The central role that NF-κB plays in a vast range of human malignant and non-malignant pathologies has catalysed an intensive effort by the pharmaceutical industry and academic laboratories over the past two and a half decades to develop a specific NF-κB inhibitor for clinical indication in these diseases [
2,
3,
11,
105]. In 2006, a survey had already counted no fewer than 750 candidate therapeutics designed to target the NF-κB pathway, and this number is certain to have grown considerably over the past decade [
3]. These compounds comprise a disparate variety of chemical classes, including peptidomimetics, small molecules, small interfering RNAs and microbial products and their derivatives. Many of them function as general inhibitors of NF-κB signalling, while for many others the mode of action is poorly understood [
3,
21,
105]. Yet, notwithstanding the tremendous progress made in recent years towards unravelling the intricate signalling networks governing the NF-κB pathway and its functions, and the intensive efforts and mighty investments committed to develop a specific, clinically useful NF-κB inhibitor, the output of pharmacological interventions has been disappointingly scant, with a continuing dismaying absence of a specific NF-κB or IKK inhibitor in the clinical anticancer armamentarium. The insurmountable challenge, having precluded the clinical success of these NF-κB-targeting approaches, has been to achieve contextual selective inhibition of the NF-κB pathogenetic activity, while preserving the essential physiological functions of NF-κB. By contrast, the efforts so far have resulted in drug candidates often causing the global suppression of NF-κB and its pleiotropic and ubiquitous functions, leading to severe dose-limiting toxicities.
3.1. Inhibitors Operating Upstream of the IKK Complex
NF-κB can be activated by a multitude of stimuli, which initiate distinct signalling pathways that converge on the IKK complex (
Figure 1 and
Figure 2). From a clinical standpoint, this paradigm provides a significant opportunity for therapeutic interventions aimed at interfering with pathogenic NF-κB activation (
Figure 3). An important area of drug discovery in this context has been the TNF-R superfamily, which governs diverse physiological processes, including inflammation, cell survival and lymphoid organogenesis, and drives the pathogenesis of chronic inflammatory diseases as well as multiple cancer types [
108,
109]. TNF-Rs comprise a family of 29 structurally-related receptors, which are bound by 19 ligands of the TNF superfamily [
108]. Due to the important roles that TNF-Rs play in widespread human pathologies, there has been a great deal of interest over the past few decades in developing therapeutics targeting these receptors or their ligands [
110]. The first such a therapeutic has been infliximab, a TNF-α-specific neutralising antibody, which was approved in 1998 for the treatment of Crohn’s disease, followed by the approval in the same year of etanercept [
108]. Both drugs are currently in clinical use for the treatment of rheumatoid arthritis, psoriasis, psoriatic arthritis, and other forms of chronic arthritis, and infliximab is also in clinical use for Chron’s disease and ulcerative colitis [
108]. However, patients treated with these drugs often experience significant side effects, including fevers, chills, nausea, shortness of breath, tachycardia and hypotension. Indeed, the dose-limiting toxicities and immunosuppressive activities of TNF-R signalling inhibitors have precluded the broader clinical development of these agents beyond chronic inflammatory diseases, in particular in the area of oncology [
109]. Notwithstanding, a few molecules interfering with TNF-R signalling have found indication in niche areas of oncology. For instance, the human TNF-α analogues, tasonermin (Beromun), has been clinically approved as an adjunct therapy to surgery for sarcoma to prevent or delay amputation and to treat unresectable soft-tissue sarcoma of the limbs [
111]. Likewise, brentuximab (Vedotin, Adcetris), a toxin-conjugated chimeric antibody targeting the TNF-R-family receptor CD30 (also known as TNFRSF8), is approved for the treatment of Hodgkin’s lymphoma and anaplastic large-cell lymphoma (ALCL) (
Table 1), two cancer types that express particularly high surface levels of CD30 [
112].
