|Home | About | Journals | Submit | Contact Us | Français|
Dysregulation of NF-κB activity contributes to many autoimmune and inflammatory diseases. At least nine pathways for NF-κB activation have been identified, most of which converge on the IκB kinases (IKKs). Although IKKs represent logical targets for potential drug discovery, chemical inhibitors of IKKs suppress all known NF-κB activation pathways, and thus lack the selectivity required for safe use. A unique NF-κB activation pathway is initiated by protein kinase C (PKC) that is stimulated by antigen receptors and many growth factor receptors. Using a cell-based high throughput screening (HTS) assay and chemical biology strategy, we identified a 2-aminobenzimidazole compound, CID-2858522, which selectively inhibits the NF-κB pathway induced by PKC, operating downstream of PKC but upstream of IKKβ, without inhibiting other NF-κB activation pathways. In human B cells stimulated through surface immunoglobulin, CID-2858522 inhibited NF-κB DNA-binding activity and expression of endogenous NF-κB-dependent target gene, TRAF1. Altogether, as a selective chemical inhibitor of the NF-κB pathway induced by PKC, CID-2858522 serves as a powerful research tool, and may reveal new paths towards therapeutically useful NF-κB inhibitors.
Members of the nuclear factor-kappa B (NF-κB) family of transcription factors play crucial roles in the control of many physiological and pathological processes, including host-defense, immune responses, inflammation, and cancer 1. In mammals, at least nine pathways leading to NF-κB activation have been elucidated, including; (i) a “classical” pathway induced by Tumor Necrosis Factor (TNF) and many TNF-family cytokine receptors, involving degradation of Inhibitor of NF-κB-alpha (IκB-α) and release of p65-50 NF-κB heterodimers 2; (ii) an “alternative” pathway activated by selected TNF-family receptors (e.g. CD40, Lymphotoxin-β Receptor, BAFF Receptor) involving p100 NF-κB2 proteolytic processing to generate p52, a preferred heterodimerization partner of NF-κB-family member RelB; (iii) the Toll-like receptor pathway for NF-κB induction, involving TIR domain-containing adapters and IRAK-family protein kinases 3; (iv) a pathway activated by exogenous RNA, involving Helicard/Mda5, RIG-I and mitochondrial protein MAVS, which is of importance for host defenses against viruses 4; (v) a DNA-damage pathway involving PIDD, a target of p53 5; (vi) NLR/NOD-family proteins, cytosolic proteins that oligomerize in response to microbial-derived molecules, forming NF-κB-activating protein complexes; (vii) Ultraviolet (UV) irradiation and some DNA-damaging drugs, which stimulates NF-κB activation via mechanism involving C-terminal phosphorylation of IκB-α 6, 7 (viii) oncogenic fusion proteins comprised of portions of cIAP2 and mucosa-associated lymphoid tissue-1 (MALT1), which drive NF-κB activation via interactions with TRAF2 and TRAF6 8 and (ix) a pathway induced by ligation of B-cell or T-cell antigen receptors, as well as many growth factor receptors, involving a cascade of interacting proteins that includes caspase recruitment domain-containing membrane-associated guanylate kinase protein-1 (CARMA1, Bimp3), Bcl-10, and MALT (Paracaspase), Caspase-8, and other proteins (reviewed in 9). The core event upon which most of these NF-κB activation pathways converge is activation of Inhibitor of κB Kinases (IKKs), typically comprised of a complex of IKK-α, IKK-β, and the scaffold protein, IKK-γ/NEMO 2. In all but the “alternative” NF-κB pathway, IKK activation results in phosphorylation of IκB-α, targeting this protein for ubiquitination and proteasome-dependent destruction, thus releasing p65/p50 NF-κB heterodimers from IκB-α in the cytosol, and allowing their translocation into the nucleus where they initiate transcription of various target genes.
The NF-κB pathway activated by antigen receptors is critical for acquired (as opposed to innate) immunity, contributing to T- and B-lymphocyte activation, proliferation, survival, and effector functions. Dysregulated NF-κB activation in lymphocytes can contribute to development of autoimmunity, chronic inflammation, and lymphoid malignancy 9, 10. The NF-κB activation pathway linked to antigen receptors is initiated by certain PKCs and involves the aforementioned CARMA/Bcl-10/MALT complex. Formation of this complex is stimulated by PKC-mediated phosphorylation of CARMA proteins. Contributions to the PKC-activated NF-κB activation mechanism are also made by Caspase-8, apparently forming heterodimers with c-FLIP and inducing proteolytic processing of c-FLIP 11. In T and B cells, this pathway is initiated by Protein Kinase C (PKC)-theta and PKC-beta, respectively, leading ultimately to IKK activation through a mechanism possibly involving lysine 63-linked polyubiquitination of IKK-gamma 12. In addition to antigen receptors, many growth factor receptors also initiate NF-κB activation via stimulation of various PKCs.
Although IKKs represent logical targets for potential drug discovery, chemical inhibitors of IKKs suppress all known NF-κB activation pathways, and thus lack the selectivity required to inhibit antigen receptor and growth factor receptor responses without simultaneously interfering with innate immunity and creating broad immunosuppression with considerable risk of infection 13. We therefore devised a chemical biology strategy for identification of small molecule chemical probes that selectively inhibit antigen receptor and growth factor receptor-mediated NF-κB activation, and describe herein 2-aminobenzimidazole compounds that inhibit at a point between PKCs and IKKs, without blocking other NF-κB activation pathways. These compounds thus provide unique research tools for interrogating the PKC-initiated pathway for NF-κB induction and may represent a starting point for eventually generating pathway-selective drugs with utility for autoimmunity and cancer.
