colon lysates of mice left uninfected (UI) or infected with C. rodentium or C. rodentium-Stx at day 2 post infection. Each lane represents an individual mouse and each blot has lysates from separate groups of one uninfected, five C. rodentium-infected, and five C. rodentium-Stx-infected mice.

(M-N) Image J-based densitometry analysis of relative density of gasdermin D p30 (G) and caspase-1 p20 (H) bands normalized to the density of β-actin bands from immunoblots shown in figure 7L. n=2 for uninfected, n=10 for C. rodentium and C. rodentium-Stx.

In A-K and M-N, each symbol represents one mouse and the horizontal bars indicate mean. Data analyzed by unpaired two tailed t-test (A-F) or one-way ANOVA with Tukey’s post-test (G-K and M-N); *=p<0.05; **=p<01; ***<0.001. See also fig. S7.

Stx inhibits inflammasome responses against Citrobacter rodentium infection in mice

Next, we examined if Stx mediated suppression of inflammasome responses occurs during the bacterial infection in vivo. As EHEC do not normally colonize mice, C. rodentium, a natural murine pathogen, is commonly used to model human EHEC infection in mice. However, C. rodentium does not encode for Stx, and therefore to make C. rodentium a better EHEC-mimic, a Stx-expressing strain of C. rodentium was generated by lysogenizing it with Stx2-encoding phage (C. rodentium-Stx) (46). A murine infection model with C. rodentium-Stx recapitulates the critical features of human EHEC disease (46). Our previous studies have shown that like EHEC, C. rodentium activates the caspase-11-dependent cytosolic LPS sensing pathway (47). To investigate the impact of Stx on inflammasome responses to an intestinal bacterial infection, we orally infected mice with C. rodentium-Stx or isogenic wild-type C. rodentium that lacks Stx and assessed inflammasome responses at an early stage of infection. Inflammasome activation in colon and kidney, two major organs affected by Stx, was assessed. Similar to previous reports, both the strains colonized mice at similar levels (Fig. 7G). Remarkably, while the C. rodentium strain induced a strong IL-18 response in the colon compared to uninfected mice, this response was significantly decreased in C. rodentium-Stx infected mice (Fig. 7H, S7A and S7C). A similar decrease in IL-18 levels was observed in the kidney of mice infected with C. rodentium-Stx compared to those infected with C. rodentium (Fig. 7I). On the other hand, there was no significant difference in colon and kidney IL-6 levels between C. rodentium-Stx- and C. rodentium-infected mice (Fig. 7J–K). Furthermore, immunoblot analysis of colon lysates demonstrated a considerable reduction in the cleavage of gasdermin D and caspase-1 into their active forms in mice infected with C. rodentium-Stx (Fig. 7L–N). Notably, in vivo infection with Stx expressing bacteria did not affect the level of caspase-11 or pro IL-18 levels in the colon (Fig. 7L, S7A–B). Overall these data further confirm that the suppressive role of Stx on inflammasome responses exists in vivo as well.

Discussion

A highly conserved bacterial molecule, LPS, is detected in the cytosol by a family of inflammatory caspases (caspase-11 in mice, and caspase-4 and caspase-5 in humans). This cytosolic LPS sensing by the noncanonical inflammasome and the ensuing responses are emerging as a key host defense strategy against Gram-negative bacterial infections (47). However, less is known about the ways by which bacteria antagonize this pathway. In this study, we provide key insights into a way by which a successful enteric pathogen, EHEC, employs its primary virulence factor, Stx, to inhibit gasdermin D cleavage and limit pyroptotic and IL-1 responses driven by cytosolic LPS sensing. We show that Stx deficiency results in increased inflammasome activation by EHEC, whereas Stx-overexpression or treatment with purified Stx protein leads to a significant suppression of noncanonical inflammasome responses. This Stx-mediated inflammasome suppression was dependent on its catalytic activity but not on translational inhibition of inflammasome components. Notably, Stx is also able to restrict inflammasome responses against purified LPS and C. rodentium intestinal infection in mice.