In B-cell lymphoma and leukaemia, the NF-κB pathway can be inhibited by therapeutic agents targeting proximal signalling events downstream of the BCR. Upon antigen engagement, the BCR triggers a signalling cascade that involves receptor-associated CD79A/B heterodimers, SRC-family protein tyrosine kinases and Burton tyrosine kinase (BTK), which, as in the case of TCR-induced signalling, leads to the downstream the assembly of a PKC signalosome (mediated by PKCβ, rather than PKCθ, as in T cells) and the CBM signalling complex (
Figure 1c) [
65]. BTK is an essential signalling intermediate in the BCR-induced pathway leading to NF-κB activation and B-cell survival [
113]. Notably, while BTK is expressed in virtually all cells of the haematopoietic lineage, except for T cells and plasma cells, its functions in NF-κB activation and cell survival appear to be dispensable outside of B-cell lineage [
114]. Consequently, BTK has been developed into an effective therapeutic target upstream of IKK in various types of B-cell malignancy, including chronic lymphocytic leukaemia (CLL), mantle-cell lymphoma (MCL), follicular lymphoma (FL), DLBCL and acute lymphoblastic leukaemia (ALL) [
113]. The first-in-class oral BTK inhibitor, ibrutinib (PCI-32765), has demonstrated impressive clinical responses in clinical trials as a single agent and has been subsequently approved by the FDA for the treatment of refractory MCL (November 2013), CLL (February 2014) and Waldenström’s macroglobulinemia (WM; January 2015) [
115]. Ibrutinib irreversibly binds to cysteine residue, C481, in the active site of BTK, thereby inhibiting BTK phosphorylation on T223 and resulting in loss of BTK function, NF-κB inhibition and induction of tumour-cell apoptosis [
114]. Owing to BTK’s restricted pattern of expression, ibrutinib is generally relatively well tolerated, causing for the most part only transient adverse effects, such as diarrhoea, nausea, vomiting, hypertension, urinary and upper respiratory tract infections, fatigue, arthralgia, pyrexia, and peripheral oedema [
116]. However, secondary resistance is almost inevitable, and tumours with oncogenic NF-κB-pathway alterations affecting signalling events downstream of BTK, such
CARD11 mutations and
A20 mutations and deletions, are naturally refractory to this agent, thereby limiting its clinical utility, especially in patients with aggressive NF-κB-pathway mutated lymphomas (
Table 1) [
113].
The gene encoding the adaptor protein, MYD88, which mediates NF-κB and MAPK activation downstream of all TLRs, with the exception of TLR3 [
117], is recurrently mutated in haematological malignancies, such as DLBCL, WM and CLL, where it induces constitutive NF-κB and STAT3 activation, thereby promoting cancer-cell survival and oncogenesis [
117,
118]. These findings provide a strong rationale for therapeutically targeting TLR signalling in certain types of lymphoma and leukaemia. Congruently, preclinical studies have demonstrated that the antisense oligonucleotide TLR inhibitor, IMO-8400, which specifically targets TLR7, TLR8 and TLR9, is effective in diminishing the growth of WM and DLBCL xenografts, driven by gain-of-function MYD88 mutations [
118]. A phase I/II trial of IMO-8400 is ongoing in patients in WM and DLBCL (
Table 1), and second generation TLR 7/TLR 8/TLR9 inhibitors are currently in development. Another strategy aimed at therapeutically targeting TLR signalling in cancer is directed at the downstream protein kinases, IRAK1/4. This strategy has demonstrated preclinical efficacy in tumours harbouring MYD88 mutations, and preclinical studies of IRAK1/4 small-molecule inhibitors have reported encouraging preliminary results in melanoma [
119], myelodysplastic syndrome (MDS) [
120] and T-cell acute lymphoblastic leukaemia (T-ALL) [
121,
122].
The non-canonical NF-κB pathway plays an important role in the pathogenesis of multiple myeloma, where it is constitutively activated by recurrent genetic alterations, including
NIK amplifications and
c-IAP1/2 and
TRAF3 deletions [
10,
12,
14]. Recent studies suggest an additional role for non-canonical NF-κB signalling in other types of malignancy, such as DLBCL [
45]. Within this signalling pathway, NIK is an especially attractive target, owing to its central role in controlling IKKα phosphorylation and p100 processing (
Figure 3). Two NIK small-molecule inhibitors, which were recently developed by Amgen, AM-0216 and AM-0561 [
123], have demonstrated significant therapeutic activity in multiple myeloma cells, in vitro [
124,
125,
126]. However, further studies are required to evaluate their potential therapeutic efficacy, in vivo.
c-IAP proteins play an important role in tipping the balance between canonical NF-κB activation and inhibition of non-canonical signalling. Upon TNF-R1 stimulation, receptor-associated c-IAP proteins contribute to recruit the LUBAC complex by ubiquitinating various signalling intermediates, resulting in NF-κB activation [
127]. By contrast, in the non-canonical pathway, c-IAP-mediated ubiquitination reactions result in constitutive NIK degradation, thereby dampening non-canonical NF-κB activation [
45]. Underscoring these opposing functions of c-IAPs in canonical and non-canonical NF-κB activation,
c-IAP genes are subject to both amplification and deletion in human cancer [
127]. Consequently, the aim of the therapeutic interventions targeting c-IAP proteins in cancer is to inhibit canonical NF-κB activation, without interfering with non-canonical NF-κB signalling. The endogenous c-IAP antagonist, second mitochondria-derived activator of caspases (SMAC), provides an attractive target for achieving this aim. In response to apoptosis-inducing stimuli, SMAC is released from mitochondria into the cytoplasm, where it binds to the conserved c-IAP domain, baculovirus IAP repeat (BIR), via its N-terminal AVPI tetrapeptide, thereby neutralising the prosurvival activity of c-IAPs [
128]. This paradigm has served as a reference point for the development of drug mimetics that act as selective c-IAP inhibitors. At least two such compounds, i.e., LCL161 and birinapant (TL32711), have now entered clinical trials in patients with solid cancers, such as ovarian serous carcinoma (
Table 1) [
129]. Although these compounds have been shown to promote cancer-cell death by stimulating an increase in cytokine production, they have also been reported to have severe dose-limiting toxicities, in particular the onset of cytokine-release syndrome, which significantly restrict their clinical application [
130]. Other common, but less severe, side effects include vomiting, nausea, fatigue, and anorexia [
131].
