GD-NT, but not GD-FL, possesses K63-linked polyubiquitination

To investigate the ubiquitination of GSDMD, the vectors expressing GD-FL, GD-NT, GD-CT, or caspase-11 were constructed (Fig. 1A, B). We transfected Flag-GD-FL/caspase-11 into HEK293 cells and performed IP & immunoblot assay. Interestingly, the result showed that GSDMD was slightly ubiquitinated when it was introduced into the cells alone. However, the ubiquitination level of GSDMD was significantly increased when it was co-transfected with caspase-11 (Fig. 1C). To further ensure it, we adopted a more rigorous ubiquitination detection method, in which we denatured the cellular proteins using guanidine and subjected to NI/NTA pulldown (Fig. 1D). Immunoblot (IB) analysis of the precipitants revealed the same ubiquitination pattern as shown in Fig. 1C. We supposed that the alteration of ubiquitination may take place during GSDMD processing by caspase-11. To test this hypothesis, we independently transfected Flag-GD-FL, Flag-GD-NT, or Flag-GD-CT into HEK293 cells. 18 hours later, the cells were subjected to IP assay. IB analysis of the precipitants revealed that GD-NT had a distinct ubiquitination status from GD-FL (Fig. 1E). Later on, we performed GST-pulldown and NI/NTA pulldown assay under denatured conditions. Consistently, these assays also indicated that GD-NT, but not GD-FL or GD-CT, was massively ubiquitinated (Fig. 1F, G).

To further identify the conjugation form of ubiquitin on GD-NT, we transfected Flag-GD-NT along with either Ub-WT or Ub-K63 only into HEK293 cells and then performed an NI/NTA pulldown assay. The result showed that GD-NT was mainly conjugated with K63-linked polyubiquitin chains (Fig. 1H). Thus, our data suggest that GD-NT, but not GD-FL, is massively modified with K63-ubiquitination.

The K63-linked polyubiquitination of GD-NT is regulated by TRAF1-OTUB1 axis

Ubiquitination is a dynamic and reversible process in which E3 ligases and deubiquitinases (DUB) catalyze the conjugation and removal of ubiquitin on substrate proteins, respectively [39, 40]. In addition, the E3 ligase-DUB axis exhibits its specificity by determining the target proteins and their ubiquitination forms. To identify the E3 ligase responsible for the K63-linked ubiquitination of GD-NT, we co-transfected Flag-GD-NT with various E3 ligases into HEK293 cells and conducted an IP assay. IB analysis of the precipitants indicated that Flag-GD-NT interacted with TRAF1 (Fig. 2A, B). To assess the influence of TRAF1 on GD-NT ubiquitination, we transfected either TRAF1 or shTRAF1 together with Ub-K63 only and Flag-GD-NT into HEK293 cells, followed by an NI/NTA pulldown assay. We found that TRAF1 significantly enhanced the K63-linked ubiquitination of GD-NT, whereas the ubiquitination level was downregulated by shTRAF1 (Fig. 2C, D). Interestingly, although the interaction of GD-NT with TRAF2 was not observed in our earlier experiments (Fig. 2A, B), TRAF2 still enhanced the K63-linked ubiquitination of GD-NT (the last two lanes in Fig. 2C). This phenomenon, we believe, may be attributed to the formation of a heterodimer between TRAF1 and TRAF2 [41], or to the involvement of another factor that has yet to be identified.

A, B Interaction between GD-NT and E3 ligases. HEK293T cells were transfected with Flag-GD-NT and different HA-E3. 24 hours later, the cells were harvested for Flag-IP assay (A) and HA-IP assay (B). The preciptants were subjected to immunoblot with the indicated antibodies. C, D Deubiquitylation of GD-NT by DUBs. HEK293T cells were transfected with Flag-GD-NT, His-Ub-K63 only, and HA-TRAF/2 (C) or shTRAF1 (D). 24 hours later, the cells were harvested for Ni-NTA pull-down assays. The precipitants and WCL were immunoblotted with the indicated antibodies. E GD-NT interaction with DUBs. HEK293T cells were transfected with HA-GD-NT and different Flag-DUB. 24 hours later, the cells were harvested for IP assay. The preciptants were subjected to immunoblot with the indicated antibodies. F Deubiquitylation of GD-NT by DUBs. HEK293T cells were transfected with HA-GD-NT, His-Ub-WT, and Flag-OTUB1 or Flag-OTUB2. 24 hours later, the cells were harvested for Ni-NTA pull-down assays. The precipitants and WCL were immunoblotted with the indicated antibodies. G HEK293T cells were transfected with HA-GD-NT, His-Ub-K63 only, and Flag-OTUB1 or Flag-OTUB2. 24 hours later, the cells were harvested for Ni-NTA pull-down assays. The precipitants and WCL were immunoblotted with the indicated antibodies. H Similar to (G), except that shOTUB1 was used to knock down the expression of OTUB1. I TRAF1 and TRAF2 enhance the cytolytic activity of GD-NT. HEK293 cells were transfected with 3xFlag-GD-NT with or without HA-TRAF1 and HA-TRAF2. 48 hours later, morphological changes were observed using phase-contrast imaging (lower panel). Dying cells were stained with PI and photographed using fluorescent microscope (upper panel). J Similar to (I), except that the cells were subjected to LDH-based Cytotoxicity Assay. K Similar to (I), except that the WCL were immunoblotted with the indicated antibodies.

