Mul1 suppresses Nlrp3 inflammasome activation through ubiquitination and degradation of Asc

Activation of the Nlrp3 inflammasome consisting of three major components, Nlrp3, Asc, and pro-caspase-1, results in the activation of caspase-1 and subsequent proteolytic cleavage of pro-IL-1β and pro-IL-18. To avoid excessive inflammatory response, the Nlrp3 inflammasome has to be precisely controlled. In this study, we show that the mouse mitochondrial E3 ubiquitin protein ligase (Mul1) suppresses Nlrp3 inflammasome activation through ubiquitination and degradation of Asc. In J774A.1 cells, Mul1 overexpression attenuated Nlrp3 activation, whereas Mul1 knockdown augmented Nlrp3 activation in terms of IL-1β secretion and cleavage of pro-caspase-1 and pro-IL-1β. Mul1 interacted with Asc, and ubiquitinated it at K21, K22, K26, and K55 residues, in a K48-linked manner, leading to proteasomal degradation. Convincingly, Mul1-mediated suppression of Nlrp3 activation was inhibited by K21R-, K22R-, K26R-, K52R-Asc mutants in RAW264.7 cells, when compared with the wild-type Asc. Furthermore, Aim2 inflammasome activation was also inhibited by Mul1 in the wild-type Asc-, but not in mutant Asc-expressing RAW264.7 cells. Taken together, these data suggest that Mul1 suppresses Nlrp3 inflammasome activation, through Asc ubiquitination and degradation.


Introduction
Nlrp3 (NOD-, LRR-, and pyrin domain-containing protein 3) is an intracellular sensor that detects a broad range of microbial pathogens, endogenous danger signals, and environmental irritants [1][2][3][4]. Typically, priming by "Signal 1" involves transcriptional and post-transcriptional regulation of related components, such as IL-1β and Nlrp3, through pattern recognition receptors and NFκB signaling. In the following step, recognition of Nlrp3 activator, "Signal 2," induces the formation of the Nlrp3 inflammasome complex that consists of three major components: Nlrp3, Asc (apoptosis-associated speck-like protein containing a Card; also known as Pycard, Pyd And Card domain containing), and pro-caspase-1, acting as sensor, adaptor, and effector molecules, respectively. Nlrp3, after sensing stimuli, engages multiple molecules of Asc, which, through protein-protein interaction, form oligomers and finally "bird nest"-like supramolecular protein complexes known as Asc specks [5][6][7][8]. The supramolecular protein complexes recruit and activate procaspase-1, culminating in the production of active IL-1β and IL-18 through proteolytic cleavage by activated caspase-1.
Furthermore, cAMP induced through the dopamine receptor D1, binds to Nlrp3 and promotes its K48-linked ubiquitination and phagosomal degradation through Membrane Associated Ring-CH-Type Finger 7 (MARCH7) [20]. IL-1β is also the target for ubiquitination and proteasomal degradation by the endogenous E3 ligase E6AP in human papilloma virus infection [21]. In contrast to Nlrp3 and IL-1β, ubiquitinationmediated negative regulation of Asc remains largely unknown, whereas positive regulation of Asc through linear ubiquitination by Linear Ubiquitin Chain Assembly Complex LUBAC [22] and K63-linked ubiquitination by TNF Receptor Associated Factor 3 (TRAF3) [23] have been reported.
Mitochondrial E3 ubiquitin protein ligase (Mul1 [24], also known as MAPL [25], MULAN [26], or GIDE [27]), is a multifunctional protein that is embedded in the outer mitochondrial membrane with its C-terminal RING finger domain facing the cytoplasm.
In addition to involvement in mitochondrial dynamics and mitophagy, Mul1 controls innate immune response. Mul1 induces SUMOylation and activation of Retinoic Acid-Inducible Gene I (RIG-1), an essential innate immune receptor that binds doublestranded RNA in the cytosol, leading to initiation of antiviral signaling [31]. However, whether and how Mul1 regulates innate immune response in the context of the Nlrp3 inflammasome remains largely unknown.
Here we report a novel regulatory mechanism for the Nlrp3 inflammasome, in which mouse Mul1 suppresses Nlrp3 inflammasome activation through ubiquitination and proteasomal degradation of Asc. Through Mul1 overexpression and knockdown experiments, we demonstrate that Mul1 suppressed Nlrp3 inflammasome activation.
Molecularly, Mul1 bound Asc, and mediated K48-linked ubiquitination at multiple sites and subsequent proteasomal degradation. Furthermore, Asc mutants at Mul1mediated ubiquitination sites inhibited Mul1-mediated suppression of Nlrp3 inflammasome activation. Finally, activation of Aim2 inflammasome, which also uses Asc as an adaptor, was inhibited by Mul1 in J774A.1 cells, while Mul1 did not affect Aim2 inflammasome activation in the mutant Asc-expressing RAW264.7 cells. Taken together, these novel data support the hypothesis that Mul1 may act as a suppressor of the Nlrp3 inflammasome activation through ubiquitination and degradation of Asc.