Our strategy for compound library screening entailed using phorbol ester ([PMA]) and Ca2+-ionophore ionomycin to achieve PKC activation. For convenience, we used HEK293 epithelial cells, in which it has previously been shown by siRNA-mediated gene silencing and transfection of dominant-negative mutants that PMA/Ionomycin-induced NF-κB activation is dependent on CARMA1, Bcl-10, and MALT-1 14-17. HEK293 were stably transfected with a luciferase-reporter gene driven by a NF-κB-responsive promoter, and the responsiveness of this integrated promoter to various NF-κB inducing stimuli was confirmed, including PMA/Ionomycin and TNF (Supplemental Figure 1). Using these cells, 96- and 384- well plate-based high throughput screening (HTS) assays were established, with good assay performance characteristics (Z’ >0.5) (PubChem AID = 1384) (http://pubchem.ncbi.nlm.nih.gov/assay/assay.cgi?aid=1384) (http://pubchem.ncbi.nlm.gov/assay/assay.cgi?aid-465&loc=ea_ras). We initially screened 53,280 chemical compounds at an average concentration of 5 μM, of which 519 primary hits were obtained (based on cut-off of 50% inhibition). Of these, 248 were confirmed upon repeated testing (Figure 1). Counter-screening for compounds that inhibit TNFα-induced activation of the reporter gene eliminated 202 compounds and testing for cytotoxicity of the HEK293 reporter cell line discounted 2 additional compounds, leaving 46 candidates. Fresh stocks of these chemicals were obtained from the vendor, of which 11 showed suppression of PMA-induced NF-κB reporter gene activity. Finally, these 11 candidates were tested in an orthogonal assay in which PMA-induced secretion of Interleukin-8 (IL-8) by the HEK293 reporter cell line was measured, thus examining an endogenous NF-κB target gene, leaving only 1 candidate compound, CID-2858522 (SID-17450324 or ChemBridge-5653914), a substituted 2-aminobenzimidazole (Figure 1; Figure 2A). CID-2858522 also inhibited NF-κB activation induced by another PKC activator, phorbol dibutryate (PDBu), with cellular potency of ≤ 0.1 μM (IC50) using NF-κB reporter gene assays and using assays where secretion of NF-κB-induced cytokine IL-8 was measured (Figure 2E). While inhibiting NF-κB reporter gene activity, CID-2858522 did not interfere with PKC-induced AP-1 or NFAT reporter gene activity in 293 cells (Supplemental Figures 2 and 3). Nor did the compound inhibit luciferase (http://pubchem.ncbi.nlm.nih.gov/assay/assay.cgi?aid=1384)(http://pubchem.ncbi.nlm.nih.gov/assay/assay.cgi?aid=1269&loc=ea_ras), thus excluding non-specific activity.
To characterize the selectivity of the hit, CID-2858522 was also tested in seven other NF-κB pathways. CID-2858522 did not inhibit the NF-κB activation induced by over expression of CD40, CD4, NOD1, NOD2, XIAP/TAB, IAP2/MALT1 or induced by either Doxorubicin (an inducer of DNA damage) or retinoic acid (an inducer of RIG-1) (Supplemental Figure 4), confirming the specificity of CID-2858522 for the PKC-initiated pathway.
To further investigate the activity and specificity of CID-2858522, we then tested the compound in other cell lines stimulated by various stimuli. CID-2858522 also slightly but reproducibly inhibited IL-2 production in a Jurkat T-cell line (Figure 3) and consistently suppressed proliferation of mouse B-cell splenocytes (Figure 4) induced by anti-IgM but failed to inhibit the NF-κB induced by lipopolysaccharide (LPS) (IL-6 secretion measured in THP.1 cell cultures), anti-Lymphotoxin-β receptor (NF-κB luciferase measured in HeLa cells), γ-Tri-DAP (IL-8 secretion was measured in MCF-7 cell cultures) and MDP (IL-6 measured in THP.1 cell cultures) (Supplemental Figures 5-8).
Testing >250 analogs of CID-2858522 using the HEK293-NF-κB-luc reporter cell line revealed the fundamental structure-activity relationship (SAR) of this compound, in which various moieties within the compound structure were interrogated, resulting in numerous structurally related analogs that lost all activity or had markedly decreased activity, and approximately 10 analogs with comparable or equipotent activity, but no analogs with clearly superior activity (data to be published elsewhere).
Compound CID-2858522 is a substituted 2-aminobenzimidazole (Figure 2A). Representative data are provided in Figure 2, contrasting the activity of CID-2858522 with another compound derived from library screening, CID-2998237 (Figure 2A). In the HEK293 cell line used for primary screening, CID-2858522 suppressed NF-κB reporter gene activity in a concentration-dependent manner, with IC50 ~70 nM and with maximum inhibition achieved at 0.25-0.5 μM (Figure 2B). In contrast, this compound did not inhibit TNF-induced NF-κB-reporter gene activity at concentrations as high as 4 μM, thus demonstrating selectivity for the NF-κB pathway activated by PMA/Ionomycin (Figure 2B). Cell viability assays indicated that CID-2858522 was not toxic to HEK293 cells at concentrations ≤ 8 μM. Moreover, CID-2858522 also potently inhibited PMA/Ionomycin-induced NF-κB reporter gene activity in transient transfection assays, where the NF-κB-luciferase reporter gene activity was measured from an episomal plasmid (not shown), thus excluding an impact of the chromosomal integration site on measured activity. Similar results were obtained with another “hit” compound CID-2998237, though this compound was less potent at suppressing PMA/Ionomycin-induced reporter gene activity and had also modest inhibition of TNF-induced NF-κB activity (Figure 2C).
