Inhibition of NF-κB DNA binding suppresses myeloma growth via intracellular redox and tumor microenvironment modulation

Abstract

Multiple myeloma (MM) is a plasma cell malignancy that is still largely incurable, despite considerable progress in recent years. NF-κB is a well-established therapeutic target in MM, but none of the currently available treatment options offer direct, specific pharmacological targeting of NF-κB transcriptional activity. Thus, we designed a novel direct NF-κB inhibitor (IT848) as a drug candidate with strong potential for clinical translation and conducted comprehensive in vitro and in vivo mechanistic studies in MM cell lines, primary MM cells, xenograft models and immunocompetent mouse models of MM. Here we show that IT848 inhibits NF-κB activity through inhibition of DNA binding of all five NF-κB subunits. IT848 treatment of MM cell lines and patient samples inhibited proliferation and induced caspase-dependent and independent apoptosis. In addition to direct NF-κB inhibitory effects, IT848 treatment altered the redox homeostasis of MM cells through depletion of the reduced glutathione pool, selectively inducing oxidative stress in MM but not in healthy cells. MM xenograft studies confirmed the efficacy of IT848 as single agent and in combination with bortezomib. Furthermore, IT848 significantly improved survival when combined with PD-1 inhibition, and correlative immune studies revealed that this clinical benefit was associated with suppression of regulatory T cell infiltration of the bone marrow microenvironment. In conclusion, IT848 is a potent direct NF-κB inhibitor and inducer of oxidative stress specifically in tumor cells, displaying significant activity against MM cells in vitro and in vivo, both as monotherapy as well as in combination with bortezomib or immune checkpoint blockade.

Introduction

Multiple Myeloma (MM) is the second most common hematologic malignancy, with an incidence of about 32,000 new cases and almost 13,000 deaths per year in the United States. Current frontline therapies include proteasome inhibitors, immunomodulators such as lenalidomide, alkylating agents, monoclonal antibodies such as daratumumab, and steroids. Several phase III trials have established the use of high-dose chemotherapy followed by autologous stem cell transplantation as standard of care in eligible patients (1). Second-generation proteasome inhibitors as well as third-generation immunomodulators have contributed to improving outcomes in the previous decade (2), and B cell maturation antigen (BCMA)-targeted agents have shown promise in relapsed or refractory MM (3). However, almost all patients eventually become refractory to currently available therapies. Therefore, there is an urgent need for novel agents with enhanced efficacy in this setting.

The NF-κB family of transcription factors is found in all nucleated cells. This pathway plays an important role in B cell proliferation, T cell activation, regulation of inflammatory responses, as well as the pathogenesis of cancer (4). The NF-κB family comprises of 5 structurally related monomeric subunits: Rel A, Rel B, c-Rel, p50 and p52. The combinatorial association of these subunits produces transcription factors with dimerization and DNA binding domains. In latent form, NF-κB is bound to an inhibitory protein (IκB). When stimulated extracellularly (via ligands such as TNFα, IL-6, RANKL etc.), NF-κB dimers translocate into the nuclei and mediate the expression of a plethora of immune/ stress response genes and pro-survival factors(4). NF-κB is constitutively active in MM cells (5,6) promoting cell proliferation, survival and resistance to chemotherapy (7). Consequently, NF-κB is an attractive therapeutic target in MM, with both canonical and non-canonical NF-κB pathways contributing to the total NF-κB activity. NF-κB also regulates transcriptional signatures involved in inflammation and immunity, including the programmed death-ligand 1 (PD-L1)/programmed death protein 1 (PD-1) axis (8). Inhibition of NF-κB DNA binding in MM might therefore not only counteract upstream activating mutations but also modulate the immunosuppressive tumor microenvironment (TME). PD-L1/PD-1 interaction results in suppression of T cell-mediated cytotoxicity (9), promoting an immunosuppressive milieu in tumor microenvironment (10). Disruption of the PD-L1/PD-1 axis by monoclonal antibody therapy has been highly effective in several solid tumors but has shown only limited activity against MM in initial trials (11). This has led to the additional investigation of protocols combining immune checkpoint blockade with lenalidomide or dexamethasone (12,13). After initial setbacks, immunotherapeutic agents have now emerged as an attractive option for the treatment of MM, but additional preclinical and clinical investigations are warranted to establish safe and effective treatment regimens (14) such as the combination of NF-κB inhibition with immune checkpoint blockade. Here we introduce IT848, a small molecule inhibitor of NF-κB DNA binding with drug-like physicochemical properties, as a candidate for future clinical development of an NF-κB inhibitor drug. We show that IT848 combines NF-κB inhibition with dysregulation of the redox homeostasis in MM cells as well as favorable immunomodulatory properties, translating into significant in vitro and in vivo efficacy against MM.

