USP3 plays a critical role in the induction of innate immune tolerance

Abstract

Microbial products, such as lipopolysaccharide (LPS), can elicit efficient innate immune responses against invading pathogens. However, priming with LPS can induce a form of innate immune memory, termed innate immune “tolerance”, which blunts subsequent NF‐κB signaling. Although epigenetic and transcriptional reprogramming has been shown to play a role in innate immune memory, the involvement of post‐translational regulation remains unclear. Here, we report that ubiquitin‐specific protease 3 (USP3) participates in establishing “tolerance” innate immune memory through non‐transcriptional feedback. Upon NF‐κB signaling activation, USP3 is stabilized and exits the nucleus. The cytoplasmic USP3 specifically removes the K63‐linked polyubiquitin chains on MyD88, thus negatively regulating TLR/IL1β‐induced inflammatory signaling activation. Importantly, cytoplasmic translocation is a prerequisite step for USP3 to deubiquitinate MyD88. Additionally, LPS priming could induce cytoplasmic retention and faster and stronger cytoplasmic translocation of USP3, enabling it to quickly shut down NF‐κB signaling upon the second LPS challenge. This work identifies a previously unrecognized post‐translational feedback loop in the MyD88–USP3 axis, which is critical for inducing normal “tolerance” innate immune memory.

USP3 controls innate immune signaling and inflammatory responses by regulating the NF‐κB pathway by deubiquitinating MyD88. LPS‐priming‐induced cytoplasmic translocation and retention of USP3 are essential for establishing normal “tolerance” innate immune memory.

Introduction

The innate immune system is the first line of defense against invading pathogens, which senses and responds to pathogen‐associated molecular patterns (PAMPs) through a variety of pattern recognition receptors (PRRs) (Janeway & Medzhitov, 2002; Takeuchi & Akira, 2010; Duan et al, 2022). Upon recognition of their specific ligands, PRRs initiate various downstream signaling cascades, including NF‐κB, MAPK, and type I interferon (IFN) pathways, which ultimately result in the production of cytokines and chemokines that are essential for effective antimicrobial responses (Takeuchi & Akira, 2010; Zhang et al, 2017a; Xing et al, 2022). Activation of NF‐κB signaling is crucial to induce an immune response and pathogen clearance (Hayden & Ghosh, 2008; Duan et al, 2022). Despite the importance of the NF‐κB signaling pathway in inflammation, immune signaling, cancer, and diverse cellular responses, the precise regulatory mechanism of the NF‐κB signaling pathway is still not fully understood (Ghosh & Hayden, 2012; Newton & Dixit, 2012; Liu et al, 2017; Zhang et al, 2017b; Taniguchi & Karin, 2018; Duan et al, 2022).

MyD88 is a common adaptor protein of Toll‐like receptor (TLR)‐ and interleukin‐1 receptor (IL‐1R)‐induced NF‐κB signaling (Muzio et al, 1997; Takeuchi & Akira, 2010; Brubaker et al, 2015; Duan et al, 2022), which can recruit IRAK4 and IRAK2/1 to form a MyD88/IRAKs complex (also known as myddosome) through their death domains (DD) and then activate the E3 ligase TRAF6 and downstream kinase cascade (Muzio et al, 1997; Lin et al, 2010; Gay et al, 2011; Duan et al, 2022). In turn, the TAK1 and IκB kinase (IKK) complex is activated, triggering the phosphorylation and degradation of IκBα, which releases the transcription factor p65, which enters the nucleus and initiates the transcription of downstream genes (Hayden & Ghosh, 2008; Duan et al, 2022). TLR‐induced activation of the NF‐κB signaling pathway is crucial for pathogen clearance. However, the aberrant immune response is also harmful and may even cause the death of the host (Taniguchi & Karin, 2018; Xing et al, 2021, 2022; Duan et al, 2022). Thus, tight regulation of NF‐κB signaling is critical for maintaining the homeostasis of both innate and adaptive immunity (Feng et al, 2017; Duan et al, 2022).

While immune memory has traditionally been associated with the acquired immune system, recent studies have revealed that the innate immune system can also exhibit a form of memory, leading to hyper‐ or hypo‐responsiveness upon subsequent stimulation. This phenomenon is referred to as “innate immune memory” (Netea et al, 2020). Due to the dual features of innate immune memory, it can be further defined as “training” (trained immunity) or “tolerance” innate immune memory (innate immune tolerance) (Divangahi et al, 2021; Zubair et al, 2021). Endotoxin tolerance is considered a type of “tolerance” innate immune memory (Seeley & Ghosh, 2017; Seeley et al, 2018; Vergadi et al, 2018; Weisheit et al, 2020), which is defined as the reduction of a cell's capacity to respond to gram‐negative bacterial lipopolysaccharide (LPS) after initial exposure to the same stimulus, also characterized as innate immune hypo‐responsiveness or “immunoparalysis” (Beeson, 1946; Biswas & Lopez‐Collazo, 2009; Seeley et al, 2018; Vergadi et al, 2018). The downregulation of inflammatory mediators, such as tumor necrosis factor α (TNFα), interleukin‐1β (IL‐1β), and C‐X‐C motif chemokine 10 (CXCL10), is the main characteristic of endotoxin tolerance (Liu et al, 2019). Importantly, innate immune tolerance is a crucial homeostatic mechanism that prevents the excessive activation of innate immune responses upon sustained TLR stimulation and protects the organism against a lethal LPS challenge during severe pathogen infection (Cavaillon & Adib‐Conquy, 2006; Vergadi et al, 2018). However, reduced inflammatory cytokine output (Ertel et al, 1995), increased secondary infection (Otto et al, 2011), and an increased risk of organ failure and mortality (Boomer et al, 2011) were also observed in some sepsis patients with immunoparalysis. Although endotoxin tolerance is involved in multiple cellular signal pathways, receptor alterations, and biological molecules (Vergadi et al, 2018; Liu et al, 2019), the exact mechanism underlying tolerance, immunoparalysis, and other forms of innate immune memory is not fully understood (Seeley et al, 2018; Liu et al, 2019).

Ubiquitin is a highly conserved protein of 76 amino acids, which is ubiquitously expressed in all eukaryotes. Ubiquitin could be covalently attached to the lysine of other proteins through an isopeptide bond, a post‐translational modification termed ubiquitination (Liu & Chen, 2011). One ubiquitin can be further conjugated to the seven lysines (K6, K11, K27, K29, K33, K48, and K63) or N‐terminal methionine on the other ubiquitin, thus forming a polyubiquitin chain with different types of linkages (Zinngrebe et al, 2014). The different types of polyubiquitin chains could determine the different modes of substrate regulation (Chen & Chen, 2013). Since the critical role of K63‐linked ubiquitination in NF‐κB signaling activation has been demonstrated (Liu & Chen, 2011; Hu & Sun, 2016; French et al, 2021), it is crucial to understand how these K63‐Ub chains are regulated in response to pathogenic stimuli. In particular, it remains unknown whether the regulation of these K63‐Ub chains is also involved in innate immune tolerance.

The ubiquitination status of proteins is counterbalanced by 99 currently identified deubiquitinases (DUBs) (Clague et al, 2019), including USP3, which was first characterized in 1999 (Sloper‐Mould et al, 1999). Because of its nuclear localization, USP3 was initially identified as a chromatin modifier that helps to maintain genomic integrity and regulate ubiquitin‐mediated DNA damage response by deubiquitinating H2A and γH2AX (Nicassio et al, 2007; Lancini et al, 2014; Sharma et al, 2014; Cheng & Shieh, 2018). Recently, its deubiquitination function in promoting cancer cell proliferation and metastasis has been reported in relation to multiple cancers (Fu et al, 2017; Wang et al, 2017; Wu et al, 2019, 2021; Das et al, 2020; Liao et al, 2020; Tu et al, 2020; Alkhizzi et al, 2021; Nagy & Seneviratne, 2021; Shi et al, 2021). However, the involvement of USP3 in innate immune signaling has not been completely elaborated. Our previous studies showed that USP3 can act as a negative regulator of the type I IFN signaling pathway (Cui et al, 2014), and it may also play a regulatory role in IL‐1β‐mediated NF‐κB activation, as recently reported in chondrocytes (Zhou et al, 2019). Here, we identified ubiquitin‐specific protease 3 (USP3) as a critical participant in establishing innate immune tolerance in immune cells. Upon NF‐κB activation, we found that USP3 was stabilized and exited the nucleus, functioning as a key negative regulator in TLRs/IL‐1R‐induced NF‐κB signaling by specifically cleaving K63‐linked polyubiquitin chains on MyD88. LPS priming resulted in the cytoplasmic retention of USP3, which enabled it to remove the K63‐linked polyubiquitin chains more efficiently on MyD88 and attenuate subsequent NF‐κB signaling upon a second LPS challenge. Therefore, the spatial regulation of USP3 is a critical factor in the formation of innate immune tolerance. Collectively, our works reveal a feedback system mediated by the MyD88‐USP3 axis, which is essential for the establishment of innate immune tolerance and the maintenance of immune homeostasis.