The orthogonal assay for PMA/Ionomycin-stimulated NF-κB reporter gene activity in the HEK293 engineered cell line used for primary screening proved to be a key differentiator of true-positive versus false-positive compounds, and demonstrated the importance of not relying exclusively on luciferase-based reporter genes. Figure 2D compares CID-2858522 with a false-positive compound, CID-2998237, showing that CID-2858522 inhibited PMA/Ionomycin-stimulated IL-8 production in a concentration-dependent manner, with IC50 <0.1 μM and maximum suppression achieved at ~1 uM, whereas CID-2998237 had minimal effect on IL-8 production at concentrations as high as 4 μM. While several compounds derived from library screening demonstrated similar characteristics with respect to suppression of NF-κB reporter-gene activity induced by PMA/Ionomycin but not TNF (n = 11 of 114,889 total compounds screened), only CID-2858522 suppressed PMA/Ionomycin-induced IL-8 secretion.
Similar results for CID-2858522 were obtained when phorbol dibutryate (PDB) was substituted for PMA (Figure 2E), thus extending the observations to an alternative PKC-activating phorbol ester. The IC50 values for suppression of PDB-induced NF-κB reporter gene activity and PDB-induced IL-8 production by HEK293 cells were ~70 nM and ~100 nM, respectively.
CID-2858522 also suppressed PMA/Ionomycin-stimulated NF-κB DNA-binding activity (Figure 2F), as measured by an immunoassay wherein nuclear NF-κB-family proteins were captured on beads displaying oligonucleotides with NF-κB-binding sites and p65 Rel-A was detected using a specific antibody. Suppression was evident at concentrations as low as 60 nM, with IC50 ~ 125 nM (Figure 2F). Similar results were obtained by electromobility gel-shift assay (EMSA) (Supplemental Figure 9). PMA/Ionomycin-induced p65-RelA DNA-binding activity was also inhibited by the PKC inhibitor, Bisindolylmaleimide I, used here as a control. CID-2858522 did not block p65-DNA binding activity induced by TNF (data not shown), thus demonstrating pathway selectivity.
Because NF-κB can be activated by at least 9 known pathways, we next triggered each of these pathways in HEK293 cells by either stimulation with appropriate cytokines, transfection with plasmids, or stimulation with various agents that initiate each NF-κB activation pathway (Summarized in Figure 1 [right side]; data shown in Supplemental Figures 4-8). The activity of CID-2858522 was compared with an IKK inhibitor, BMS-345541 as a control, relying on the ability of chemical inhibitors of IKKs to block nearly all NF-κB activation pathways. First, we stimulated the TLR-pathway by transfection with a CD4-TLR4 fusion protein, in which the extracellular domain of CD4 is fused with the transmembrane and cytosolic domain of TLR4, and anti-CD4 antibody (rather than the natural ligand, lipopolysaccharide [LPS]) is used to activate TLR4. TLR4 induced robust NF-κB reporter gene activity (> 50 fold increase), which was suppressed by IKK inhibitor BMS-345541, but not by CID-2858522 or by PKC inhibitor, Bisindolylmaleimide I. Second, the “alternative” NF-κB pathway was stimulated by over-expressing CD40 in HEK293 cells. CD40-induced NF-κB reporter gene activity was potently suppressed by the IKK inhibitor but not by CID-2858522 or by the PKC inhibitor. Third, the NLR-dependent NF-κB pathway was stimulated by over-expressing NOD1 (NLRC1) or NOD2 (NLRC2) in the HEK293-NFκB-luc cells. NOD1 and NOD2 induced 6-7 fold increases in NF-κB-luciferase reporter gene activity, which were inhibited by the IKK inhibitor but not by CID-2858522. Fourth, IAP-initiated pathways for NF-κB activation were induced by transfecting 293-NF-κB-luciferase cells with plasmids encoding either cIAP2/MALT oncoprotein or XIAP plus TAB. While an IKK inhibitor effectively suppressed these IAP-driven pathways, CID-2858522 did not. Note that XIAP has recently been implicated in the NF-κB pathway activated by NOD1 and NOD2 18. Fifth, the DNA-damage-inducible pathway for NF-κB activation was triggered by stimulating HEK293-NF-κB-luc cells with doxorubicin, which induced a ~12-fold increase in NF-κB activity in these cells. Again, the IKK inhibitor but not CID-2858522, suppressed NF-κB activity. Finally, the retinoic acid (RA)-inducible pathway involving RIG-I 19 was induced by treating cells with all-trans-retinoic acid. RA induced a modest ~3-fold increase in NF-κB activity in HEK293 cells, which was significantly suppressed by the IKK inhibitor but not by CID-2858522 (Supplemental Figure 3). Thus, when taken together with the data showing that the “classical” NF-κB pathway activated by TNF was not inhibited by CID-2858522 (Figure 2), the above results demonstrate CID-2858522 uniquely suppresses the NF-κB pathway initiated by PKC activators.