Materials and Methods

Small Molecule NF-κB Inhibitor compound

We previously identified thiohydantoin (IT603) and naphthalenethiobarbiturate (IT901) derivatives as conformation-disrupting direct NF-κB inhibitors. After additional structure activity relationship analysis and pharmacokinetics studies, the novel small molecule IT848 (9-chloro-8-(hexyloxy)-2H-chromeno[2,3-d]pyrimidine-2,4(3H)-dione), MW = 348.78 g/mol is a was selected as our drug candidate. The IT848 synthesis method is described in detail in our patent (see example # 21, Table 1) (15).

Mouse models of multiple myeloma, in vivo imaging and NF-kB inhibitor treatment

NOD/scid/IL2Rγ(null) (NSG) mice were obtained from the Jackson Laboratory for xenograft experiments of multiple myeloma. MM.1S human multiple myeloma cells were transduced with dtomato-firefly luciferase and 5×10⁶ cells were administered to NSG mice via tail vein injection. C57Bl/KaLwRij mice were originally obtained from Envigo (Horst, the Netherlands) and maintained at the Center for Discovery and Innovation. All animal studies were conducted in accordance with Institutional Animal Care and Use Committee (IACUC) standards and with approval of the IACUC of the CDI (protocol#291).

Mice used for these experiments were 6 to 9 weeks old. For the study in immunocompetent mice, 5TGM1 mouse multiple myeloma cells (16) were transduced with dtomato-firefly luciferase and 1.5×10⁶ cells were administered to C57Bl/KaLwRij mice via tail vein injection. Bone marrow engraftment of MM cells was confirmed by in vivo bioluminescence imaging (BLI) prior to initiation of treatment, and MM progression was monitored at the indicated time points by in vivo BLI. Briefly, mice were anesthetized using isoflurane inhalant anesthetic and administered 3 mg/kg firefly luciferin intraperitoneally and imaged using the IVIS Lumina X5 optical imaging platform. We superimposed pseudocolor images showing the whole-body distribution of bioluminescent signal intensity on grayscale photographs and quantified total flux (photons s⁻¹) for individual mice using Living Image 4.7.3 software (Perkin Elmer, Waltham, MA). Additionally, coelenterazine was used to detect changes in regional reactive oxygen species production in vivo in response to IT848 administration (17). For in vivo efficacy studies, IT848 was dissolved in PBS with 30 % (w/w) cyclodextrin and 5 % PEG 1000 to prepare a 1mg/ml solution. Mice received either empty vehicle solution or 10mg/kg body weight of IT848 solution via intraperitoneal injection on either three days/week for four weeks or on five days/week for two weeks.

Cell lines and primary cells

Vendor-authenticated human MM cell lines U266, MM.1S, and NCI-H929 were originally obtained from the American Type Culture Collection (ATCC, Manassas, VA) in 2018. The human DLBCL-derived cell lines U2932, HBL1 and TMD8 were kindly provided by Dr. Anas Younes (Memorial Sloan Kettering Cancer Center) in 2013 and authenticated before using them by the MD Anderson Characterized Cell Lines Core Facility. These cells were cultured in RPMI 1640 medium supplemented with 10 to 20% heat-inactivated fetal bovine serum (GIBCO BRL, Gaithersburg, MD), 1% l-glutamine, and penicillin-streptomycin in a humid environment of 5% CO₂ at 37°C. The 5TGM1 cell line was kindly provided by Dr. Rena Feinman (Center for Discovery and Innovation, Nutley, NJ). Vendor-authenticated NIH 3T3 cells were purchased from ATCC in 2019.

A vendor-authenticated NF-κB/Jurkat/GFP transcriptional reporter cell line (18) was purchased from System Biosciences (Mountain View, CA) in 2015. Cells were stimulated with human TNF-alpha (10ng/ml) (Gemini Bioproducts, West Sacramento, CA) and incubated in the presence of IT848 or empty vehicle. NF-κB activity (GFP fluorescence intensity) was analyzed by flow cytometry. Vendor-authenticated ARE/HepG2/Luciferase reporter cells and NFAT/Jurkat/luciferase transcriptional reporter cells were purchased from BPS Bioscience (San Diego, CA) in 2017. ARE/Nrf2 and NFAT transcriptional activity was determined using the One-Step Luciferase Assay System (BPS Biosciences, San Diego, CA). Luminescence was measured with a multimodal plate reader (Infinite 200 pro from Tecan, Switzerland). Human EBV-transformed B lymphoblastoid cells were generated from PBMC of a healthy donor and kindly provided by Richard O’Reilly (Memorial Sloan Kettering Cancer Center, NY) in 2018. All cell lines were used for experiments within 3 weeks of thawing (up to 4 passages) and were annually tested for mycoplasma negativity using a mycoplasma PCR kit (MycoAlert kit from Lonza).

