An IκB Kinase-regulated feed-forward circuit prolongs inflammation

Summary

Loss of NF-κB signaling causes immunodeficiency while inhibition of NF-κB can be efficacious in treating chronic inflammatory disease. Inflammatory NF-κB signaling must therefore be tightly regulated, and while many mechanisms to downregulate NF-κB have been elucidated, there have only been limited studies demonstrating positive feed-forward regulation of NF-κB signaling. In this work, we use a bioinformatic and proteomic approach to discover that the IKK family of proteins can phosphorylate the E3 ubiquitin ligase, ITCH, a critical downregulator of TNF-mediated NF-κB activation. Phosphorylation of ITCH by IKKs leads to impaired ITCH E3 ubiquitin ligase activity and prolongs NF-κB signaling and pro-inflammatory cytokine release. Since genetic loss of ITCH mirrors IKK-induced ITCH phosphorylation, we further show that the ITCH^−/− mouse’s spontaneous lung inflammation and subsequent death can be delayed when TNF signaling is genetically deleted. This work thus identifies a new positive feed-forward regulation of NF-kB activation that drives inflammatory disease.

Introduction

Inflammatory signaling pathways are among the most tightly regulated in the body. Inflammatory responses must be strong enough to combat pathogens, but they must also be measured to avoid unnecessary host damage. As a major driver of inflammatory signaling, NF-κB transcriptional activation is a prime example of such tight pathway regulation. While the main products of classically activated NF-κB include potent cytokines and chemokines, the initial NF-κB transcriptional program is also designed to limit NF-κB activity (Hayden and Ghosh, 2012). Negative feedback proteins such as IκBα and A20 are among the first NF-κB-driven genes transcribed (Krikos et al., 1992; Shembade and Harhaj, 2012; Sun et al., 1993). However, transcriptional reprogramming takes time, and a more a rapid response is essential to properly tailor the strength and duration of inflammation (Perkins, 2006). To this end, posttranslational modifications such as phosphorylation and ubiquitination can quickly alter cellular activity for tight signaling regulation (Hunter, 2007; Tigno-Aranjuez et al., 2013). Posttranslational modifiers such as kinases and ubiquitin ligases can quickly change the activity of enzymes, affect localization of proteins, alter stability in regard to degradation and promote interaction of diverse proteins to fine-tune signaling such that the body can eradicate an offending pathogen while also limiting host damage (Gallagher et al., 2006; Prabakaran et al., 2012).

As a central hub for the NF-κB signaling pathway, the IκB Kinases (IKKs) are positioned to provide such a rapid response. While being most famous for the phosphorylation of the IκB proteins to activate the NF-κB transcription factors (Chen et al., 1996), they are also now recognized to be fast activators of feedback inhibition of this pathway. For example, IKKα phosphorylates TRAF4 to promote its stability and activity as an inhibitor of NOD2-mediated NF-κB (Marinis et al., 2012). IKKα also phosphorylates TAX1BP1 to recruit A20, RNF11, and ITCH, allowing formation of a ubiquitin-remodeling complex that downregulates TNF-mediated NF-κB activation (Shembade et al., 2008; Shembade et al., 2011). Similarly, IKKβ phosphorylates A20 to promote downregulation of NF-κB by enhancing its deubiquitination function (Hutti et al., 2007). However, this rapid response is not limited to feedback inhibition (Liu et al., 2012). It has recently been shown that IKKβ can phosphorylate IRF5 to induce an interferon response (Lopez-Pelaez et al., 2014; Ren et al., 2014), and it is also known that NF-kB can undergo a positive feed forward regulation through the paracrine and autocrine expression of IL-17 and IL-6 (Ogura et al., 2008). Thus, identification of novel IKK phosphorylation sites and study of the regulation of those sites continue to help us understand inflammatory responses.