Results

USP3 negatively regulates TLR‐induced NF‐κB signaling

To identify potential DUBs that have a negative effect on the NF‐κB signaling pathway, we screened 41 candidate genes encoding USPs and identified USP3 as one of the potent negative regulators of the NF‐κB signaling pathway (Fig 1A). Using USP3‐transfected cells treated with IL‐1β (a ligand for IL‐1R), TNFα (a ligand for TNFR), or 293T‐TLR4 cells treated with LPS (a ligand for TLR4), we showed that USP3 potently inhibited the NF‐κB activation induced by LPS and IL‐1β (Fig 1B). We next generated an USP3‐overexpressing human monocyte cell line (THP‐1^USP3) (Fig EV1A). As shown in Fig 1C, the phosphorylation of IKK (p‐IKK) and JNK (p‐JNK) was inhibited at least threefold compared to the control, and the phosphorylation of IκBα (p‐IκBα), ERK (p‐ERK), and p38 (p‐p38) was also markedly inhibited (Fig EV1B). On the contrary, we did not observe any appreciable differences in the phosphorylation of IKK (p‐IKK) in the presence of TNFα (Fig EV1C ). Consistent with this result, the transcription levels of IL6, IL‐1β, and TNFα in THP‐1^USP3 cells were significantly decreased compared to that in control cells after LPS treatment (Fig 1D), while there was no difference in the expression of those genes in THP‐1^USP3 and control cells after TNFα treatment (Fig EV1D). To determine whether USP3 could inhibit different types of TLRs‐mediated NF‐κB activation, we tested the secretion of NF‐κB‐responsive cytokines after stimulation with different TLR ligands: Pam₃CSK₄ (a ligand for TLR1/2), LPS, and CL‐097 (a ligand for TLR7/8). The cytokine levels of IL‐6 and TNFα decreased in THP‐1^USP3 cells compared to control cells (Fig 1E). Remarkably, we found that a C168S mutant, which is enzymatically inactive, did not reduce LPS‐mediated NF‐κB signaling (Fig EV1E) or the LPS‐induced expression of proinflammatory genes (Fig EV1F) and key NF‐κB downstream genes (Fig 1F). These results suggest that ectopic expression of USP3 specifically inhibits TLRs‐ and IL‐1β‐ but not TNFα‐induced NF‐κB signaling activation through its enzyme activity.

Figure 1. USP3 negatively regulates TLR‐induced NF‐κB signaling^† .

1. Screen of the regulators of NF‐κB signaling in USP proteins. HEK293T cells were transfected with an individual plasmid encoding one of 41 DUBs along with MyD88 and the NF‐κB promoter‐luciferase (NF‐κB‐luc) reporter plasmid. Cell extracts were used for measuring NF‐κB activity in a reporter assay.
2. Luciferase activity in 293T‐TLR4 cells or 293T cells transfected with a plasmid encoding an NF‐κB‐luc reporter, together with empty vector (no wedge) or an expression vector for USP3 (50 and 100 ng; wedge), followed by treatment with LPS (10 μg/ml), IL‐1β (20 ng/ml), or TNFα (20 ng/ml).
3. Empty vector (EV) or USP3‐transduced THP‐1 cells were treated with LPS (200 ng/ml). The cell extracts were harvested at indicated time points and used for the immunoblotting of various kinases and signaling proteins.
4. Quantitative real‐time PCR analysis of NF‐κB target genes in EV (blue)‐ or USP3 (red)‐transduced THP‐1 cells treated with LPS (200 ng/ml) for the indicated lengths of time.
5. Productions of TNFα and IL‐6 in the culture medium of EV‐ or USP3‐transduced THP‐1 cells were measured after treatment with different TLR ligands (Pam₃CSK, LPS, and CL‐097) for 24 h.
6. Quantitative real‐time PCR analysis of NF‐κB target genes in EV (blue)‐, USP3 (red)‐, or USP3 (C168S) (green)‐transduced THP‐1 cells treated with LPS (200 ng/ml) for the indicated lengths of time.

Data information: Data in (A–F) are representative of multiple replicates. Data in (A, B, D, E, and F) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test (B, D, and E) or two‐way ANOVA followed by the Sidak post‐test (F). *P < 0.05, **P < 0.01, ***P < 0.001; ns: not significant, N.D.: not detected.

Source data are available online for this figure.

^†Correction added on 28 November 2023, after first online publication: Figure 1D has been corrected. Similarly, Figures 2B, 3E and 6H have been corrected.

Figure EV1. The inhibitory function of USP3 in NF‐κB signaling depends on its enzyme activity.

1. Immunoblot or mRNA analysis of the USP3 in EV‐ or USP3‐transduced THP‐1 cells.
2. Density scanning analysis of Fig 1C.
3. EV‐ or USP3‐transduced THP‐1 cells were treated with TNFα. The cell extracts were harvested at the indicated time points and used for western blotting of p‐IKKα/β and IKKβ. The quantitative comparison involved density scanning of the blots.
4. Real‐time PCR analysis of TNFα and IL‐1β mRNA in EV‐ or USP3‐transduced THP‐1 cells treated with TNFα for 2 h.
5. EV‐, USP3‐, or USP3 (C168S)‐transduced THP‐1 cells were treated with LPS (200 ng/ml). The cell extracts were harvested at the indicated time points and used for immunoblotting various kinases and signaling proteins.
6. Quantitative real‐time PCR analysis of proinflammatory genes in EV (blue)‐, USP3 (red)‐, or USP3 (C168S) (green)‐transduced THP‐1 cells treated with LPS (200 ng/ml) for the indicated lengths of time.

Data information: Data in (A, C, D, E, and F) are representative of multiple replicates. Data in (A, D, and F) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test (A and D) or two‐way ANOVA followed by the Sidak post‐test (F). *P < 0.05, **P < 0.01, and ***P < 0.001; ns: not significant.

The deficiency of USP3 enhances TLR‐mediated proinflammatory responses

To determine the function of endogenous USP3 under physiological conditions, we generated USP3 knockout (USP3 ^−/−) THP‐1 cells using CRISPR/Cas9 technology (Fig EV2A and B) and USP3 knockdown (USP3 ^KD) THP‐1 cells using USP3 shRNA‐2 (Fig EV2C and D). We found that USP3 deficiency resulted in enhanced and prolonged phosphorylation of IKK, IκBα, and MAPK (p38, JNK, ERK) in USP3 ^−/− cells treated with LPS compared with LPS‐treated wild‐type (WT) cells (Fig 2A). Similar results were obtained in THP‐1 USP3 ^KD cells compared with reference THP‐1 (THP‐1^Ref) cells (Fig EV2E). In contrast, THP‐1^Ref and THP‐1 USP3 ^KD exhibited comparable levels of IKK phosphorylation when stimulated with TNFα (Fig EV2F). We next sought to determine whether the enhanced IKK, IκBα, and MAPK phosphorylation caused by USP3 deficiency promotes NF‐κB‐dependent gene expression. Consistent with our observations (Fig 2A), USP3 deficiency resulted in a drastic elevation of mRNA proinflammatory cytokines and other NF‐κB target genes (Fig 2B). Furthermore, the expression of the NF‐κB target genes was also substantially increased in THP‐1 USP3 ^KD with LPS treatment (Fig EV2G) but not with TNFα treatment (Fig EV2H). To further demonstrate the effect of USP3 deficiency on the secretion of proinflammatory cytokines, we treated the different THP‐1 cells with multiple TLR ligands and observed markedly higher secretion levels of IL‐6 and TNFα protein in USP3 ^−/− and USP3 ^KD cells (Figs 2C and EV2I). To exclude the possibility of clone‐specific artifacts contributing to the observed augmentation of NF‐κB signaling activation in USP3 ^−/− cells, we transduced the polyclonal THP‐1 cells with control (Ctrl) or USP3‐sgRNA (USP3 ^sgRNA), treated the cells with LPS, and obtained consistent results (Appendix Fig S1A and B). To confirm the function of USP3 in vivo, we generated Ctrl or Usp3 knockout chimeric mice (Usp3 ^cKO mice) through bone marrow transfer (Fig 2D and E). We intraperitoneally injected Ctrl or Usp3 ^cKO mice with a lethal dose of LPS. Importantly, Usp3 ^cKO mice were significantly less resistant to a lethal dose of LPS in terms of overall survival compared with the control mice (Fig 2F). Collectively, these results suggest that USP3 is a potent negative regulator of the TLR‐induced NF‐κB signaling pathway.

Figure EV2. Knockdown of USP3 enhances TLRs‐induced NF‐κB signaling.

• A, B
  Sequence alignment analysis (A) and immunoblot analysis (B) of WT and USP3 ^−/− THP‐1 cells. β‐actin was used as a loading control.
• C
  Immunoblot analysis of the knockdown efficiency of endogenous USP3 in HEK293T.
• D
  Immunoblot or mRNA analysis of the knockdown efficiency of endogenous USP3 in the THP‐1 stable cell line.
• E
  Reference or USP3 knockdown THP‐1 cells were treated with LPS. The cell extracts were harvested at the indicated time points and used for the western blot analysis of various kinases and signaling proteins.
• F
  Reference or USP3 knockdown THP‐1 cells were treated with TNFα. The cell extracts were harvested at the indicated time points and used for the western blot analysis of p‐IKKα/β and IKKβ. The quantitative comparison involved density scanning of the blots.
• G
  qPCR analysis of NF‐κB target genes (IL6, IL‐1β, TNFα, CCL‐20, ICAM‐1, Bcl‐xL, MCP‐1, and MnSOD) in reference or USP3 knockdown THP‐1 cells treated with LPS at the indicated time points.
• H
  qPCR analysis of TNFα and IL‐1β mRNA in reference or USP3 knockdown THP‐1 cells treated with TNFα for 2 h.
• I
  Productions of cytokine TNFα and IL‐6 in the culture medium of reference or USP3 knockdown THP‐1 cells were measured after treatment with different TLR ligands for 24 h.

Data information: Data in (B–I) are representative of multiple replicates. Data in (D, G–I) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. *P < 0.5, **P < 0.01, and ***P < 0.001; ns: not significant, N.D.: Not detected.

Figure 2. USP3 deficiency significantly enhances TLR‐induced NF‐κB activation and inflammatory responses.

1. Wild‐type (WT) or USP3 ^−/− THP‐1 cells were treated with LPS (200 ng/ml). The cell extracts were harvested at the indicated time points and used for the immunoblotting of various kinases and signaling proteins.
2. Quantitative real‐time PCR analysis of NF‐κB target genes in WT (blue) or USP3 ^−/− (red) THP‐1 cells treated with LPS (200 ng/ml) for the indicated lengths of time.
3. Productions of TNFα and IL‐6 in the culture medium of WT or USP3 ^−/− THP‐1 cells were measured after treatment with different TLR ligands (Pam₃CSK, LPS, and CL‐097) for 24 h.
4. Schematic overview of the generation of Usp3 KO chimeric mice.
5. Immunoblotting or mRNA analysis of the KO efficiency of bone marrow from chimeric mice with an antibody against USP3 or Usp3 KO site‐specific primer.
6. The survival (Kaplan–Meier curve) of mice (n = 6 per group) injected intraperitoneally with LPS (20 mg/kg) was assessed every hour for 3 days.