Next, to identify the DUB responsible for the K63-linked ubiquitination of GD-NT, we transfected Flag-GD-NT along with different DUBs into HEK293 cells and performed an IP assay. The result showed that Flag-GD-NT interacted with both OTUB1 and OTUB2 (Fig. 2E). Subsequently, either Ub-WT or Ub-K63 only was co-transfected with HA-GD-NT and/or OTUB1/2, followed by NI/NTA pulldown. IB analysis of the precipitants showed that, although both OTUB1 and OTUB2 affected the ubiquitination of GD-NT, only OTUB1 significantly reduced the K63-linked ubiquitination of GD-NT (Fig. 2F, G). Consistently, shRNA-mediated silencing of OTUB1 significantly increased the K63-linked ubiquitination of GD-NT (Fig. 2H). Moreover, we found that 3 × Flag-GD-NT, which loses its pore-forming activity due to the addition of triple Flag-tag on its N-terminal [3], regained the ability to mediate pyroptosis with the assistance of TRAF1 and TRAF2 (Fig. 2I–K). Therefore, our findings suggest that the K63-linked ubiquitination of GD-NT is regulated by the TRAF1-OTUB1 axis, which in turn influences the pyroptotic activity of GD-NT.

GD-NT possesses the K63-linked polyubiquitination through its Lys237

To identify the ubiquitin conjugation sites on GD-NT, we searched the online PhosphoSitePlus® database (PSP, https://www.phosphosite.org), which provides comprehensive information for the study of ubiquitination as well as other post-translational modifications (PTMs) [42]. Based on Low ThroughPut and High ThroughPut datasets obtained from the PSP database [38, 43], several amino acids were considered as the potential ubiquitination sites, including Lys204, Lys205, and Lys237 in mouse GSDMD (refer to Lys203, Lys204, and Lys236 in humans). Sequence alignment of GSDMD among different species revealed an evolutionary conservation of these lysine residues (Fig. 3A).

A Lys237 of GD-NT is evolutionarily conserved among various organisms, including H. sapiens, P. troglodytes, M. mulatta, C. lupus, B. taurus, M. musculus, and R. norvergicus. B Structural damage analysis of GSDMD site mutation. Structural damage was assessed using the online tool Missense3D (http://www.sbg.bio.ic.ac.uk/missense3d). No structural damage was detected during the creation of GD-NT mutants via site-directed mutagenesis. C HEK293 cells were transfected with Flag-GD-NT, K237R, K204/205R. 24 hours later, the cells were harvested for Flag-IP assay. The precipitants and WCL were immunoblotted with the indicated antibodies. D His-Ub-WT was transfected with Flag-GD-NT, K237R, K204/205R into HEK293 cells. 24 hours later, the cells were harvested for Flag-IP assay. The precipitants and WCL were immunoblotted with the indicated antibodies. E Similar to (D), except that His-Ub-K63 only was used instead of His-Ub-WT. F His-Ub-WT was transfected with Flag-GD-NT, K237R, K204/205R into HEK293 cells. 24 hours later, the cells were harvested for Ni/NTA pulldown assay. The precipitants and WCL were immunoblotted with the indicated antibodies. G Similar to (F), except that His-Ub-K63 only was used instead of His-Ub-WT. H HEK293T cells were transfected with Flag-GD-NT and HA-TRAF1/2 or HA-OTUB1. 16 hours later, the cells were harvested for Ni-NTA pull-down assays. The precipitants and WCL were immunoblotted with the indicated antibodies. I Similar to (F), except that His-Ub-K48 only was used instead of His-Ub-WT. J HEK293 cells were transfected with Flag-GD-NT-WT or K237R. 12 hours later, the cells were treated with cycloheximide (CHX, 50 μg/mL) and then harvested at different time points for Flag-IP assay. The WCL were immunoblotted with the indicated antibodies. In C, D, and E, to keep the integrity of the ubiquitin chain attached to substrate proteins, NEM (10 mM) was added to inhibit the activity of cysteine peptidases. To prevent the influence of GSDMD interactome, protein samples for IP were denatured via heating at 100 °C for 5 minutes to break protein-protein interactions.