Given this, we hypothesized that there are additional IKK substrates that might prolong NF-κB signaling. We therefore developed a dual bioinformatic/proteomic approach to rapidly identify novel substrates of the IKKs. Coupling data from peptide array analysis with a novel IKK-substrate motif antibody, we describe an approach to rapidly identify novel IKK substrates. We used this approach to identify and confirm phosphorylation of a novel IKK substrate ITCH, a HECT E3 Ubiquitin ligase. ITCH is present in endo-lysosomal compartments, an area shown to contain active IKK complexes, and functions with the E2 ubiquitin ligases, UbcH5, UbcH6 and UbcH7 to catalyze the formation of a wide variety of polyubiquitin chains including K11, K29, K48 and K63-linked polyubiquitination (Angers et al., 2004; Chastagner et al., 2006; Perry et al., 1998; Shembade et al., 2008; Tao et al., 2009; Wei et al., 2012). Biologically, ITCH is known to downregulate a variety of inflammatory signaling pathways including the RIG-I/MAVS pathway in antiviral signaling (You et al., 2009), the NOD/RIP2 pathway in intracellular bacterial signaling (Tao et al., 2009) and TNFα-mediated NF-κB activation in inflammatory signaling (Shembade et al., 2008). In this work, we find that IKK phosphorylation of ITCH inhibits its ability to downregulate signaling pathways and inhibit cytokine responses. Lastly, given that TNF is a major activator of IKKs, and that ITCH serves to downregulate of TNF signaling, we show that genetic loss of TNFR1 attenuates the pulmonary inflammatory phenotype of the ITCH^−/− mice. This work thus identifies a novel phosphorylation-dependent signaling cascade that serves to prolong inflammatory signaling.

Results & Discussion

IκB Kinases phosphorylate the HECT E3 Ubiquitin ligase ITCH on Serine 687

To identify novel substrates of IκB Kinases (IKKs) using a joint proteomics-bioinformatics approach, peptide substrate array data from our previous studies (Hutti et al., 2009; Hutti et al., 2007; Marinis et al., 2012) was used to generate polyclonal antibodies designed against the preferred IKK phospho-peptide motif (F/Y/M–X–pS–L/I/M) (Hutti et al., 2009; Hutti et al., 2007). We then quantified the relative positively and negatively selected amino acid preferences from the peptide substrate array data to generate a matrix such that proteomic databases could be searched for amino acid sequences matching these phosphorylation motifs (Obenauer et al., 2003). This paired approach allowed us to identify a large number of potential substrates via bioinformatic searches while also allowing us to quickly screen those substrates through the use of the IKK phospho-motif antibody. In initial testing, 11 substrates were screened including the known IKK substrate, TANK. Of these initial substrates tested, positive IKK phosphorylation was identified for 7 substrates. Among these seven substrates, ITCH, a WW domain-containing HECT E3 ligase family shown previously by our lab to downregulate the NOD2: RIP2 signaling pathway was identified (Fig. 1a). To determine if the IKKs directly phosphorylate ITCH, we conducted in vitro kinase assays with γ³²P-labeled ATP and recombinant ITCH. ³²P transfer by both IKKα and IKKβ to both ITCH and ligase-dead (C830A) ITCH was detected, indicating direct phosphorylation by IKKα and IKKβ that did not require the catalytic activity of ITCH (Fig. 1b; generation of recombinant bacterial ITCH and variants is shown in Supplemental Fig. 1a). Furthermore, co-transfection of ITCH or C830A ITCH with IKKα also revealed phosphorylation of ITCH, but did require the C2 domain and subsequent membrane localization of ITCH (Fig. 1c, upper blots). IKKβ and kinase inactive K44A IKKβ were also tested and showed similar results (Fig. 1c, lower blots). Additionally, as the IKK-related kinases, IKKε and TBK1, share similarities in kinase structure and preferred phosphorylation motif (Hutti et al., 2012; Hutti et al., 2009), both of these kinases induced phosphorylation of ITCH (Fig. 1d). Thus, all four IKKs phosphorylate ITCH. We further investigated IKK-mediated phosphorylation of ITCH in an endogenous setting. ITCH is known to have a strong function in T cells (Jin et al., 2013). For this reason, Jurkat T cells were treated with TNF. Samples were collected at the indicated time points and the IKK substrate antibody was used to immunoprecipitate proteins containing the phosphomotif. Probing for ITCH revealed the presence of phosphorylated ITCH that correlated both with IKK activation and with the phosphorylation of the known IKK substrates p105 and IκBα (Fig. 1e). Similarly, treatment of the lung epithelial cell line, A549, showed TNF-inducible phosphorylation of endogenous ITCH (Fig. 1f). Inhibiting both IKKα and IKKβ through the CRISPR mediated deletion of the IKK scaffolding protein, NEMO, showed loss of TNF-induced ITCH phosphorylation (Supplemental Fig. 1b). This IKK-induced phosphorylation can be detected in an endogenous setting, and in the case of TNF, is likely mediated by the IKKα/IKKβ/NEMO scaffolding complex.