Stx is a well characterized bacterial virulence factor in terms of its toxic activity. However, it is not clear if Stx executes any additional functions in the context of EHEC-immune cell interactions. This is particularly important given the cell specificity of Stx-mediated cytotoxicity; only cells expressing high levels of Gb3 receptor such as renal tubular epithelial cells and microvascular endothelial cells undergo translational inhibition and apoptosis upon Stx interaction. In contrast, immune cells such as macrophages and monocytes are not sensitive to Stx-mediated protein synthesis inhibition and cell death (21, 23). Therefore, it is possible that Stx has other biological effects on these cells. Although previous studies suggested that Stx activates the caspase 4/5-mediated NLRP3 inflammasome in THP1 cells, it was later revealed that this inflammasome activation was in fact mediated by the LPS contamination of the Stx preparation and not Stx itself (48, 49).

Our study assigns a new biological function for Stx that aids in the bacterial evasion of the caspase-11-dependent noncanonical inflammasome. Consistent with the previous studies that showed the refractory nature of primary macrophages to Stx-mediated protein synthesis inhibition (21–23), the inflammasome-suppressive function of Stx revealed in this study is not mediated by the global protein synthesis inhibition. Numerous proteins including TNF, IL-6, pro IL-1β, and pro IL-18 as well as noncanonical inflammasome components such as caspase-11 and gasdermin D were synthesized normally in the presence of Stx both in vitro and in vivo. Additionally, while Stx disrupted inflammasome responses elicited by multiple activators of caspase-11 including, EHEC, E. coli K12, and LPS transfection, it did not impede the activation of other inflammasome pathways such as AIM2, canonical NLRP3, and NLRC4 demonstrating that inflammasome-associated proteins are synthesized and are functioning normally in the presence of Stx.

How does Stx suppress gasdermin D activation and the downstream responses to cytosolic LPS? Caspase-11 activates gasdermin D via a unique mechanism; the exosite recognition of the C-terminal domain of gasdermin D triggers the tetrapeptide sequence-independent cleavage of gasdermin D by caspase-11 (50). Taking this new insight and our collective observations into account, we speculate that Stx, via its N-glycosidase activity, interferes with the hydrophobic interaction between the exosite and gasdermin D’s C-terminal domain, thus impairing gasdermin D recognition by caspase-11. Our data from an in vitro gasdermin D cleavage assay and the HEK293T co-expression system indicate that Stx may not be directly binding to caspase-11 or gasdermin D to inhibit the interactions between the two proteins. Further biochemical studies are required to explore the possibility of Stx inhibiting caspase-11-gasdermin D interaction through an indirect mechanism.

Pathogens employ a variety of direct and indirect mechanisms to prevent the activation of innate immune pathways including inflammasomes. Although a recent study found that Shigella flexneri effector protein, OspC3, inhibits caspase-4-mediated cell death in human epithelial cells (37), bacterial proteins specifically targeting the cytosolic LPS sensing pathway still remain largely unknown. On the other hand, suppression of canonical inflammasomes such as NLRC4 and AIM2 by bacterial pathogens is relatively well characterized (51, 52). Notably, this inflammasome suppressive function of Stx is quite unique compared to other bacterial exotoxins such as anthrax lethal toxin, pertussis toxin, C3 toxin of Clostridium botulinum, and pneumolysin of Streptococcus pneumoniae. Many of these toxins are known to activate rather than suppress various inflammasome pathways; anthrax lethal toxin activates the NLRP1b inflammasome and the pore-forming toxins pneumolysin and listeriolysin O activate the NLRP3 inflammasome (53). Similarly, enzymatic activity of numerous bacterial toxins leads to Rho GTPase inactivation and subsequent activation of the pyrin inflammasome; examples include ADP-ribosyltransferase activity of pertussis toxin and C3 botulinum toxin, adenylyl transferase activity of VopS, and glucosyltransferase activity of TcdB (53, 54). In contrast to the enzymatic activity of these toxins, Stx possesses N-glycosidase activity. Notably, this enzymatic activity is required for Stx-mediated inflammasome suppression suggesting that the distinct biological activity of Stx could be the underlying reason for Stx functioning differently from other bacterial toxins in the context of inflammasome signaling.

