Dimethyl itaconate inhibits TNF-α induced NF-κB signaling pathway in human epithelial cells

Background: Dimethyl itaconate (DMI), a membrane-permeable derivative of itaconate, was found to moderate IL-17-IκBζ-induced skin pathology including psoriasis in mouse experiments . TNF-α induced NF-κB pathway, which controls a variety of immune and inammatory responses, was also proven to play a crucial role as mediator in psoriasis. However, whether DMI interacts with the TNF-α induced NF-κB pathway remains unclear. Results: Here we show that DMI inhibits TNF-α induced NF-κB transcriptional activities in dose-dependent manner in several human cell lines using dual luciferase assay and blocks the NF-κB nuclear entry. Moreover, DMI potently inhibits IKKβ dependent phosphorylation and degradation of IκBα in TNF-α induced activation of NF-κB pathway. We also demonstrate that DMI covalently binds to cysteine residue in IKKβ, a key regulator in NF-κB pathway, to suppress IKKβ activation and inhibit the canonical NF-κB pathway. Conclusion Our study presents a new mechanism for DMI as an anti-inammatory agent that may have therapeutic potentials in treating NF-κB related human inammatory diseases. Our results also suggest that itaconate produced by endogenous IRG1 may regulate NF-κB at post translation modication level, and the IRG1-itaconate-NF-κB axis could be targeted as a novel strategy for the treatment of IRG1-NF-κB mediated diseases. a the were by Western blotting using phospho-IKKα/β (Ser176/180) antibody. One representative experiment of three was shown (A). Quantitation of the level of phospho-IKKα/β was shown (B). (C) Western blot analysis of phospho-IKKα/ β(Ser176/180) in lysates of Hela cells untreated or different concentrations of DMI-pretreated and then stimulated with TNF-α (10 ng mL-1, 10 min). β-actin was used as loading control. Blot shown is representative of three independent experiments. (D) Histogram of intracellular phospho-IKKα/ β(Ser176/180) from (C).

DMI as an anti-in ammatory agent that may have therapeutic potentials in treating NF-κB related human in ammatory diseases. Our results also suggest that itaconate produced by endogenous IRG1 may regulate NF-κB at post translation modi cation level, and the IRG1-itaconate-NF-κB axis could be targeted as a novel strategy for the treatment of IRG1-NF-κB mediated diseases.