Figure 1. ITCH is a novel substrate of IκB Kinases.

A. A schematic outline for finding novel IKK substrates is shown. IKK phospho-peptide array data revealed a consensus IKK phosphorylation motif. The motif generated was used for a Scansite bioinformatic search and coupled with a custom antibody generated against the phosphorylated motif and identified ITCH as a candidate IKK substrate.

B. Recombinant ITCH was subjected to ³²P in vitro kinase assay with recombinant IKKα and IKKβ in separate experiments. Both IKKα and IKKβ could phosphorylate ITCH.

C, D. HEK 293 cells were transfected with FLAG-tagged ITCH or C830A (catalytically inactive ITCH) and either active or kinase-inactive IKKα (C, upper blots), active IKKβ or kinase-inactive IKKβ (C, lower blots), active or inactive IKKi or active or inactive TBK1 (D). All IKK family members caused ITCH phosphorylation.

E, F. Jurkat T cells (E) or A549 lung epithelial cells (F) were treated with TNFα as indicated. Immunoprecipitation of IKK phospho-motif-containing proteins was performed and immunoblotting was performed. As a control irrelevant rabbit polyclonal antibody was used (F). Phosphorylation of ITCH matched the time course of IKK activation and IKK phosphorylation of the known substrates IκBα and p105.

To determine if the predicted phosphoacceptor on ITCH corresponded to the actual phosphoacceptor, we immunopurified IKK-phosphorylated ITCH from Flag-tag ITCH-transfected cells. Purified ITCH was subjected to mass spectrometric analysis (Fig. 2a). 251 peptides were analyzed encompassing 69% of the amino acids of ITCH. Phosphorylation occurred on 33% of peptides containing S687 (DLESIDPEFYNpSLIWVK). Mass Spec sequence analysis also identified an additional phosphoacceptor (S13) occurring within an IKK peptide motif (F/Y/M–X–pS–L/I/M). To further determine which of these two sites were detected by our IKK phospho-substrate antibody, site-directed mutagenesis was employed. While single mutations did not result in loss of detected phosphorylation, we found that mutation of both sites (S13A/S687A ITCH) abolished the signal (Fig. 2b). To differentiate the importance of these two potential phosphorylation sites, we conducted further sequence analysis and molecular modeling. While S687 was conserved through zebrafish, S13A was not conserved (Fig. 2c). Molecular modeling showed that Ser 687 lies within the N-terminal lobe of the HECT domain. This region of the HECT domain is known to make contact with E2 conjugating enzymes (Hatakeyama et al., 1997) and this region of ITCH has been published to make contact with the E2 ubiquitin ligase UBCH7 (Ingham et al., 2004; Schwarz et al., 1998). These findings lead to the testable hypothesis that IKK-mediated phosphorylation of ITCH on S687 affects its E3 ubiquitin ligase activity while phosphorylation of S13 does not.

Figure 2. ITCH is phosphorylated on Ser 687.

A. ITCH was co-transfected with IKKβ and immunopurified. Phosphorylated ITCH was excised from the Coomassie-stained gel and analyzed by mass spectrometry. 251 peptides from ITCH were identified with 69% coverage obtained. Time of flight histogram shows a mass to charge shift in 33% of peptides DLESIDPEFYNpSLIWVK, indicating phosphorylation at residue Ser 687.