In T cells, the antigen receptor stimulates several signal transduction pathways that converge on the IL-2 gene promoter, including NF-κB, NF-AT, and AP-1 (reviewed in 20). To evaluate the effects of CID-2858522 on TCR-initiated, NF-κB-driven events in lymphocytes, we employed Jurkat T-leukemia cells, which have been utilized extensively as a model for studying TCR-signaling leading to IL-2 gene expression 21. For these experiments, Jurkat cells were stimulated with either anti-CD3 (to activate the TCR complex) and anti-CD28 (co-stimulator) or with PMA/Ionomycin, in the presence or absence of CID-2858522, IKK inhibitor, or PKC inhibitor, then IL-2 production was measured 24 hrs later in culture supernatants. Both anti-CD3/CD28 and PMA/ionomycin stimulated marked increases in IL-2 production by Jurkat T-cells, with CD3/CD28 more potent than PMA/Ionomycin (Figure 3A). The IKK inhibitor partially suppressed PMA/Ionomycin-induced IL-2 production, and potently inhibited (~90% suppression) anti-CD3/CD28-induced IL-2 production at concentrations ≤ 10 μM (Figure 3B). The PKC inhibitor suppressed IL-2 production by 80-90% in Jurkat cell stimulated with either CD3/CD28 or PMA/Ionomycin at concentrations < 0.5 μM (Figure 3C). In contrast, CID-2858522 only partially suppressed IL-2 production by CD3/CD28- and PMA/Ionomycin-stimulated Jurkat cells (Figure 3D). The suppression of IL-2 production by Jurkat cells by CID-2858522, IKK inhibitor, or PKC inhibitor was not due to cytotoxicity (Figure 3E).
We also assessed effects of CID-2858522 on cytokine production induced by other stimuli. CID-2858522 did not suppress IL-6 production by THP.1 monocytes stimulated with TLR4 agonist LPS, IL-8 production stimulated by NOD1 agonist γ-TriDAP in MCF7 breast cancer cells, or NF-κB luciferase activity induced by anti-Lymphotoxin-β Receptor antibody in HeLa cells (Supplemental Figures 5-8), all of which involve other NF-κB activation pathways.
NF-κB plays roles in antigen receptor-driven lymphocyte proliferation 10. We therefore tested the effect of CID-2858522 on mouse lymphocyte proliferation induced by anti-CD3/CD28 or anti-IgM antibodies, measuring 3H-Thymidine incorporation. Anti-CD3/CD28 and anti-IgM significantly induced ~80-fold and ~8-fold increases, respectively, in DNA synthesis in cultures of murine lymphocytes (Figure 4A). The IKK and PKC inhibitors suppressed lymphocyte proliferation in a concentration-dependent manner, inhibiting B-cells (anti-IgM) (IC50 ~2 μM for IKK inhibitor; ~0.2 μM for PKC inhibitor) more potently than T-cells (anti-CD3/CD28) (IC50 ~4 μM for IKK inhibitor; ~1.5 μM for PKC inhibitor) (Figures 4B, C). In contrast, CID-2858288 inhibited anti-IgM-induced lymphocyte proliferation in a concentration-dependent manner, with an IC50 ~2 μM, while having minimal effect on anti-CD3/CD28, suggesting that the NF-κB inhibitory mechanism of CID-2858288 was more prominent in B-cells than in T-cells. However, because CD3/CD28 stimulation induced stronger proliferative responses than anti-IgM, we cannot exclude a quantitative rather than qualitative explanation for this observation.
To further evaluate the effect of CID-2858522 on antigen receptor signaling in lymphocytes, we examined its effect on leukemia B-cells from patients with chronic lymphocyte leukemia (CLL). Stimulation with biotinylated anti-IgM (crosslinked using Streptavidin) resulted in expression of TRAF1 (Figure 4E), an endogenous target of NF-κB 22. Adding CID-285252 to cultures of anti-IgM-stimulated CLL cells inhibited TRAF1 induction, in 3 of 3 cases measured at 24 hr after stimulation (Figure 4E and Supplemental Figure 10). Levels of Actin and TRAF6, which are not regulated by NF-κB, did not show any change (Supplemental Figure 10), thus showing selectivity and confirming equivalent protein loading. As a control, CLL cells were also treated by a structurally related but inactive 2-aminobenzimidazole analog, MLS-0292123, which did not inhibit PMA/Ionomycin-induced NF-κB luciferase activation or IL-8 production in HEK293 cells (Supplemental Figure 11), showing that MLS-0292123 did not inhibit TRAF1 expression (Figure 4E, 4F and Supplemental data). As a positive control, CLL cells were also treated with a PKC inhibitor, Bisindolylmaleimide I, which also inhibited TRAF1 expression. The effects of CID-2858522 on the capacity of anti-IgM to induce CLL-cell expression of TRAF1 were not due to cytotoxicity during the time-frame analyzed, as confirmed by measuring ATP levels (data not shown). In addition to the indirect evidence of NF-κB activation using TRAF1 expression, CID-2858522 also showed direct suppression on p65-DNA binding activity in human CLL cells induced by anti-IgM, while its inactive analog, MLS-0292123 did not (Figure 4F). Thus, CID-2858522 inhibited antigen receptor-stimulated NF-κB activation in primary leukemia B cells.
Protein kinases play critical roles in NF-κB activation. PKCs are proximal kinases in the NF-κB pathways activated by PMA/Ionomycin and by T-cell and B-cell antigen receptors, while the IKKs are distal kinases operating in the terminal segments of these and most other NF-κB activation pathways 23. We therefore tested whether CID-2858522 inhibits members of these kinase families using in vitro kinase assays. For these experiments, we tested PKC-beta and PKC-theta (the PKC family members implicated in TCR/BCR signaling), and IKK-beta (a component of the IKK complex) and IKK-epsilon (not shown). At concentrations up to 8 μM, CID-2858522 failed to suppress these kinases, while known PKC and IKK inhibitors and the broad-spectrum kinase inhibitor staurosporin (STS) afforded potent inhibition (Supplemental Figure 12). Thus, CID-2858522 did not directly inhibit PKC-beta, PKC-theta, or IKK-beta. Recognizing that in vitro kinase assays do not always detect the activity of chemical inhibitors, we also explored the effect of CID-2858522 on endogenous PKC activity by analyzing effects on cellular PKC substrates. For these experiments, 293 cells were pre-incubated with CID-2858522 or a PKC inhibitor, then cells were stimulated with PMA and Ionomycin for 1 hr before preparing cell lysates and analyzing PKC substrates using a phospho-specific antibody. While the PKC inhibitor Bisindolmaleimide I suppressed PKC-induced phosphorylation events, our compound CID-2858522 did not (Supplemental Figure 13).