Therefore, although targeting upstream NF-κB signalling components has yielded tangible clinical results in the treatment of certain cancers and represents an attractive therapeutic strategy from the standpoint of achieving a degree of tissue- and context-specificity, this approach has so far been limited by the onset of dose-limiting adverse effects, inherent cancer recalcitrance and/or an early onset of secondary drug resistance [
109,
113,
130]. Nevertheless, as the clinical efforts in this area are relatively new and the underlying research is constantly advancing the understanding the intricate upstream signalling networks leading to IKK activation, targeting these networks as a means of therapeutic intervention holds promise for the future development of safe and effective anticancer therapeutics. While strictly speaking, most therapies interfering with upstream signalling intermediates will lack NF-κB-selective specificity, as they will inevitably also affect pathways beyond the NF-κB pathway, this limitation may be overcome at least in part by targeting protein-protein interactions involved in IKK activation, an area that remains largely unexplored and is likely to deliver effective and more specific NF-κB-targeting therapeutics. Although developing molecules that affect protein–protein interactions presents clear challenges, the heavy reliance of NF-κB activation pathways upon adaptor molecules and inducible formation of multimeric signalling complexes, coupled with the flourishing progress being made in unravelling new interactions and their regulation, is certain to catalyse the translational research efforts to develop NF-κB inhibitors in the future.
3.2. IKK Inhibitors
Owing to its central role as the signal integration hub for NF-κB activation pathways (
Figure 1 and
Figure 2), the IKK complex has been the focus of significant drug discovery efforts since its discovery in 1996. However, while a vast spectrum of inhibitors has been developed throughout the years, only a few of these agents have ever been entered into clinical trials, and none has been clinically approved (
Figure 3). A seminal paper by Michael Karin and colleagues in 2007 irreversibly tempered the initial enthusiasm over these agents as candidate NF-κB-targeting therapeutics [
132]. This paper demonstrated that pharmacological IKKβ inhibition results in elevated systemic levels of IL-1β, owing to increased pro-IL-1β processing and IL-1β secretion by macrophages and neutrophils upon bacterial infection or exposure to endotoxin, leading to overt systemic inflammation and lethality in mice [
132]. These severe on-target toxicities of IKKβ inhibitors have exposed the serious consequences of long-term IKK inhibition, which combined with the associated immunodeficiency and increased risks of malignancies arising from the liver, skin and other tissues [
9,
11,
107,
132,
133,
134], have irrevocably undermined any research efforts to clinically develop IKKβ-targeting therapeutics.
Despite their amino acidic sequence similarity, IKKα and IKKβ play largely distinct roles in NF-κB activation [
107,
135], whereby canonical NF-κB signalling strictly relies upon IKKβ-mediated phosphorylation of IκBs, while non-canonical NF-κB activation exclusively relies on IKKα. Nonetheless, IKKα also contributes to canonical NF-κB-dependent transcriptional responses by modulating the nuclear activities of RelA, histones and various transcriptional co-activators and co-repressors [
2,
136]. While the focus of the translational research has been on developing specific IKKβ inhibitors, these molecules often also target IKKα [
3,
136], owing to the high degree of similarity between these kinases.
IKKα/IKKβ inhibitors can be broadly classified into three major groups on the basis of their mode of action: ATP analogues, allosteric modulators, and agents interfering with the kinase activation loops [
3,
136]. ATP analogues are the largest group and comprise both natural products such as β-carboline, and small-molecule inhibitors such as SPC-839 (Celgene). SPC-839 is a synthetic quinazoline analogue, which exhibits an approximately 200-fold selectivity for IKKβ over IKKα [
2,
3,
105,
137]. The imidazoquinoxaline derivative, BMS-345541, binds to an allosteric pocket present on both IKKα and IKKβ and is an example of an allosteric modulator, having a 10-fold higher selectivity for IKKβ than IKKα. Curiously, while BMS-345541 does not interfere with ATP binding to IKKβ, it disrupts ATP binding to IKKα [
2,
3,
11,
105,
137,
138]. BMS-345541 has so far mainly been tested in vitro, and demonstrated some efficacy against collagen-induced arthritis in mouse models [
2,
3,
11,
105,
137,
138,
139,
140,
141]. Tiol-reactive compounds, including parthenolide, arsenite and epoxyquinoids, inhibit IKKβ by interacting with T-loop cysteine residue, C179 [
3,
105]. Although their exact mode of action is not well understood, this class of compounds appears to induce post-translational modifications that curtail IKKβ activity [
3,
105]. Parthenolide displays poor bioavailability and therefore has limited therapeutic application [
142]. A new IKKβ inhibitor more recently developed by Sanofi-Aventis, SAR-113945, has been investigated in four different clinical trials in non-oncological patients, with initial promising results [
143]. However, SAR-113945 has failed to demonstrate clinical efficacy in follow-on clinical trials, underscoring the challenge of striking an acceptable balance between efficacy and adverse side-effects [
143]. Overall, the disappointing clinical performance of IKKβ inhibitors has considerably dampened the interest in this class of agents, and this trend is likely to continue in the future, as shown by the dramatic decline in the number of patent applications filed on these agents in recent years [
107].