Interestingly, increasing evidence from recent studies show that proteins encoded by bacteriophages inserted into the chromosome of pathogenic bacteria play an important role in modulating immune responses (25). Pf phage of P. aeruginosa has been shown to suppress proinflammatory cytokine production as well as phagocytosis by macrophages (55). Similarly, TarP, a phage-encoded glycosyltransferase expressed by methicillin resistant Staphylococcus aureus (MRSA), modifies the bacterial cell wall lipoteichoic acid resulting in reduced immunogenicity and thereby aids in immune evasion (56). Furthermore, a recent study identified a phage-encoded protein, Ankyrin (ANKp), that suppresses innate immune responses elicited by macrophages against bacteria such as E. coli and Bacillus subtilis (57). Collectively, these studies along with our findings on Stx reveal a common theme where bacterial pathogens employ bacteriophage-encoded proteins to subvert immune responses. This provides an evolutionary advantage to both the bacteria as well as the phage as protecting its bacterial symbiont from the host defense attack will be beneficial for the survival of the phage. In summary, this study illustrates that the phage-encoded virulence factor of EHEC, Stx, is indeed a multifunctional protein with both pathogenic and immunosuppressive activity.

Materials and Methods

Study design

The objective of this study was to characterize a mechanism of inhibition of the cytosolic LPS sensing pathway by the human pathogen, EHEC. Stx, a major virulence factor of EHEC, was found to be inhibiting inflammasome responses. We further characterized this Stx-mediated inflammasome inhibition extensively in mouse macrophages and human epithelial cells using Stx-mutant of EHEC, E. coli BL21 expressing Stx, and purified Stx. We also studied this Stx-mediated inflammasome inhibition in vivo using LPS intraperitoneal injection and C. rodentium-Stx infection models in mice.

Mice

C57BL/6 mice from NCI Charles River, Casp11^−/− mice (kind gift of Vishva Dixit and Kate Fitzgerald), and Gsdmd^−/− mice (kind gift of Vishva Dixit) were bred and maintained in specific pathogen–free conditions at the UConn Health animal facility. Both male and female mice were used for this study. Animal protocols were carried out in accordance with the guidelines set forth by the Institutional Animal Care and Use Committee at UConn Health.

Bacterial strains and growth conditions

Bacterial strains used in this study include the EHEC strain E. coli O157:H7 EDL933, various isogenic EHEC mutants including ΔStx that lacks both Stx1 and Stx2, E. coli K12, E. coli BL21 and its derivatives, Francisella tularensis subsp. novicida strain Utah 112 (F. novicida; BEI resources, NIAID, NIH), and Salmonella enterica serovar Typhimurium (S. Typhimurium) strain SL1344. E. coli strains were grown overnight at 37°C in Luria Bertani (LB) broth unless otherwise mentioned. F. novicida was grown in Mueller Hinton broth overnight at 37°C. S. Typhimurium was sub-cultured in fresh LB broth from an overnight culture and grown for 3 h at 37°C to reach exponential phase of growth.

Construction of E. coli BL21 strain expressing Stx

A PCR amplicon encoding the whole sequence of Stx2 from EHEC EDL933 strain was cloned into a pET21a expression vector carrying an isopropyl-β-D-thiogalactoside (IPTG)-inducible promoter through standard molecular biology techniques. pET21a harboring Stx2 (pStx2) or empty pET21a were then transformed into E. coli BL21(DE3)/pLysS strain generating E. coli BL21/pStx2 (BL21/pStx2) and E. coli BL21/pEmpty (BL21/pEmpty) respectively. To induce Stx expression, overnight grown BL21/pStx2 or BL21/pEmpty were re-inoculated into fresh LB media containing ampicillin (for pET21a) and chloramphenicol (for pLysS). After 2 h of growth the cultures were treated with 0.5 mM IPTG or water for 5 h and were used for infecting macrophages at MOI 25.