Background
Nuclear factor-κB (NF-κB) signaling pathway plays a critical role in immune and in ammatory responses, cell survival/proliferation, and cell/tissue homostasis maintenance. It controls the expression of hundreds of genes, including cytokines, chemokines, adhesion molecules and anti-apoptosis proteins 1 .
Its key element, NF-κB, is a transcriptional factor consisting of a family of ve proteins, including p50, p52, p65 (RelA), c-Rel and RelB. Each of them can form homodimers or heterodimers with another protein of a different type within the protein family 2 . In resting cell, NF-κB is arrested by IκB inhibitory proteins (for example, IκBα, IκBβ, IκB) in cytoplasm. Upon stimulation by various pro-in ammatory stimuli such as TNF-α, microbial infection, and some of chemotherapeutic agents, IκB kinase complex (IKK) is activated.
This complex is composed of IKKα and IKKβ catalytic subunits and a regulatory subunit, IKKγ/NEMO.
IKKβ is considered as the major subunit responsible for phosphorylation of IκBs. The activated IKK then phosphorylates IκBs and leads to its ubiquitination and degradation by 26S proteasome. NF-κB is then liberated and translocates to the nucleus and activates gene transcription 3 . Hence, the activation of IKKβ and phosphorylation of IκBs are the central points in NF-κB pathway. Under normal condition, NF-κB pathway is activated whenever needed and is terminated properly. However, under abnormal condition, it could be constantly activated, leading to a wide range of chronic in ammatory and autoimmune diseases, such as psoriasis, rheumatoid arthritis and Crohn's disease [4][5][6] .
The NF-κB pathway is considered as a crucial mediator in psoriasis, an in ammatory skin disease featured by cell hyperproliferation and differentiation. The level of active NF-κB is signi cantly elevated in psoriasis 7 . Many chemokines and cytokines involved in psoriasis, such as IL-17, TNF-α and IL-12/23, depend on NF-κB signaling. Therefore, several anti-psoriasis therapies, including TNF-α blockers like in iximab, etanercept, adalimumab and IL-12/23 inhibitors like ustekinumab are used to reduce the activity of NF-κB and related downstream elements 8 . However, the use of these antibody drugs is limited by treatment resistance, potential risk for adverse events and high cost. These limitations emphasize the need for additional treatment options for psoriasis. Small molecule drug always represents a more costeffective way for treatment of many diseases.
Recently, Bambouskova et al. discovered that dimethyl itaconate (DMI), a derivative of itaconate which is a newly discovered metabolite with signi cantly upregulated concentration under the in ammatory condition and has been reported to have anti-in ammatory property, has a signi cant effect on psoriasis by inhibiting the production of IκBζ induced by IL-17 9 . This indicates the potential therapeutic application of DMI on psoriasis. Recent studies have shown that itaconate and its derivatives can inhibit in ammation through NRF2 and also act via ATF3 to block in ammation driven by the IL-17-IκBζ axis 2,13,17 . More recently, they were found to attenuate the in ammatory response by impairing glycolysis 21,22 .
Nevertheless, none of them focused on the anti-in ammatory mechanism of itaconate and its derivatives on NF-κB signaling pathway which is important in in ammatory diseases including psoriasis.
DMI was rstly used as a membrane-permeable non-ionic form of itaconate to probe itaconate metabolism [10][11] . Treatment of macrophages with DMI has an inhibitory effect on in ammatory responses induced by LPS or LPS plus IFNγ 11 . Nevertheless, ElAzzouny et al. found that itaconate produced from DMI was not detectable in cells 12 , and the itaconate-mediated in ammatory inhibition effects may be directly induced by DMI. These results demonstrated the potential role of DMI in in ammatory diseases, especially in psoriasis. However, currently DMI was only shown to function on IL-17-IκBζ-mediated psoriasis 9 , whether it has effect on other model of psoriasis driven by cytokines, such as TNF-α remains unclear. As NF-κB pathway is the key element in pathogenesis of psoriasis, we hypothesize that it may also inhibit NF-κB pathway. Based on the differential gene expression analysis of bone marrow-derived macrophages (BMDMs) pretreated with DMI or vehicle and then stimulated with LPS from Lampropoulouon et al.'s work 11 , we found that DMI-pretreatment modulated the expression of several NF-κB-regulated genes, including IL-6, IL-1β and Cxcl11. This provides a strong evidence that DMI can affect NF-κB pathway. Moreover, recently, DMI was reported to ameliorate LPS-induced mastitis by activating MAPKs and Nrf2 and inhibiting NF-κB signaling pathways 13 . These indicate that DMI has strong relationship with NF-κB signaling pathway.
Here we show that DMI inhibits TNF-α-induced NF-κB signaling pathway which is crucial in the pathogenesis of psoriasis in several epithelial cells. The inhibitory effect of DMI on NF-κB signaling is exerted via targeting IKKβ by covalent binding to C412 in IKKβ both in vitro and in cells and inhibiting the activation process of IKKβ. This discovery not only provides the evidence that DMI can inhibit TNFαinduced NF-κB pathway, but also provides a novel probe to investigate the molecular mechanism of IKKβ. Furthermore, we found that itaconate can regulate NF-κB pathway at post translational modi cation level. We con rmed that itaconate could covalently modify several cysteines in IKKβ. Moreover, we demonstrated that itaconate produced in the cell can inhibit the NF-κB signaling pathway by overexpressing IRG1 that can produce itaconate. Therefore, our study provides new insights of itaconate and its derivative DMI's anti-in ammatory function.

Results
Dimethyl itaconate inhibits TNF-α induced NF-κB signaling pathway in epithelial cells DMI has been successfully used to moderate IL-17-IκBζ-driven skin in ammation in mouse model of psoriasis. As NF-κB plays crucial role in psoriasis, we investigated if DMI also acts as an antiin ammatory agent in psoriasis by inhibiting NF-κB signaling pathway. We rst used NF-κB luciferase assay to evaluate the effects of DMI on NF-κB-dependent transcriptional activity upon TNF-α stimulation.