Human peripheral blood mononuclear cells (PBMC) were purchased from United States Biological (Salem, MA). Human multiple myeloma patient cells were collected after informed written consent was obtained, using Institutional Review Board (IRB) protocol Pro00004259 at Hackensack Meridian Health - John Theurer Cancer Center. All studies of patient materials were conducted in accordance with the Declaration of Helsinki.

MM.1S and 5TGM1 cells were transduced using 3^rd generation lentiviral vector with bi-cistronic expression of dtomato and luciferase (pUltra-Chili-Luc) (Addgene #48688). Transduced cells were expanded and individual clones/population with high dtomato expression were sorted using a BD FACSMelody™. Bioluminescence signal intensity of single clones was determined, and the brightest clones were used for subsequent experiments.

RNA sequencing

RNA was purified using an AllPrep DNA/RNA extraction kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. RNA quantity was measured with a Nanodrop spectrophotometer (DeNovix, Wilmington, DE). RNA was submitted to Genewiz (South Plainfield, NJ) for cDNA library creation and sequencing. RNA-seq was performed in triplicate for each treatment condition on an Illumina Hi-Seq 2×150bp configuration, single index, per lane with estimated data output as ~350M raw paired-end reads per lane (19). Initial sample QC involved assessing RNA integrity and library size by Agilent Tapestation and concentration by Qubit assay, and final quantitation by qPCR. Differential expression analysis between sets of conditions was performed using the DESeq2 pipeline of Basepair analysis platform (Basepair, New York, NY). GSEA was performed as previously described (20) and a visualization of the canonical NF-kB pathway from the KEGG database is presented.

Data availability statement

The data generated in this study are available within the article and its supplementary data files. The RNA-seq data generated in this study are publicly available in Gene Expression Omnibus (GEO) at GSE200385.

Results

IT848 is a potent inhibitor of NF-κB DNA binding

We previously described a novel small molecule compound mediating anticancer properties by combining inhibition of NF-κB transcriptional activity with the capacity to modulate the redox balance of B cell lymphoma cells (21). After additional structure activity relationship analysis and in vivo drug metabolism and pharmacokinetics (DMPK) testing of our lead molecules, IT848 emerged as the most suitable candidate for drug development (Fig. 1A). In silico molecular docking analysis (22) identified IT848 binding sites on all NF-κB subunits (Fig. 1B), and homology modeling indicated that IT848 interacts with thiol groups of cysteines of the Rel/NF-κB subunits. IT848 elicited strong and dose-dependent inhibition of global NF-κB transcriptional activity both in lymphoma (Jurkat, TMD8) and myeloma (U266) cell lines or primary cells (Epstein-Barr virus-transformed B lymphoblastoid cells) (Fig. 1C and D). Moreover, we confirmed dose-dependent target-specific in vivo activity of IT848 (Supplementary Figure 1). Western blot analysis of NF-κB complex proteins (IKΚBeta, IκBalpha, interferon regulatory factor 4 (IRF4)) further validated IT848 as a potent NF-κB inhibitor with consistent, inhibitory effects across several Diffuse Large B cell lymphoma (DLBCL) and MM cell lines (Fig. 1E). Analysis of the NF-κB target gene and pro-tumorigenic cytokine interleukin 6 (IL-6) (23,24) in representative DLBCL and MM cell lines by quantitative reverse transcription PCR (RT-qPCR) (Fig. 1F) and enzyme-linked immunosorbent assay (ELISA) (Fig. 1G) demonstrated IT848 mediated supression of IL-6 expression, an effect that might not only be useful therapeutically, but also as a biomarker for IT848 activity in MM cells. In order to further elucidate the mechanism of action of IT848 and confirm its capacity to act as indended as a direct NF-κB inhibitor, we assessed DNA binding of individual NF-κB subunits in response to IT848 treatment. We used MicroScale Thermophoresis to characterize binding affinities of IT848 to c-Rel and p50, and to analyze DNA binding of c-Rel and p50 following IT848 treatment. We found that IT848 did not bind to DNA, but it showed binding affinity to both c-Rel and p50 in the low µM range (K_d of 3.9 and 7.0, respectively). In addition, IT848 treatment disrupted DNA binding of c-Rel and decreased p50 DNA binding 30 to 100-fold (Supplementary Figure 2). In line with this, NF-κB DNA binding ELISA of nuclear extracts from IT848 treated representative DLBCL and MM cell lines showed decreased DNA binding of all five NF-κB subunits (although with cell type-specific variability) (Fig. 1H). We next assessed the specificity of IT848 and found that inhibition of transcriptional activity was specific to NF-κB but spared unrelated transcription factors that were tested as controls (NFAT and Nrf2) (Fig. 1I). Taken together, our findings identify IT848 as a potent and specific small molecule NF-κB inhibitor that decreases the affinity of NF-κB subunits to their DNA binding sites and disrupts downstream biological events.