In addition to performing conventional in vitro kinase assays for PKCs and IKKs, we also performed a kinome screen using a high throughput screening method, KINOMEscan™, which is an active-site dependent competition binding assay in which human kinases of interest are fused to a proprietary tag (Ambit). The amount of kinase bound to an immobilized, active-site directed ligand was measured in the presence and absence of the test compound 24. Of 353 protein kinases surveyed, CID-2858522 at 10 μM suppressed only 3 protein kinases by more than 50%: Raf (57% inhibition), TLK1 (70% inhibition), and JAK2 (53 % inhibition) (Supplemental Table 1), none of which are clearly implicated in NF-κB regulation. Thus, we were unable to find evidence that CID-2858522 inhibits protein kinases previously implicated in regulating NF-κB.
Based on these kinase screens, we deduced that CID-2858522 operates somewhere between PKCs and IKK to inhibit the NF-κB pathway involved in signaling by antigen receptors and many growth factor receptors, which is known to include CARMA-family proteins, Bcl-10, MALT, TRAF6 (which binds Ubc13 to induce lysine 63-linked polyubiquitination of IKKγ/NEMO), IKKγ, and Caspase-8 9, 12. To characterize the effects of CID-2858522 on these possible targets of the antigen receptor pathway for NF-κB activation, we first evaluated PMA-induced phosphorylation of Carma1, by phospho-specific antibody immunoblotting, finding no effect of CID-2858522 on this molecular event that initiates formation of the CBM complex (Figure 5A). This result provided further evidence that CID-2858522 does not inhibit PKC in intact cells. Next, we performed co-immunoprecipitation (co-IP) experiments, assessing the interactions of Bcl-10, MALT, TRAF6, IKKγ, and Caspase-8 with either CARMA1 or CARMA3 in transfected HEK293 cells, before and after stimulation with PMA. PMA induced or increased association of CARMA1 or CARMA3 with each of these proteins, which was inhibited by a PKC inhibitor, Bisindolylmaleimide I, but not by CID-2858522 (Figure 5B-E and Supplemental Figure 14). Thus, CID-2858522 did not disrupt the formation of CARMA/MALT1/Bcl-10 (CMB) complex induced by PMA/Ionomycin in either cells or lysates.
Caspase-8 plays an essential role in antigen receptor-mediated NF-κB activation 25. It was recently reported that MALT1 interacts with Caspase-8 and activates this protease upon antigen receptor activation 11. We confirmed that, in HEK293 cells, caspase-8 activation is required for NF-κB activation, as z-ITED-fmk (a peptidyl caspase-8 inhibitor), or caspase-8 siRNA significantly inhibited NF-κB luciferase activation induced by PMA/Ionomycin (data not shown). We then examined whether CID-2858522 interferes with caspase-8 participation in this pathway. PMA induced caspase-8 recruitment to the CARMA complex in HEK293 cells. The interaction was inhibited by a PKC inhibitor but not by CID-2858522 (Supplemental Figure 15). The caspase-8 p43/41 processing intermediate was generated in HEK293 cells after PMA/Ionomycin treatment. Proteolytic processing of caspase-8 was suppressed by a PKC inhibitor but not by CID-2858522 (Supplemental Figure 16). We then assessed the effects of CID-2858522 on PMA-induced proteolytic processing of c-FLIP, a Caspase-8-mediated event recently shown to be required for antigen receptor-mediated NF-κB activation 11. Immunoblot analysis of lysates from HEK293 cells following stimulation with PMA/Ionomycin showed processing of c-FLIP (Figure 5F), which was completely inhibited by the PKC inhibitor but not affected by CID-2858522. Thus, CID-2858522 failed to inhibit Caspase-8 activation and c-FLIP processing.
Finally, we examined IKK-β phosphorylation 26. Phosphorylation of IKK-β was induced by PMA/Ionomycin in HEK293 cells and was significantly inhibited by CID-2858522, but not by its inactive analog, MLS-0292123 (Figure 5G). In contrast, CID-2858522 failed to inhibit TNF-α-induced IKK-β phosphorylation, indicating pathway selectivity. We conclude from these studies that CID-2858522 inhibits PMA/Ionomycin-induced NF-κB at a point downstream of CBM complex formation, caspase-8 activation and c-FLIP processing, but upstream of IKK-β phosphorylation.
Chemical inhibitors of NF-κB have been widely sought for potential use as therapeutics for autoimmunity, inflammation, and cancer 13. However, the most pharmaceutically tractable of the NF-κB-activating targets, the IKKs, represent a shared component of nearly all known NF-κB activation pathways and thus lack selectivity. In this regard, NF-κB activity is required for innate immunity and host-defense against microorganisms and various viral and bacterial pathogens. In addition to impaired host defense, broad-spectrum suppression of NF-κB pathways may reduce basal NF-κB activity and interfere with the function of NF-κB as a survival factor, leading to potentially toxic side effects. For example, IKK-β knockout mice die at mid-gestation from uncontrolled liver apoptosis 27. In addition to potentially providing for novel therapeutic agents, development of pathway-selective inhibitors could lead to highly useful research tool compounds for interogating which pathways are important for specific cellular responses.