Mul1 negatively regulates Nlrp3 inflammasome activation
First, we addressed whether Mul1 might affect Nlrp3 inflammasome activation in mouse macrophage J774A.1 cells by overexpressing and knocking down Mul1. When J774A.1 cells were infected with the wild-type Mul1-lentivirus, and primed and stimulated with LPS and ATP, respectively, overexpressed Mul1 dramatically suppressed the secretion of IL-1β and cleavage of pro-IL-1β and pro-caspase-1, in a dose-dependent manner, compared with the control lentivirus-infected group (Fig. 1A).
In contrast, Mul1 overexpression did not significantly change TNF-α secretion from J774A.1 cells treated with LPS or LPS+ATP, compared with the control lentivirusinfected group (Appendix Fig. S1A). Moreover, infection of Mul1-lentivirus by itself did not induce secretion of IL-1β and cleavage of pro-IL-1β and pro-caspase-1, in J774A.1 cells treated with LPS alone. Similar to the inhibitory effects on Nlrp3 inflammasome activation induced by LPS and ATP, Mul1 overexpression also significantly suppressed inflammasome activity in LPS-primed J774A.1 cells, stimulated by other Nlrp3 Signal 2 stimuli including nigericin (potassium ionophore), and alum crystal without altering TNF-α secretion ( Fig. EV1A and EV2A).
To confirm the inhibitory action of Mul1 on Nlrp3 inflammasome activation, we assessed the effects of Mul1 knockdown using Mul1 shRNA-lentivirus (shMul1).
Compared to the control lentivirus, shMu1 significantly augmented IL-1β secretion and cleavage of pro-IL-1β and procaspase-1, in a dose-dependent manner in LPS-primed and ATP-stimulated J774A.1 cells, without any significant changes in TNF-α secretion As Mul1 was reported to induce ubiquitination of several proteins, such as p53 [32] and Akt [33], we investigated whether the inhibitory effects of Mul1 on Nlrp3 inflammasome activation might be mediated through the ubiquitin E3 ligase activity by using a RING domain-mutant Mul1 (C302/305S-Mul1, Mt-Mul1, defective in E3 ligase activity) [33]. As shown in Fig. 1C, Mt-Mul1 did not attenuate secretion of IL-1β and cleavage of pro-IL-1β and pro-caspase-1 in LPS-primed-and ATP-stimulated J774A.1 cells. Instead, Mt-Mul1 slightly but significantly increased IL-1β secretion in a dosedependent manner without any significant change in TNF-α secretion ( ligase activity of its RING domain, thus prompting us to search for a potential substrate within the Nlrp3 inflammasome components.

Mul1 reduces Asc levels by K48-linked poly-ubiquitination and subsequent proteasomal degradation
To find a potential substrate for Mul1, we assessed whether Mul1 alters the levels of major Nlrp3 inflammasome components Nlrp3, Asc, and pro-caspase-1 in J774A.1 cells. Interestingly, WT-Mul1 significantly reduced Asc levels in a dose-and timedependent manner, without noticeable changes in levels of Nlrp3 and pro-caspase-1 ( Fig. 2A). In contrast to WT-Mul1, Mt-Mul1 did not reduce Asc levels (Fig. 2B). As Mul1 was reported to be involved in protein ubiquitination/proteasomal degradation as well as mitophagy, we investigated the possible pathway that might contribute to the As these data implied Mul1 facilitated Asc degradation via the proteasome, we addressed Mul1-mediated Asc ubiquitination. Using an in vivo ubiquitination assay in HEK293 cells, we found that Asc was poly-ubiquitinated by WT-Mul1 in a complete reaction mixture (Fig. 3A), while Mt-Mul1 was unable to do so (Fig. 3B). In the experiment to determine ubiquitin linkage mode, Asc poly-ubiquitination was seen only with the wild-type and K48-ubiquitin (all Lys residues replaced with Arg except for K48) ( Fig. 3C). Asc poly-ubiquitination was not seen with K63-ubiquitin (all Lys residues replaced with Arg except for K63), K48R-ubiquitin (K48 replaced with Arg), and KOubiquitin (all Lys residues replaced with Arg). Additionally, WT-Mul1 but not Mt-Mul1 recombinant protein induced poly-ubiquitination of Asc in vitro (Fig. 3D).
Next, we determined the site of Mul1-mediated poly-ubiquitination in Asc. For this aim, we generated ten KR single mutant Asc plasmids, in which each Lys residue (K21, 22,24,26,55,109,139,158,161, and 174) was replaced with Arg. When the wildtype (WT-) and mutant Asc plasmids were transfected into RAW264.7 cells that do not to express Asc [34], all Asc mutants were expressed to a similar extent (Fig. 3E, upper).
To observe the interaction between Mul1 and Asc, both Mul1-FLAG and Asc-Myc were expressed in HEK293 cells and their interaction was checked by a reciprocal coimmunoprecipitation assay (Fig. EV3). After immunoprecipitation with anti-Myc and anti-FLAG antibodies, Mul1 and Asc were detected by western blotting, respectively.
In addition, the interaction of endogenous Mul1 with endogenous Asc in J774A.1 cells was detected by co-immunoprecipitation using an anti-Asc antibody (Fig. 3G, left).