Figure 1. IT848 inhibits NF-κB transcriptional activity as a result of decreased DNA binding of NF-κB subunits.

A, Chemical structure of IT848. B, Pose view interaction of IT-848 (red) with NF-κB subunits (cyan): (A) RelB, (B) p52, (C) RelA, (D) p50, and (E) c-Rel. The white regions represent NF-κB residues which exhibit nonpolar interactions with IT848, and the yellow regions represent NF-κB residues which exhibit polar interactions with IT848. C, Jurkat/GFP/NF-κB transcriptional reporter cells were stimulated with TNF-α and treated for 22 hours in the presence of empty vehicle or 1, 3, 6 μM of IT848. NF-κB transcriptional activity (GFP median fluorescence intensity) was analyzed at 8 and 22 hours by flow cytometry. Mean and SEM of normalized values are presented (percentages of vehicle-treated baseline activity). One of two independent experiments is shown. D, U266 MM cells, TMD8 DLBCL cells, and an EBV-transformed B lymphoblastic cell line (BLCL) were engineered to constitutively express Renilla luciferase as well as NF-κB-driven firefly luciferase. Cells were incubated for 24 hours in the presence of empty vehicle or 2, 4, and 6 μM of IT848. NF-κB transcriptional activity was analyzed by dual luciferase assay. Mean and SEM of normalized values are presented (percentages of vehicle-treated baseline activity). One of two independent experiments is shown. E, DLBCL and MM cell lines were cultured in the presence of IT848 (2, 4 or 6 µM) or DMSO for 12 hours. NF-κB activity was analyzed by Western Blot of the NF-κB pathway signaling transduction molecules phospho-IKKb, phospho-IκBa, and IRF. Beta Actin was included as an internal control. F, TMD8 and U266 cells were treated with IT848 (4 μM) or empty vehicle for 4 hours, followed by RNA extraction and qRT-PCR of interleukin 6 (IL-6). Mean and SEM of relative gene expression is presented, * P < 0.001; ** P = 0.002. One of three independent experiments is shown. G, TMD8 and U266 cells were incubated for 24 hours in the presence of empty vehicle or IT-848 (4 and 6 μM). IL-6 concentrations in the culture media were analyzed after 24 hours by ELISA. Mean and SEM of normalized values are presented (percentages of vehicle-treated baseline expression), * P < 0.01. H, MM.1S and U266 human MM and HBL-1 human DLBCL cells were treated with IT848 (4uM) or empty vehicle for 4 hours, followed by nuclear extraction and analysis of DNA binding of all five NF-κB subunits by ELISA. Mean and SEM of normalized values (percentages of vehicle-treated baseline DNA binding) are presented. One of two independent experiments is shown. I, Jurkat/GFP/NF-κB transcriptional reporter cells were stimulated with TNF-α, Jurkat/Luciferase/NFAT transcriptional reporter cells were stimulated with PMA/ionomycin, and HepG2/Luciferase/Nrf2-Antioxidant response element (ARE) transcriptional reporter cells were treated with tert-butylhydroquinone. The respective cell lines were incubated for 12 hours in the presence of empty vehicle or 1, 3, 6 and 10 μM of IT848. NF-κB transcriptional activity was analyzed by flow cytometry, NFAT and Nrf2 transcriptional activities were analyzed by luminescence measurement. Mean and SEM of normalized values are presented (percentages of vehicle-treated baseline activity). One of two independent experiments is shown.