Data information: Data in (A–C) are representative of multiple replicates. Data in (B, C, and E) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test (B, C, and E). Analysis of the survival curve for mice was performed using a log‐rank test (F). *P < 0.05, **P < 0.01, ***P < 0.001; ns: not significant.

Source data are available online for this figure.

USP3 inhibits NF‐κB signaling by targeting MyD88

Since IL‐1R and most TLRs share the MyD88‐TRAF6 axis to activate the NF‐κB signaling pathway, although TNFR uses different upstream molecules (e.g., TRADD, RIP, TRAF2, TRAF5) to trigger TAK1 complex activation (Hayden & Ghosh, 2008), we reasoned that USP3 might function upstream of the TAK1 complex in the NF‐κB signaling pathway. To identify the molecular target of USP3, we co‐transfected USP3 with different signaling proteins in the NF‐κB pathway and found that USP3 mainly inhibited MyD88‐induced NF‐κB activation and slightly but significantly inhibited TRAF6‐induced NF‐κB activity only at a high USP3 concentration (Fig 3A). Next, we generated USP3 knockout (USP3 ^−/−) 293T cells (Fig EV3A and B). We also obtained a cell clone with the deletion of nine base pairs in the exon of USP3 (USP3 ^NFS), which did not affect the open reading frame of USP3 and served as another control for USP3 ^−/− cells (Fig EV3A and B). In contrast with previous results regarding USP3‐overexpressing 293T cells, we found that IL‐1β‐induced NF‐κB activation was significantly enhanced in USP3 ^−/− 293T cells (Fig 3B). Furthermore, we observed that this enhancement could be reversed by re‐introducing USP3 (Fig 3B). Using the USP3 ^−/− and USP3 ^NFS cells, we found that USP3 deficiency enhanced the NF‐κB activation induced by MyD88 and TRAF6, but not by IKKα or IKKβ (Fig 3C). There was no detectable difference in NF‐κB activity between WT and USP3 ^NFS cells (Fig 3C). Ectopic expression of USP3 inhibited the NF‐κB activation induced by MyD88 in USP3 ^−/− cells (Fig 3D). Similar results were obtained with the cells transfected with USP3‐specific shRNA (Fig EV3C and D). Collectively, our results suggest that although USP3 could moderately reduce TRAF6‐induced NF‐κB activation (Zhou et al, 2019), it specifically inhibits TLR‐ and IL‐1β‐induced NF‐κB activation, mainly through the adaptor molecule MyD88.

Figure 3. USP3 inhibits NF‐κB signaling by targeting MyD88.

• A
  293T cells were transfected with an NF‐κB promoter‐driven luciferase reporter (NF‐κB‐luc), together with vectors for TRIF, MyD88, TRAF2, TRAF5, TRAF6, TAK1 + TAB1, IKKα, IKKβ, or p65, along with an empty vector (no wedge) or increasing amounts (wedge) of USP3 expression vectors. Data were collected and plotted as the percentage of empty vector control (100%).
• B
  Luciferase activity in WT or USP3 ^−/− 293T cells transfected with a plasmid encoding an NF‐κB‐luc reporter, together with an empty vector or an expression vector for USP3, followed by treatment with IL‐1β (20 ng/ml) for 24 h.
• C
  Luciferase activity in WT, USP3 ^−/−, and USP3 ^NFS 293T cells transfected with an NF‐κB‐luc together with MyD88, TRAF6, IKKα, or IKKβ.
• D
  Luciferase activity in WT, USP3 ^−/−, and USP3 ^NFS 293T cells transfected with NF‐κB‐luc together with MyD88, along with empty vector or expression vector for USP3.
• E
  Whole‐cell lysis (WCL) of 293T cells co‐transfected with Flag‐MyD88 and Myc‐USP3 was subjected to immunoprecipitation with anti‐Flag beads.
• F
  The recombinant His‐MyD88‐Flag protein purified from E. coli was incubated with recombinant Myc‐USP3 (synthesized using an insect cell‐free in vitro protein expression system) overnight at 4°C, followed by immunoprecipitation with anti‐Flag beads.
• G, H
  Co‐immunoprecipitation and immunoassay of extracts of THP‐1 cells (G) and PBMCs (H) treated with LPS for the indicated lengths of time.
• I
  Localization of GFP‐USP3 and DsRed2‐MyD88 in 293T cells was determined by confocal microscopy. Scale bars, 10 μm.
• J
  WCL of 293T cells co‐transfected with Myc‐USP3 and Flag‐MyD88 full‐length (FL) or its deletion mutants were subjected to immunoprecipitation with anti‐Flag beads.
• K
  WCL of 293T cells co‐transfected with Flag‐MyD88 and Myc‐USP3 full‐length (FL) or its deletion mutants were subjected to immunoprecipitation with anti‐Myc beads.

Data information: Data in (A–K) are representative of multiple replicates. Data in (A, B, C, and D) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. *P < 0.05, **P < 0.01, and ***P < 0.001; ns: not significant.

Source data are available online for this figure.

Figure EV3. USP3 specifically targets and removes the K63‐linked polyubiquitin chains on MyD88.

• A
  Sequence alignment analysis of WT 293T, USP3 KO (USP3 ^−/−) 293T and USP3 not‐ORF (open reading frame) shift (USP3 ^NFS) 293T cells.
• B
  Immunoblot analysis of WT 293T, USP3 ^−/− 293T, and USP3 ^NFS 293T cells. β‐actin was used as a loading control.
• C
  Immunoblot analysis of the knockdown efficiency of exogenous Myc‐USP3 in HEK293T.
• D
  Luciferase activity in reference (Ref), USP3 shRNA1, and USP3 shRNA2‐transfected 293T cells, co‐transfected with an NF‐κB promoter‐driven luciferase reporter, MyD88, and either an EV or a vector expressing USP3.
• E
  The recombinant His‐MyD88‐Flag protein purified from E. coli was incubated with recombinant HA‐USP3 full‐length (FL), HA‐USP3 ZnF domain (ZnF) or HA‐USP3 UCH domain (UCH) overnight at 4°C, followed by immunoprecipitation with anti‐Flag beads.
• F
  Whole‐cell lysis (WCL) of 293T cells co‐transfected with HA‐K63‐Ub and full‐length (FL) Flag‐MyD88 or its deletion mutants were subjected to immunoprecipitation with anti‐Flag beads.
• G
  WCL of 293T cells co‐transfected with Flag‐MyD88 (DD) and HA‐USP3 (ZnF) or HA‐USP3 (UCH) were subjected to immunoprecipitation with anti‐HA beads.
• H
  WCL of 293T cells co‐transfected with Flag‐MyD88 (TIR) and HA‐USP3 (ZnF) or HA‐USP3 (UCH) were subjected to immunoprecipitation with anti‐HA beads.
• I
  Co‐immunoprecipitation (Co‐IP) and immunoassay of extracts of 293T Ref or USP3 knockdown cells transfected with plasmids for Flag‐MyD88 and HA‐K63‐Ub.
• J
  Co‐IP and immunoassay of extracts of 293T WT or USP3 ^KO cells transfected with plasmids for Flag‐MyD88 and HA‐K63‐Ub.
• K
  Co‐IP and immunoassay of extracts of 293T cells transfected with plasmids of Flag‐IRAK4 and HA‐tagged K63‐linked ubiquitin (HA‐K63‐Ub), together with the empty vector (EV) or USP3.
• L, M
  Co‐IP and immunoassay of extracts of 293T cells transfected with plasmids of HA‐MyD88, together with EV, Flag‐TRAF6 (L), or Flag‐TAK1 (M), followed by EV or Myc‐USP3.

Data information: Data in (B–M) are representative of multiple replicates. Data in (D) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. ***P < 0.001.

We next sought to determine whether USP3 could interact with MyD88. Co‐immunoprecipitation (Co‐IP) experiments showed that USP3 interacted with MyD88 (Fig 3E). Specifically, using recombinant proteins, we showed that USP3 and MyD88 interacted directly in vitro (Fig 3F). To determine the physiological relevance of these findings, we stimulated the THP‐1 cells or peripheral blood mononuclear cells (PBMCs) with LPS and found that the interaction between MyD88 and USP3 was induced by LPS treatment and peaked at 60 min after LPS treatment, which correlated with p‐IKK activation (Fig 3G and H). The confocal microscopic analysis also revealed that the USP3 protein was extensively colocalized with MyD88 (Fig 3I).

To determine the interaction domain of MyD88 with USP3, we generated two deletion mutants containing MyD88 death domain (DD) or MyD88 Toll‐interleukin 1 receptor domain (TIR), respectively, and found that both the TIR and DD domains could interact with USP3. However, the TIR domain exhibited stronger binding ability (Fig 3J). Furthermore, to identify the region of USP3 binding to MyD88, we generated two deletion mutants of USP3 containing the zinc‐finger Ub‐binding domain (ZnF‐UBP) (referred to as ZnF) or ubiquitin‐specific protease catalytic domain (UCH). Our results showed that both USP3 domains could bind to MyD88, but the ZnF domain exhibited a stronger binding ability with MyD88 (Fig 3K). However, an in vitro binding assay using recombinant MyD88 showed that neither of the individual domains of USP3 could directly interact with non‐ubiquitinated MyD88 purified from E. coli (Fig EV3E). This indicates that the intact structure of USP3 is critical for its direct interaction with MyD88, and the interaction between individual domains of USP3 with MyD88 shown in Fig 3K may be attributed to the ubiquitination of MyD88.