To assess the ubiquitination of Lys204, Lys205, and Lys237 of GD-NT (but not GD-FL), we created the non-ubiquitinatable mutants GD-NT-K237R and GD-NT-K204/205R. First, we performed Missense3D analysis and found no evidence of artificial structural damage resulting from these “K to R” substitutions, indicating that these constructs could be used for subsequent ubiquitination studies (Fig. 3B). Then, we transfected HEK293 cells with GD-NT-WT, GD-NT-K237R or GD-NT-K204/205R and conducted IP assay. The results showed that K237R, but not K204/205R, significantly impaired the K63-linked polyubiquitination of GD-NT (Fig. 3C–E). This difference was even more obvious in the subsequent NI-NTA pulldown assay (Fig. 3F, G). Consistently, we found that the TRAF1-OTUB1 axis affected GD-NT-WT, but not the mutant GD-NT-K237R, for K63-linked polyubiquitination (Fig. 3H). These findings indicate that K63-linked polyubiquitination of GD-NT primarily occurs at Lys237, and this event is regulated by the TRAF1-OTUB1 axis.

Intriguingly, we observed that GD-NT-K237 also possessed K48-linked polyubiquitination, which is primarily associated with proteasomal degradation (Fig. 3I). However, this modification did not affect GD-NT stability (Fig. 3J), suggesting that the K48-linked polyubiquitination on GD-NT-K237 represents a non-functional modification.

The K63-linked polyubiquitination controls GD-NT pyroptotic activity

As a specific pattern of PTMs, the K63-linked polyubiquitination generally controls various properties of the proteins, including protein-protein interaction, translocation, and activation. To assess whether the K63-linked polyubiquitination of Lys237 influences GD-NT pore-forming and cytolytic activity, we transfected HEK293 cells with GD-NT-WT, GD-NT-K237R or GD-NT-K204/205R. 24 hours later, culture medium and whole cell lysate were harvested for immunoblot analysis, respectively. Interestingly, GD-NT-K237R was not detected in the culture medium, suggesting that this GD-NT mutant can not permeabilize the plasma membrane and translocate into the culture medium (Fig. 4A).

A Detection of GD-NT in culture medium. HEK293 cells were transfected with the indicated constructs. 16 hours later, the culture medium and WCL were harvested and immunoblotted with the indicated antibodies. B Distribution pattern of GD-NT-WT or its mutants. HeLa cells were transfected with the indicated constructs. Representative confocal microscopy images showing distribution of ectopic Flag-GD-FL, GD-NT-WT, GD-NT-K237R, or GD-NT-K204/205R (green) co-stained with DAPI (blue). C, D Subcellular localization of GD-NT-WT or its mutants. HEK293 cells were transfected with the indicated constructs. 16 hours later, cells were harvested and subjected to sample preparation using a Thermo Fisher Subcellular Fractionation Kit. Membrane fractions (C), and other cellular fractions (D), were immunoblotted with the indicated antibodies. WCL was included as an input control. E Oligomerization of GD-NT. HEK293 cells were transfected with the indicated constructs. 16 hours later, cells were harvested and subjected to IB analysis with the indicated antibodies under reducing or non-reducing conditions. F Fractionation via SEC/gel filtration. HEK293 cells were transfected with the indicated constructs. 16 hours later, cells were harvested for fractionation with SEC/gel filtration. Eluent protein fractions were immunoblotted with the indicated antibodies. Input represents 10% of proteins used for SEC/gel filtration. Data are representative of at least two independent experiments.

To determine how the K63-linked polyubiquitination of Lys237 affects GD-NT cytolytic activity, we further investigated the subcellular localization and oligomerization of GD-NT-K237R using various strategies [26]. First, we transfected HeLa cells with GD-FL, GD-NT-WT, GD-NT-K237R, or GD-NT-K204/205R and then performed confocal immunofluorescent microscopy to visualize the localization of these constructs. Our result showed that GD-NT-K237R clustered near the plasma membrane, which was different from GD-NT’s even distribution along the plasma membrane (Fig. 4B).