IT848 inhibits proliferation and induces apoptosis in MM cells

NF-κB is known to be constitutively active in all MM cells (5,6,25). Indeed, NF-κB DNA binding analysis of two MM cell lines (MM.1S and U266) confirmed constitutive activity of all five NF-κB subunits (Supplementary Figure 3). To assess the anti-myeloma activity of IT848 we treated MM.1S, U266 and NCI-H929 cells with increasing doses of the compound. This resulted in a dose-dependent decrease in cell numbers over a 72-hour period as measured by luciferase assay (Fig. 2A) and direct cell counting (Fig. 2B). Importantly, peripheral blood mononuclear cells (PBMCs) of several healthy control subjects were not negatively affected by IT848 treatment except for a 10 percent decrease in metabolic viability in the presence of the highest tested concentration (6 µM) (Fig. 2C). To further evaluate the mechanism of action associated with the cytotoxic effects of IT848, we examined its impact on cell cycle phase distribution after 24 hours of treatment. The percentages of IT848-treated cells in S phase were increased while the frequency of cells in G2/M phase was lowered in a dose-dependent manner (Fig. 2D). This suggests that IT848 effectively affects cell-cycle checkpoints leading to S phase arrest and inhibition of cell division, which was confirmed by concordant decreased levels of Ki-67 expression, a nuclear protein present in proliferating cells (Fig. 2E). Again, normal PBMCs were insensitive to the anti-proliferative effect of IT848 (Fig. 2E). Furthermore, Annexin-V and 7-AAD staining revealed that IT848 induced a shift from early apoptosis (Annexin V+, 7-AAD−) to MM cell death (Annexin V+, 7-AAD+) after 24 hours of treatment (Fig. 2F). Therefore, the IT848-induced decrease in cell numbers is likely due to a dual effect, including inhibition of proliferation and induction of apoptosis. IT848 treatment resulted in a dose-dependent increase in caspase activity in MM cells (using a kit detecting activity of caspase 1, 3, 4, 5, 6, 7, 8 and 9; Fig. 2G). Furthermore, IT848 treatment enhanced nuclear translocation of apoptosis-inducing factor (AIF), a factor involved in a caspase-independent pathway of apoptosis (Fig. 2H) (26–28). AIF is a mitochondrial pro-apoptotic factor that is released from mitochondria upon decrease of the mitochondrial membrane potential and subsequent increased permeability of the mitochondrial membrane (26,27). We found that while low-dose IT848 treatment (2 and 4 μM) did not alter the mitochondrial membrane potential of MM cells, higher doses triggered a loss of the mitochondrial membrane potential (Fig. 2I), consistent with bioenergetic stress and release of pro-apoptotic factors such as AIF.

Figure 2. IT848 inhibits growth and induces apoptosis of MM cells.

MM.1S, U266 and NCI-H929 cells and normal peripheral blood mononuclear cells (PBMC) were incubated in the presence of IT848 or empty vehicle (control) for 72 hours (A-C) or 24 hours (D-F). A, Cell numbers after 0, 24, 48, 72 hours were quantified by luciferase assay. Mean and SEM are presented. B, Cell numbers after 0, 24, 48, 72 hours were quantified by manual cell counting. Mean and SEM are presented. C, metabolic viability of PBMC (n=4) was measured by MTS assay. Mean and SEM of normalized values (percentages of vehicle-treated baseline activity) are presented. D, Cell cycle analysis of live MM.1S cells treated with IT848 (4uM) or DMSO for 24 hours. Percentage of cells differentially stained with Hoechst 33342 and Ki-67 were quantified to show DNA content distribution/changes of affected S and G2/M phases. Representative dot plots as well as mean and SEM are presented. E, proliferation of the indicated cell lines and normal PBMC was analyzed by Ki-67 assay. Representative histograms and mean and SEM are presented. F, Apoptosis of MM.1S and U266 cells was analyzed by Annexin V/7AAD staining. Mean and SEM are presented. G, MM.1S, U266 and NCI-H929 cells were incubated in the presence of 2, 4, and 6 μM of IT848 or empty vehicle (control) for 24 hours. Activity of caspase 1, 3, 4, 5, 6, 7, 8 and 9 was quantified using a caspase activity detection kit as described in Materials and Methods. Mean and SEM of normalized data (percentage of control) are presented. H, U266 and NCI-H929 cells were incubated in the presence of 4 μM of IT848 or empty vehicle (control) for 24 hours and stained with DAPI and a monoclonal antibody for microscopic detection of AIF. Representative fluorescence microscopy images are shown (blue: DNA, green: AIF; for more experimental details see Materials and Methods). I, MM.1S, U266 and NCI-H929 cells were incubated in the presence of 2, 4, and 6 μM of IT848 or empty vehicle (control) for 24 hours. The mitochondrial membrane potential was quantified using a mitochondrial transmembrane potential detection kit as described in Materials and Methods. Mean and SEM of normalized data (percentage of control) are presented.