A significant effort has also been made towards developing molecules that target the IKKγ/NEMO scaffold (
Figure 3), using one of three main targeting strategies aimed at perturbing either the IKKγ/NEMO interaction with catalytic IKK subunits, IKKγ/NEMO dimerization, or IKKγ/NEMO ubiquitination. The IKKγ/NEMO interactions with IKKα and IKKβ involve the kinase C-terminal NEMO-binding domain (NBD), which consists of the hexapeptide amino acid sequence, LDWSWL [
135,
136]. Peptidomimetics of this sequence have been shown to disrupt the IKKγ/NEMO–IKKα/IKKβ interaction and accordingly inhibit NF-κB activation by various upstream signals. However, the therapeutic utility of these agents is limited by their poor bioavailability and significant instability, in vivo [
135]. Notwithstanding, small-molecule inhibitors of the IKKγ/NEMO–IKKα/IKKβ interaction may have the potential to provide useful agents to pharmacologically target the NF-κB pathway [
136]. Following publication of the crystal structure of the IKKγ/NEMO–IKKβ interface, four phenothiazine derivatives were developed and demonstrated to have inhibitory effects on NF-κB activation in macrophages, in vitro [
135]. However, further studies are required to determine whether these agents retain their inhibitory activity, in vivo, and crucially whether they are suitable for further development, especially in consideration of the safety concerns raised by IKKβ inhibitors [
135]. Peptidomimetics were also developed to interfere with IKKγ/NEMO dimerisation [
144]. However, to our knowledge, no significant preclinical development has ever been reported on any of these agents. Moreover, there has been significant interest in generating agents targeting the IKKγ/NEMO UBAN (i.e., UBD in the ABIN proteins and NEMO) domain, which transduces IKK activation signals by mediating the inducible interactions between IKKγ/NEMO and the polyubiquitin scaffolds of other signalling components [
21,
145]. Several agents have been shown to disrupt these ubiquitin-dependent IKKγ/NEMO interactions, including peptidomimetics [
57], as well as small molecules such as anthraquinone derivatives of emodin [
145]. However, despite the ability of some IKKγ/NEMO-targeting agents to modulate NF-κB activation without interfering with the IL-1β release, there appear to have been no clinical trials ever initiated to investigate the clinical safety and efficacy of these molecules.
The IKK-related kinase, TBK1 was originally discovered due to its involvement in NF-κB activation by PKCε-mediated signals [
146]. As well as IKKε, TBK1 primarily regulates the activation of IRF-family factors, such as IRF3, IRF5 and IRF7, by RLRs and other receptors [
147,
148,
149], but also phosphorylates RelA to enhance its transcriptional activity [
136,
146,
150]. TBK1 and IKKε share a 64% sequence identity, but display only 27% identity with classical IKKs, and are both involved in inflammatory responses, oncogenesis, and insulin resistance [
107]. Moreover, IKKε has been found overexpressed in breast and ovarian carcinoma, while TBK1 cooperates with RAS in promoting malignant transformation, in vitro, and is overexpressed in the carcinoma of the lung, colon and breast [
151,
152,
153,
154,
155,
156]. The antiinflammatory agents, BX765 and CYT387 (momelotinib), were originally developed as inhibitors of 3-phosphoinositide-dependent protein kinase 1 (PDK1) [
157] and Janus kinases (JAKs), respectively, but were subsequently shown to also have potent inhibitory activity against TBK1 and IKKε [
158]. Consequently, BX765 and CYT387 have been used as starting points for the molecular design of new TBK1 and IKKε antagonists. However, similar to these prototypes, the resulting compounds also demonstrated broad kinase target specificity [
159]. Domainex and Myrexis have since generated more selective TBK1 and IKKε inhibitors, such as DMXD-011 and MPI-0485520, respectively [
159]. DMXD-011 displays good drug-like properties, is orally bioavailable and has shown promising therapeutic activity in in vivo inflammatory disease models [
160]. Likewise, MPI-0485520 displays good oral bioavailability and has been investigated in mouse models of systemic lupus erythematous, rheumatoid arthritis and other autoimmune disorders, with significant therapeutic activity demonstrated, especially in the context of especially in the context of rheumatoid arthritis [
161]. However, further investigations are required to determine the potential clinical benefit resulting from these agents and whether their preclinical efficacy translates to the oncological context [
159].