To determine which domains of MyD88 are responsible for its ubiquitination, we assessed the K63‐linked ubiquitination on the two domains of MyD88 purified from 293T cells. We detected strong K63‐linked ubiquitination on the TIR domain but only minimal ubiquitination on the DD domain (Fig EV3F). Next, we tested whether the individual domains of USP3 interact selectively with specific ubiquitinated domains of MyD88. Our results revealed that the ZnF domain of USP3 strongly interacted with both domains of MyD88, while the UCH domain of USP3 interacted with the TIR domain of MyD88 but not the DD domain (Fig EV3G and H). Taken together, these data suggest that only full‐length USP3 is able to directly interact with MyD88 without being ubiquitinated, indicating that an intact USP3 structure is critical for its specific binding capability with MyD88. However, the ubiquitination of MyD88 might be required to facilitate its binding with individual domains of USP3. Notably, the ZnF domain of USP3 interacted with both ubiquitinated domains of MyD88, whereas the UCH domain of USP3 only interacted with the highly ubiquitinated TIR domain of MyD88.

USP3 inhibits NF‐κB activation by deubiquitinating MyD88

As a DUB, USP3 can remove the ubiquitin from the histones H2A, γH2AX, H2B, and RIG‐I (Sloper‐Mould et al, 1999; Nicassio et al, 2007; Cui et al, 2014; Lancini et al, 2014; Sharma et al, 2014). Since we demonstrated that the inhibitory function of USP3 in NF‐κB signaling depends on its enzyme activity (Figs 1F and EV1E and F), we hypothesized that USP3 may regulate MyD88 function through its deubiquitinating activity. To test this possibility, we performed ubiquitination assays and found that USP3 markedly reduced both total ubiquitination and the K63‐linked ubiquitination of MyD88. However, it showed little or no effect on the K6‐, K11‐, K27‐, K29‐, K33‐, and K48‐linked ubiquitination of MyD88 (Fig 4A and B). USP3‐mediated inhibition of the K63‐linked ubiquitination of MyD88 was dose‐dependent (Fig 4C). Importantly, overexpression of USP7 failed to produce the same effect as USP3 in cleaving the K63‐linked polyubiquitin chains on MyD88, ruling out the possibility that the observed effects of USP3 were merely due to the overriding of DUB specificity by overexpression (Appendix Fig S2A–I). Moreover, knockdown or knockout of endogenous USP3 resulted in increased K63‐Ub chains on MyD88 (Fig EV3I and J). Furthermore, MyD88 underwent K63‐linked ubiquitination after LPS treatment in THP‐1 cells, and USP3 deficiency dramatically enhanced the K63‐linked ubiquitination of MyD88 (Fig 4D). We further examined whether USP3 could cleave the K63‐Ub chains on MyD88 in a cell system, finding that USP3 colocalized with MyD88 and K63‐Ub in the cytoplasmic punctum. And the fluorescence intensity of K63‐Ub was negatively correlated with USP3, as depicted in the enlarged image shown in Fig 4E. Interestingly, we found that only the USP3 (UCH) domain is responsible for removing K63‐Ub chains on MyD88 and inhibiting MyD88‐dependent NF‐κB activation (Fig 4F and G). To further understand how USP3 deubiquitinates the MyD88, we generated an inactive catalytic mutant of USP3 (C168S) and a USP3 mutant (H56A) with a disrupted zinc‐finger structure. Our results showed that both USP3 mutants abolished the ability of USP3 to affect the ubiquitination of MyD88. However, the USP3 (C168S) mutant retained a stronger interaction with MyD88 (Fig 4H), indicating that both an intact zinc‐finger domain and protease activity are required for the full‐length USP3 to deubiquitinate MyD88. Consistent with these observations, we found that both of these two mutants failed to inhibit the NF‐κB signaling activation induced by MyD88 (Fig 4I). These results suggest that USP3 hinders MyD88's function by cleaving K63‐Ub chains on MyD88. This action involves the ZnF domain, which binds to polyubiquitin chains, and the UCH domain, which is responsible for the cleavage process.

Figure 4. USP3 inhibits NF‐κB activation by removing the K63‐linked polyubiquitin chains on MyD88.

1. Co‐immunoprecipitation (Co‐IP) and immunoassay of extracts of 293T cells transfected with plasmids of Flag‐MyD88 and HA‐tagged ubiquitin (HA‐Ub), HA‐tagged K48‐linked ubiquitin (HA‐K48‐Ub), or HA‐tagged K63‐linked ubiquitin (HA‐K63‐Ub), together with empty vector (EV) or USP3.
2. Co‐IP and immunoassay of extracts of 293T cells transfected with plasmids of Flag‐MyD88 and HA‐Ub, HA‐tagged K6‐linked ubiquitin (HA‐K6‐Ub), HA‐tagged K27‐linked ubiquitin (HA‐K11‐Ub), HA‐tagged K29‐linked ubiquitin (HA‐K29‐Ub), or HA‐tagged K33‐linked ubiquitin (HA‐K33‐Ub), together with empty vector (EV) or USP3.
3. Co‐IP and immunoassay of extracts of 293T cells transfected with plasmids for Flag‐MyD88 and HA‐K63‐Ub, together with empty vector or increasing amounts (wedge) of an expression vector for Myc‐USP3.
4. Co‐IP and immunoassay of extracts of wild‐type (WT) or USP3 ^−/− THP‐1 cells treated with or without LPS (200 ng/ml) at the indicated time points. The lysis buffer was added with a final concentration of 10 mM N‐ethylmaleimide (NEM) to inhibit DUBs.
5. Localization of DsRed2‐MyD88, HA‐K63‐Ub, and GFP‐USP3 in 293T cells determined by confocal microscopy. Scale bars, 10 μm.
6. Co‐IP and immunoassay of extracts of 293T cells transfected with plasmids for Flag‐MyD88 and HA‐K63‐Ub, together with EV or Myc‐USP3 or its domain deletion mutants.
7. Luciferase activity of 293T cells transfected with an expression vector for MyD88 and an NF‐κB luciferase reporter (NF‐κB‐luc) together with EV or USP3 or its domain deletion mutants.
8. Co‐IP and immunoassay of extracts of 293T cells transfected with plasmids for Flag‐MyD88 and HA‐K63‐Ub, together with EV or Myc‐USP3 or its point mutants.
9. Luciferase activity of 293T cells transfected with an expression vector for MyD88 and an NF‐κB luciferase reporter (NF‐κB‐luc) together with EV or USP3 or its point mutants.

Data information: Data in (A–I) are representative of multiple replicates. Data in (G, I) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. **P < 0.01, and ***P < 0.001; ns: not significant.

Source data are available online for this figure.

To exclude the possibility that the USP3‐deubiquitinated polyubiquitin chains derive from other signaling complex proteins (e.g., IRAK4, IRAK1, and TRAF6) rather than MyD88, we checked whether USP3 could remove the K63‐linked polyubiquitin chains on these proteins. Our data showed that USP3 cannot decrease the K63‐linked polyubiquitin chains on IRAK4 or IRAK1, and it only slightly reduces the K63 ubiquitination of TRAF6 (Fig EV3K and Appendix Fig S3A and B), suggesting that USP3‐cleaved K63‐linked polyubiquitin chains are indeed located on MyD88. In addition, we investigated whether MyD88 deubiquitination influenced the TRAF6‐recruitment or signaling complex assembly. Co‐IP data revealed that USP3 co‐expression cannot disrupt the interaction between MyD88 and TRAF6, while the interaction between MyD88 and TAK1 is severely interrupted (Fig EV3L and M), indicating that the deubiquitination of MyD88 by USP3 may interrupt TAK1 recruitment.

Cytoplasmic translocation is required for USP3 to inhibit NF‐κB signaling

USP3 is exclusively localized in the nucleus (Nicassio et al, 2007), whereas MyD88 is an adaptor in the cytoplasm, raising the question of how USP3 could regulate MyD88. Interestingly, we found that GFP‐USP3 translocated from the nucleus to the cytoplasm in HeLa/TLR4 cells upon LPS stimulation (Figs 5A and EV4A). Next, we analyzed the cell fractionation of the cytoplasm and nucleus to determine whether endogenous USP3 could also translocate to the cytoplasm after TLR‐induced NF‐κB activation. Consistently, the amount of USP3 protein in the cytoplasm increased significantly after LPS treatment and reached a peak at 120 min post‐stimulation, while USP3 in the nucleus gradually reduced and then recovered 180 min post‐stimulation (Fig 5B). Importantly, USP3 was exported from the nucleus to the cytoplasm when p65, but not STAT1, translocated to the nucleus (Fig EV4B). Together, these results suggest that the activation of NF‐κB signaling leads to the cytoplasmic translocation of USP3.

Figure 5. USP3 is stabilized and exits the nucleus upon NF‐κB signaling activation.

1. HeLa cells transfected with GFP‐USP3 and TLR4 were treated with LPS (50 μg/ml), and the cells were observed continuously by fluorescence microscope at the indicated time points. The cells from five microscope fields were analyzed to calculate the ratio of USP3 export. Scale bars, 20 μm.
2. Cytoplasmic and nuclear extracts of THP‐1 cells treated with LPS (200 ng/ml) for the indicated lengths of time were analyzed by immunoblotting with anti‐USP3 antibody. β‐tubulin or lamin A/C served as the loading control.
3. 293T cells were transfected with an NF‐κB luciferase reporter (NF‐κB‐luc), together with the vector for MyD88, along with EV or expression vector for USP3. The cells were treated with leptomycin B (LMB, 50 nM) or ethanol (solvent for LMB) for 12 h before harvest.
4. Luciferase activity in 293T cells transfected with NF‐κB‐luc together with MyD88, along with expression vector for WT USP3, USP3 nucleus location sequence (NLS) mutant, USP3 nucleus export sequence (NES) mutant, or EV.
5. Localization of GFP‐USP3^NLS and DsRed2‐MyD88 in 293T cells determined by confocal microscopy. Scale bars, 10 μm.
6. Localization of GFP‐USP3^NES and DsRed2‐MyD88 in 293T cells determined by confocal microscopy. Scale bars, 10 μm.
7. Localization of DsRed2‐MyD88, HA‐K63‐Ub, and GFP‐USP3^NLS or GFP‐USP3^NES in USP3 ^−/− 293T cells determined by confocal microscopy. Scale bars, 10 μm.
8. Immunoassay of extracts of 293T cells transfected with Myc‐USP3, together with empty vector (EV) or increasing amounts of plasmids for Flag‐MyD88.
9. immunoassay of extracts of 293T cells transfected with Myc‐USP3, together with EV or plasmids for Flag‐MyD88, followed by DMSO or bafilomycin A1 (0.2 μM) treatment for 6 h before harvest.
10. Immunoassay of the endogenous USP3 in THP‐1 cells pretreated with DMSO or bafilomycin A1 (0.2 μM), followed by LPS stimulation (200 ng/ml) for indicated lengths of time.