Dimethyl itaconate inhibits TNF-α-induced nuclear translocation of NF-κB
We then tested whether DMI can abolish the TNF-α-induced nuclear translocation of p65. In this experiment, we used a SK-N-AS monoclonal stable cell line transfected with NF-κB/p65 labeled by red uorescent protein and nuclear H 2 B labeled by green uorescent were pre-incubated with 0.25 mM DMI for 1 h before stimulation with 10 ng mL -1 TNF-α at 37℃ (5% CO 2 ). The movement of NF-κB/p65 was captured every 5 min for 4 h as described before 14 . We measured the nuclear amounts of uorescent protein-RelA by time-lapse imaging adding DMI or not. About 120 live cells were quanti ed for each condition. We found that addition of DMI signi cantly reduced the percentage of activated cells (Supplementary Figure 3) and nuclear NF-κB intensity (Figure 2A, B). Collectively, these data suggested that DMI inhibited TNF-α-induced nuclear translocation of p65.

DMI inhibits TNF-α-induced NF-κB Activation by Targeting IKK
From NF-κB luciferase assay and time-lapse confocal imaging experiment, we showed that DMI inhibited the NF-κB signaling pathway. We next identi ed the molecular target of DMI. Because IKKβ plays central role in NF-κB signaling pathway, we rst evaluated the effect of DMI on IKKβ-dependent phosphorylation and degradation of IκBα in TNF-α-induced activation of NF-κB in Hela cells. As shown in Figure 3, phosphorylated IκBα and IκBα were detected in Hela cells pretreated with or without DMI for 12 h and then stimulated with TNF-α for indicated time by Western blot analysis using antiphospho-Ser32/Ser36 IκBα antibody and no phosphorylated IκBα antibody. As shown in Figure 3A-C, TNF-α induced phosphorylation and degradation of IκBα were in time-dependent manner as described before 15 . For the control group, in Figure 3A and 3B, it is clear that when the cells were stimulated with TNFα, the level of IκBα rapidly decreased, while the level of pIκBα increased to the highest level at 5 minutes and then slowly decreased along with time due to the low level of IκBα. We further calculated the ratio of pIκBα/IκBα (shown in Figure 3D) and the control group showed a tendency of continuing increase with time. For the DMI group, when the cells were stimulated with TNFα, the level of IκBα almost kept constant (shown in Figure 3A, 3B), while the levels of pIκBα were signi cantly lower compared with the control group at corresponding time points. For the ratio of pIκBα/IκBα (shown in Figure 3D), the ratio of the DMI group kept at a low level at all time points. These results suggest that DMI inhibits TNF-α-induced NF-κB activation by preventing IκBα phosphorylation and degradation by targeting IKKβ.
We further tested whether DMI can directly interact with IKKβ. We generated N-10xHis-tagged IKKβ recombinant protein from HEK293T cells transfected with a pCMV-N-His-IKKβ expression plasmid for further analysis. The N-10xHis-tagged IKKβ was captured by NTA agarose beads. The puri ed recombinant IKKβ protein was incubated with or without DMI for 12 h at 4 ℃, and then the mixtures were resolved by SDS-PAGE. As shown in Supplementary Figure Table  1). Neither of these modi cations was observed at any cysteine in vehicle-treated IKK-β.
To test whether the DMI can modify IKK-β in cell, we treated HEK293T cells overexpressing recombinant IKK-β with 0.25 mM DMI for 3h. After incubation, we extracted whole-cell lysate using NTA agarose beads and digested the protein with trypsin. As shown in Figure  Da mass shift occurred starting from y 5 to the y 12 fragment, indicating that the C412 residue was covalently modi ed by DMI. However, the modi ed molecule was itaconate rather than DMI. It was shown that DMI was rapidly degraded in cells without releasing itaconate 12 , hence it was possible that DMI covalently binds to IKKβ rst and the complex undergoes further esterase digestion. As the MMI modi ed IKKβ was also observed in vitro, the other possibility is the fragmentation may happen in the mass spectrometry detection. In addition to this, we also hypothesized that the inhibitory effect of NF-κB signaling pathway from DMI treatment was comprehensive results of both DMI and itaconate. We supposed that only a few DMI may convert to itaconate, and this part of itaconate alkylate cysteine residues on proteins rapidly. It has been discovered that IRG1, an enzyme that produces itaconate in cells, can suppress NF-κB signaling pathway by decreasing IKKα/β activation upon LPS stimulation 16 . We then tested whether itaconate can bind to IKKβ covalently in vitro. As shown in Supplementary  Figure 7A, B). Not surprisingly, itaconate produced by IRG1 inhibited TNF-α-induced NF-κB signaling pathway (Supplementary Figure 7B), which was consistent with the results before 16 . To eliminate the effect of IRG1 itself, we also used 4-octyl itaconate (OI), a recently described cell-permeable derivative of itaconate which can release itaconate in cells 17 , to treat cells and then detected the activation of NF-κB pathway. Cell luciferase assay and Western blot analysis showed that OI can inhibit TNF-α-stimulated NF-κB signaling pathway (for detail, see the Supplementary Figure 6 and Supplementary Figure 8). However, the inhibitory effect of OI on NF-κB signaling pathway was weaker than that of DMI on NF-κB signaling pathway. These results indicated that the inhibitory effect of DMI on NF-κB signaling pathway may from both DMI and itaconate. Despite of these possibilities, our results suggested that DMI treatment indeed modi ed IKKβ in cells.
Having identi ed that DMI can directly modify recombinant IKKβ in cells, we next explored how this modi cation affected IKKβ structure and activity. IKKβ protein exists as a dimer in solution. Each IKKβ monomer has a trimodular linear architecture: the N-terminal kinase domain (KD, 1-309), the central ubiquitin-like domain (ULD, 310-404), and the C-terminal dimerization domain (SDD, 408-664) 18 . IKKβ activity can be described as a kinase cycle of three states: poised, active, and inactivated. In the absence of stimulation, IKKβ was in the poised state 19 . After stimulation by TNF-α, the poised IKKβ is activated by phosphorylation on Ser177 and Ser181 in its activation loop. Then the activated IKKβ phosphorylates its substrate IκBα. In addition to this, the activated IKKβ also phosphorylates its own C terminus, thereby inducing conformation change that results in the decrease of kinase activity. Then the inactive hyperphosphorylated IKKβ becomes available by dephosphorylation. Among all these steps in the IKK kinase cycle, activation of IKKβ is the most important one. Furthermore, IKKβ autophosphorylation seems to be the key step in activation of IKKβ. Thus, we tested whether DMI can directly inhibit the activation of IKKβ. We monitored the level of active IKKβ in Hela cells stimulated with TNF-α pre-treated with or without DMI using Western blot analysis. As shown in Figure 5, DMI dose-dependently and time-dependently inhibited phosphorylation of Ser177/181 on IKKβ, the activated state of IKKβ. This result indicated that DMI indeed inhibits the activation of IKKβ.
Recent crystal structure studies revealed that IKKβ formed higher order homo-oligomers to phosphorylate its dimer partners. The oligomerization surface was a "V shaped" interface including the N-terminal KD-ULD portions and the proximal SDD from the two promoters. The double mutant I413A/L414A within the V-shaped interface was found to disrupt IKKβ activation in cells 20 . We found the modi ed residue, C412, in IKKβ was right in this "V-shaped" interface. Moreover, mutation of Cys-412 (C412A) in IKKβ could inhibit its activation. And then DMI had no obvious effect on C412A mutant comparing with wile type IKKβ (Supplementary Figure 9). These evidences all support that DMI inhibits the autophosphorylation of IKKβ by modifying C412 to disrupt the oligomerization surface of IKKβ. Our results also suggest that C412 in IKKβ provides a new and potent druggable binding site for IKKβ based drug design.