3.3. Ubiquitin and Proteasome Pathway Inhibitors
The ubiquitin pathway provides a highly versatile and tightly regulated system of signal propagation and protein degradation conserved throughout the eukarya dominion [
162,
163,
164]. Over the past decade, this system has witnessed a burgeoning interest, not only because of its central importance in signal transduction and protein degradation, but also due to its potential for serving as a treasure trove of drug targets for therapeutic intervention in a wide range of malignant and inflammatory pathologies [
162,
165]. The ubiquitin pathway consists of a three-step enzymatic process, catalysed by as many protein complexes, whereby a ubiquitin-activating enzyme (E1) first binds to and activates a ubiquitin molecule in an ATP-dependent reaction, followed by the sequential transfer of the activated ubiquitin molecule to the active site of a ubiquitin-conjugating enzyme (E2) and, finally, to the target protein through the site-specific recognition and correct positioning by a ubiquitin-protein ligase (E3), which operates in conjunction with the E2 ubiquitin-conjugating enzyme to catalyse the covalent attachment of an 8-kDa ubiquitin molecule to the target site(s) of the protein substrate [
25,
33,
162,
165,
166,
167]. In addition to catalysing the attachment of the first ubiquitin molecule onto the protein substrate, the same three-step enzymatic process conjugates further ubiquitin molecules to form polyubiquitin chains [
167].
Ubiquitin-mediated signalling pathways play a central role in the regulation of both canonical and non-canonical NF-κB activation (
Figure 1 and
Figure 2) [
21,
24]. They also regulate oncogenesis by participating in either tumour suppression or tumour promotion [
21,
24,
35,
162]. Therefore, interfering with the ubiquitin system can affect cancer development and progression in many different ways (
Figure 3). Notwithstanding, tumour cells that depend on constitutive NF-κB signalling for survival often display sensitivity to inhibitors of the ubiquitin-proteasome pathway (UPP), owing to the essential role of this pathway in the proteolytic degradation of IκB proteins [
6,
25,
162]. The active proteasome consists of a large 2.4-MDa protein complex, which comprises a 20S catalytic core of a cylindrical shape and two regulatory 19S components forming a lid-like structure at both extremities of the 20S cylinder, and catalyses the proteolysis of substrate proteins via an ATP-dependent process [
167].
The first proteasome inhibitors developed were molecules of the class of peptide aldehydes and were subsequently extensively used for preclinical research [
168]. The best characterised of these molecules is MG132, which also served as a prototype for the generation of the next classes of proteasome inhibitors that later found common use in the clinical practice [
169]. The first proteasome inhibitor ever tested in humans was bortezomib (velcade; formerly known as PS-341), a reversible boronic-acid inhibitor of the 20S catalytic subunit [
137,
162]. Bortezomib received the approval of the FDA in 2003 for the treatment of refractory multiple myeloma, and is currently in clinical use as a front-line therapy in combination with other agents for the treatment of multiple myeloma and mantle cell lymphoma (MCL) [
162]. A number of clinical trials are ongoing to assess the efficacy of bortezomib in further oncological indications, including solid cancers (
Table 1). An especially promising area of development for bortezomib and other proteasome inhibitors is their use as part of combination therapies aimed at overcoming radio- and chemo-resistance in cancer. Indeed, bortezomib has been shown to have a strong synergistic activity when used in combination with radiotherapy and/or chemotherapy in various types of haematological and solid cancer [
170,
171,
172,
173,
174,
175]. The second-generation proteasome inhibitor, carfilzomib (Kyprolis), was approved in 2012 as a single agent and in 2016 in combination with dexamethasone, with or without lenalidomide, for the treatment of patients with relapsed or refractory multiple myeloma who have received at least one line of prior therapy. Carfilzomib is an epoxyketone compound, acting as an irreversible proteasome inhibitor to afford prolonged therapeutic inhibition [
162], and is currently being investigated in multiple myeloma in combination with other agents, as well as other oncological indications (
Table 1). Since both bortezomib and carfilzomib can be only administered intravenously or subcutaneously, the boronic proteasome inhibitor, ixazomib (MLN-9708), was recently developed as an oral therapy and was approved by the FDA in 2015 in combination with lenalidomide and dexamethasone for the treatment of patients with multiple myeloma who have received at least one line of prior therapy [
176]. Ixazomib is currently being evaluated in patients with other cancer types and as part of other combination regimens (
Table 1).
The human genome encodes two E1 ubiquitin-activating enzymes, UBA1 and UBA5, an estimated fifty E2 ubiquitin-conjugating (UBC) enzymes, and more than six hundred E3 ligases, conferring a large degree of substrate specificity to the ubiquitin cascade [
11,
25,
166]. It follows that therapeutically targeting individual E3s, which selectively bind to the recognition sites of protein substrates, will achieve maximal target specificity, as compared to targeting any E1s or E2s. The inhibitor of the ubiquitin-like protein, neural precursor cell-expressed developmentally down-regulated 8 (NEDD8), MLN4924 (pevonedistat), was developed to inhibit Cullin-RING E3 ubiquitin ligases, the largest family of E3s, which require activation by E1/E2-mediated NEDDylation [
165]. As a result of this mode of action inhibiting multiple E3 ligases, MLN4924 has broad target specificity. Although most of its adverse effects are mild or moderate, several higher-grade toxicities been reported, including febrile neutropenia, thrombocytopenia and elevated circulating levels of aspartate transaminase [
165,
166,
177]. Notwithstanding, clinical trials of MLN4924 in combination with 5-azacytidine are underway in patients with acute myeloid leukaemia (AML) and are yielding promising initial results (
Table 1) [
177].