Data information: Data in (A–J) are representative of multiple replicates. Data in (A) are plotted as means ± SD, representing five technical replicates (n = 5). Data in (C and D) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. ***P < 0.001; ns: not significant.

Source data are available online for this figure.

Figure EV4. USP3 is stabilized by NF‐κB signaling.

1. Confocal microscopy images of GFP‐USP3 in HeLa/TLR4 cells treated or not treated with LPS. UT: Untreated; DAPI, DNA‐intercalating dye. Scale bars, 20 μm.
2. Confocal microscopy of GFP‐USP3 in 293T cells transfected with either Flag‐p65 or Flag‐STAT1 with IFN‐α (STAT1 transfected cell) treatment. Scale bars, 20 μm.
3. Cytometer analysis of 293T cells transfected with GFP‐USP3, together with DsRed2‐EV or DsRed2‐MyD88.
4. Immunoassay of extracts of 293T cells transfected with Myc‐USP3, together with EV or plasmids for Flag‐MyD88, followed by BAY (20 μM) treatment after 12 h.
5. qPCR analysis of the endogenous USP3 in THP‐1 cells after LPS stimulation for the indicated lengths of time.
6. Immunoassay of the endogenous USP3 in THP‐1 cells after LPS stimulation for the indicated lengths of time.
7. Immunoassay of extracts of 293T cells transfected with Myc‐USP3, together with EV or plasmids for Flag‐MyD88, followed by MG132 (5 μM) treatment for the indicated lengths of time before harvest.
8. Immunoassay of extracts of 293T cells transfected with Myc‐USP3, together with EV or plasmids for Flag‐MyD88, followed by MG132 treatment at the indicated concentrations for 6 h before harvest.

Data information: Data in (A–H) are representative of multiple replicates. Data in (E) are plotted as means ± SD, representing three technical replicates (n = 3).

We employed three approaches to determine whether cytoplasmic translocation is indispensable for the inhibitory function of USP3: (i) using leptomycin B (LMB) to block the cytoplasmic translocation of USP3; (ii) generating the USP3 nucleus location sequence mutant (USP3^NLS) to abolish its nuclear location; and (iii) generating the USP3 nucleus export sequence mutant (USP3^NES) to abolish its cytoplasmic translocation. We found that both LMB treatment and USP3 nucleus export mutation reduced the inhibitory function of USP3 in NF‐κB activation, while the USP3^NLS mutant (the released USP3 in the cytoplasm) maintained a stronger inhibitory function than WT USP3 (Fig 5C and D). These results suggest that cytoplasmic translocation is essential for USP3 to inhibit NF‐κB signaling.

The confocal microscopic analysis also revealed that the USP3^NLS mutant alone exhibited a diffused localization pattern throughout the cytoplasm and nucleus, but co‐expression of USP3^NLS and MyD88 formed a colocalized punctum (Fig 5E). In contrast, it was hard for the USP3^NES mutant (only localized in the nucleus) to form colocalized puncta with MyD88 when they were co‐expressed (Fig 5F). Furthermore, we directly observed that USP3 colocalized with MyD88 and K63‐Ub in the cytoplasmic punctum, and the fluorescence intensity of K63‐Ub colocalized with MyD88, as the cytoplasmic punctum completely disappeared in the USP3 ^−/− cells transfected with USP3^NLS (localized in both cytoplasm and nuclei) but not in the cells transfected with USP3^NES (localized in the nuclei) (Fig 5G). These results suggest that the translocation of USP3 from the nucleus to the cytoplasm is an essential step in its function of removing the K63‐Ub chains on MyD88.

USP3 can be stabilized by NF‐κB signaling at the protein level

Besides cytoplasmic translocation, we found that USP3 could be stabilized by MyD88 in a dose‐dependent manner (Fig 5H). Flow cytometry analysis further showed that MyD88 significantly increased the fluorescence intensity of GFP‐USP3 when co‐expressing with it in the cells (Fig EV4C). To determine whether this MyD88‐induced stabilization of USP3 is dependent on the activation of NF‐κB signaling, we used an NF‐κB signaling inhibitor, BAY 11‐7082 (BAY), to block the MyD88‐induced NF‐κB signaling pathway activation at the IKK/IκB level. BAY is a commonly used NF‐κB signaling pathway inhibitor that has been reported to potently suppress NF‐κB signaling at the IKK/IκBα level and block the translocation of NF‐κB subunit (p65) without altering cell viability at a dose of 20 μM (Lee et al, 2012). Interestingly, we found that MyD88‐induced stabilization of USP3 was reduced by BAY treatment (Fig EV4D), indicating that USP3 is stabilized by MyD88‐induced NF‐κB signaling but not by MyD88 itself. Furthermore, we showed that LPS‐induced NF‐κB activation could stabilize endogenous USP3 at the protein level without increasing its mRNA abundance (Fig EV4E and F). These data suggest that the activated NF‐κB signaling might stabilize USP3 at the protein level.

To investigate whether the stabilization of USP3 upon activation of NF‐κB signaling can be attributed to altered proteasomal degradation, we treated cells with the proteasome inhibitor MG132 and evaluated USP3 expression levels with or without MyD88‐induced NF‐κB signaling activation. However, we found that MG132 treatment did not alter USP3 expression, indicating that the stabilization of USP3 by NF‐κB signaling is not due to altered proteasomal degradation (Fig EV4G and H). To further understand the mechanism behind the stabilization of USP3, we treated the cells with autophagic/lysosomal degradation inhibitor bafilomycin A1. Interestingly, bafilomycin A1 treatment significantly enhanced the expression of USP3, both exogenously and endogenously (Fig 5I and J). Autophagosome/lysosome inhibition leads to increased USP3 expression while proteasome inhibition does not, suggesting that changes in autophagy (during LPS stimulation) mediate increased USP3 expression.

USP3 is involved in the establishment of innate immune tolerance

Since nucleus exportation is a prerequisite step for USP3 to target MyD88 and inhibit NF‐κB signaling activation, we hypothesized that LPS priming could result in the retention of USP3 in the cytoplasm. This cytoplasmic retention of USP3 might be critical for the formation of innate immune tolerance since the cytoplasmic USP3 can quickly shut down NF‐κB signaling upon the second LPS challenge (Fig 6A). To test this hypothesis, we first primed the THP‐1 cells with low‐dose LPS and challenged them with high‐dose LPS 12 h later (Fig 6B). Consistent with our hypothesis, we found that LPS priming resulted in the cytoplasmic retention of USP3 and induced a faster and stronger cytoplasmic translocation of USP3 upon the second LPS challenge (Fig 6C). Importantly, we observed that the augmented phosphorylation of downstream protein kinases in USP3 ^−/− or USP3 ^KD cells were markedly enhanced after LPS pretreatment (Figs 6D and EV5A) compared with the non‐pretreatment results shown in Figs 2A and EV2E (Appendix Fig S4A and B). On the contrary, LPS priming significantly enhanced the function of USP3 in inhibiting the phosphorylation of downstream protein kinases (Fig EV5B) compared with the non‐pretreatment results shown in Fig 1C (Appendix Fig S4C). Consistently, LPS priming resulted in the cytoplasmic retention of USP3 in control THP‐1 cells (Fig EV5C). Although we observed the existence of cytoplasmic USP3 even without LPS stimulation in USP3‐expressing THP‐1 cells, possibly due to the overexpression of USP3, LPS pretreatment still increased the amount of cytoplasmic USP3 in these cells (Fig EV5C), indicating that LPS‐pretreatment induces the cytoplasmic retention of both endogenous and exogenous USP3 in THP‐1 cells. As expected, priming with LPS also dramatically increased the difference of proinflammatory cytokines at the mRNA level, triggered by USP3 overexpression, USP3 knockdown, or USP3 knockout (Figs 6E and EV5D and E) compared with non‐pretreatment results shown in Figs 1D, 2B, and EV2G (Appendix Fig S5A–C). Furthermore, the productions of IL‐6 and TNFα greatly enhanced in USP3‐deficient cells or reduced in USP3‐overexpressing cells, respectively, re‐challenged with LPS compared to that in WT or control cells, as shown in Figs 6F and EV5F and G. After normalizing the data based on WT or Ctrl cells, significantly increased differences in protein levels were observed in LPS‐pretreatment group compared with non‐pretreatment (Appendix Fig S6A–C). Next, we sought to determine the physiological significance of USP3‐dependent innate immune tolerance. Therefore, we directly stimulated THP‐1 cells with Salmonella typhimurium (flagellated Gram‐negative bacterium) or Staphylococcus aureus (Gram‐positive bacterium) and found that the productions of TNFα and IL‐6 were elevated by USP3 deficiency (Fig 6G and H). Importantly, this elevation was further enhanced by LPS priming (Appendix Fig S6D and E).

Figure 6. USP3 is involved in the establishment of innate immune tolerance.