B. HEK 293 cells were co-transfected with IKKβ and each of the following ITCH variants: WT, S13A, S687A, S13A/S687A. Mutation of both Ser 13 and Ser 687 abolished the phosphorylation signal in the presence of IKKβ.

C. Sequence alignment indicates conservation of S687 in zebrafish, while S13 is not conserved.

D. The HECT domain’s predicted molecular structure is shown with Ser 687 highlighted. Ser 687 lies within the N-terminal lobe of the HECT domain.

Ubiquitin ligase activity and downstream signaling is modulated in an ITCH S687D phosphomimetic mutant

To determine the effect of Ser 687 phosphorylation on ITCH’s E3 ligase activity, we generated a phosphomimetic mutant and used this mutant to perform in vitro ubiquitination assays. Multi-ubiquitin chains were detected in the presence of WT ITCH while no ubiquitin chains were detected in the presence of either ligase-dead C830A ITCH or phospho-mimetic S687D ITCH (Fig. 3a), suggesting that this phosphorylation site might negatively influence ITCH’s E3 ubiquitin ligase activity. We then performed cellular ubiquitination assays. HEK293 cells were transiently transfected with HA-tagged ubiquitin and ITCH variants as indicated. Lysates were generated and ITCH was immunoprecipitated. Western blotting indicated that both ITCH and S687A ITCH could autoubiquitinate while S687D ITCH or ligase-dead C830A ITCH could not (Fig. 3b). To then test if ITCH has impaired ubiquitin ligase activity on known substrates, we performed substrate ubiquitin assays with two known substrates of ITCH, DVL2 and RIP2. ITCH variants were co-transfected with HA-Ubiquitin and Omni-tagged RIP2 or DVL2 in separate experiments. ITCH and S687A ITCH were shown to have similar ubiquitination activity on both RIP2 and DVL2. Catalytically inactive C830A ITCH and phosphomimetic S687D ITCH showed greatly reduced ubiquitin chains on both of these proteins (Fig. 3c and 3d, upper blots). In a similar experiment without addition of HA-Ubiquitin, ITCH and S687A ITCH catalyzed the formation of ubiquitin chains using endogenous ubiquitin, while again S687D and C830A showed greatly reduced ubiquitination of RIP2 (Supp. Fig. 2). Both the TNF pathway and the WNT/β-catenin pathway are known to be downregulated by ITCH (Shembade et al., 2008; Wei et al., 2012). To determine the role of ITCH phosphorylation on these pathways, luciferase assays were performed with NF-κB and TCF/LEF reporter constructs. TNFα and WNT gene expression activity was strongly reduced with increasing amounts of ITCH, but remained largely unaffected in the presence of S687D ITCH (Fig. 3d). Comparable experiments were performed to investigate the effect of phosphorylation at the S13 phosphomotif of ITCH, however no differences were detected (Supplemental Fig. 3a,b). Likewise, an ITCH construct lacking the C2 domain and therefore lacking the ability to be phosphorylated had no effect on TNF-induced NF-kB activation (Supplemental Fig. 3c). Taken together, these findings suggest that the IKKs phosphorylate ITCH at residue S687 to inhibit ITCH’s ability to downregulate signaling pathways.

Figure 3. Ubiquitin ligase activity and downstream signaling is modulated in ITCH S687D phospho-mimetic mutant.

A. Recombinant ITCH, S687D ITCH, and C830A ITCH were added to a reaction mix that contained Omni-UBCH7, HA-ubiquitin, E1 activating enzyme and ATP as indicated. Reactions were incubated at 37° C for 30 min, then subjected to SDS/PAGE and immunoblotting. S687D ITCH showed loss of ubiquitination activity that was comparable to the ligase-dead (C830A) ITCH mutant.

B. HEK 293 cells were transfected with FLAG-tagged ITCH, S687D ITCH, S687A ITCH, or C830A ITCH with HA-tagged ubiquitin. Cell lysates were generated. ITCH was immunoprecipitated and immunoblotting was performed. S687D ITCH showed greatly decreased autoubiquitination activity.