To precisely define GD-NT-K237R subcellular localization, HEK293 cells were transfected with these constructs and then fractionated into five compartments (soluble cytoplasmic, membrane, soluble nuclear, chromatin-bound nuclear, and insoluble cytoskeletal content) for immunoblot analysis. These results showed that GD-NT-K237R was less abundant in the membrane and insoluble cytoskeletal fractions in comparison with GD-NT-WT (Fig. 4C, D). Next, to assess the oligomer formation of GD-NT-K237R, HEK293 cells were transfected with these constructs and subsequently subjected to immunoblot under non-reducing conditions. We found that GD-NT-K237R failed to form the oligomers (Fig. 4E). In addition, these transfected cells were fractionated through size exclusion chromatography (SEC)/gel filtration under native conditions and then subjected to immunoblot under reducing conditions. The result revealed that GD-NT-K237R exhibited a diffuse distribution pattern, whereas GD-NT-WT formed a single huge oligomer (≥440 kDa) (Fig. 4F). Collectively, these findings suggest that GD-NT loses pyroptotic activity when it is modified with K63-linked polyubiquitin chains at Lys237.

GD-NT-K237R fails to mediate pyroptosis in vitro and in vivo

To directly evaluate the influence of ubiquitination on GD-NT-mediated pyroptosis, HEK293 cells were transfected with GD-NT-WT, GD-NT-K237R, or GD-NT-K204/205R and then subjected to cell death/survival assessment. Phase-contrast imaging indicated that, unlike GD-NT-WT and GD-NT-K204/205R that exhibited cellular toxicity, the mutant GD-NT-K237R caused no obvious morphological changes (Fig. 5A). Consistently, subsequent LDH-based cell death assay and ATP-based cell viability assay suggested that GD-NT-K237R was incapable of mediating pyroptosis (Fig. 5B, C). To further investigate the pyroptotic activity of GD-NT-K237R, we employed a xenograft model in NSG mice using HeLa cells that have doxycycline (Dox)-inducible expression of GD-NT-WT, GD-NT-K237R or GD-NT-K204/205R. Dox was administered via intraperitoneal injection to induce the expression of GD-NT. Similar to in vitro findings, the xenografts expressing GD-NT-WT had a slower growth rate than those expressing the empty vector. In contrast, GD-NT-K237R did not efficiently suppress xenograft growth (Fig. 5D–F). These data suggest that the K63-linked polyubiquitination on Lys237 prevents GD-NT from mediating pyroptosis.

A HEK293 cells were transfected with Flag-GD-NT-WT or its mutants. Phase-contrast images were taken 24 hours after transfection (left panel). In addition, the cells were harvested for IB analysis 16 hours after transfection (right panel). B Similar to (A), except that the cells were harvested for LDH-based Cytotoxicity Assay 24 hours after transfection. C Similar to (A), except that the cells were harvested for ATP-based Cell Viability Assay 24 hours after transfection. D–F HeLa xenograft in NSG mice. HeLa cells expressing Dox-inducible GD-NT-WT or its mutants (0.5 × 10⁶ cells per mouse) were subcutaneously implanted into the right flank of NSG mice (n = 6 mice per group). Dox (50 mg/kg, i.p.) was administrated on day 6, 8, 10, 12, 14 and 16. Tumor growth was recorded every other day. Shown are tumor growth (D), tumor image (E), and tumor weight (F). In B, C, D, and F, error bars represent a variation range of duplicated experiments. In B, C, and F, differences among groups were analyzed by two-tailed Student’s t-test (means ± s.e.m). In D, the areas under the growth curves were compared by a two-tailed Student’s t-test (means ± s.e.m). Data are representative of at least two independent experiments.

UBA1 inhibitor PYR-41 suppresses GD-NT-mediated pyroptosis in vitro and in vivo

As K63-linked ubiquitination plays a positive role in GD-NT pyroptotic activity, we sought to evaluate the effects of ubiquitination-targeting small molecules on GD-NT-mediated pyroptosis. Due to the absence of selective molecules targeting TRAF1 or OTUB1, we instead evaluated the universal ubiquitination modulators, including UBA1 inhibitors PYR-41 and MLN7243, NAE inhibitor MLN4924, SAE inhibitor TAK-981 and the proteasome inhibitor Bortezomib.

First, to test the influence of PYR-41 on GD-NT ubiquitination status, HEK293 cells were transfected with His-Ub and Flag-GD-NT and then treated with PYR-41. Ni/NTA pulldown assay indicated that PYR-41 significantly reduced the ubiquitination of GD-NT (Fig. 6A).

