E3 Ligase WWP2 Negatively Regulates TLR3-Mediated Innate Immune Response by Targeting TRIF for Ubiquitination and Degradation

E3 ligase WWP2 negatively regulates TLR3-mediated innate immune response by targeting TRIF for ubiquitination and degradation

Yan Yanga,b,1, Bing Liaoc,1, Suyun Wangb, Bingru Yana, Ying Jinc, Hong-Bing Shua, and Yan-Yi Wangb,2

aCollege of Life Sciences, Wuhan University, Wuhan 430072, China; bWuhan Institute of Virology, State Key Laboratory of Virology, Chinese Academy of Sciences, Wuhan 430071, China; and cDepartment of Molecular Developmental Biology, Shanghai JiaoTong University School of Medicine, Shanghai 20025, China

Edited by Philippa Marrack, Howard Hughes Medical Institute, National Jewish Health, Denver, CO, and approved February 15, 2013 (received for review November 20, 2012)

Recognition of viral double-stranded RNA by Toll-like receptor 3 Ccl5/Rantes and Ifnb is abolished, indicating that LPS-induced (TLR3) triggers activation of the transcription factors NF-κB and inter- activation of IRF3 and subsequent expression of type I IFNs and feron regulated factor 3, leading to induction of type I interferons and IFN-inducible genes are dependent on TRIF (8, 9). proinflammatory cytokines. TIR-domain–containing adapter-inducing TRIF contains an N-terminal proline-rich region, a middle interferon-β (TRIF) is an adapter protein required for TLR3-mediated TIR domain and a C-terminal receptor-interacting protein (RIP) signaling. Here we identified the E3 ubiquitin ligase WW domain- homotypic interaction motif (RHIM). The N-terminal domain of containing protein 2 (WWP2) as a TRIF-associated protein by biochem- TRIF is responsible for its association with TBK1, which is ical purification. WWP2 mediated K48-linked ubiquitination and deg- a downstream kinase required for TRIF-mediated IRF3 activation (10–12). The N-terminal domain of TRIF also contains a consen- radation of TRIF upon TLR3 activation. Overexpression of WWP2 κ sus TRAF-binding motif (250-PEEMSW-255) that is required for inhibited TLR3-mediated NF- B and interferon regulated factor 3 ac- the recruitment of TRAF6 for NF-κB activation (13). The C-ter- tivation, whereas knockdown of WWP2 had opposite effects. We Wwp2 fi minal RHIM motif of TRIF can recruit RIP1 via RHIM homo- generated -de cient mice to further investigate the roles of typic interaction, which may account for the NF-κB activation in Wwp2 in innate immune responses. Consistently, production of TRAF6-deficient macrophages (14, 15). Moreover, TRIF also IMMUNOLOGY IFN-β, CCL5, TNFα, and IL-6 in response to the TLR3 ligand poly(I:C) – – −/− mediates apoptosis through the RIP FADD caspase-8 pathway was elevated in Wwp2 macrophages and Wwp2-deficient mice (12). These studies suggest that TRIF plays divergent roles in exhibited increased susceptibility to poly(I:C)-induced death than TLR3-mediated signaling. However, the mechanisms by which the control littermates. Our findings suggest that WWP2 negatively TRIF are regulated remain unclear. regulates TLR3-mediated innate immune and inflammatory In this study, we identified a HECT-domain containing E3 responses by targeting TRIF for ubiquitination and degradation. ubiquitin ligase WW domain-containing protein 2 [WWP2, also called atrophin-1 interacting protein 2 (AIP2)] as a TRIF-associ- fi oll-like receptors (TLRs) are evolutionarily conserved pattern ated protein by biochemical puri cation experiments. WWP2 was Trecognition receptors (PRRs) that are critically involved in associated with TRIF upon TLR3 activation and targeted TRIF host defense from plants to humans. So far, 13 TLRs (named for K48-linked ubiquitination and degradation. Knockdown of fi WWP2 resulted in compromised TRIF ubiquitination, elevated TLR1 to TLR13) have been identi ed in humans and mice, each IFNΒ of which recognizes a distinct set of pathogen-associated molec- TRIF protein level and enhanced expression of after TLR3 activation. Furthermore, gene knockout of Wwp2 in mice resulted ular patterns (PAMPs) (1). TLRs contain an extracellular domain fl consisting of leucine rich repeats which is responsible for PAMP in increased expression of proin ammatory cytokines and type I recognition, a transmembrane domain, and a conserved cyto- IFNs in macrophages and susceptibility to poly(I:C)-induced plasmic toll/IL-1 receptor (TIR) domain which is able to mediate death in vivo. These results reveal a previously uncharacterized homotypic protein–protein interactions (2). The TIR domains of mechanism by which TLR3-mediated innate immune and inflammatory responses are regulated at the TRIF adapter level. TLRs are responsible for their homo- or hetero-dimerization, and upon ligand stimulation, they also act as platforms to recruit Results downstream TIR domain-containing adaptor proteins and sig- fi naling molecules, leading to the activation of transcription factors Identi cation of WWP2 as a TRIF-Associated Protein. Previous studies κ have shown that TRIF is a critical adapter protein in TLR3- such as NF- B and interferon regulated factor 3 (IRF3) (1, 3, 4). κ These transcription factors act alone or in collaboration to in- mediated NF- B and IRF3 activation pathways (16). To un- fl ambiguously identify potential new proteins associated with duce transcription of proin ammatory cytokines and/or type I fi fi interferons (IFNs). TRIF, we performed tandem af nity puri cation (TAP) assays Among the