C. HEK 293 cells were transfected with ITCH variants, HA-tagged ubiquitin and Omni-tagged DVL2 or RIP2 in separate experiments. DVL2 or RIP2 were immunoprecipitated in each respective experiment and subjected to Western blotting. S687D ITCH showed greatly decreased substrate ubiquitination.

D. HEK 293 cells were transfected with NF-κB Luciferase and CMV-renilla reporter constructs with either ITCH or ITCH S687D mutant expression constructs. After 16 hours, cells were treated with 10 ng/mL TNFα for 7 hours and harvested for a luciferase assay (upper graph). Separately, HEK 293 cells were transfected with TCF/LEF Luciferase and CMV-Renilla reporter constructs with ITCH or S687D ITCH ± DVL2. Cells were harvested for a luciferase assay 24 hours post-transfection. (Data are represented as mean ± SEM with *P<0.05.) ITCH could inhibit both TNF and β-catenin signaling, however, this activity was lost in the S687D ITCH variant.

ITCH phosphorylation causes increased TNF-induced pro-inflammatory cytokine release and loss of TNF signaling delays the pulmonary inflammatory phenotype in ITCH^−/− mice

To further investigate the downstream effects of phosphorylation at S687, ITCH expression was stably inhibited through shRNA knockdown in A549 cells (Fig. 4a left panels). These cells were subsequently transduced with RNAi-resistant lentiviral FLAG-tagged ITCH variants using mutagenesis of the 3^rd nucleotide in a codon in such a way that does not alter the amino acid sequence (Fig. 4a, right panels). Each reconstituted cell line was stimulated with TNFα to determine induced expression of IL-6 and IL-8 (Fig. 4b). Levels of IL-6 and IL-8 mRNA were similarly high in cells with vector only and S687D ITCH while greatly reduced in cells reconstituted with WT ITCH. To then determine if this finding were true in vivo, a genetic approach was utilized. ITCH^−/− mice develop pulmonary interstitial pneumonitis and consolidated peripheral inflammation of the alveolar space that ultimately leads to their death at approximately 6–8 months of age (Matesic et al., 2008). Because TNF-mediated phosphorylation of ITCH is functionally similar to genetic ITCH loss, we hypothesized that unrestrained TNF signaling might be influencing the inflammatory lung phenotype in the ITCH^−/− mice. To test this hypothesis, ITCH^−/− mice were bred with TNFR1^−/− mice to yield ITCH^−/−TNFR1^−/− mice. Lungs were harvested from mice aged 2, 4, and 6 months, and histopathological analysis was performed. Consolidated inflammation was most prominent at the peripheral, pleural alveolar space in the ITCH^−/− mice. This is significantly delayed and decreased in the ITCH^−/−TNFR1^−/− mice (Fig. 4c). WT, TNFR1^−/−, ITCH^−/− and ITCH^−/−TNFR1^−/− lung histology slides were then blindly scored to indicate differences in histology scores at 4 months and 6 months of age (Fig. 4d). ITCH^−/−TNFR1^−/− showed a significant decrease in lung pathology. To further show the delay in pathology, mice were aged for up to 300 days and survival was plotted. ITCH^−/−TNFR1^−/− mice had significantly greater survival time compared to ITCH^−/− mice (Fig. 4e, lower graph). These findings suggest that unrestrained TNF signaling partially underlies pathology in the ITCH^−/− mouse. Coupled with the fact that TNF-activated IKKs phosphorylate ITCH to halt ITCH’s inhibitory effect on TNF signaling (Figs. 1–3), these data suggest a positive feedback loop centered on ITCH phosphorylation.

Figure 4. Loss of TNFR1 partially complements the inflammatory phenotype in ITCH^−/− mice.

A. A549 cells were stably transduced with lentiviral shRNA constructs directed against ITCH. ITCH shRNA1 cell lines were then stably reconstituted with vector only or RNAi-resistant FLAG-ITCH and FLAG-S687D ITCH retroviral constructs. Western blots showed both ITCH expression knockdown as well as reconstitution.