Discussion
The roles of metabolic regulation in diseases and human health-related processes have received increasing attention. In immune system, extracellular and intracellular signals regulate the activity of metabolic pathways, thus allowing cells to synthesize metabolites to facilitate their growth and to enable their functions. Different metabolic pathways, which correspond to different cellular functions, produce a large number of small molecule metabolites. These metabolites not only provide material and energy supplies for cells, but also serve as signal molecules to transmit information. In addition, because of their endogenous characteristics, they could be potentially applied in drug discovery. For example, dimethyl fumarate (DMF), a derivative of fumarate, is an U.S. Food and Drug Administration and European Medicines Agency approved immunomodulatory drug used to treat multiple sclerosis and psoriasis.
DMI is a derivative of itaconate which is discovered recently as an important metabolite. It is rstly used as the permeable itaconate surrogate to study the function of itaconate in cells. DMI was reported to reduce the in ammatory responses in LPS or LPS plus IFNγ stimulated macrophages 11 and to moderate the IL-17-Iκζ-driven psoriasis 9 . This highlights the potentials of DMI as a novel anti-in ammation agent. Recent research found that DMI, like DMF with electrophilic α, β-unsaturated moieties, could induce electrophilic stress by covalent conjugation with GSH and subsequently induce Nrf2 responses to inhibit the production of cytokines associated with in ammation 9 . It can also block the IκBζ expression through ATF3 9 . More recently, DMI has been reported to inhibit NF-κB signaling pathway in LPS-induced mastitis but without precise mechanism 13 . Most recently, another itaconate derivate OI was reported to exert antiin ammatory effects by limiting aerobic glycolysis 21 . In this study, we demonstrated that DMI blocks the nuclear translocation and transcriptional activity of NF-κB upon TNF-α stimulation. Particularly, DMI inhibited NF-κB-dependent responses by directly modifying Cys412 of IKKβ. This modi cation suppressed the activation of IKKβ, which demonstrated the essential role of Cys412 in IKKβ function.
Accumulating evidences show that the anti-in ammatory effect of DMI is complex progress integrating the metabolism and signaling pathway. In the present work, we studied the effect of DMI on IKKβ, a hub in the NF-κB pathway. Whether DMI also bind to other proteins related to NF-κB pathway need further study.
The molecular forms that DMI function in the cells are still controversial. Synthesis of methyl or ethyl ester analogues of polar carboxylate metabolites is a common approach used to deliver these metabolites intracellularly. These methyl or ethyl ester would be hydrolyzed by esterases in cells. This prodrug strategy has been applied in glutamate and fumarate intracellular delivery. Therefore, DMI was supposed to presumably deliver intracellular itaconate. But ElAzzouny et al. found that DMI was not metabolized into itaconate intracellularly 12 . This indicated the in ammatory inhibition effect may be directly induced by DMI. But no matter what form of DMI in cells to exert anti-in ammatory effect, it is true that DMI indeed functions. In our study, we demonstrated that DMI may be converted to itaconate to some extent. We have shown that another itaconate derivate OI which can release itaconate in cells also has the inhibitory effect of NF-κB signaling pathway. In addition, overexpression of IRG1 has the similar effect. So we demonstrated that the inhibitory effect of DMI on NF-κB signaling pathway was a comprehensive results of both DMI and itaconate.