TLRs, TLR3 has been reported to recognize viral with full-length TRIF as bait. An expression plasmid for TRIF dsRNA as well as its analog poly(I:C). Recognition of these tagged with streptavidin binding peptide (SBP) and calmodu- ligands by TLR3 activates signaling pathways leading to the acti- lin binding peptide (CBP) was transfected into 293 cells, and κ TRIF-associated proteins were purified by the mammalian TAP vation of NF- B and IRF3 and subsequent production of type I fi IFNs and proinflammatory cytokines (5). TLR3-mediated signal- system. The eluted proteins were identi ed by a shotgun mass ing critically depends on the TIR domain-containing adapter TIR- domain–containing adapter-inducing interferon-β (TRIF; also −/− called TICAM-1) (6). It has been shown that Trif lung fibro- Author contributions: Y.Y. and Y.-Y.W. designed research; Y.Y., S.W., and B.Y. performed blasts are defective in poly(I:C)-induced activation of NF-κBand research; B.L. and Y.J. contributed new reagents/analytic tools; Y.Y., H.-B.S., and Y.-Y.W. analyzed data; Y.-Y.W. wrote the paper. IRF3 as well as production of type I IFNs, demonstrating that fl TRIF is indispensable for TLR3-mediated signaling (7). In addi- The authors declare no con ict of interest. tion, the TLR4 ligand LPS has been shown to signal through This article is a PNAS Direct Submission. TRIF-dependent and independent [myeloid differentiation factor 1Y.Y. and B.L. contributed equally to this work. −/− 88 (MyD88)-dependent] pathways (7, 8). In Trif macrophages, 2To whom correspondence should be addressed. E-mail: [email protected]. although LPS-induced expression of proinflammatory cytokines This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. such as Tnf, Cxcl1,andCxcl2 is normal, LPS-induced expression of 1073/pnas.1220271110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1220271110 PNAS Early Edition | 1of6 Downloaded by guest on September 26, 2021 spectrometry analysis. By comparing with other nonrelated between WWP2 and TRIF in RAW264.7 and A549 cells (Fig. 1C). purifications with the same method, we identified WWP2 as These results suggest that poly(I:C) binding to TLR3 induces the a candidate protein associated with TRIF. WWP2 is an E3 association between WWP2 and TRIF. ubiquitin ligase containing a C-terminal HECT domain, four WW Because WWP2 is associated with TRIF after TLR3 activa- repeats, and an N-terminal C2 domain that may be important for tion, we determined the effects of WWP2 on TRIF-mediated its subcellular localization (17). It has been reported that WWP2 signaling. In reporter assays, WWP2 inhibited TRIF-mediated is involved in inhibition of activation-induced T-cell death by activation of NF-κB, interferon stimulated response element ubiquitinating EGR2 (18) as well as in early mammalian de- (ISRE), and the IFN-β promoter in a dose-dependent manner. velopment by targeting the octamer-binding transcription factor 4 In the same experiments, WWP2 did not inhibit MyD88-mediated (Oct-4) (19). In addition, recent studies have suggested that NF-κB activation (Fig. 1D). These results suggest that WWP2 WWP2 can facilitate the budding process of retrovirus (20). specifically inhibits TRIF-mediated signaling. Whether and how WWP2 is involved in TRIF-mediated signaling is unknown. WWP2 Targets TRIF for Ubiquitination and Degradation. Because To confirm the association between WWP2 and TRIF, we WWP2 is an E3 ubiquitin ligase, and our earlier results indicated performed transient transfection and coimmunoprecipitation ex- that TRIF was down-regulated following TLR3 activation and periments in 293 cells. The results indicated that WWP2 inter- TRIF-mediated signaling was inhibited by WWP2, we reasoned that acted with TRIF in mammalian overexpression system (Fig. 1 A WWP2 may mediate TRIF ubiquitination and degradation. In and B). Domain mapping experiments indicated that the second mammalian overexpression system, WWP2 but not its enzymatic WW domain of WWP2 and the N-terminal and TIR domains of inactive mutant WWP2(C838A) markedly increased the ubiquiti- TRIF are important for their interaction (Fig. 1 A and B). nation and degradation of TRIF (Fig. 2A). Using K48-only or K63- We next determined whether endogenous WWP2 could asso- only ubiquitin mutant, we found that WWP2 enhanced K48-linked ciate with TRIF. Endogenous coimmunoprecipitation experi- ubiquitination of TRIF, whereas it had minimal effect on K63-linked ments indicated that WWP2 was not associated with TRIF under ubiquitination of TRIF (Fig. 2B). In vitro ubiquitination assays physiological conditions. However, their association was readily further confirmed that WWP2 could directly ubiquitinate TRIF detected following poly(I:C) stimulation of 293 cells stably (Fig. 2C). In these in vitro experiments, WWP2 dramatically in- expressing TLR3 (293-TLR3) (Fig. 1C). Poly(I:C) stimulation also creased K48-linked ubiquitination of TRIF but had little effect on induced endogenous association between WWP2 and TRIF in K63-linked ubiquitination of TRIF (Fig. 2C). These results suggest macrophage cells RAW264.7 and epithelial cells A549 (Fig. 1C), that WWP2 preferably mediates K48-linked ubiquitination of TRIF. which have previously been shown to express TLR3 (21). In similar We next determined whether WWP2 would regulate ubiq- experiments, LPS stimulation did not induce the interaction uitination and degradation of TRIF in untransfected cells. As