Cell culture and stimulations

Bone-marrow derived macrophages (BMDM) were generated as described previously (58). Cells used to assess the inflammasome and cell death responses were left unprimed or were primed with Pam3CSK4 (InvivoGen) for at least 3 h prior to infection. For E. coli strains and F. novicida, cells were infected with overnight grown bacteria at an MOI of 50 or the indicated MOI. After 1 h the media was replaced with gentamicin (100 μg/ml) containing media and the supernatants and/or lysates were collected 16 h post-infection. For S. Typhimurium, cells were infected with MOI=1 of exponential phase bacteria for 4 h as described before (59). For transfection experiments, ultrapure LPS (1 μg/10⁶ cells) or poly(dA:dT) (1 μg/10⁶ cells) were transfected using lipofectamine 2000 (Thermo Fisher Scientific) and supernatants and lysates were collected after 16 h of treatment. Cell were treated with nigericin (Sigma-Aldrich) at 10 μM for 1 h. In experiments with Stx treatments, BMDMs were treated with purified Stx subtypes, Stx2a, Stx2d, or Stx1 (List Labs or Tufts Phoenix Laboratory) or catalytically inactive Stx2a or Stx1 (Tufts Phoenix Laboratory) (38), or purified B Subunit of Stx (BEI resources, NIAID, NIH), at the time of infection or treatment with indicated stimuli at a dose of 4 μg/ml unless otherwise indicated. LAL assay (Associates of Cape Cod Inc) was used to confirm lack of LPS contamination in all Stx preparations. For LPS electroporation, BMDMs were electroporated with 1 μg per 1×10⁶ cells of LPS using the Neon Transfection System (Thermo Fisher Scientific) according to the manufacturer’s instructions and supernatants were collected at 6 h later. For experiments with Caco-2 cells, cells were primed with human IFNγ for 16 h prior to transfection with ultrapure LPS (1 μg/10⁶ cells) in the presence of PBS or multiple variants of Stx (4 μg/ml). Caco-2 cells were also infected with EHEC, ΔStx, or E. coli BL21 strains at MOI=100. After 2 h, the media was replaced with gentamicin (100 μg/ml) containing media and cell death was assessed at 16 h post-infection. For experiments with A431 cells, cells were infected with EHEC or ΔStx at MOI=100. After 2 h, the media was replaced with gentamicin (100 μg/ml) containing media and cell death was assessed at 16 h post-infection. A431 cells were also treated with 2, 4, or 8 μg/ml of Stx2a and cell death was assessed at 16 h after treatment.

Vero cell cytotoxicity assay

Vero cells were treated with the indicated preparations of purified Stx or supernatants or pellets from water- or IPTG-treated BL21/pStx2 or BL21/pEmpty strains for 24 h and cell death was assessed as described below.

ELISA and cell death assay

IL-1β, IL-1α, TNF, and IL-6 levels were assessed by Ready-Set-Go!® ELISA kits (Thermo Fisher Scientific) according to manufacturer’s instructions. IL-18 ELISA was performed as described before (60). Cell death was assessed by LDH cytotoxicity detection kit (Clontech) or CellTiter-Glo® assay (Promega) or propidium iodide (PI) staining.

Immunoblotting and antibodies

Immunoblotting was performed on cell lysates or precipitates from cell supernatants as described before (2, 47) with the following the antibodies; caspase-1 p20 (clone casper1; Adipogen), IL-1β (AF-401-NA; R&D Systems), gasdermin D (EPR19828; Abcam), mouse caspase-11 (17D9; Cell Signaling), Stx2 A subunit (11E10; Thermo Scientific), human gasdermin D (EPR19829; Abcam), IL-18 (Biovision), and β-actin (Sigma-Aldrich).

Isolation of cytosol fraction from BMDMs

BMDMs were infected with the indicated E. coli strains or treated with Stx for 5 h or 16 h as indicated and subcellular fractionation of BMDM was conducted by a digitonin-based fractionation method as described previously (2). Briefly, the cells were washed and treated with 0.005% digitonin extraction buffer for 8 min to collect the supernatant containing cytosol. The residual cell fraction containing cell membrane, organelle, and nucleus was collected in 0.1% CHAPS buffer. Dilutions of these fractions were used for LAL assay to determine LPS quantity.

Gentamicin killing assay

BMDMs were infected with the indicated E. coli strains or treated with Stx as indicated. At 30 min post infection, the media was replaced with gentamicin (100 μg/ml) containing media. At the indicated time points, the cells were washed with PBS, lysed with 0.1% Triton-X, and the lysates were serially diluted and plated on LB agar to enumerate the intracellular bacterial count.

Immunoprecipitation of caspase-11 with biotinylated-LPS

BMDMs were transfected with biotinylated- or untagged-LPS in the presence or absence of water, Stx2a, Stx2d, or Stx1. The cells were lysed 6 h after the treatments, biotin-LPS was pulled down using streptavidin beads, and the level of capase-11 bound to the LPS was assessed by immunoblotting with anti-caspase-11 antibody.

Measurement of caspase-11 activation with fluorescent substrate

Gsdmd^−/− BMDMs were left untreated or transfected with 1 μg per 1×10⁶ cells of LPS in the presence or absence of Stx variants. Cells were lysed after 5 h of transfection and the lysates were incubated with Ac-YEVD-AMC (Santa Cruz Biotechnology) as described before (43). The cleavage of the substrate was monitored by the release of AMC fluorescence at 460 nm using 385 nm excitation.