Using a chemical biology strategy, we devised chemical library screens for inhibitors that selectively inhibit the NF-κB activation pathway induced by PKCs. This pathway is uniquely involved in acquired immunity (rather than innate immunity), and has been linked to numerous autoimmune diseases and some types of lymphomas and lymphocytic leukemia 28. NF-κB is also induced via PKC by many growth factor receptors. In this reagard, PKC hyperactivity has been associated with some solid tumors 29, and thus the pathway interrogated here may also be relevant to a variety of malignancies. The NF-κB activation pathway linked to PKCs is known to involve proteins unique to this pathway among the nine known NF-κB activation pathways – namely, CARMA (Bimp)-family proteins, Bcl-10, and MALT (reviewed in 12). Upon phosphorylation of CARMA1 by PKC in the context of antigen receptor signaling, these proteins form a complex, which recruits TRAF6, an E3 ligase that binds Ubc13, resulting in lysine 63-linked poly-ubiquitination of IKKγ/NEMO, resulting in IKK activation 30. Caspase-8 is also recruited, resulting in proteolytic processing of c-FLIP, an event required for antigen receptor-induced activation of NF-κB 11. The components of this complex required for IKK activation may not be completely known and an active complex has not been reconstituted in vitro using purified components, thus making biochemical screens difficult. For this reason, a cell-based strategy for chemical library screening was the only practical option.
Using HEK293 cells containing an NF-κB-driven reporter gene stimulated by PMA/Ionomycin, followed by an orthogonal screen in which we measured levels of the protein product of an endogenous NF-κB target gene (e.g. IL-8) secreted by these same cells, we screened 114,889 compounds, finding only one that had the desired properties, namely CID-2858522. This substituted 2-aminobenzimidazole compound potently inhibits NF-κB reporter gene activity and IL-8 production induced by PKC activators in HEK293 cells, with IC50 < 0.1 uM, while failing to inhibit NF-κB reporter gene activation by agonists of the other NF-κB activation pathways (Figure 1). CID-2858522 also suppressed anti-IgM-stimulated proliferation of murine B-lymphocytes, as expected for an antagonist of the NF-κB activation pathway activated by B-cell antigen receptors. Because CID-2858522 inhibits NF-κB activation induced by phorbol esters and antigen receptors, it cannot be argued that the compound somehow interferes with uptake of PMA or other PKC-activating phorbol esters. Also, CID-2858522 did not inhibit PKC-mediated phosphorylation of various endogenous substrates in intact cells, arguing against a direct or indirect inhibitory effect on PKCs.
The observation that CID-2858522 only partially suppressed CD3/CD28- or PMA/Ionomysin-induced production of IL-2 by Jurkat T cells is consistent with the fact that NF-κB is only one of several transcriptional regulators of the IL-2 gene, which also include NF-κB, NFAT, and AP-1 20. We documented that CID 2858522 inhibited NF-κB while failing to suppress AP-1 or NFAT reporter gene activity induced by PKC. Furthermore, given that a variety of NF-κB-activating cytokines were elaborated upon stimulation of cultured lymphocytes with antibodies cross-linking CD3 (TCR) or surface IgM (BCR), it is perhaps not surprising that CID-2858522 only partially suppressed anti-IgM-induced proliferation of primary B-cells and had little effect on anti-CD3/CD28-induced T-cell proliferation. In contrast, an IKK inhibitor essentially completely suppressed lymphocyte proliferation at concentrations of ~ 5 μM, consistent with its ability to neutralize nearly all known NF-κB activation pathways. CID-2858522 also inhibited anti-IgM-induced expression of the endogenous NF-κB target gene, TRAF1, in CLL B-cells. In this regard, the TRAF1 gene promoter contains four NF-κB target sites and a TATA-box, but essentially no other recognizable transcriptional elements 22, thus making it a good surrogate marker of NF-κB activity in primary cells.
Although the mechanisms involved in antigen receptor-mediated NF-κB activation (upstream of PKC activation) in T cells and B cells are distinct, the downstream events following PKC activation share great similarity. Knockout mice models showed that CARMA1, Bcl-10 and MALT1 are required for antigen receptor-induced NF-κB activation and proliferation of both T cells and B cells 16, 31. However, CARMA1 mutant mice exhibited normal T but impaired B cell development 32 and MALT1 deficiency has only mild effects on B cell activation MALT1 33, indicating that the signal transduction apparati by which antigen receptors stimulate NF-κB downstream of PKC activation in T cells versus B cells are not identical. In this regard, it is also possible that antigen receptors and other upstream activators of PKCs induce NF-κB activation by more than one pathway, with CID-2858522 inhibiting only one of them. In this regard, it will be interesting to explore whether various lymphocyte subsets differ in their reliance on the NF-κB-activation pathways targeted by CID-2858522.
The mechanism by which CID-2858522 suppresses PKC-induced NF-κB activity remains to be determined. We mapped at least one site of action of this compound downstream of PKCs and upstream of IKK-β. PKCs induce phosphorylation of CARMA1, an event that was not inhibited by CID-2858522. This compound also neither inhibited PMA-induced recruitment of Bcl-10, MALT, TRAF6, Caspase-8, or IKKγ to CARMA1/CARMA3, nor did it inhibit caspase-8 or FLIP proteolytic processing. The active compound however selectively inhibited IKK-β phosphorylation induced by PkC activators but not TNFα, suggest CID-2858522 acts upstream of IKK-β. However, we cannot exclude the possibility that CID-2858522 was more than one site of action within the PKC-driven pathway for NF-κB activation, including acting at steps downstream of IKKβ.