A WWP2 and mutants TRIF interaction TRIF and mutants WWP2 interaction C838A WWP2(1-870) C2 WW WW WW WW HECT + TRIF(1-712) TIR +

WWP2(1-335) C2 WW - TRIF(1-394) -

WWP2(1-363) C2 WW WW + TRIF(1-532) TIR + Fig. 1. WWP2 interacts with TRIF. (A) A schematic WWP2(1-437) C2 WW WW WW + TRIF(394-712) TIR - presentation of full-length WWP2, TRIF and their WWP2(1-477) C2 WW WW WW WW + TRIF(348-580) TIR - mutants. (B) Domain mapping of TRIF and WWP2 in- WWP2(478-870) HECT - - TRIF(532-712) teraction. The 293 cells (2 × 106) were transfected with the indicated expression plasmids (5 μg each). Coim- B HA-TRIF HA-WWP2 munoprecipitation were performed with anti-Flag F-WWP2: 1-870 1-335 1-363 1-437 1-477 478-870 F-TRIF: 1-394 394-712 1-532 532-712 348-580 1-712 (αF) or control mouse lgG (Ig). (Upper) The immuno- IP: Ig F Ig F Ig F Ig F Ig F Ig F IP: Ig F Ig F Ig F Ig F Ig F Ig F 117- -HA-TRIF 117- -HA-WWP2 precipitates were analyzed by immunoblot with anti- 85- 85- HA. (Lower) Expression of the transfected proteins were analyzed by immunoblots with anti-Flag or anti- 117- 117- --F-TRIFF-TRIF -F-WWP2 85- HA. (C) Endogenous association of WWP2 with TRIF. -F-TRIF(1-532)-F-TRIF(1-532 The 293-TLR3 (5 × 107), RAW264.7 (1 × 108), or A549 85- -F-TRIF(1-394)-F-TRIF(1-394 × 8 -F-WWP2(1-477) 49- (1 10 ) cells were left untreated or treated with poly -F-WWP2(1-437) Lysate LysateLysate (I:C) (50 μg/mL) or LPS (50 ng/mL) for the indicated -F-WWP2(1-363) -F-TRIF(358-580)-F-TRIF(358-5 -F-WWP2(1-335) 34- -F-TRIF(394-712)-F-TRIF(394-7 times. Cell lysates were immunoprecipitated, and 49- -F-WWP2(478-870)-F-WWP2(478-870) immunoprecipitates were analyzed by immunoblots 25- -F-TRIF(532-712)-F-TRIF(532-7 with the indicated antibodies. For experiments with -HA-TRIF -HA-WWP2-HA-WWP2 293-TLR3 cells, coimmunoprecipitation was performed with mouse preimmune serum (Pre) or mouse 293-TLR3 RAW264.7 A549 anti-WWP2 serum. The blots were detected by rabbit C Poly(I:C) LPS IP: Pre WWP2 - Poly(I:C) LPS 0 0.5 1.0 0 0.5 1.0 h anti-TRIF or rabbit anti-WWP2. For experiments Poly(I:C): 000.5 1 2 h IP: Ig W Ig W Ig W IP: Pre T Pre T Pre T Pre T Pre T Pre T with RAW264.7 cells, coimmunoprecipitation was -TRIF-TRIF --TRIFTRIF --WWP2 performed with rabbit IgG control (Ig) or rabbit -WWP2-WWP2 anti-WWP2 (αW). The blots were detected by mouse -TRIF-TRIF --TRIFTRIF -TRIF--TRIF- Lysate LysateLysate -WWP2--WWP2- anti-TRIF. For experiments with A549 cells, coimmuno- - -actin-actin LLysateysate --WWP2WWP2 ---actin-actin --actin-- -actin- precipitation was performed with mouse preimmune serum (Pre) or mouse anti-TRIF (αT) serum. The blots

293-NF- B 293-ISRE 293-IFN- were detected by rabbit anti-WWP2. (Lower)Theex- D pression levels of the proteins were examined by 200 600 600 immunoblots with the indicated antibodies. (D)WWP2 t. c 150 A 400 400 inhibits TRIF- but not MyD88-mediated signaling. c. 5 100 The 293 cells (1 × 10 ) were transfected with TRIF or el. Lu

R 200 200 MyD88 expression plasmid (0.2 μg each), increased 50 amounts of WWP2 expression plasmid and the in- 0 0 0 μ vector TRIF MyD88 vector TRIF vector TRIF dicated reporter plasmid (0.1 g). Reporter assays F-WWP2: were performed 20 h after transfection.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1220271110 Yang et al. Downloaded by guest on September 26, 2021 ABMyc-Ub: + + + F-TRIF: ++ + HA-WWP2: - + - F-TRIF: ++++++ HA-WWP2(C838A): -- + HA-Ub: + + - - - - IP: Ig F Ig F Ig F HA-Ub-K48: - - - + + - Fig. 2. WWP2 targets TRIF for K48-linked ubiquiti- HA-Ub-K63: - - - - + + nation and degradation. (A) Overexpression of wild- F-WWP2: - + - + - + - type but not mutant WWP2 promotes ubiquitination -Ub-TRIF IP: TRIF of TRIF. The 293 cells (1 × 107) were transfected with IB: Myc -Ub-TRIF IB: HA the indicated plasmids. Twenty hours after trans- 117- 117- fection, cell lysates were immunoprecipitated with -F-TRIF anti-Flag (αF) or control mouse lgG (Ig). The immu- 85- Lysate noprecipitates were analyzed by immunoblot with -F-WWP2 -TRIF anti-Myc (Upper). The expression levels of the pro- Lysate teins were examined by immunoblots with the in- -WWP2/ dicated antibodies (Lower). (B)WWP2markedly WWP2(C838A) enhances K48-linked ubiquitination of TRIF. The 293 cells (2 × 106) were transfected with the indicated E1+UbcH5c C plasmids. Twenty hours after transfection, cell lysates TRIF: - ++ - ++ - ++ - ++ were immunoprecipitated with anti-TRIF. The im- WWP2: + - + + - ++- ++- + munoprecipitates were re-extracted in lysis buffer containing 1% SDS and denatured by heating for Ub-K48: ------++ + --- 5 min at 95 °C. Supernatants were diluted with lysis Ub-K63: ------+++ buffer until the concentration of SDS was decresed Ub-WT: ++ + ------to 0.1%, followed by reimmunoprecipitation with anti-TRIF and then analyzed by immunoblot with anti-HA (Upper). The expression levels of the proteins were examined by immunoblots with the indicated antibodies (Lower). (C) WWP2 mediates K48- 117- linked ubiquitination of TRIF in vitro. TRIF, WWP2, IB: Ub ubiquitin, and ubiquitin mutants were translated in IMMUNOLOGY 85- vitro, and E1 and UbcH5c were added for ubiquiti- IB: TRIF nation assays. (Upper) Ubiquitin-conjugated TRIF