Mul1 suppresses Aim2 inflammasome activation
As Asc also acts as an adaptor for other types of inflammasome complexes including the Aim2 inflammasome [35], we addressed whether Mul1 might play a suppressive role in Aim2 inflammasome activation by Asc ubiquitination. Similar to the suppressive effects of Mul1 upon Nlrp3 inflammasome activation seen in Fig Asc is composed of an N-terminal Pyd, an unstructured linker region, and a C-terminal Card [1-4, 6, 7]. Once inflammasome sensors, that typically contain a Pyd, are activated by recognizing inflammasome activators, they act as a seed for the recruitment of Asc through Pyd:Pyd homotypic interactions between inflammasome sensors and Asc, leading to the formation of Asc oligomers. Asc oligomers, in turn, nucleate filamentous polymers of Asc through inter-strand Card:Card interactions in Asc, forming Asc specks that are micrometer-sized supramolecular protein complexes resembling a bird's nest. Asc specks are the sites of pro-caspase-1 recruitment and activation through Card:Card interactions between Asc and pro-caspase-1, and proteolytically activate pro-IL-1β, pro-IL-18, and pore-forming gasdermin D that induces pyroptosis, a type of pro-inflammatory cell death. In addition to the intracellular role of Asc specks, these complexes are released into the extracellular space, and amplify the inflammatory response by activating extracellular pro-caspase-1 and pro-IL-1β and acting as a seed for new Asc speck formation after being engulfed by bystander macrophages [36,37].
In this context, Asc plays a key role in inflammasome activation, and therefore may be a critical target for precise and appropriate control of inflammasome activation. Posttranscriptional modifications including phosphorylation and ubiquitination have been reported to regulate Asc. Asc is phosphorylated during Nlrp3 and Aim2 inflammasome activation by JNK and Syk, and Syk-mediated phosphorylation at Tyr144 residue in the Card domain of mouse Asc is critical for Asc speck formation and caspase-1 activation [38]. IKKα negatively regulates Asc by restraining Asc in the nuclei, and Signal 2 of Nlrp3 activation recruits protein phosphatase 2 and inhibits IKKα, thus allowing Asc to participate in the Nlrp3 inflammasome activation [39]. In addition to phosphorylation, ubiquitination of Asc is a known regulatory mechanism. LUBAC, consisting of HOIL-1L, HOIP, and SHARPIN, induces linear ubiquitination of Asc that enables Nlrp3 inflammasome activation [22]. Additionally, mitochondrial antiviral signaling protein (MAVS) was reported to stabilize Asc by recruiting TRAF3 that promoted K63-linked ubiquitination of Asc, and enhance Asc specks and IL-1β secretion upon infection with the RNA virus vesicular stomatitis virus [23]. However, the MAVS-TRAF3-Asc axis did not work with other Nlrp3 activators including monosodium urate and calcium pyrophosphate dehydrate. In contrast, Shi et al.
reported that Asc aggregates are poly-ubiquitinated in a K63-linked manner and suggested that poly-ubiquitinated inflammasomes can be trafficked by p62 into the autophagic pathway, thus limiting Aim2 inflammasome activation [40]. However, they did not characterize the ubiquitin E3 ligase involved, the ubiquitination sites in Asc, and the causal relationship between Asc ubiquitination and inflammasome activation.
To the best of our knowledge, K48-linked ubiquitination and proteasomal degradation-mediated negative regulation of Asc has not been reported yet.
In this study, we demonstrated that Mul1 negatively regulated both Nlrp3 and Aim2 inflammasome activation. We envisage that additional kinds of inflammasomes including NLRC4 and pyrin inflammasomes may be regulated by Mul1, because they use Asc as an adaptor for the inflammasome complex formation as well [41,42]. In the case of NLRP1b inflammasome activation, while Asc is dispensable for IL-1β secretion and pyroptosis, it is required for Asc speck formation and subsequent activation of caspase-1, which augments IL-1β cleavage [43]. Hence, NLRP1b Mul1 is a multifunctional protein ligase that conjugates ubiquitin or SUMO to substrate proteins, playing distinct biological roles. Mul1 retards cell growth by K48-linked polyubiquitination and degradation of Akt [33], while it attenuates apoptosis by K48-linked poly-ubiquitination and degradation of exonuclear p53 and subsequent suppression of p53-dependent transcription [32]. Also, Mfn, which causes mitochondrial hyperfusion, was reported to be a substrate for Mul1 by two research groups. Even though both groups reported that Mfn is poly-ubiquitinated by Mul1, they proposed different However, this data is contradictory to the previous report by Jenkins et al. that suggested that Mul1 negatively regulated the antiviral response to Sendai virus [31].
Even though the authors proposed Mul1-mediated SUMOylation of RIG-1 as a mechanism for the inhibitory role of Mul1, further study seems to be needed to support that claim.