The CARMA family proteins includes 3 members in mammals, which each contain a N-terminal CARD domain followed by a coiled-coil domain, a PDZ domain, a SH3 domain, and a C-terminal guanylate kinase-like (GUK) domain 34-36. Predominantly expressed in spleen, thymus, and peripheral blood leukocytes (PBL), CARMA1 has been definitively implicated in antigen receptor signaling. In contrast, CARMA3 is expressed in broad range of tissues but not in spleen, thymus, or PBL 37 and CARMA2 is expressed only in placenta. Suppression of selected members of the CARMA family could provide another plausible explanation for partial inhibition by CID-2858522 of events such as IL-2 production by CD3/CD28- or PMA/Ionomycin-stimulated Jurkat cells.
In summary, using a chemical biology approach, we have identified the first selective chemical inhibitor of the PKC-initiated NF-κB activation pathway. This compound and its active analogs provide novel research tools for elucidating the role of this NF-κB pathway in cellular responses, and may pave the way for future therapeutic applications of specific inhibitors of selective pathways involved in pathogenic activation of NF-κB.
Phorbol myristic acetate (PMA), Ionomycin, muramyl dipeptide (MDP), Retinoic Acid (RA), Doxorubicin and γ-Tri-DAP were from Sigma-Aldrich (St. Louis, MO), phorbol dibutryate (PDBu), PKC inhibitor (Bisindolylmaleimide I), and IKK inhibitor (BMS-345541) were from Calbiochem (Gibbstown, NJ). Anti-mouse-CD3, anti-mouse-CD28, anti-mouse-IgM were obtained from Biomeda (Foster City, CA). Anti-human CD3, anti-human CD28 and anti-mouse-IgG antibody were from R&D System (Minneapolis, MN). Anti-human TRAF6 antibody has been described 38.
Plasmids encoding HA-IKK-γ 39, XIAP 40, HA-TAK1, TAB1 41, CD4-TLR4 42, CD40 43, NOD1, NOD2 44, cIAP1/MALT 45, Caspase-8 and Caspase-8 (C360S) 46 and TRAF6 47 have been previously described. Myc-CARMA1 and CARMA3 were gifts from Dr Xin Lin (University of Texas, M. D. Anderson Cancer Center).
HEK293 cells were co-transfected with pUC13-4xNFκB-Luc and p-TK-puromycin-resistance plasmids. Stable clones were selected by culture in Dulbecco’s Modified Eagle’s Media (InVitrogen) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Hyclone), 1% v/v penicillin-streptomycin (InVitrogen) containing 1 μg/mL puromycin. Individual clones were tested for responsiveness to PMA/Ionomycin- and to TNF-induced NF-κB reporter gene activity, and a clone was selected for HTS.
Chemical libraries were screened using the cellular NF-κB luciferase reporter assay, including a ChemBridge library (San Diego, CA) having 50,000 compounds, Microsource Spectrum collection (Gaylordsville, CT) having 2,000 compounds, the LOPAC library (Sigma) having 1,280 compounds, and NIH library having 61,609 compounds at the time of screening.
NF-κB-luciferase expressing HEK293 cells were seeded at 105 per well in white 96 well plates (Greiner Bio-One) in 90 μl of DMEM and incubated overnight. 10 μl of compound-containing solution was added to each well (final 1.5 μg/ml in 1% DMSO) using a liquid handler (Biomek™ FX; Beckman Coulter). After 2 h incubation, cells were stimulated using 11 μl of a PMA (final 100 ng/ml; Calbiochem) and ionomycin (final 50 ng/ml; Calbiochem) in DMEM. Cell plates were incubated for 16 h before media was removed and 40 μl of 0.5x passive lysis buffer (Promega Corp.) was added to cell plates. Plates were allowed to stand at room temperature for at least 15 minutes before adding 40 μl of 0.125x luciferin substrate (Promega Corp.) to each well. Plates were analyzed within 30 seconds with a Criterion Analyst™ using the luminescence method (0.1 second read/well).
To counter-screen for inhibitors of the TNF pathway, HEK293-NFκB-Luc cells were seeded at 105 cells per well in 90 uL medium in white 96-well plates (Greiner Bio-One) and cultured overnight, before adding compounds (5 μL in medium) to cells. After 2 h incubation, 5 μL TNF (200 ng/ml) (R&D Systems) was added (final concentration 10 ng/ml) and cells were incubated for 16 h. Luciferase activity was measured using Britelite kit (Perkin Elmer). To counter-screen for inhibitors of NF-κB induced by TLR4, CD40, NOD1, NOD2, cIAP2/MALT, or XIAP/TAB, the 293-NF-κB-luc cells cultured in 96-well plates as above were pretreated with compounds for 2 h and then transfected using Lipofectamine 2000 with various plasmids including pcDNA3 (“empty vector” control) or plasmids encoding CD4-TLR4, CD40, NOD1, NOD2, cIAP2/MALT, XIAP/TAB, using 0.2 uL of transfection reagent containing 100 ng DNA per well. Cells were cultured in medium containing CID-2858522 or other compounds and luciferase activity was measured 48 h later. Alternatively, 293-NF-κB-luciferase cells were cultured with 16 μM all-trans-retinoic acid for 48 hrs or 2 μM doxorubicin for 48 hrs before measuring luciferase reporter gene activity.