IB: WWP2 was detected by immunoblot with anti-ubiquitin antibody. (Lower) The levels of input proteins were D examined by immunoblots with the indicated 293-TLR3 A549 antibodies. (D) Effects of WWP2-RNAi on poly(I:C)- Myc-Ub: ++++ Pre TRIF Pre TRIF induced ubiquitination and degradation of endog- Poly(I:C): --++ WWP2-RNAi: - ++- enous TRIF. The 293-TLR3 (5 × 107) or A549 (5 × 107) WWP2-RNAi: - + - + Poly(I:C): 0 0 1 0 1 LPS: 0 0 110 h cells were transfected with a control or WWP2-RNAi IP: TRIF plasmid. The cells were left untreated or treated with IB: Myc -Ub-TRIF poly(I:C) (50 μg/mL) or LPS (100 ng/mL) for 1 h. The cell IP: TRIF 117- 117- IB: Ub lysates were immunoprecipitated with mouse pre- 85- 85- immune serum control or mouse anti-TRIF serum, -TRIF and the immunoprecipitates were analyzed by Lysate -WWP2 -TRIF immunoblot with anti-Myc (Upper Left)oranti- --actin ubiquitin (Upper Right). (Lower) The expression Lysate -WWP2 levels of the related proteins were examined by --actin immunoblots with the indicated antibodies.

shown in Fig. 2D, in both 293-TLR3 and A549 cells, knockdown Knockdown of WWP2 Potentiates TLR3-Mediated Signaling. Because of WWP2 had minimal effects on basal ubiquitination and pro- overexpression of WWP2 inhibits TLR3-mediated signaling, tein level of TRIF. However, knockdown of WWP2 inhibited we next determined the roles of endogenous WWP2 in TLR3- poly(I:C)-induced ubiquitination and degradation of TRIF (Fig. mediated signaling by RNAi-mediated knockdown experiments. 2D). In contrast, knockdown of WWP2 did not inhibit LPS- As shown in Fig. 3D, knockdown of WWP2 markedly potentiated induced ubiquitination of TRIF in A549 cells (Fig. 2D). These poly(I:C)-induced activation of NF-κB, ISRE, and the IFN-β pro- results suggest that WWP2 mediates ubiquitination and degra- moter in reporter assays. Consistently, knockdown of WWP2 dation of TRIF after TLR3 activation in untransfected cells. also potentiated poly(I:C)-induced expression of IFNB1, CCL5, and TNFα genes in 293-TLR3 cells (Fig. 3E). In similar experi- Overexpression of WWP2 Inhibits TLR3-Mediated Signaling. Because ments, however, knockdown of WWP2 had no marked effect on WWP2 interacted with TRIF following poly(I:C) stimulation in TNF-induced NF-κB activation or IFN-γ–induced IRF1 activa- 293-TLR3 cells, we determined whether WWP2 was involved in tion (Fig. 3 F and G). These results demonstrate that WWP2 is TLR3-mediated signaling. As shown in Fig. 3A, overexpression a negative regulator of TLR3-mediated signaling pathways. of WWP2 but not its enzymatic inactive mutant WWP2(C838A) inhibited poly(I:C)-induced activation of NF-κB, ISRE, and the Wwp2 Negatively Regulates TLR3-Mediated Induction of Cytokines in IFN-β promoter in 293-TLR3 cells in reporter assays. In similar Macrophages and Mice. Because human WWP2 and its mouse experiments, WWP2 did not inhibit IL-1β- or TNFα-induced NF- ortholog share 96% sequence identity at the amino acid level, we κB activation (Fig. 3B). Results from RT-PCR analysis indicated hypothesize that they share similar functions. To further eluci- that overexpression of WWP2 but not WWP2(C838A) inhibited date the physiological role of WWP2 in TLR-mediated innate poly(I:C)-induced transcription of endogenous IFNB1, CCL5, immune responses, we generated Wwp2-deﬁcient mice by stan- and ISG15 genes in 293-TLR3 cells (Fig. 3C). These data suggest dard conditional gene knockout technology (Fig. S1A). The that WWP2 inhibits TLR3-mediated signaling and such an in- heterozygosity and nullizygosity were veriﬁed by PCR and im- −/− hibitory function requires its E3 ubiquitin ligase activity. munoblot analysis (Fig. S1 B and C). The Wwp2 mice were

Yang et al. PNAS Early Edition | 3of6 Downloaded by guest on September 26, 2021 Fig. 3. WWP2 regulates TLR3-mediated signaling 293-TLR3 293-TLR3 293-TLR3 293-TLR3 A NF- B ISRE IFN- pathways. (A) WWP2 inhibits poly(I:C)-induced sig- IFN- × 5 5 9 9 20 - naling in 293-TLR3 cells. The cells (1 10 ) were Poly(I:C) 4.5 8 8 18 transfected with the indicated luciferase reporter 4 7 7 16 μ 3.5 6 6 14 plasmid (0.1 g) and increased amounts of empty 3 12 5 5 2.5 10 control vector (empty bars) or WWP2 expression 4 4 2 8 3 3 Rel. Luc. Act. Luc. Rel. Rel. Luc. Act. Luc. Rel. ﬁ Act. Luc. Rel.

plasmid ( lled bars). Twenty hours after transfection, 1.5 Act. Rel. Luc. 6 μ 1 2 2 4 the cells were treated with poly(I:C) (50 g/mL) or left 0.5 1 1 2 0 untreated for 8 h before luciferase assays were per- 0 0 0 g 0 0.1 0.2 g 0 0.1 0.2 g 0 0.1 0.2 g 0 0.1 0.2 formed. (B) WWP2 has no marked effects on TNFα-and WWP2 WWP2 WWP2 WWP2(C838A) IL-1β- induced NF-κB activation. The 293 cells (1 × 105)

were transfected with the indicated luciferase re- B 293-NF- B C Mock Poly(I:C) porter plasmids (0.1 μg) and empty control vector 12 r r Vector o o (empty bars) or WWP2 expression plasmid (ﬁlled 10 WWP2 Vect WT C838A Vect WT C838A -IFNB1 bars). Twenty hours after transfection, cells were 8 treated with TNFα (20 ng/mL) and IL-1β (20 ng/mL) 6 -CCL5