In contrast to antibody-based targeted therapy, chemical approaches utilizing small molecules are expected to be less specific and off target effects are common. Therefore, to broadly assess the effects of IT848 on cellular processes in MM cells we performed RNA-seq of MM.1S cells treated with IT848 for six hours (Fig. 3A and B). This timepoint was chosen because significant cell death has not occurred yet and it therefore allows to assess the impact of IT848 treatment on the transcriptome of MM cells. As expected, IT848 treatment of MM.1S cells resulted in downregulation of a variety of NF-κB target genes (Supplementary Table 1). However, we also found highly significant upregulations of reactive oxygen species (ROS) and apoptosis pathway genes, including heme oxygenase 1 (HMOX1) and oxidative stress-induced growth inhibitor 1 (OSGIN1), following exposure to IT848 (Fig. 3A and B). Gene set enrichment analysis (GSEA) revealed that oxidative stress and apoptosis signatures were among the most significantly enriched gene sets in myeloma cells treated with IT848 (Fig. 3C and Supplementary Figure 4). We validated RNA-seq findings regarding HMOX1 and OSGIN1 using RT-qPCR (Fig. 4A and B). Remarkably, we showed that the effect of IT848 on oxidative stress was selective for MM cells while only minimal changes were observed in normal PBMCs (Fig. 4A and B). Specifically, IT848 treatment induced an increase of cellular superoxide (Fig. 4C) but not of hydrogen peroxide levels. Glutathione is one of the major ROS-scavenger systems protecting cells from oxidative stress. We therefore compared reduced glutathione (GSH) levels in MM.1S and U266 cells with and without IT848 treatment and found up to 70% depletion of the GSH pool with increasing drug concentrations (Fig. 4D). Control cells (splenocytes and NIH-3T3 cells) showed a negligible inhibition pattern, reinforcing our previous observation that myeloma cells are selectively sensitive to oxidative stress induction by IT848. Repletion of GSH partially reversed the anti-myeloma effect of IT848 in assays monitoring cell growth (Fig. 4E) and apoptosis (Supplementary Figure 5), validating GSH depletion in cancer cells as a critical component of the mechanism of action of our drug. For in vivo confirmation of its pro-oxidant properties, we tested IT848-mediated ROS induction in NGS mice inoculated with MM.1S cells. Administration of the light emitting molecule coelenterazine resulted in enhanced chemiluminescent detection of superoxide anion (29) in the treatment group, suggesting that IT848 treatment can achieve in vivo modulation of the cellular redox homeostasis (Fig. 4 F).

Figure 3. IT848 treatment induces distinctive gene expression patterns in multiple myeloma cells.

U266 and MM.1S cells were treated with IT848 (4uM) or DMSO for 6 hours and gene expression was analyzed by RNA-seq (n=3). A+B. Volcano plot (left panel) and principal component analysis (right panel) showing significantly upregulated genes (p<0.05) and a clearly distinguishable pattern in IT848 versus vehicle treated U266 cells (A) and MM.1S cells (B) as determined by RNA-seq. Statistical significance was determined by linear modelling and Bayesian statistics after correcting for multiple testing using the Benjamini–Hochberg procedure. C. Gene set enrichment analysis of MM.1S cells. Enrichment plots of categories that are enriched in genes correlated with IT848 treatment and representative genes from each category are shown. Statistical significance was determined by Wald test with Benjamini-Hochberg’s multiple-comparison correction.

Figure 4. IT848 mediates oxidative stress selectively in MM cells.

A+B, Myeloma cell lines U266 and MM.1S and human PBMC (as control) were treated with IT-848 (4 uM) or control vehicle for 6 hours. HMOX1 and OSGIN1 gene expression levels by qPCR are shown in A and B, respectively. C, U266 and MM.1S cells were further analyzed by Cellular Superoxide Detection Assay Kit to monitor real time superoxide production in live cells post-IT-848 treatment and normalized to the vehicle treatment. Cells were treated with Tert-Butyl hydroperoxide (TBHP, 200 uM) as a positive control to generate high levels of oxidative stress. D, Myeloma cell lines (U266 and MM.1S) and control cells (splenocytes and NIH3T3 cells) were cultured in presence of IT848 (1, 2, 4 and 6 uM) or empty vehicle for 24 h. Cellular levels of reduced glutathione levels in response to oxidative stress caused by IT-848 treatment were measured using ThiolTracker Violet GSH detection reagent as described in Materials and Methods. Mean and SEM of normalized values are presented (percentages of vehicle-treated baseline levels). E, MM.1S cells were cultured with empty vehicle, IT848 (4 uM), L-glutathione reduced (1 mM GSH) and a combination of IT848 and GSH and luciferase assay was performed over a 72-h period to demonstrate the antioxidant effect of GSH repletion on IT848 action and anti-myeloma potency. Values represent mean ± SEM (n=3, technical replicates) at each timepoint. One of two independent experiments is presented. F, Coelenterazine administration in vivo results in chemiluminescent detection of superoxide anion (n=3 per group). Changes in regional reactive oxygen species production (localization and magnitude) in response to IT-848 and vehicle administration over 4.5 h period are presented. All data are representative of three independent experiments.