B. Reconstituted ITCH shRNA1 cells were stimulated with 10 ng/mL TNFα for the indicated times. RNA was isolated and subjected to qRT-PCR to determine IL-6 and IL-8 gene expression. (Data are represented as mean ± SEM with *P<0.05.)

C. Representative images of ITCH^−/− and ITCH^−/−TNFR1^−/− lung histology at the pleural surface at 2, 4, and 6 months of age. While the ITCH^−/− mouse shows significant distal inflammation, consolidation and chronic inflammation, the TNFR1^−/−ITCH^−/− shows a significant delay in lung pathology.

D. Histopathology of the lungs was blindly scored at 4 (WT n=4; TNFR1^−/− n=4; ITCH^−/− n=7; ITCH^−/−TNFR1^−/− n=5) and 6 (WT n=4; TNFR1^−/− n=4; ITCH^−/− n=5; ITCH^−/−TNFR1^−/− n=6) months. TNFR1^−/−ITCH^−/− mice show significantly less lung pathology.

E. Survival of the mice was plotted on a Kaplan Meier curve. (ITCH^−/−: n=24; ITCH^−/−TNFR1^−/−: n= 22; TNFR1^−/−: n=22.) Survival time varied significantly between ITCH^−/− mice and ITCH^−/−TNFR1^−/− mice (median survival ratio=0.6863; log rank p-value of 0.0009).

As central kinases in inflammatory signaling cascades, the IKKs are poised to be central regulators of a diverse set of cellular responses. Through a proteomic-bioinformatic approach designed to identify novel IKK substrates, we have identified a novel IKK phosphorylation site on the E3 ubiquitin ligase, ITCH, which increases the duration of NF-κB signaling. This phosphorylation site lies in a region of ITCH that is required for ITCH’s enzymatic activity and IKK-mediated phosphorylation of this site greatly decreases ITCH’s E3 ubiquitin ligase activity. Biochemically, we hypothesize that since S687 is a conserved residue in the E2 binding region of the HECT domain of ITCH, phosphorylation may impair E2 binding or limit transfer of ubiquitin from the E2-ubiquitin complex to the HECT domain of ITCH.

The WW-HECT domain E3 ubiquitin ligase family, of which ITCH is a member, is subject to extensive regulation. These family members lie in a catalytically inactive state that requires phosphorylation or Ca²⁺ binding by the C2 domain to be activated (Mund and Pelham, 2009; Wang et al., 2010). Our data suggest that phosphorylation of S687 is not required to activate ITCH, but rather to deactivate it. Thus, it is not surprising that the ΔC2 domain mutant, a 200+ amino acid deletion mutant that relieves both auto-inhibition and cellular localization is neither phosphorylated nor inactivated by the IKKs (Fig. 1c, Supp. Fig. 3c). More surprising is the fact that a mutant that lacks the ability to be phosphorylated (S687A) is not hyperactive. This could be due to a number of factors, including an increased basal inhibitory state in the mutant and/or a limiting amount of E2-ubiquitin conjugates in the cell such that the rate of ubiquitin transfer can’t increase. Additionally, since Serine-687 is completely conserved in all WW-HECT family members and lies in an invariant region of the E2 binding domain, there may be subtle structural changes that make the E2-HECT interaction less effective. What is clear is that the activity of this family of proteins is heavily regulated in the cell and that when activated, ITCH downregulates a variety of signaling pathways including the WNT signaling pathway and the TNF signaling pathway. The work adds a new dynamic to the signaling pathway by which the IKKs can more globally influence diverse signal transduction cascades.