The F-box protein, β-TrCP, is involved in the signal-dependent recognition and degradative K48-linked ubiquitination of canonical IκBs and p100, as well as in the recognition and ubiquitination of a large group of other target substrates, including β-catenin, Snail, Emi1, Wee1, Cdc25A and Claspin [
178,
179]. From a theoretical standpoint at least, targeting NF-κB signalling via β-TrCP would represent a more specific and, possibly, safer alternative to proteasome inhibition [
178,
179], although a limitation of the approach would remain the accumulation of protein substrates outside the NF-κB pathway, such as β-catenin, which contributes to colorectal carcinogenesis [
178,
179]. Notwithstanding this caveat, small-molecule inhibitors of β-TrCP, such as GS143, have been developed and shown to inhibit signal-induced IκBα ubiquitination [
180]. However, little information is available on the mode of action of this agent, although it seemingly involves an interaction with both β-TrCP and IκBα, or its further characterisation [
165].
While proteasome inhibitors, such as bortezomib, can provide significant clinical benefit in multiple myeloma and a few other indications, these agents inhibit NF-κB and many other essential cellular pathways that rely on the proteasome function [
6,
162,
166], and are therefore by no means specific for the NF-κB pathway. Moreover, proteasome inhibitors target these pathways in normal and cancer cells alike, thereby resulting in a low therapeutic index and significant dose-limiting toxicities [
137,
166,
181,
182]. Indeed, despite their indisputable commercial and clinical success, proteasome inhibitors as a therapy present several limitations, which remain largely unaddressed, including their broad cellular activities, dose-limiting side effects and the relatively rapid onset of secondary drug resistance [
6,
162,
165,
166]. In particular, treatment with bortezomib is often associated with peripheral neuropathy, which is a dose-limiting adverse effect, as well as trombocytopenia, neutropenia, nausea, diarrhea, and fatigue [
162]. Furthermore, drug resistance is inevitable and generally develops within a year from the start of treatment [
166]. While carfilzomib administration is associated with different side effects and the onset of peripheral neuropathy is significantly less frequent than with bortezomib, patients can nonetheless develop renal impairment and cardiovascular complications [
162]. Likewise, treatment with Ixazomib can induce peripheral neuropathy, gastrointestinal adverse effects and skin rash [
162].
Importantly, from a mechanistic standpoint, it is also unclear that the clinical response to proteasome inhibitors in patients with multiple myeloma and other B-cell malignancies results from the inhibition of NF-κB signalling, as it is becoming increasingly clear that the cellular accumulation of undigested proteins in these immunoglobulin-producing tumours can activate the unfolded protein response (UPR), thereby accelerating tumour-cell death [
162]. Therefore, there remains a need for novel therapeutic strategies capable of selectively targeting the NF-κB pathway in these oncological indications [
137,
166,
179]. While targeting upstream components of the UPP, such as NF-κB-activating E3 ligases, would be a preferable strategy than targeting the proteasome, as it could mitigate at least some of the limitations of proteasome inhibition, and, more broadly, represents a highly promising area for future drug discovery and development in oncology, this strategy would nevertheless be unlikely to ever yield a specific NF-κB-targeting agent [
137,
162,
165,
166].
3.4. Inhibitors of NF-κB Nuclear Activities
Once liberated from IκBs, NF-κB dimers migrate into the nucleus where they regulate gene expression. This activation step offers a significant opportunity for therapeutic intervention, since, at least from a theoretical standpoint, a number of nuclear activities of NF-κB could be targeted with drug agents, including the post-translational modification of NF-κB proteins and their ability to dimerise, translocate into the nucleus, bind to DNA and interact with chromatin components, coactivators and corepressors and other transcription factors [
3,
105]. While the drug development output in this area has been relatively limited, a few examples of candidate therapeutics targeting the nuclear translocation and DNA-binding activity of NF-κB complexes are worthy of consideration (
Figure 3). In particular, several strategies have been developed to inhibit the NF-κB nuclear translocation, including small peptidomimetics, such as SN-50, which encompasses the NLS of p50 [
3,
105,
183]. At high concentrations, SN-50 has been shown to saturate the transport machinery importing p50-containing dimers into the nucleus [
3,
105,
183]. However, apart from the high peptide concentrations required to achieve this effect, a drawback of SN-50 is its non-specific inhibition of transcription factors other than NF-κB complexes, such as AP-1-family factors [
3,
105,
183]. Additionally, dehydroxymethylepoxyquinomicin (DHMEQ), a derivative of the antibiotic epoxyquinomicin C, isolated from
Amycolatopsis, has been shown to exhibit a potent and specific inhibitory activity on NF-κB nuclear import by means of its ability to directly bind to NF-κB dimers [
3,
105,
183,
184]. Studies in mouse models have demonstrated a therapeutic effect of DHMEQ in prostate carcinoma and an immunomodulatory effect in ovarian carcinoma. Despite these promising preclinical results, to our knowledge, no clinical trials have been initiated to evaluate DHMEQ in cancer patients [
185,
186,
187].