IT848 is effective against MM xenografts and patient cells

Next, we investigated the pharmacokinetics (PK) of IT848 by analyzing its whole blood or plasma levels following oral (PO), intraperitoneal (IP), or intravenous (IV) administration (Supplementary Figure 6). We found that the maximum plasma concentration (c_max) and half-life (t_1/2) of IT848 following all three routes of administration was sufficient to achieve sustained plasma concentrations equivalent to the 4 to 6 µM range, the therapeutic window associated with in vitro efficacy (see Fig. 2A and B). Additionally, toxicology studies demonstrated the absence of organ toxicities during a 12-day course of IT848 treatment.

For in vivo efficacy testing of IT848 we first utilized a xenograft model of MM and included the proteasome inhibitor bortezomib (30,31). We chose bortezomib as a reference drug since it is a well-established anti-myeloma agent with NF-κB inhibitory properties but a different mechanism of action than IT848, hypothesizing that combination therapy should be safe and may producde additive effects. Mice were treated with either vehicle (control), bortezomib monotherapy, IT848 monotherapy, or the combination of the two. Monitoring disease progression by in vivo bioluminescence imaging, we observed significant efficacy with both bortezomib (p = 0.008) and IT848 monotherapy (p < 0.001) (Fig. 5A and B). Notably, IT848 showed superior tumor growth inhibition compared to bortezomib (p = 0.032), and combination treatment resulted in improved efficacy when compared with either monotherapy (p < 0.001, p = 0.029). RNA-seq comparing MM.1S cells treated with bortezomib alone versus bortezomib plus IT848 (Supplementary Figure 7) revealed a more significant downregulation of seven NF-κB related genes with combination treatment, including the NF-κB essential modulator (NEMO) (P = 0.02), which is a validated target for molecular cancer therapy (32–34).

Figure 5. IT848 is effective as monotherapy, and it provides additive effects in combination with bortezomib.

NSG mice received 2.5×10⁶ luciferase expressing MM.1S cells intravenously. After 10 days engraftment was confirmed and mice were assigned to the following groups: 1. empty vehicle I.P. 3x/week x 4 weeks, 2. bortezomib 0.5mg/kg I.P. 2x/week x 4 weeks, 3. IT848 10mg/kg I.P. 3x/week x 4 weeks, 4. Bortezomib + IT848 × 4 weeks. A, MM progression was analyzed by in vivo bioluminescence imaging at the indicated time points. Pseudocolor images superimposed on conventional photographs are presented. One of two independent experiments is shown. B, Mean and SEM of bioluminescence intensities are presented (n=7–10). The area under the curve (AUC) was used to summarize intensity values for each subject and a permutation test was performed to determine of there was a significant difference between groups. One of two independent experiments is shown. C, Cause-specific survival (death from disease) is presented. Differences between groups were analyzed by Log-rank test. Combined data from two independent experiments are presented (n=10–20). D+E, CD138-selected cells from patients with newly diagnosed MM (n=2) and plasma cell leukemia (n=1) were incubated in the presence of IT848 or empty vehicle (control) for 24 hours. D, apoptosis of patient cells was analyzed by Annexin V/7AAD staining. Representative FACS plots and mean and SEM are presented. E, proliferation of patient cells was analyzed by Ki-67 assay. Mean and SEM are presented.

However, frequent intraperitoneal administration, which is a clincially irrelevant drug devivery route but commonly used in preclinical animal models, of IT848 for more than two weeks was associated with considerable toxicity (such as development of ascites and diarrhea) and concordant treatment-associated mortality in the range of 20 to 30 percent (Fig. 5A and Supplementary Figure 8). Analysis of cause-specific survival revealed that despite this mouse model-specific drawback, the compound still mediated therapeutically beneficial in vivo efficacy resulting in significantly improved disease-specific survival (Fig. 5C). These observations suggest the need for the development of more clinically relevant better tolerated treatment protocols based on intravenous or oral delivery routes. Importantly, we were recently able to demonstrate that less frequent dosing of intravenously delivered IT848 was efficacious in the absence of toxicity (Supplementary Figure 9), and additional studies with the goal to refine and optimize our protocols are currently underway.

Encouraged by the effectiveness of IT848 against MM cell lines both in vitro and in vivo we sought to assess the response of primary MM cells to IT848. We therefore performed apoptosis and proliferation assays with MM samples derived from two patients with newly diagnosed MM and one patient with highly aggressive advanced disease (plasma cell leukemia). These assays confirmed that IT848 induced cell death (Fig. 5 D) and inhibited proliferation in a dose-dependent manner (Fig. 5 E) not just in MM cell lines but also primary MM cells.