or left untreated for 8 h before luciferase assays 4 -ISG56 Rel. Luc. Act. Luc. Rel. were performed. (C) WWP2 inhibits poly(I:C)-induced 2 -GAPDH transcription of endogenous IFNB1, CCL5, and ISG15 0 - IL-1 genes. The 293-TLR3 cells (2 × 105) were transfected TNF with the indicated expression plasmids (1 μg each) for WWP2-RNAi 293-TLR3 293-TLR3 293-TLR3 20 h. The cells were treated with poly(I:C) or left D Con NF- B ISRE IFN- #1 #2 untreated for the indicated times before RT-PCR for 9 - 117- 18 30 -WWP2 8 Poly(I:C) the indicated genes were performed. (D) Effects of 16 25 85- -MPP5 7 14 WWP2-RNAi plasmids on expression of WWP2 and on 6 12 20 5 poly(I:C)-induced signaling. For the blots, the 293 cells 10 15 × 5 WWP2-RNAi 4 8 (2 10 ) were transfected with control or WWP2- Con 3 6 Act. Luc. Rel. 10 #1 #2 Act. Luc. Rel. RNAi plasmid (1 μg each) for 24 h. Cell lysates were 2 Act. Luc. Rel. 4 -WWP2 5 analyzed by immunoblots with the indicated anti- 1 2 --actin 0 0 0 bodies. For the reporter assays, 293-TLR3 cells (1 × 105) Con #1 #2 Con #1 #2 Con #1 #2 were transfected with the indicated luciferase re- WWP2-RNAi WWP2-RNAi WWP2-RNAi porter (0.1 μg) and RNAi (0.5 μg each) plasmids. 293-TLR3 Twenty-four hours after transfection, cells were left E F 293-NF- B G 293-IRF1 untreated or treated with poly(I:C) (50 μg/mL) for 8 h 50 20 15 18 - 8 - Con 16 TNF IFN- before luciferase assays were performed. (E) Effects 40 15 WWP2-RNAi 14 6 10 of WWP2-RNAi on poly(I:C)-induced transcription of 30 12

mRNA 10 5 mRNA 10

× mRNA downstream genes. The 293-TLR3 cells (2 10 ) were 20 8 4 5 Ifnb Tnf transfected with a control or WWP2-RNAi plasmid Ccl5 5 6 Rel. Luc. Act. Luc. Rel. 10 Act. Luc. Rel. μ 4 2 (1 g each). Twenty hours after transfection, cells were 0 0 0 2 ck treated with poly(I:C) or left untreated for 1 h before o ck ck 0 0 y(I:C) o o (I:C) m m oly(I:C) m ly Con WWP2-RNAi Con WWP2-RNAi qRT-PCR was performed. (F) Effects of WWP2-RNAi Pol P Po on TNFα-induced NF-κB activation. The experiments were similarly performed as in A except that TNFα (20 ng/mL) was used for stimulation. (G) Effects of WWP2-RNAi on IFN-γ-induced activation of the IRF1 promoter. The 293 cells (1 × 105) were transfected with IRF1 promoter reporter (0.1 μg) and WWP2-RNAi (0.5 μg each) plasmids. Twenty-four hours after transfection, cells were left untreated or treated with IFN-γ (100 ng/mL; ﬁlled bars) or left untreated (open bars) for 6 h before reporter assays were performed.

born at the normal Mendelian ratio and grew healthily. The total (Fig. 4 A and B). These results suggest that deficiency of Wwp2 cell numbers of thymocytes, splenocytes, and cells in peripheral fi Ifnb1 −/− speci cally promotes TLR3-mediated induction of and its lymph nodes were comparable between wild-type and Wwp2 Ccl5 fl + + + + downstream gene as well as the proin ammatory cytokines mice. The compositions of CD4 , CD8 , or CD4 CD8 double TNFα and IL-6. positive thymocytes, and T cells and B cells in spleen were −/− comparable between wild-type and Wwp2 mice (Fig. S1 D–G). Wwp2−/− Mice Were More Susceptible to Poly(I:C)-Induced Death. To In addition, Wwp2-deficiency did not affect the composition of + + examine whether Wwp2 regulates TLR3-mediated innate im- naïve (CD62L ) and memory (CD44 ) T cells in spleen and mune response in vivo, we monitored poly(I:C)-induced pro- G peripheral lymph nodes (Fig. S1 ), indicating that WWP2 was duction of type I IFNs, proinflammatory cytokines, as well as dispensable for animal or immune cell development. poly(I:C)-induced inflammatory death of wild-type and Wwp2- −/− To determine whether mouse Wwp2 is involved in innate im- deficient mice. Age- and sex-matched wild-type and Wwp2 mice mune response, we collected bone marrow-derived macrophages Wwp2 fi were injected with poly(I:C) or LPS through i.p. route. The in- (BMDMs) from wild-type and -de cient mice and stimu- duced cytokines in the serum were measured 2 h after injection, lated these cells with poly(I:C), LPS (ligand for TLR4), R848 (li- and the survived animals were recorded every 12 h. As shown in gand for TLR7 and TLR8), PGN (ligand for TLR2), and Sendai Fig. 5A, poly(I:C)-induced production of TNFα, IL-6, and IFN-β virus (SeV) (stimulating RIG-I-like receptors). The results showed fi Wwp2−/− that poly(I:C)- and LPS-induced transcription of Ifnb1 and Ccl5 was signi cantly enhanced in the sera from mice com- fi Wwp2−/− pared with that in their wild-type counterparts. In contrast, LPS- was signi cantly increased in BMDMs in comparison α with their wild-type or heterozygous counterparts (Fig. 4A). In- induced production of TNF or IL-6 in the sera was not markedly Wwp2−/− A terestingly, although both poly(I:C)- and LPS-induced transcrip- increased in mice (Fig. 5 ). Tnf Il6 Wwp2−/− In consistent with the results of cytokine production in the sera, tion of and genes was enhanced in BMDMs −/− compared with their wild-type or heterozygous counterparts (Fig. Wwp2 mice showed an early death onset (12 h v.s. 36 h) and 4A), secreted TNFα and IL-6 cytokines were increased only in poly a higher percentage of lethality (100% v.s. 50%) within 60 h in − − (I:C)- but not LPS-induced Wwp2 / BMDMs (Fig. 4B). In these comparison with their wild-type counterparts after injection of Ifnb1 Ccl5/Rantes poly(I:C) plus D-galactosamine (Fig. 5B). In contrast, wild-type experiments, the induction of and mRNA −/− or TNFα and IL-6 cytokines by R848, PGN, or SeV was not and Wwp2 mice responded similarly to LPS-induced death − − discernable among wild-type, heterozygous, and Wwp2 / BMDMs (Fig. 5B). These results suggest that Wwp2 negatively regulated