Experimental Procedures

Cell culture, transient transfection, immunoprecipitation, and Western blotting

HEK 293T, Jurkat and A549 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum with antibiotic/antimycotic solution (Invitrogen). Calcium phosphate transfections were carried out as previously described (Abbott et al., 2004). Cell harvest and immunoprecipitations (IPs) were conducted in lysis buffer (50 mM Tris [pH 7.5], 150 mM NaCl, 1% Triton X-100, 1 mM EGTA, 1 mM EGTA, 2.5 mM sodium orthophosphate, 1 mM β-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1 mM Na₃VO₄, 10 nM calyculin A in the presence of protease inhibitor cocktail [Sigma]). Protein G Sepharose beads (Invitrogen) or FLAG® agarose beads were added to lysates for immunoprecipitation. Beads were washed five times in lysis buffer and boiled in sample buffer prior to Western blotting on nitrocellulose membranes (Bio-Rad), as described (Abbott et al., 2007). Phospho IKK Substrate motif polyclonal antibody was generated against the peptide sequence, F/Y/M-X-pS-L/I/M, as previously described (Hutti et al., 2009). ITCH H110, Actin, and Omni antibodies (Santa Cruz Technology), Glutathione S-transferase (GST), Phospho IKKα/β, and DVL2, phospho-IkBa, IkBa and phospho-p105 antibodies (Cell Signaling Technology), anti-hemagglutinin (HA-11) (Covance), anti-Xpress™ antibody (Invitrogen), anti-FLAG® antibody and FLAG® agarose beads (Sigma) were all used according to manufacturer’s directions. 3XFlag DVL2 (WT) was purchased from Addgene (plasmid # 24802) (Narimatsu et al., 2009) and was placed in the pcDNA6His Max (Invitogen) Omni vector using standard restriction digest. FLAG-ITCH, HA-Ubiquitin, Omni-RIP2, GST-IKKα, GST-IKKα GST-IKKβ, GST-IKKε, GST-TBK1 and mutants were used as previously described (Hutti et al., 2009; Hutti et al., 2007; Tao et al., 2009). Mutant constructs S687D ITCH, S687A ITCH, and C830A ITCH were generated by QuikChange site-directed mutagenesis (Stratagene) and verified by sequence analysis. Recombinant 6x-His-UBCH7 GST-ITCH, and GST-ITCH mutants were grown in E. coli and purified using standard methodology. Recombinant full-length human GST-IKKα and GST-IKKβ was purchased from Millipore. Recombinant HA-Ubiquitin and XP-UBE1 were purchased from Boston Biochem.

In vitro kinase assay

Recombinant GST-ITCH or GST-C830A ITCH was added to GST-IKKα and 10 μCi of [γ-³²P] ATP per reaction. Reaction mix was incubated at 30°C for 30 min in kinase buffer (25 mM Tris [pH 7.5], 1 mM β-glycerophosphate, 1 mM DTT, 1 mM Na₃VO₄, 10 mM MgCl₂, and 100 μM ATP). Boiling and SDS-PAGE was followed by autoradiography.

Mass spectrometry

Mass spectrometry analysis was performed by the Beth Israel Deaconess Mass spec core facility (Boston, MA) under the direction of Dr. John Asara as described (Tigno-Aranjuez et al., 2010).

In vitro ubiquitin ligase assay

Recombinant GST-ITCH, GST-ITCH variants and 6XHis-UbcH7 were purified as described above, and the reaction was performed as previously described (Wilkins et al., 2004). Ube1 and HA-Ubiquitin from BostonBiochem. The reaction was stopped by addition of 2× sample buffer and boiling, followed by SDS PAGE and immunoblotting.

Luciferase assays

HEK 293T cells were calcium phosphate-transfected with CMV-Renilla and either a TCF/LEF luciferase reporter construct (M50 Super 8× TOPFlash, (Veeman et al., 2003) or an NF-κB luciferase reporter) construct along with varying amounts of FLAG-ITCH. M50 Super 8× TOPFlash was a gift from Randall Moon (Addgene plasmid #12456). Cells were lysed in 1× passive lysis buffer and assayed for Luciferase and Renilla activities (Promega Dual-Luciferase® Reporter Assay System). Luciferase values were normalized to Renilla and fold change was calculated relative to unstimulated cells in the absence of ITCH.