Recently, Barry et al. clearly demonstrated that Nlrp3 is SUMOylated by Mul1 and
Signal 2-dependent de-SUMOylation of Nlrp3 by the SUMO-specific proteases Senp2 and Senp7 promote Nlrp3 activation [48]. In contrast to the inhibitory action of Mul1 on Nlrp3 inflammasome activation, the authors showed that Mul1 did not suppress Aim2-  [49], and Mul1 expression was inhibited by transfection of peroxisome proliferator-activated receptor γ-coactivator 1α (PGC-1α), which not only suppressed FOXO1 and FOXO3, but also attenuated Mfn2 ubiquitination and degradation [50]. As FOXO3 is directly and indirectly activated by AMP-activated protein kinase (AMPK), a master regulator of metabolism and mitochondrial homeostasis [51], the AMPK/FOXO pathway may be a good candidate for Mul1 induction. In support of this idea, the anti-diabetic drug metformin that activates AMPK [52] and reduced Nlrp3 inflammasome activation [13] induced Mul1 expression [53].
Damaged mitochondria play a key role in inflammasome activation not only by generating harmful Nlrp3 activators including oxidized mitochondrial DNA [54] and reactive oxygen species [44], but also by acting as a platform for the Nlrp3 inflammasome assembly through binding of Nlrp3, Asc, and pro-caspase-1 to mitochondrial cardiolipin, which is externalized to the outer mitochondrial membrane at priming [55,56]. Hence, mitophagy is considered to be an important process for limiting the Nlrp3 inflammasome activation. Drp1 and Mfn control mitochondrial dynamics by inducing mitochondrial fission and fusion, respectively, are suggested to play a role in the Nlrp3 inflammasome activation. RNA virus infection promotes activation of Drp1 by the RIP1-RIP3-dependent cascade resulting in mitochondrial damage and Nlrp3 inflammasome activation [57], whereas Park et al. reported that Drp1 knockdown led to increased Nlrp3 inflammasome activation by ATP and nigericin [58]. As Drp1 is a substrate for Mul1-mediated SUMOylation, it needs to be assessed whether Mul1 may regulate inflammasome activation through Drp1 SUMOylation. In addition to SUMOylation, Ichinohe et al., found that Mfn2 is required for Nlrp3 inflammasome activation, proposing a model wherein Nlrp3 and MAVS associate with Mfn2 and recruit Asc and pro-caspase-1 to the outer mitochondrial membrane [59]. Cleavage of pro-IL-1β and pro-caspase-1 was assessed by western blot of 50 μg of protein extracts of cell lysates, based on our previous report [62]. Images of cleaved caspase-1 (p20) and IL-1β (p17) in the western blot were quantified using a software (ImageJ, https://imagej.net).

Plasmids and lentiviruses
Plasmids of the wild-type mouse Mul1 (pMul1-FLAG) and Asc (pAsc-Myc) were

In vivo and in vitro Asc ubiquitination
Both in vivo and in vitro Mul1-mediated Asc ubiquitination assays were performed, based on our previous report [62].