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1220271110 Yang et al. Downloaded by guest on September 26, 2021 BMDM BMDM deficiency increased TLR3-, but not TLR2- or TLR7/8-mediated A 2000 200 400 production of the proinflammatory cytokines TNFα and IL-6 in 1500 macrophages. Moreover, Wwp2-deficiency did not affect SeV- 1000 ˆˆ ˆ 150 300 ˆ β α 500 induced expression of IFN- , TNF , and IL-6 in mouse macro- 0 phages, suggesting that WWP2 does not regulate innate immune mock SeV mRNA mRNA 100 200 ˆˆ response mediated by RIG-I-like receptors in these cells. Ccl5 Ifnb Previous studies have demonstrated that TRIF is also involved in 50 100 TLR4-mediated signaling (8). However, TRIF plays distinct roles in TLR3- and TLR4-mediated signaling pathways. TRIF is an es- 0 0 mock PGN Poly(I:C) LPS R848 mock SeV PGN Poly(I:C) LPS R848 sential adapter in TLR3-mediated signaling because all TLR3- −/− BMDM BMDM Trif Wwp2+/+ mediated innate immune responses were abolished in mice 400 1500 ˆˆ Wwp2+/- (7). In contrast, TLR4 uses both MyD88 and TRIF for signaling. Wwp2-/- These two adapter molecules play both redundant and distinct 300 ˆˆ 1000 roles in LPS-induced gene expression. It is demonstrated that ˆˆ MyD88 is primarily involved in TLR4-mediated activation of NF-

mRNA 200 mRNA κ ﬂ

Il6 B and transcription of numerous proin ammatory cytokines such Tnf − − 500 ˆˆ α Myd88 / 100 as TNF and IL-6 (9). Consistently, mice are highly resistant to LPS-induced death (23). On the other hand, TRIF is

0 0 mostly required for TLR4-mediated expression of type I IFNs and mock SeV PGN Poly(I:C) LPS R848 mock SeV PGN Poly(I:C) LPS R848 their downstream genes. Interestingly, our studies suggest that Wwp2 deficiency increased poly(I:C)-induced production of B BMDM BMDM fl α Wwp2+/+ proin ammatory cytokines TNF and IL-6 in the serum, as well as 2500 4000 sensitized mice to poly(I:C)-induced systematic inflammation and Wwp2+/- Wwp2 fi 2000 Wwp2-/- death. However, de ciency did not elevate LPS-induced 3000

(pg/ml) 1500 2000 IL-6 (pg/ml)

TNF 1000 ˆˆ ˆ ˆ ˆˆ 1000 600 1000 A 3000 500 ˆˆ

800 IMMUNOLOGY 0 0 mock SeV PGN Poly(I:C) LPS R848 mock SeV PGN Poly(I:C) LPS R848 2000 400 600 Fig. 4. Effects of Wwp2-deﬁciency on expression of various inﬂammatory

+ + + − − − TNF (pg/ml) and innate immune genes in macrophages. Wwp2 / , Wwp2 / ,orWwp2 / 400

IL-6 pg/ml IL-6 1000

IFN-200 pg/ml) BMDMs were treated with poly(I:C) (20 μg/mL), LPS (25 ng/mL), PGN (50 μg/mL), 200 and R848 (40 nM) or infected with SeV for 3 h before qRT-PCR was performed (A), or the cells were treated for 18 h and then culture supernatants were 0 0 0 -/- -/- collected for measurement of TNFα and IL-6 by ELISA (B). **P < 0.01; *P < 0.05 +/+ +/+ +/+ -/- Wwp2 (t test). Wwp2 Wwp2 Poly(I:C)

TLR3-mediated cytokine production, inﬂammatory and innate 600 immune responses in vivo. 2500 30000

Discussion 2000 400 20000 The adapter protein TRIF plays a critical role in TLR3-mediated 1500 signaling, which activates the transcription factors NF-κB and (pg/ml) IRF3 and leads to induction of proinﬂammatory cytokines and 1000 IL-6 (pg/ml) IL-6