HEK293T cell expression and co-immunoprecipitation of proteins

We generated three HEK293T cell lines stably expressing (i) twin-strep-tagged caspase-11 and HA-2x-FLAG-tagged murine gasdermin D (ii) untagged caspase-11 and HA-2x-FLAG-tagged murine gasdermin D, and iii) pMSCV puro vector alone as control. 20×10⁵ cells were transfected with 4 μg of each construct using Lipofectamine 2000 as per the manufacturer’s protocol. Media was replaced with fresh media 24 h post transfection. Cells were harvested 48 h post transfection, seeded at a concentration of 1×10⁵ cells in 10 ml of 10% FBS containing DMEM, and selected with 0.4 μg/ml of puromycin. Cells stably expressing the proteins were maintained in puromycin. Cells expressing caspase-11 and gasdermin D as described in (i) and (ii) were electroporated with PBS or 1 μg per 1×10⁶ cells of LPS. LPS-electroporated cells were left untreated or treated with 4 μg/ml Stx2 (2a variant). All cells were lysed 2 h after electroporation. Caspase-11 and gasdermin D were immunoprecipitated using Strep-Tactin (MagStrep “type3” XT Beads, Strep-Tactin® XT coated magnetic beads – IBA-Lifesciences) or FLAG magnetic beads (Anti-FLAG® M2 Magnetic Beads – Sigma) respectively and co-immunoprecipitation of Stx was assessed by immunoblotting with anti-Stx2 A subunit antibody.

In vitro gasdermin D cleavage assay

To examine cleavage of gasdermin D, approximately 7.5 μM of purified murine gasdermin D was incubated with 2.5 μM purified caspase-11(p22/p10 form) (33) at 37°C for 30 min in buffer A (20mM Tris-HCl, pH8.0; 150mM NaCl) with or without 2 μg or 4 μg purified Stx2. The cleavage was stopped by adding SDS loading buffer to the reaction mixture followed by boiling at 95°C for 5 min. The samples were analyzed by SDS-PAGE and Coomassie brilliant blue staining.

In vivo stimulation and infection

Eight to 12 weeks old C57BL/6 mice were intraperitoneally (i.p.) injected with PBS or 20 ng of purified Stx2a or 50 ng of Stx1 30 min prior to i. p injection with 100 μg of purified LPS (Sigma). Cytokine levels in the plasma were analyzed at 6 h post-LPS injection.

For C. rodentium infection studies, 8–12 weeks old C57BL/6 mice were orally gavaged with 1×10⁸ CFU of either C. rodentium-Stx or the isogenic wild-type C. rodentium. Mice mock infected with PBS were used as controls. Mice were weighed daily and fecal shedding of each strain was monitored by fecal colony counts. Tissues such as colon and kidneys were harvested at 2 days post infection. Tissue homogenates were prepared in lysis buffer (1% NP-40 in PBS + protease inhibitor cocktail) and the levels of cytokines such as IL-1β, IL-18, and IL-6 were measured in these homogenates using ELISA. Protein quantity in the homogenates was measured using Pierce™ BCA protein assay kit (Thermo Fisher Scientific). Cleavage of caspase-1 and gasdermin D in tissue homogenates was assessed by running equal protein quantity of samples on polyacrylamide gels followed by transfer onto nitrocellulose membranes and immunoblotting with corresponding antibodies.

Statistical analysis

In vitro data were analyzed for statistical significance by one-way or two-way analysis of variance (ANOVA) followed by the Tukey’s or Sidak’s post-test with Prism Software. Data from in vivo experiments were analyzed by unpaired two tailed t-test or one-way ANOVA followed by Tukey’s post-test. P values of less than 0.05 were considered significant.

Supplementary Material

Supplementary figures

Figure S1. Stx suppresses caspase-11-dependent inflammasome responses

Figure S2. In trans Stx expression impairs gasdermin D-dependent inflammasome responses elicited by E. coli BL21

Figure S3. Purified Stx does not induce cell death nor inhibit TLR-mediated responses in macrophages

Figure S4. Multiple variants of Stx inhibit gasdermin D cleavage in response to LPS transfection

Figure S5. Catalytic activity of Stx is essential for inflammasome inhibition

Figure S6. Stx inhibits inflammasome responses elicited by LPS electroporation

Figure S7. Stx inhibits cleavage of IL-18 in colon upon infection with C. rodentium

Table S1

Table S1. Raw data