Furthermore, sesquiterpene lactone (SL) compounds have been shown to inhibit the DNA-binding activity of RelA-containing NF-κB dimers by interacting with cysteine residue, C38, within RelA’s DNA-binding loop 1 (L1) [
3,
105]. By binding to homologous cysteine residues within p50 and c-Rel, certain SL compounds have been shown to have the additional capacity to inhibit NF-κB complexes containing these subunits [
3,
105]. Interestingly, some SLs, such as parthenolide, also have the ability to inhibit IKKβ [
3,
105]. However, while this dual effect of parthenolide on the NF-κB pathway has attracted some interest, the poor bioavailability of this agent has precluded any further drug development effort [
142,
188]. However, the search for a functionally equivalent compound has resulted in the generation of the amino acid-analogue, dimethylaminoparthenolide (DMAPT), exhibiting enhanced bioavailability [
189]. A phase I clinical trial of DMAPT was initiated in 2009 in patients with AML, but was then suspended later in the same year [
189,
190]. Another class of agents designed to interfere with the NF-κB DNA-binding activity consists of decoy oligodeoxynucleotides, which can compete for binding to NF-κB complexes with κB DNA sites on specific gene promoters [
2]. Most therapeutic decoy oligodeoxynucleotides were further modified to increase their in vivo stability, as well as their affinity for NF-κB complexes [
3,
105]. A clinical trial of the NF-κB decoy oligodeoxynucleotide, anesiva, was initiated in 2005 to test its efficacy as a local ointment for the treatment of atopic dermatitis, but its clinical use was subsequently discontinued in 2014 (NCT00125333, [
190]).
3.5. Inhibitors of NF-κB Downstream Effectors
Since NF-κB induces transcriptional programmes that affect all hallmarks of cancer [
79,
191], an attractive alternative to therapeutically targeting NF-κB in malignant disease would be to inhibit the non-redundant, cancer cell-specific downstream effectors of the NF-κB oncogenic functions (
Figure 3). Given that the insurmountable challenge with conventional NF-κB or IKKβ inhibitors has been to achieve cancer-cell specificity, due to the pleiotropic and ubiquitous functions of NF-κB [
11], agents targeting these effectors, having functional restriction to cancer cells or their microenvironment, could provide safer and more selective anticancer therapeutics, lacking the dose-limiting toxicities of global NF-κB inhibitors.
Our group recently sought to obtain proof-of-concept for this principle in multiple myeloma, the paradigm of NF-κB-driven malignant diseases. Since a key pathogenetic function of NF-κB in multiple myeloma is to upregulate genes that block apoptosis and, despite its ubiquitous nature, NF-κB signalling induces highly tissue- and context-specific transcriptional programs [
6,
11,
12,
14,
79,
192], we targeted an essential downstream effector of this pathogenically critical activity of NF-κB, in order to achieve cancer cell-selective therapeutic specificity and thereby circumvent the limitations of conventional IKK/NF-κB-targeting drugs. Several years ago, we identified the immediate-early gene, growth arrest and DNA damage 45B (
GADD45B), as a novel transcriptional target of NF-κB and effector of the NF-κB-dependent inhibitory activity on JNK signalling and apoptosis in response to TNF-α and other cues [
193,
194]. GADD45β is a member of the GADD45 family of proteins, also comprising GADD45α and GADD45γ, which play distinct roles in multiple cellular functions, including cell-cycle regulation, DNA repair, apoptosis, senescence and DNA demethylation [
195,
196,
197]. Interesting, GADD45β is the only member of this family that is largely regulated downstream of NF-κB signalling [
193,
194]. A number of studies, including some from our own group, have demonstrated that GADD45β and other members of this family play many of their important biological roles by regulating the activity MAPK pathways, such as the JNK and p38 pathway [
195,
198,
199]. Indeed, more recently, we have identified the complex formed by GADD45β and the JNK kinase, MKK7, as a functionally critical survival module downstream of NF-κB and novel therapeutic target in multiple myeloma [
194,
198,
200,
201,
202,
203]. As most normal cells do not constitutively express
GADD45B [
204], and, unlike mice lacking
RelA or any
IKK subunit [
11],
Gadd45b−/− mice are viable, fertile and die of old age [
205,
206], we reasoned that, in contrast to global NF-κB blockade, pharmacological GADD45β inhibition would be well tolerated in vivo.