TNF 200 type I IFNs. However, little is known about how TRIF is regu- 10000 IFN- pg/ml) lated, particularly, at a posttranscriptional level. In this study, we 500 N.D. fi fi N.D. performed biochemical puri cation experiments and identi ed 0 0 0 WWP2 as a TRIF-associated protein. Although WWP2 consti- +/+ -/- +/+ -/- +/+ -/- tutively interacted with TRIF in mammalian overexpression Wwp2 Wwp2 Wwp2 system, endogenous WWP2 only interacted with TRIF following LPS poly(I:C) stimulation. Overexpression of WWP2 inhibited poly B κ (I:C)-triggered activation of NF- B and IRF3, whereas knock- 10 10 down of WWP2 had opposite effects, suggesting that WWP2 neg- Wwp2+/+ Wwp2-/- atively regulates TLR3-mediated signaling in mammalian cells. 8 8 −/− In addition, Wwp2 BMDMs and mice produced higher levels of proinflammatory cytokines and type I IFNs after poly(I:C) 6 6 Wwp2 fi treatment and -de cientmiceweremoresusceptibleto 4 Live mice 4 poly(I:C)-induced death. Therefore, our data demonstrate that Live mice WWP2 regulates TLR3-mediated inflammatory and innate immune 2 2 responses both in vitro and in vivo. 0 0 WWP2 is an E3 ubiquitin ligase. Our results suggest that 0 122436486072 0 12 24 36 48 60 72 84 96 WWP2 mediates K48-linked ubiquitination and degradation of TRIF following TLR3 activation, which might be responsible for Time after Poly(I:C) injection (h) Time after LPS injection (h) its negative regulation of TLR3-mediated signaling. It has been Fig. 5. Wwp2−/− mice are more susceptible to poly(I:C)- but not LPS-induced reported that the TIR-containing protein SARM specifically death. (A) Serum cytokine concentrations in Wwp2+/+ and Wwp2−/− mice regulates TRIF-mediated signal transduction by interacting with (n = 6 or 7) injected intraperitoneally with poly(I:C) plus D-galactosamine or + + TRIF, but the underlying molecular mechanism is still unclear LPS for 2 h. **P < 0.01; *P < 0.05 (t test). (B) Sex- and age-matched Wwp2 / − − (22). Interestingly, overexpression of WWP2 inhibited TRIF- but and Wwp2 / mice (n = 8) were injected intraperitoneally with poly(I:C) plus not MyD88-mediated NF-κB activation. Consistently, Wwp2- D-galactosamine or LPS, and their survival was monitored every 12 h for 4 d.

Yang et al. PNAS Early Edition | 5of6 Downloaded by guest on September 26, 2021 production of TNFα and IL-6 in the serum, nor had any marked by standard conditional knockout strategy that replaced the exon 3 effects on LPS-induced death. These observations are consistent with a Neomycin cassette. Because WWP2 is constitutively ex- with the previous studies showing that TRIF is required for all pressed in most tissues, it is possible that WWP2 plays divergent TLR3-mediated responses but dispensable for TLR4-induced and roles in various biological and pathological processes. Our study MyD88-mediated production of certain proinflammatory cytokines clearly establishes a role for WWP2 in regulation of TLR3- such as TNFα and IL-6. Taken together, our findings suggest that mediated innate immune and inflammatory responses. WWP2 specifically targets TRIF and plays a critical regulatory role in TLR3-mediated inflammatory and innate immune responses. Materials and Methods − − In this study, we found that Wwp2 deficiency up-regulated Mice and Cells. Wwp2 / mice were generated by a conditional gene-targeting mRNA levels of Ifnb, Ccl5, Tnfa, and cretion following LPS fl −/− strategy. In the targeting vector, exon 3 of Wwp2 is anked by two loxP sites stimulation in BMDMs was not markedly changed in Wwp2 and a neomycin-resistance gene cassette. An HSV-tk cassette was introduced cells in comparison with their wild-type counterparts. It is possible in the targeting vector at the end of 3′ arm for negative selection of clones that in addition to a negative regulatory role in LPS-induced with random integration in the genomic DNA. Successful targeting would transcription of cytokine genes, Wwp2 is also involved in posi- lead to deletion of exon 3, which corresponds to nucleotide 71–218 of the tively regulating TNFα and IL-6 protein expression and/or se- mouse Wwp2 cDNA coding sequence. Deletion of exon 3 would result in an cretion at a posttranscriptional level following LPS stimulation in out of frame mutation that leads to an early stop codon which gives rise to BMDMs. The combination of these functions may lead to in- a premature mRNA encoding only 34 aa if exon 2 and 4 are spliced together. creased mRNA levels of Tnfa and Il6 but no marked changes of After electroporation of the Wwp2 targeting vector into embryonic stem secreted TNFα and IL-6 cytokines following LPS stimulation in cells (129SV/EV), G418 and gancyclovir double-resistant colonies were se- these cells. In our experiments, we found that WWP2 did not lected and screened by PCR. The selected Wwp2 ES clones were injected interact with TRIF or mediate ubiquitination of TRIF following into the blastocysts of C57BL/6. Chimeric mice were mated with C57BL/6 LPS stimulation in A549 epithelial cells, which seems to contra- female mice to produce heterozygotes. The heterozygotes were interbred dict with our observation that Wwp2 deficiency caused increased to produce homozygous mice. The homozygous mice were then mated with transcription of cytokine genes following LPS stimulation in mice [FVB/N-TgN(EIIa- Cre)C5379Lmgd] that expresses Cre recombinase in BMDMs. It is possible that Wwp2 is differentially involved in nearly all tissues to produce Wwp2-deficient mice. All animal experiments LPS-induced signaling in different cell types. Alternatively, Wwp2 were performed in accordance with the Wuhan University animal care and targets a different component rather than TRIF in LPS-induced use committee guidelines. signaling pathways. Obviously, more rigorous studies are needed BMDMs were generated from mouse bone marrows as described (26). to decipher the roles of Wwp2 in LPS-induced signaling in dif- The cells (1 × 107) were cultured in RPMI medium 1640 containing 10% FBS ferent cell types in the future. and 10 ng/mL recombinant murine M-CSF (Peprotech) in a 100-mm dish for Previously, several studies have demonstrated that WWP2 is 5 d before experiments. involved in regulating transcription, embryonic stem-cell fate, The 293 cells stably expressing TLR3 (293-TLR3) cells were generously cellular transport, tumorigenesis, and T-cell activation processes provided by Katherine Fitzgerald (University of Massachusetts Medical by targeting distinct substrates (18–20). Recently, a gene knock- School, Worcester, MA) and Tom Maniatis (Department of Molecular and out study revealed a developmental role for WWP2 in chondro- Cellular Biology, Harvard University, Cambridge, MA). genesis via mechanisms involving cartilage-specific transcription factors (24). Mice deficient in Wwp2 develop mild malformations Other Materials and Methods. Other materials and methods used in this study of the craniofacial region (25). However, we have not observed are described in SI Materials and Methods. such a phenotype in our Wwp2-deficient mice. It is possible that different gene knockout strategies may contribute to this minor ACKNOWLEDGMENTS. We thank Drs. Katherine Fitzgerald and Tom Maniatis difference. Whereas the Zou et al. (25) used a gene-trap tech- for 293-TLR3 cells. This work was supported by Chinese Ministry of Science Wwp2 and Technology Grants 2012CB910201 and 2013CB530500; National Natural nology to generate -null mice by using embryonic stem cells Science Foundation of China Grants 31221061, 31170792, 31270932, and that contained the bacterial β-galactosidase gene inserted in in- 31000639; and an Academic Award for Excellent PhD Candidates funded tron 3–4 of the Wwp2 locus, our study produced Wwp2-null mice by Ministry of Education of China.