Lentiviral shRNA and retroviral RNAi-resistant ITCH reconstitution/CRISPR NEMO^−/− cells

A549 cells were stably transduced with ITCH shRNA lentiviral constructs in the pLKO vector (Sigma). ITCH sh1 and sh2 were directed at the sequences 5′-GCAGCAGTTTAACCAGAGATT-3′ and 5′-CCAGGAGAAGAAGGTTTAGAT-3′ within ITCH, respectively. ITCH shRNA knockdown was confirmed through Western blotting after puromycin selection. ITCH sh1-resistant FLAG-ITCH and FLAG-S687D ITCH were generated through point mutagenesis within the pBABE (hygro)-FLAG-ITCH vector (mutagenesis primers: forward, 5′-CTTCAAGGAGCAATGCAGCAGTTCAATCAGAGATTCATTTATGG-3′; reverse 5′-CCATAAATGAATCTCTGATTGAACTGCTGCATTGCTCCTTGAAG-3′). sh1-resistant constructs were stably transduced in the ITCH sh1 stable cell line and selected with hygromycin. NEMO^−/− CRISPR cell lines were generated via lentiviral transduction of a LentiCRISPRv2 construct (Sanjana et al., 2014) modified to contain the guide sequence 5′-GAGCGCCCTGTTCTGAAGGC-3′. lentiCRISPR v2 was a gift from Feng Zhang (Addgene plasmid #52961). Cells were selected with puromycin, then cloned to select for colonies with validated knockout. Six validated clones were pooled. Similarly, negative CRISPR control cells were virally transduced with a LentiCRISPRv2 construct containing the irrelevant guide sequence 5′-CGCGATAGCGCGAATATATT-3′.

Quantitative RT-PCR

RNA was isolated using RNEasy Qiagen kit and cDNA was generated using Qiagen’s Quantitect Reverse Transcription kit, according to manufacturer’s instructions. Real-time PCR was performed with primers generated to detect human IL-6 (forward, 5′-TCCACAAGCGCCTTCGGTCC-3′; reverse, 5′-GTGGCTGTCTGTGTGGGGCG-3′), IL-8 (forward, 5′-CCTGATTTCTGCAGCTCTGTG-3′; reverse, 5′-CCAGACAGAGCTCTCTTCCAT-3′), and GAPDH (forward, 5′-GACCTGACCTGCCGTCTA-3′; reverse, 5′-GTTGCTGTAGCCAAATTCGTT-3′), using iQ SYBR Green Supermix (Bio-Rad). The data shown are normalized to GAPDH.

Mice and histology scoring

ITCH^−/− mice (MRC Harwell; backcrossed over 10 generations onto the C57BL/6 mice from Jackson Laboratories) were crossed with C57BL/6 TNFR1^−/− mice (Jackson Laboratories) Lung tissue was harvested from mice at 2, 4, and 6 months of age from each genotype and scored blindly by a board certified Anatomic Pathologist (D.W.A.) The scoring system was based on the following parameters: arteriolar inflammation, bronchiolar inflammation, protein, pleural inflammation, and inducible bronchus-associated lymphoid tissue (iBALT) with a scoring range was 0–4 for each parameter. Animal studies were performed within a specific-pathogen-free animal facility under IACUC-approved protocols, Case Western Reserve University.

Graphs and statistical analyses

GraphPad Prism software was used to create graphs and perform statistical analyses. Significance on luciferase assays and quantitative RT-PCR were determined using a Paired Student’s t-test (NF-κB n=3; TCF/LEF n=4). Significance on histology scoring was determined performing multiple comparisons of One-way ANOVA.

Structural and sequence analyses

Clustal W2 was used for multiple sequence alignment of protein sequences obtained from NCBI Protein. PyMOL was used to highlight residue S687 within the ITCH HECT domain from NCBI Structure (MMDB ID: 94340).

Highlights.

• IKKs phosphorylate ITCH in a highly conserved region of the HECT domain

• E3 ubiquitin ligase activity is impaired in IKK-phosphorylated ITCH

• Impaired ITCH results in heightened TNF signaling

• Activated IKKs can affect multiple pathways through inhibitory ITCH phosphorylation