We therefore developed the D-tripeptide inhibitor of the GADD45β/MKK7 complex, DTP3, which effectively disrupts this complex at nanomolar concentrations, in vitro, by binding to MKK7, and as a result, selectively kills multiple myeloma cells by inducing MKK7/JNK-dependent apoptosis, without toxicity to normal tissues [
200,
201]. We showed that, due to this target-selective mode of action, DTP3 displays an excellent cancer-cell selective specificity in multiple myeloma cell lines and primary cells from patients, in vitro, and eradicates multiple myeloma xenografts in mice, with excellent tolerability and no apparent side effects at the therapeutic doses [
200,
201]. The first-in-human phase I/IIa clinical study of DTP3 has recently been initiated in patients with refractory or relapsed multiple myeloma, and initial results from this study preliminarily demonstrate the clinical safety of this agent, alongside a cancer-selective pharmacodynamic response (
Table 1).
Further investigations will be required to determine the potential clinical benefit resulting from this approach in multiple myeloma patients, as well as its potential side-effects, propensity to develop drug resistance, and therapeutic efficacy in combination with other agents [
207,
208]. Nevertheless, the available body of preclinical data and the encouraging initial clinical results preliminarily demonstrate that cancer-selective inhibition of the NF-κB pathway is possible and promises to provide an effective therapeutic strategy, with no preclusive toxicity, that could profoundly benefit patients with multiple myeloma and, potentially, other cancers where NF-κB is a driver of pathogenesis. Indeed, the same principle we developed of targeting an axis of the NF-κB pathway with cancer-restricted function, rather than NF-κB globally, could be similarly applied to also selectively inhibit NF-κB oncogenic functions beyond the suppression of cancer-cell apoptosis [
200,
201], such as functions in governing tumour-associated inflammation.
Several antiapoptotic genes, in addition to
GADD45B, have been shown to be transcriptionally regulated by NF-κB, including those encoding various BCL-2-family members such as B-cell lymphoma-extra large (
Bcl-XL), myeloid cell leukaemia sequence 1 (
MCL1),
A1 (also known as BFL-1) and, at least in certain tissue contexts, B-cell lymphoma 2 (
BCL-2) itself [
191,
209]. These proteins are also frequently deregulated in certain cancer types, often as a result of chromosomal translocations or other genetic abnormalities, and promote oncogenesis by means of their ability to suppress cancer-cell apoptosis [
210,
211,
212,
213]. BCL-2-family members have also been shown to contribute to NF-κB-dependent radio- and chemo-resistance in cancer [
80]. Accordingly, the therapeutic inhibition of these proteins has also been pursued as a combination therapy with radiation and chemotherapeutic drugs to overcome resistance to these agents in recalcitrant forms of malignancy [
214]. Congruently, certain NF-κB-dependent multiple myeloma cell lines express low levels of GADD45β and are completely refractory to DTP3-induced killing [
200], confirming the existence of GADD45β-independent mechanisms for NF-κB-dependent survival in certain subtypes of multiple myeloma, and such mechanisms are certain to also exist in other types of malignancy. The BCL-2 family of proteins comprises an evolutionarily conserved group of 20 members, which share one or more of four so-called BCL-2 homology (BH) domains, referred to as BH1, BH2, BH3 and BH4, and are involved in either suppressing or promoting apoptosis [
211,
212,
213]. The members of this family can be classified into three functionally and structurally distinct groups: prosurvival proteins such as BCL-X
L, MCL1, A1/BFL-1 and BCL-2 itself; multidomain proapoptotic effectors such as BCL-2-associated X protein (BAX) and BCL-2 antagonist/killer (BAK); and BH3-only proteins, which convey apoptosis-initiating signals, such as BCL-2-interacting mediator of cell death (BIM), BH3 interacting-domain death agonist (BID), Bcl-2-associated death promoter (BAD), and p53 upregulated modulator of apoptosis (PUMA) [
211,
212,
213]. Historically, the main strategy utilised to inhibit the antiapoptotic activity of BCL-2-family members has been to generate cell-permeable drug mimetics of the BH3 domains of proapoptotic BCL-2-like proteins, which neutralise prosurvival BCL-2 proteins by binding to their surface hydrophobic groove [
211,
212,
213,
215]. However, many of these agents have shown limited therapeutic activity as single agents and/or significant side-effects due to their low specificity [
211,
215]. A notable exception in this group of molecules has been ABT-199 (venetoclax, RG7601, GDC-0199), a potent and selective first-in-class BCL-2 inhibitor [
211,
213,
215]. ABT-199 has demonstrated significant anticancer activity in various models of lymphoma and leukaemia, both in vitro and in vivo, and has entered clinical trials in patients with non-Hodgkin’s lymphoma, CLL, AML and T-ALL, both as a single agent and in combination with other drugs [
211,
216]. ABT-199 was subsequently granted breakthrough status designation by the FDA in 2016 for the treatment of patients with relapsed or refractory CLL with 17p deletion (
Table 1).