1. Kawai T, Akira S (2011) Toll-like receptors and their crosstalk with other innate re- 14. Cusson-Hermance N, Khurana S, Lee TH, Fitzgerald KA, Kelliher MA (2005) Rip1 ceptors in infection and immunity. Immunity 34(5):637–650. mediates the Trif-dependent toll-like receptor 3- and 4-induced NF-kappaB activation 2. Jin MS, Lee JO (2008) Structures of the toll-like receptor family and its ligand com- but does not contribute to interferon regulatory factor 3 activation. JBiolChem plexes. Immunity 29(2):182–191. 280(44):36560–36566. 3. Wang YY, Ran Y, Shu HB (2012) Linear ubiquitination of NEMO brakes the antiviral 15. Meylan E, et al. (2004) RIP1 is an essential mediator of Toll-like receptor 3-induced response. Cell Host Microbe 12(2):129–131. NF-kappa B activation. Nat Immunol 5(5):503–507. 4. Nie Y, Wang YY (2013) Innate immune responses to DNA viruses. Protein Cell 4(1):1–7. 16. Yamamoto M, et al. (2002) Cutting edge: A novel Toll/IL-1 receptor domain-containing 5. Alexopoulou L, Holt AC, Medzhitov R, Flavell RA (2001) Recognition of double-stranded adapter that preferentially activates the IFN-beta promoter in the Toll-like receptor RNA and activation of NF-kappaB by Toll-like receptor 3. Nature 413(6857):732–738. signaling. J Immunol 169(12):6668–6672. 6. Oshiumi H, Matsumoto M, Funami K, Akazawa T, Seya T (2003) TICAM-1, an adaptor 17. Rotin D, Kumar S (2009) Physiological functions of the HECT family of ubiquitin li- molecule that participates in Toll-like receptor 3-mediated interferon-beta induction. gases. Nat Rev Mol Cell Biol 10(6):398–409. Nat Immunol 4(2):161–167. 18. Chen A, et al. (2009) The HECT-type E3 ubiquitin ligase AIP2 inhibits activation-induced 7. Yamamoto M, et al. (2003) Role of adaptor TRIF in the MyD88-independent toll-like T-cell death by catalyzing EGR2 ubiquitination. Mol Cell Biol 29(19):5348–5356. receptor signaling pathway. Science 301(5633):640–643. 19. Xu H, et al. (2009) WWP2 promotes degradation of transcription factor OCT4 in hu- 8. Hoebe K, et al. (2003) Identification of Lps2 as a key transducer of MyD88-independent man embryonic stem cells. Cell Res 19(5):561–573. TIR signalling. Nature 424(6950):743–748. 20. Martin-Serrano J, Eastman SW, Chung W, Bieniasz PD (2005) HECT ubiquitin ligases 9. Hirotani T, et al. (2005) Regulation of lipopolysaccharide-inducible genes by MyD88 link viral and cellular PPXY motifs to the vacuolar protein-sorting pathway. J Cell Biol and Toll/IL-1 domain containing adaptor inducing IFN-beta. Biochem Biophys Res 168(1):89–101. Commun 328(2):383–392. 21. Tissari J, Sirén J, Meri S, Julkunen I, Matikainen S (2005) IFN-alpha enhances TLR3- 10. Fitzgerald KA, et al. (2003) IKKepsilon and TBK1 are essential components of the IRF3 mediated antiviral cytokine expression in human endothelial and epithelial cells by signaling pathway. Nat Immunol 4(5):491–496. up-regulating TLR3 expression. J Immunol 174(7):4289–4294. 11. Sato S, et al. (2003) Toll/IL-1 receptor domain-containing adaptor inducing IFN-beta 22. Carty M, et al. (2006) The human adaptor SARM negatively regulates adaptor protein (TRIF) associates with TNF receptor-associated factor 6 and TANK-binding kinase 1, TRIF-dependent Toll-like receptor signaling. Nat Immunol 7(10):1074–1081. and activates two distinct transcription factors, NF-kappa B and IFN-regulatory factor-3, 23. Kawai T, Adachi O, Ogawa T, Takeda K, Akira S (1999) Unresponsiveness of MyD88- in the Toll-like receptor signaling. JImmunol171(8):4304–4310. deficient mice to endotoxin. Immunity 11(1):115–122. 12. Han KJ, et al. (2004) Mechanisms of the TRIF-induced interferon-stimulated response 24. Nakamura Y, et al. (2011) Wwp2 is essential for palatogenesis mediated by the in- element and NF-kappaB activation and apoptosis pathways. J Biol Chem 279(15): teraction between Sox9 and mediator subunit 25. Nat Commun 2:251. 15652–15661. 25. Zou W, et al. (2011) The E3 ubiquitin ligase Wwp2 regulates craniofacial development 13. Jiang Z, Mak TW, Sen G, Li X (2004) Toll-like receptor 3-mediated activation of NF- through mono-ubiquitylation of Goosecoid. Nat Cell Biol 13(1):59–65. kappaB and IRF3 diverges at Toll-IL-1 receptor domain-containing adapter inducing 26. Sun Q, et al. (2006) The specific and essential role of MAVS in antiviral innate immune IFN-beta. Proc Natl Acad Sci USA 101(10):3533–3538. responses. Immunity 24(5):633–642.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1220271110 Yang et al. Downloaded by guest on September 26, 2021