A 293-tetO-His-Ub cells were used for ubiquitination detection. Dox (2 µg/mL) was added to induce the expression of His-Ub. 24 hours after Dox treatment, the cells were transfected with Flag-GD-NT and simultaneously treated with PYR-41 (50 µM). 24 hours after transfection, the cells were harvested for Ni-NTA Pulldown. The precipitants and WCL were immunoblotted with the indicated antibodies. B 293-tetO-GD-NT cells were treated with Dox (2 µg/mL) to induce GD-NT expression. Different concentrations of PYR-41 were added at the same time as Dox. 6 hours after Dox/PYR-41, and the cells were harvested for immunoblot with the indicated antibodies. C Similar to (B), except that phase-contrast images were taken 16 hours after Dox/PYR-41. The cells floating in the culture medium were removed, and the adherent cells (considered as living cells) were subjected to phase-contrast imaging. D 293-tetO-GD-NT cells were treated with Dox (2 µg/mL) to induce GD-NT expression. PYR-41 (50 µM) was added at the same time as Dox. 12 hours after Dox/PYR-41, the cells were harvested for flow cytometry and Annexin V/PI staining kit. E Similar to (D), except that the cells were harvested 12 hours post-Dox and subjected to LDH-based Cytotoxicity Assay. F Similar to (D), except that the cells were harvested 12 hours post-Dox and subjected to ATP-based Cell Viability Assay. G Phase-contrast images of iBMDM cells. The cells were treated with LPS (1 μg/mL) and (PYR-41 50 μM)) for 10 hours before stimulation with nigericin for 30 minutes. The cells floating in the culture medium were removed, and the adherent cells (considered as living cells) were subjected to phase-contrast imaging. H Murine septic model: LPS, with or without PYR-41, was administered according to the formula. I–J Cytokine release in septic mice. LPS (5 mg/kg), with or without PYR-41 (10 mg/kg), was intraperitoneally injected into C57BL/6 naïve mice (n = 6 mice per group). 8 hours after LPS, the sera were harvested for ELISA assay to detect IL-18 (I) and TNFα (J) levels. K Diagrammatic illustration: The ubiquitination of GD-NT mediated by TRAF1/2 and OTUB2 enables GD-NT to polymerize and form pores in the cell membrane. The UBA1 inhibitor PYR-41 can significantly block GD-NT-mediated cell apoptosis. In E, F, I, and J, differences among groups were analyzed by two-tailed Student’s t-test (means ± s.e.m). Error bars represent the variation range of duplicated experiments. Data are representative of at least two independent experiments.

Next, to evaluate the effects of PYR-41 on GD-NT-mediated pyroptosis, we established a pyroptosis model cell line, “293-tetO-GD-NT”, in which the transgene GD-NT is expressed in response to Dox. Immunoblot assay revealed a peak expression of GD-NT at 4^th hour post-Dox. Fluorescent imaging indicated that more than 95% of cells underwent pyroptosis at 24^th hour of the Dox treatment (Extended Fig. 2). We then used 293-tetO-GD-NT cells to assess the effects of PYR-41 on GD-NT-mediated pyroptosis via different methods, including Phase-contrast imaging, Flow cytometry, LDH-based cell death assay, and ATP-based cell viability assay. Notably, PYR-41 treatment significantly reduced pyroptosis in 293-tetO-GD-NT cells (Fig. 6B–F).

Sepsis is the leading cause of death in intensive care units (ICUs) across the globe [44]. It is characterized by sustained excessive inflammation and immune suppression as well as organ dysfunction. Failure of more than a hundred clinical trials for finding a possible cure is attributed to the complexity of mediators and pathways involved in sepsis. Accumulating evidence suggests that GD-NT-mediated pyroptosis of macrophages is the main pathological basis of sepsis [5, 45], and the potential therapeutic targets in pyroptosis may provide future direction for sepsis treatment. In this study, we investigated the effects of PYR-41 in LPS/Nigericin-induced pyroptosis of immortalized bone marrow-derived macrophages (iBMDM) and LPS-induced septic mice. Surprisingly, we found that PYR-41 reduced iBMDM pyroptosis as well as the release of IL-18 and TNFα in septic mice (Fig. 6G–J). Thus, we concluded that inhibition of GD-NT ubiquitination suppresses its mediated pyroptosis and reduces the severity of septic condition in mice (Fig. 6K).

On the other hand, in the context of cancer, GD-NT-mediated pyroptosis of tumor cells releases numerous immunogenic products and may enhance anti-tumor immunity [1, 4, 46, 47]. Pharmaceutical modulation of GD-NT ubiquitination in cancer requires further investigations.