• A
  Diagram depicts the hypothesis of USP3‐mediated innate immune tolerance.
• B
  Schematic overview of endotoxin tolerance assay.
• C
  Cytoplasmic extracts of THP‐1 cells primed with or without LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for indicated lengths of time were analyzed by immunoblotting with anti‐USP3 antibody. β‐tubulin and lamin A/C served as the control. The quantitative comparison involved density scanning of the blots.
• D
  WT or USP3 ^−/− THP‐1 cells were primed with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for the indicated periods. The cell extracts were harvested and used to immunoblot various kinases and signaling proteins.
• E
  Quantitative real‐time PCR analysis of WT and USP3 ^−/− THP‐1 cells primed with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for the lengths of time.
• F
  ELISA of IL‐6 or TNFα productions in the supernatants of WT and USP3 ^−/− THP‐1 cells primed without or with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for 24 h.
• G, H
  ELISA of IL‐6 or TNFα productions in the supernatants of WT and USP3 ^−/− THP‐1 cells primed with or without LPS (100 ng/ml) for 12 h and stimulated with heat‐killed Staphylococcus aureus (G) or heat‐killed Salmonella typhimurium (H) for 24 h.

Data information: Data in (C–H) are representative of repeated experiments. Data in (E, F, G, and H) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. ***P < 0.001.

Source data are available online for this figure.

Figure EV5. USP3 participates in the establishment of innate immune tolerance.

1. Reference (Ref) or USP3 knockdown THP‐1 cells were primed with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for the indicated lengths of time. The cell extracts were harvested and used for the western blot analysis of various kinases and signaling proteins.
2. THP‐1 cells transduced with empty vector (EV) or USP3 were primed with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for the indicated periods. The cell extracts were harvested and used for the western blot analysis of various kinases and signaling proteins.
3. Cytoplasmic and nuclear extracts of EV or USP3‐transduced THP‐1 cells pretreated with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) at the indicated time points were analyzed by immunoblotting with anti‐USP3 antibody. β‐tubulin and lamin A/C served as the controls.
4. Ref or USP3 knockdown THP‐1 cells were primed with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for the indicated lengths of time. The expressions of IL‐6, IL‐1β, and TNFα were determined by real‐time PCR.
5. The THP‐1 cells transduced with EV or USP3 were treated with LPS (100 ng/ml) for 12 h and challenged with LPS (200 ng/ml) for the indicated periods. The expressions of IL6, IL‐1β, and TNFα were determined by real‐time PCR.
6. Production of TNFα and IL‐6 cytokines in the culture medium of Ref or USP3 knockdown THP‐1 cells was primed with LPS for 12 h and challenged for 24 h.
7. Production of TNFα and IL‐6 cytokines in the culture medium of THP‐1 cells transduced with EV or USP3 was primed with LPS for 12 h and challenged for 24 h.
8. Schematic overview of the generation of Usp3 KO chimeric mice and the endotoxin tolerance assay of Fig 7D.

Data information: Data in (A–G) are representative of multiple replicates. Data in (D–G) are plotted as means ± SD, representing three technical replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. *P < 0.05, **P < 0.01, and ***P < 0.001.

USP3‐mediated innate immune tolerance protects against endotoxin shock in vivo

To investigate whether USP3 is involved in the TLRs‐induced cross‐tolerance between different TLRs, we isolated the bone marrow from WT mice and knocked out Usp3 using CRISPR/Cas9 technology ex vivo. Subsequently, we differentiated the bone marrow cells into bone‐marrow‐derived macrophage (BMDM) and stimulated them with various TLR ligands, with or without LPS pretreatment. Consistent with our previous findings, Usp3 deficiency significantly enhanced proinflammatory cytokine expression at the mRNA level induced by Pam₃CSK₄ (mTLR1/2 ligands), heat‐killed Listeria monocytogenes (mTLR2 ligands), flagellin (mTLR5 ligands), FSL‐1 (mTLR2/mTLR6 ligands), CL‐075 (mTLR7 ligands), and CpG (mTLR9 ligands) in BMDMs (Fig 7A–C). Importantly, the LPS pretreatment (mTLR4 ligands) further amplified this difference caused by Usp3 knockout (Fig 7A–C). Therefore, these data suggest that USP3 plays a critical role in TLR4‐induced cross‐tolerance between different TLRs ex vivo.

Figure 7. USP3‐mediated innate immune tolerance protects against endotoxin shock in vivo .

• A–C
  Quantitative real‐time PCR analysis of control (Ctrl) and Usp3 ^KO BMDM cells primed with LPS (100 ng/ml) for 12 h and challenged with Pam₃CSK₄ (500 ng/ml), heat‐killed Listeria monocytogenes (10⁸ cells/ml), flagellin (500 ng/ml), FSL‐1 (100 ng/ml), CL‐075 (1 μg/ml), and CpG (ODN1826) (5 μM) for 4 h. To compare the differences between Ctrl and Usp3 ^KO cells in the non‐pretreatment and the LPS‐pretreatment group, the data of Usp3 ^KO cells were normalized as a relative change (%) by dividing it with the data of Ctrl cells.
• D
  ELISA of TNFα, and IL‐6 productions in sera from Ctrl and Usp3 KO chimeric mice (n = 3 per group) given an intraperitoneal injection of LPS (0.3 mg/kg) every 24 h for 3 days and assessed after 6 h of injection.
• E
  The release of proinflammatory cytokines was re‐analyzed in Fig 7D. To compare the differences between the first LPS stimulation (non‐pretreatment) and second or third LPS stimulation (LPS‐pretreatment) groups, the data of Usp3 cKO mice were normalized as a relative fold by dividing them with the data of Ctrl mice.

Data information: Data in (A–C) are representative of repeated experiments. Data in (A–C) are plotted as means ± SD, representing three technical replicates (n = 3). Data in (D, E) are plotted as means ± SEM, representing three biological replicates (n = 3). Statistical analyses were performed using a two‐tailed unpaired Student's t‐test. *P < 0.5, **P < 0.01, and ***P < 0.001.

Source data are available online for this figure.

To further verify the function of USP3‐mediated innate immune tolerance in vivo, we intraperitoneally injected Usp3 ^cKO mice with a low‐dose LPS (0.3 mg/kg) for 3 days (Fig EV5H). Sera were then harvested 6 h after the first or third injection to measure TNFα and IL‐6 productions (Fig 7D). Usp3 knockout led to a nearly tenfold increase in TNFα and IL‐6 secretion in vivo compared to control mice, and this difference became even more significant during the second and third rounds of injection (Fig 7D and E). These results demonstrate that Usp3 deficiency severely disrupts the establishment of normal endotoxin tolerance and dramatically increases proinflammatory cytokine release upon endo toxin challenge. Collectively, these results further suggest that the MyD88–USP3 axis plays a pivotal role in the formation of innate immune tolerance in vivo, which could protect the host from severe septic shock.

Discussion

In this study, we unexpectedly discovered that USP3 worked as a potent negative regulator of NF‐κB signaling through the specific removal of K63‐linked polyubiquitin chains on MyD88. USP3 comprises a ZnF domain and a UCH domain, and only the full‐length USP3 has the ability to interact directly with MyD88 without undergoing ubiquitination. However, the individual domains of USP3 can interact with ubiquitinated MyD88, likely through their interaction with ubiquitin chains. Although the ZnF domain of USP3 interacted with both ubiquitinated domains of MyD88, the UCH domain of USP3 only interacted with the highly ubiquitinated TIR domain of MyD88. Importantly, although the UCH domain alone could cleave the K63‐Ub chains without the presence of the ZnF domain, it exhibited lower specificity and efficiency in inhibiting MyD88‐induced NF‐κB signaling compared to the full‐length USP3 (Fig 4G). Moreover, the inactivating mutation (H56A) on the ZnF domain severely impaired the inhibitory role of USP3 in NF‐κB signaling despite the intact UCH domain in full‐length USP3 (Fig 4H and I). These findings suggest that the presence of the ZnF domain may prevent the activation of the UCH domain in full‐length USP3 protein and that the activation of the ZnF domain is necessary to release the enzymatic activity of the UCH domain to cleave K63‐linked polyubiquitin chains. Furthermore, we found that USP3 (C168S) had a stronger interaction with MyD88 compared to WT USP3. One reasonable explanation for this is that USP3 may remove the polyubiquitin chains on MyD88 in a “cut‐and‐run” manner, and thus, its inactive catalytic mutant cannot remove K63‐linked polyubiquitin chains of the substrate, resulting in sustained interaction. Overall, these results suggest that the ZnF domain of USP3 may function as a “sensor” for specifically recognizing ubiquitinated MyD88 and then activate the UCH domain of USP3, acting as “scissors” to cleave its polyubiquitin chains.

It has been widely accepted that immunological memory is an exclusive hallmark of the adaptive immune response. However, growing evidence indicates that innate immune cells acquire a similar “innate immune memory” feature (Bowdish et al, 2007; Netea et al, 2011; Christ et al, 2018). Certain infections and vaccinations can induce broad protection against other pathogens through innate immune mechanisms, termed trained immunity (Netea et al, 2011, 2016; Divangahi et al, 2021; Zhou & Sun, 2021). On the contrary, endotoxin tolerance is another form of innate immune memory that leads to a lower inflammatory response to the second stimulation, termed innate immune tolerance (Netea et al, 2020; Divangahi et al, 2021; Zhou & Sun, 2021). Although it has been reported that epigenetic and metabolic reprogramming is responsible for the long‐term “training” innate immune memory (Novakovic et al, 2016; Seeley et al, 2018), it remains unclear whether post‐translational regulation is also involved in innate immune memory. Furthermore, although ubiquitination and deubiquitinating have been shown to play a pivotal role in controlling NF‐κB signaling (Chen & Chen, 2013; Hu & Sun, 2016), little is known about their role in innate immune tolerance. Here, we report that deubiquitinase USP3 is involved in the establishment of “tolerance” innate immune memory (Fig 7D). Although it was also recently reported that another deubiquitinase, OTUD4, negatively regulates NF‐κB signaling by cleaving the K63‐linked polyubiquitin chains on MyD88 (Zhao et al, 2018), only USP3 plays a critical role in innate immune tolerance, likely because of its unique (spatial and temporal) type of regulation after NF‐κB signaling activation. Since USP3 can be stabilized and detained in the cytoplasm upon first LPS priming, the arrested USP3 can rapidly shut down the subsequent NF‐κB signaling in the cytoplasm upon the second LPS challenge.