Conclusion
In summary, we identi ed a new mechanism for the anti-in ammatory effect of dimethyl itaconate. We showed that DMI inhibited TNF-α-induced NF-κB activation in several epithelial cells. And we con rmed this result using Western blotting experiment to monitor the IκBα phosphorylation and degradation and time-lapse confocal imaging experiment to monitor the TNF-α-induced nuclear translocation of p65.
Moreover, we showed that DMI inhibits NF-κB signaling pathway by blocking IKKβ activation via covalent binding to its Cys412 residue, which may be further explored and used to develop IKKβ covalent inhibitor. These ndings may lead to the development of DMI as a new anti-in ammatory agent for treatment of various autoimmune diseases. NF-κB/p65 nuclear translocation uorescence SK-N-AS cells with p65 labeled with mcherry and H 2 B labeled with EGFP which was constructed as previously described 14 were cultured in DMEM (Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1% Non-Essential AminoAcid (NEAA, Gibco) in a humidi ed atmosphere with 5%CO 2 at 37℃.1.5x10 5 SK-N-AS cells were seeded on 35-mm glass-bottom dishes (Invitro Scienti c Products, Inc.) in 3 mL medium. The day of experiment, the medium was changed to fresh DMEM containing indicated concentration of DMI. After 3 h incubation at 37℃ in cell culture incubator, cells were stimulated by TNFα (10 ng mL -1 ). All the images were captured by Volocity software (PerkinElmer) using an inverted TiE microscope (Nikon). All channels were collected every 5min for 4h. For image quanti cation, we used the method described before.

Western blots
Hela cells lysed in RIPA lysis buffer system(CWBIO, CW2334) and heat-denatured at 95℃ for 5 min in reducing sample buffer(CWBIO). Proteins were separated on 10% polyacrylamide gradient gels and transferred onto PVDF membranes (0.45μm pore size, Millipore). No-speci c binding was blocked with 5% BSA, and membranes were incubated with primary antibody followed by incubation with anti-rabbit-HRP (1:5000; #7074) or anti-mouse-HRP (1:5000; #7076) from Cell Signaling Technology and Western ECL substrate (Millipore). Membranes were imaged using the Fusion FX7 (Spectra) system (VILBER). βactin run on the same blot was used as a loading control.

Protein Expression and Puri cation
The HEK293T cells were grown to 80% con uency in 15-cm dished (Corning) at 37℃ in DMEM supplemented with 10% FBS. Before transfection, cells were changed to fresh medium. Viafect reagent and puri ed pCMV-N-10xHis-IKKβ plasmid were added to cells in a ratio