The counter screen for inhibitors of luciferase was performed in 96 well white plates (Greiner Bio-one) containing 45 μL per well of ATPlite solution and luciferase (Perkin Elmer). Compounds diluted in 5 μL phosphate-buffered saline (PBS) were added at 8 uM final concentration. Reactions were then initiated by addition of 50 μL 160 nM ATP (Sigma) in PBS and luciferase activity was measured 2 h later using a luminometer (LJL Biosystems, Sunnyvale, CA).
Cell viability was estimated based on cellular ATP levels, measured using ATPlite kit (Perkin Elmer). Cells at a density of 105/mL were seeded at 90 μL per well in 96-well white plates and cultured overnight. Compounds were added (5 μL in medium) to wells and cells were cultured for 16 h, Finally, 50 μL ATPlite solution was added to each well and luminescence activity was read using a luminometer (LJL Biosystems, Sunnyvale, CA).
Human IL-2 or IL-8 levels in culture medium were measured by Enzyme-Linked Immunoadsorbant Assays (ELISAs), using BD OptEIA ELISAs (BD Biosciences, San Diego, CA), according to the manufacturer’s protocol, using 96-well ELISA plates (BD Biosciences) and measuring absorbance within 30 minutes of initiating reactions using a SpectraMax 190 spectrophotometer (Molecular Devices).
Nuclear extracts were prepared from 10 cm2 plates of confluent cells using a kit (Active Motif, Carsbad, CA). The total protein content of nuclear fractions was quantified by the Bradford method, followed by storage at −80°C. NF-κB DNA-binding activity was measured in nuclear extracts (10 ug protein) using an immunoassay method (TransAM Kit [Active Motif]) employing 96 well plates coated with double-strand oligodeoxynucleotides containing NF-κB consensus binding site (5′-GGGACTTTCC-3′) and anti-p65 antibody, which was detected by secondary horseradish peroxidase (HRP)-conjugated antibody, using a colorimetric substrate with absorbance read at 450 nm within 5 minutes using a spectrophotometer, SpectraMax M5 (Molecular Devices, Sunnyvale, CA)
EMSA assays were performed using LightShift Chemiluminescent EMSA Kit (Thermo Scientific, Rockford, IL) following the manufacturer’s protocol. Briefly, nuclear extracts normalized for protein concentration (1 mg/mL) were incubated with EMSA binding buffer, poly-dI-dC and BSA for 20 min on ice. Then 2 ul biotin-NF-κB probe was added and incubated at room temperature for 30 min. The samples were then resolved by gel-electrophoresis using 4-20% precast non-denaturing polyacrylamide gels in 0.5 × TBE (Bio-Rad, Hercules, CA), then transfered to positively charged nylon membranes (Amersham Biosciences UK limited, Little Chalfont, United Kingdom). The DNA was crosslinked to membranes using UV Stratalinker 2400 (Stratagene, La Jolla, CA). The biotin-NF-κB oligo was detected using Chemiluminescent Nucleic Acid Detection Module Kit (Thermo Scientific, Rockford, IL), with exposure to x-ray film.
Splenocytes were isolated from normal Balb/c mice and red blood cells were removed using a mouse erythrocyte lysis kit (R&D Systems, Minneapolis, MN). Splenocytes were suspended in RMPI-1640 medium supplemented with 10% FBS, 1 % penicillin-streptomycin, and 1 mM L-glutamine. Cells were diluted into 2 × 106 cells/ml and 200 μL were seeded in round bottom 96-well plates (Greiner bio-one) and incubated at 37 °C in 5% CO2 and 95% relative humidity. Cells were pretreated with compounds or DMSO diluted in medium for 2 h, then treated with 0.3 μg/mL anti-CD3/anti-CD28 or 3 μg/ mL anti-IgM antibodies for 48 h, prior to adding 1 μCi [3H]-Thymidine (MP Biomedical, Solon, OH) for 12 h. Cells were transferred to fiberglass filters (Wallac, Turku, Finland) using a FilterMate Harvester (Perkin Elmer), dried, and [3H]-incorporation into DNA was quantified by scintillation counting (Betaplate Scint, Perkin Elmer) and a MicroBetaTrilux LCS and luminescence counter (Perkin Elmer).
Peripheral blood mononuclear cells from CLL patients were obtained under IRB approval from whole blood by Ficoll density gradient centrifugation and cultured with RPMI 1640 Medium supplemented with 10% FBS and antibiotics. Cells were ≥90% leukemia B cells, as assessed via flow cytometry.
PKC-beta, PKC-theta and IKK-beta in vitro kinase assays were performed using the HTScan Kinase Assay Kit (Cell Signaling, Danvers, MA) according to manufacturer’s protocols. A panel of >300 kinases was screened by Ambit, Inc 24.
This project was supported by grants from the CLL Global Research Foundation and the NIH (X01 MH077633201, U54 HG005033, P01 CA081534 and U54 HG003916). We thank The Scripps Research Institute Screening Center for screening the NIH library, Xin Lin (University of Texas M. D. Anderson Cancer Center), Guy Salvesen (Burnham), Robert Cramer (Bristol-Myers-Squibb, Princeton, NJ), and Masao Seto (RIKEN) for reagents, and Melanie Hanaii and Tessa Siegfried for manuscript preparation. Dedicated in memory of Eric Dudl, who died of cancer while researching ways to combat cancer.
Supporting Information: Supporting information is available on-line, which includes supplemental data concerning NF-kB pathway analysis and analysis of components of the Carma/Bcl-10/MALT complex.