Similar to human memory, innate immune memory should be expected to have both short‐term and long‐term effects. Compared with long‐term training innate immune memory induced by epigenetic and metabolic reprogramming, USP3‐mediated tolerance innate immune memory might be relative short term since USP3 could return to the nucleus. However, it would be interesting to further investigate how long this USP3‐mediated tolerance memory could be sustained. We intend to perform long‐term experiments to investigate the duration of innate immune tolerance involving USP3 in the future.

Although we found that USP3 inhibits MyD88‐dependent NF‐κB signaling activation by removing the K63‐linked polyubiquitin chains on MyD88, it remains unclear why K63‐linked polyubiquitin chains are so crucial for MyD88‐dependent NF‐κB signaling activation. USP3 deficiency dramatically enhances the release of TLR‐induced inflammatory cytokines, suggesting that these K63‐linked polyubiquitin chains might play a pivotal role in activating NF‐κB signaling. Recently, it was reported that hybrid polyubiquitin chains consisting of K63‐linked and Met1‐linked ubiquitin chains on MyD88, IRAK1, and IRAK4 serve as a platform for the colocalization of both TAK1 kinase and IKK kinase complexes, thus facilitating the TAK1‐catalyzed activation of IKKα and IKKβ (Emmerich et al, 2013; Cohen & Strickson, 2017). Although a positive association between the K63‐linked ubiquitination of MyD88 and NF‐κB signaling activation has been established (Emmerich et al, 2013; Lee et al, 2016; Cohen & Strickson, 2017; Zhao et al, 2018; Yu et al, 2021), the detailed molecular mechanism through which K63‐linked polyubiquitin chains on MyD88 activate NF‐κB requires further investigation. Our data suggest that these K63‐linked polyubiquitin chains on MyD88 might be necessary for TAK1 recruitment.

Finally, human patients with severe sepsis are highly likely to experience secondary infections. Although endotoxin tolerance can protect mice from sterile LPS shock, the same innate immune tolerance might be responsible for organ damage and mortality in human patients with severe sepsis (Seeley et al, 2018). Our findings suggest that USP3 may serve as a potential therapeutic target for treating patients with severe sepsis. Taken together, our results identify a previously unrecognized post‐translational feedback loop of the MyD88–USP3 axis, which is critical for the establishment of normal endotoxin tolerance (Fig 8). We are the first to characterize ubiquitination regulation as an essential part of “tolerance” innate immune memory.

Figure 8. The proposed USP3‐mediated innate immune tolerance model.

Upon the initial activation of Toll‐like receptors (TLRs), MyD88 aggregates and recruits IRAK4/1, which triggers the activation of TRAF6, subsequently activating the TAK1 and IKK complexes. This leads to the phosphorylation, ubiquitination, and degradation of IκB. Once NF‐κB is released through IκB degradation, it translocates to the nucleus, promoting the translocation of USP3 from the nucleus to the cytoplasm. In the cytoplasm, USP3 removes the K63‐linked polyubiquitin chains on MyD88, thereby inhibiting MyD88‐dependent NF‐κB activation. Although a fraction of USP3 relocalizes to the nucleus after a few hours of stimulation, a considerable amount of USP3 remains in the cytoplasm. Upon secondary TLR stimulation, the detained USP3 in cytoplasm rapidly shuts down NF‐κB signaling by directly removing the K63‐linked polyubiquitin chains on MyD88. This process leads to the establishment of a “tolerance” innate immune memory.

Materials and Methods

Cell lines

HEK 293T cells (human embryonic kidney) and THP‐1 cells (acute monocytic leukemia), both obtained from ATCC (Manassas), were cultured in Dulbecco's modified Eagle's medium (Gibco) or RPMI‐1640 medium (Gibco) supplemented with 10% fetal bovine serum (Hyclone) and 1% L‐glutamine (Corning) in an incubator (Invitrogen) at 37°C and 5% CO₂. All cell lines were regularly examined to exclude mycoplasma contamination.

Generate sgRNA‐based bone marrow chimeric mice

To produce lentiviral particles carrying sgRNAs, 293T cells were cultured in Opti‐MEM I Reduced Serum Medium supplemented with 5% FBS and 1 mM sodium pyruvate and seeded at a density of 2.8 × 10⁷ cells per 175 flask in 35 ml of medium overnight to achieve confluency between 85 and 95% at the time of transfection. HEK293Ts cells were transfected with second‐generation lentiviral packaging plasmids the following day. Specifically, 32 μg of pSpCas9(BB)‐2A‐Puro (PX459) v.2.0 plasmids containing sgRNA targeting GFP (control) (5′‐GCTTTCACGGAGGTTCGACG‐3′) or Usp3 (5′‐GCAGGTCCAACAAAAGCCCC‐3′), 23 μg of psPAX2, and 10 μg of pMD2.G were mixed and transfected into 293T cells using lipofectamine 3000 transfection reagent (112 μl P3000 and 128 μl lipofectamine 3000). After 6 h, the transfection medium was replaced with 35 ml of OPTI‐MEM supplemented with 1× ViralBoost (Alstem Bio, catalog no. VB100). Lentiviral supernatant was collected at 24 h after transfection (first harvest) and again at 48 h after transfection (second harvest). The collected medium was centrifuged at 500 g for 10 min at 4°C to clear cellular debris, and the clarified supernatant was transferred to a sterile container. To concentrate the lentiviral particles, one volume of Lenti‐X Concentrator was added to three volumes of clarified supernatant, mixed gently, and incubated at 4°C overnight. The mixture was then centrifuged at 1,500 g for 45 min at 4°C to collect an off‐white pellet. After removing the supernatant, the pellet was gently resuspended in 1/100^th of the original volume using OPTI‐MEM. WT C57BL/6 mice were purchased from Jackson Laboratory and housed in a special pathogen‐free (SPF) environment. Bone marrow (BM) cells were isolated from male donor mice (8–10 weeks old, C57BL/6), and ACK buffer was used to remove red blood cells. The isolated bone marrow cells were diluted at 1–2 × 10⁶ cells/ml and infected with 2–5% v/v concentrated Cas9 lentivirus overnight in a cell culture incubator. Polybrene was added to enhance transduction efficiency, achieving a final concentration of 8 μg/ml. The cells were then washed and resuspended in phosphate‐buffered saline (PBS) for transplantation. Recipient female C57BL/6 mice (8–10 weeks old) were irradiated with 900 cGy 1 day prior to transplantation and were randomly assigned to either the control or Usp3 knockout (KO) lentivirus‐transduced bone marrow cell group, which received 1–1.5 × 10⁷ cells per mouse via intravenous injection. The reconstituted mice were housed in an SPF animal facility and provided with water containing a 1:100 BAYTRIL injection solution (enrofloxacin) for at least 2–3 weeks. Eight to ten weeks later, bone marrow was harvested from the reconstituted mice and analyzed using western blot and quantitative real‐time PCR to confirm the KO efficiency. The mice were then ready for use in subsequent experiments. All mouse‐related procedures were performed in accordance with experimental protocols (21098) approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Southern California.

Bacterial strains

E. coli strains (Top10) were grown in Luria–Bertani (LB) broth with appropriate antibiotics. All recombinant proteins were produced in E. coli strain BL21 (DE3) at standard temperatures (37°C for growth, 25°C for protein expression induction).

PBMC and bone marrow cell isolation

Human peripheral blood samples were obtained from anonymous donors at the Gulf Coast Regional Blood Center in Houston after obtaining consent forms approved by the University of Southern California's Institutional Review Boards. PBMCs were isolated from whole blood using the standard Ficoll density gradient centrifugation method. Briefly, whole blood was collected and divided into three 50‐ml conical tubes, each containing approximately 20 ml of blood. Six new 50‐ml conical tubes were prepared, to which 20 ml of Ficoll was added. The blood was then diluted with PBS to a final volume of 40 ml and slowly added to the top of the Ficoll solution, with care taken to prevent the mixing of the two layers. The tubes were centrifuged at 800 g without a brake for 20 min at room temperature, and the upper serum layer was aspirated and discarded (leaving around 5–10 ml of serum). The white layer containing PBMCs was collected in a new 50‐ml conical tube. The PBMCs were washed twice with PBS by centrifugation at 800 g with a brake for 7 min and then resuspended in 10% dimethyl sulfoxide (DMSO) in FBS for cryopreservation or cultured at a seeding density of 1 × 10⁶ cells/ml in complete RPMI 1640 medium (RPMI‐1640 medium [Gibco] supplemented with 10% fetal bovine serum, 2 mM L‐glutamine) for the following experiments. The remaining red blood cells were further removed by ACK buffer, and PBMC cells were cultured in the incubator for 24 h before stimulation with LPS to check the interaction between USP3 and MyD88. Bone marrow cells were isolated from the tibia and femur and cultured in RPMI1640 medium with 10% FBS, 1% penicillin–streptomycin, 55 μm β‐mercaptoethanol, and 10% L929‐conditioned media containing mouse macrophage‐colony‐stimulating factor (M‐CSF) for 6 days to harvest BMDMs (Du et al, 2022).

Immunoprecipitation and immunoblot analysis

For immunoprecipitation (Du et al, 2018), the cells were extracted with low‐salt lysis buffer (50 mM HEPES, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 1.5 mM MgCl₂, and 1% Triton X‐100), supplemented with 5 mg/ml protease inhibitor cocktail (Roche) after transfection and stimulation with appropriate ligands. Next, the cell extracts were incubated with the appropriate antibodies plus protein A/G beads (Pierce) or anti‐Flag agarose (Sigma) or anti‐hemagglutinin agarose gels (Sigma) overnight. The beads were then washed five times with low‐salt lysis buffer, and immunoprecipitates were eluted with 3x SDS Loading Buffer (FuDe) and resolved using SDS–PAGE. Proteins were transferred onto polyvinylidene fluoride (PVDF) membranes (Bio‐Rad) and incubated with the appropriate primary and secondary antibodies after blocking the membrane in 5% skim milk (Bio‐Rad). Enhanced chemiluminescence (Millipore) was performed for protein detection.

Antibodies

Rabbit polyclonal antibodies against human USP3 (12490‐1‐AP) were obtained from Proteintech, and the mouse monoclonal antibodies against human MyD88 (B‐1) (sc‐136970) and human USP3 antibody (8L8) (sc‐135597) were obtained from Santa Cruz. Anti‐MyD88 (D80F5) (cat # 4283S), anti‐phospho‐IKKα/β (cat # 2697S), anti‐phospho‐JNK (cat # 9251S), anti‐JNK (cat # 9252S), anti‐phospho‐ERK (cat # 9101S), anti‐ERK (cat # 9102S), anti‐phospho‐p38 (cat # 9211S), anti‐p38 (cat # 9212S), anti‐IκBα (cat # 4814S), and anti‐phospho‐IκBα (Ser32/36) (cat # 9246S) were purchased from Cell Signaling Technology. Anti‐IKKα antibody (clone 14A231), anti‐IKKβ antibody (clone 10AG2), anti‐ubiquitin antibody, and lys63‐specific antibody (clone Apu3 [05‐1308]) were purchased from Millipore. Horseradish peroxidase (HRP)‐anti‐Flag (M2) (A8592) and anti‐β‐actin (A1978) antibodies were purchased from Sigma, while HRP‐anti‐hemagglutinin (HA) (12013819001) and anti‐c‐Myc‐HRP (11814150001) were purchased from Roche Applied Science. Goat and rabbit isotype controls were purchased from Invitrogen. Protein G agarose and protein A agarose were purchased from Pierce. Anti‐HA tag antibody [16B12] (DyLight^® 650) (ab117515) and anti‐Flag tag antibody [M2] (cat # DyLight^® 550) (ab117495) for immunofluorescence were purchased from Abcam. Donkey anti‐rabbit IgG (H + L) and Alexa Fluor 555 (Cat # A‐31572) antibodies were purchased from Invitrogen.

Plasmids

MyD88, USP3, and other plasmids used in our studies were acquired using standard PCR techniques. MyD88‐truncated mutant plasmids were constructed with the NEBuilder^® HiFi DNA Assembly Cloning Kit (NEB). Point mutations, including Myc‐USP3 (H56A), Myc‐USP3 (C163S), MyC‐USP3 (NES), and Myc‐USP3 (NLS), were generated with a site‐directed mutagenesis kit (SBSgene).

Nuclear and cytoplasmic extraction

We used the NE‐PER™ Nuclear and Cytoplasmic Extraction Reagents (Invitrogen; cat # 78833) to extract nuclear and cytoplasmic proteins from THP‐1 cells. The cells were first harvested by centrifugation at 500 g for 5 min and washed with PBS. After transferring the cells to a 1.5‐ml microcentrifuge tube, they were pelleted again, and the supernatant was discarded. The cell pellet was resuspended in ice‐cold CER I and vortexed for 15 s, followed by incubation on ice for 10 min. CER II was then added, and the tube was vortexed and incubated on ice. After centrifuging the tube for 5 min, the supernatant (cytoplasmic extract) was immediately transferred to a pre‐chilled tube and stored on ice. The pellet fraction containing nuclei was then resuspended in ice‐cold NER and vortexed for 15 s, followed by incubation on ice. This was repeated every 10 min for a total of 40 min. After centrifuging the tube for 10 min, the supernatant (nuclear extract) was immediately transferred to a pre‐chilled tube and stored on ice. The CER I: CER II: NER reagent volume ratio was maintained at 200:11:100 μl. The extracts were stored at −80°C until use.

Real‐time PCR analysis

For real‐time PCR analysis (Chu et al, 2021), total RNA was isolated from cells or tissues using TRIzol Reagent (Invitrogen), and the first‐strand cDNA was generated from total RNA using a PrimeScript™ RT Reagent Kit (Takara). Real‐time PCR was conducted with the SYBR qPCR Mix (Genestar) and specific primers using the QuantStudio 6 Flex Real‐Time PCR System (Applied Biosystems). The values of the target gene expression were normalized to hGAPDH. The gene‐specific PCR primers are shown in Appendix Table S1.

Knockdown of USP3 by RNA interference

USP3‐specific shRNA plasmids and reference shRNA plasmids (Openbiosystems) were transfected into 293T using lipofectamine 2000 (Invitrogen), following the manufacturer's instructions.

Generation of KO cells by CRISPR/Cas9 technology

KO cells were produced using CRISPR/Cas9 technology. The gRNA was inserted into pSpCas9(BB)‐2A‐Puro (PX459) v.2.0, containing a puromycin resistance gene. Each assembled pSpCas9(BB)‐2A‐Puro (PX459) v.2.0 plasmid (1 μg) was mixed with 100 μl of serum‐free Opti‐MEM and 2 μl of lipofectamine 2000 (Invitrogen). After incubation for 20 min at room temperature, the solution was added to the cells in a dropwise manner for transfection. For THP‐1 transduction, the lentiviral particles were produced by transfecting 293T cells with assembled pSpCas9(BB)‐2A‐Puro (PX459) v.2.0, VSGV, and Δ8.9. Viral supernatants were concentrated by ultracentrifugation and then used to transduce the THP‐1 cells. After two transductions at 24 and 48 h, the medium was replaced with the fresh medium containing puromycin (2.0 μg/ml) and cultured for one week. The cells were then single‐cell plated into 96‐well plates and left until colonies began to form (2–3 weeks). The individual clones were screened by PCR amplifying and sequencing genomic DNA containing the CRISPR target site. This approach allowed us to identify cells carrying the same genomic alteration in both copies of the gene, ensuring accurate selection. The sequence of the target USP3 gene was as follows: 5′‐GCCGGTCCAACAAAAGCCCT‐3′.

Luciferase reporter assay

For the luciferase report assay (Feng et al, 2017), HEK 293T cells (4 × 10⁵ cells/ml) were seeded in 96‐well plates for 12 h at 37°C. The expression plasmids encoding an NF‐κB luciferase (NF‐κB‐Luc) reporter (firefly luciferase) and a pRL‐TK reporter (Renilla luciferase plasmid) together with plasmids encoding pcDNA3.1‐Flag‐MyD88 or other indicated plasmids were transfected with lipofectamine 2000 (Invitrogen), following the manufacturer's protocol. Cells were collected at 24–36 h after transfection, and luciferase activity was examined using a Dual‐Luciferase Assay (Promega) with a Synergy 2 modular multi‐mode reader (BioTek), following the manufacturer's protocol. Reporter gene activity was determined by normalization of the firefly luciferase activity to Renilla luciferase activity.

Cytokine release assay

Concentrations of IL‐6 and TNFα in cell culture supernatants were determined using an ELISA kit (BD Biosciences), following the manufacturer's recommendations. The concentrations of the IL‐6 and TNFα in mouse sera were determined by ELISA kits (eBiosciences) according to the manufacturer's recommendations.

Immunofluorescent microscopy analysis

For the immunofluorescence experiments (Jin et al, 2017), 293T cells seeded on glass‐bottom culture dishes (Nest Scientific) were transfected with expression plasmids encoding the indicated plasmids for 24 h. HeLa cells transfected with GFP‐USP3 plasmids on glass‐bottom culture dishes (Nest Scientific) were stimulated with LPS for 1 h. Cells were fixed with 4% paraformaldehyde for 15 min and then permeabilized in methyl alcohol for 10 min at −20°C. After washing with PBS three times, the cells were blocked in 5% fetal goat serum and 0.3% Triton X‐100 for 1 h and incubated with primary antibodies (in 1% bull serum albumin) overnight. The cells were washed and stained with fluorescent‐labeled secondary antibody Alexa Fluor 568‐conjugated antibody against mouse IgG (Abcam) and Alexa Fluor 647‐conjugated antibody against rabbit IgG (Abcam). Confocal images were examined using a microscope (LSM710; Carl Zeiss, LSM780; Carl Zeiss, A1plus; Nikon). The images were processed using LSM Zen 2008 (Carl Zeiss), LSM Zen 2012 (Carl Zeiss), NIS‐Elements, and Photoshop software (Adobe Systems).

Statistical analysis

The results of all quantitative experiments are reported as mean ± SD, and an unpaired two‐tailed Student's t‐test was utilized to compare the differences between the two groups using GraphPad Prism 5.0 software. Analysis of the mice's survival curve was performed using a log‐rank test. When the P‐value was less than 0.05, differences between groups were considered significant, and the level of significance was indicated as *P < 0.05, **P < 0.01, and ***P < 0.001.

Author contributions

Tianhao Duan: Conceptualization; resources; data curation; formal analysis; validation; investigation; visualization; methodology; writing – original draft; project administration; writing – review and editing. Yanchun Feng: Validation; investigation. Yang Du: Validation; investigation. Changsheng Xing: Investigation; writing – review and editing. Junjun Chu: Investigation. Jiayu Ou: Investigation. Xin Liu: Investigation. Motao Zhu: Writing – review and editing. Chen Qian: Investigation. Bingnan Yin: Investigation. Helen Y Wang: Funding acquisition. Jun Cui: Conceptualization; resources; data curation; supervision; funding acquisition; methodology; project administration; writing – review and editing. Rong‐Fu Wang: Conceptualization; resources; data curation; supervision; funding acquisition; methodology; project administration; writing – review and editing.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

PDF+

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

Source Data for Figure 7

EMBO reports (2023) 24: e57828

Data availability

This study includes no data deposited in external repositories.