IT848 can be harnessed to overcome immunosuppression in the MM microenvironment

Finally, given the known immunomodulatory effects of NF-κB inhibition, we investigated whether IT848 treatment could be exploited to maximize immunotherapeutic approaches in MM. We previously demonstrated that small molecules inhibiting NF-κB DNA binding are active against alloimmunity (acute graft-versus-host disease) without compromising T cell-mediated protective immunity (21,35). Furthermore, we recently discovered that our drug candidate IT848 was active against autoimmunity (rheumatoid arthritis model, Supplementary Figure 10) and did not reduce the cytotoxic activity of CAR T cells (Supplementary Figure 11). Since NF-κB regulates PD-L1 expression in B cell lymphomas (8), we assessed PD-L1 levels in lymphoma and MM cells treated with IT848 and found that IT848 treatment resulted in the downregulation of PD-L1 expression in both cell types (Supplementary Figure 12), suggesting that IT848 might be a promising candidate for combination therapy with PD-1 blockade in B cell malignancies. In addition to suppression of the PD-1/PD-L1 axis, NF-κB inhibition might enhance anti-myeloma immunity by modulating either myeloid-derived suppressor cells (MDSCs) or Tregs in the BM microenvironment in the setting of MM (36–38), providing an additional rationale for combining IT848 with an immune checkpoint inhibitor strategy. To test our hypothesis we investigated the efficacy of IT848/anti-PD-1 combination both in vitro and in vivo in the immunocompetent mouse model 5TGM1 (39). In vitro co-culture experiments of BM mononuclear cells and 5TGM1 cells in the presence and absence of IT848, anti-PD-1, or their combination, revealed that only combination treatment resulted in significant anti-myeloma activity (Fig. 6A). Furthermore, PD-1 expression was decreased on all T cells of the combination treatment group (Fig. 6B). While this decrease in PD-1 expression also included Tregs (Supplementary Figure 13A), analysis of Ki-67 expression by PD-1 positive Tregs showed no evidence of Treg hyperproliferation (Supplementary Figure 13A), a phenomenon that has been associated with cancer progression following anti-PD-1 therapy (40). Combination treatment also resulted in increased CD25 expression by cytotoxic T lymphocytes (CTLs) (Fig. 6B) and an increase in the frequency of activated CTLs (CD25⁺IFNγ⁺) (Fig. 6B). We also analyzed immunosuppressive, Treg-promoting growth factors and cytokines such as TGF-β and IL-10 (41). However, intracellular cytokine analysis of M-MDSCs and PMN-MDSCs showed overall no significant alterations in TGF-β and IL-10 expression in any of the tested conditions (Supplementary Figure 13B). We next compared the same conditions (control, IT848 monotherapy, anti-PD-1 monotherapy, and combination therapy) in the 5TGM1 model in vivo. We found that a 14-day treatment course of IT848/anti-PD-1 reduced MM progression and improved progression free survival (Fig. 6C and D). Amongst immunosuppressive cell populations in the BM of MM bearing mice, the number of Tregs (but not of MDSCs) was significantly decreased in mice receiving IT848/anti-PD-1 combination treatment (Fig. 6E, Supplementary Figure 14), and CD25 expression of Tregs was lowest in Tregs of the combination therapy group (Fig. 6E), suggestive of decreased Treg-mediated immunosuppression of the tumor microenvironment. Collectively, our findings suggest that the observed therapeutic effect of IT848/anti-PD-1 combination therapy was at least partially due to modulation of the immunosuppressive BM microenvironment.

Figure 6. Modulation of the BM microenvironment by IT848 can be exploited for combination therapies.

A+B: 100,000 dtomato-expressing 5TGM1 cells were cultured with 900,000 BM mononuclear cells from C57BL/6 mice and treated with empty vehicle (control), IT848 (4uM), Anti-PD-1 antibody (10ug/ml), or IT848 + Anti-PD-1 antibody. After 72 hours of co-culture cells were analyzed by multiparameter flow cytometry. Mean and SEM of one of three independent experiments are presented (n=5). A, left panel: dtomato⁺ percentage of CD45⁺ live cells; right panel: absolute number of live 5TGM1 cells. B, left panel: PD-1 median fluorescence intensity (MFI) of CD3⁺ Tells; center panel: CD25 MFI of CD8⁺ T cells, left panel: cells were gated on CD3⁺CD8⁺ T cells and analyzed for CD25⁺interferon (IFN)-γ⁺ cells. C–F, C57Bl/KaLwRij mice received 1.5×10⁶ luciferase expressing 5TGM1 cells intravenously. After 14 days engraftment was confirmed and mice were assigned to the following groups: 1. Empty vehicle I.P. 5x/week x 2 weeks, 2. IT848 10mg/kg I.P. 5x/week x 2 weeks, 3. Anti-PD-1 200ug I.P. 3x/week x 2 weeks, and 4. IT848 + Anti-PD-1 combination therapy x 2 weeks. C, MM progression was analyzed by in vivo bioluminescence imaging. Mean and SEM of bioluminescence intensities are presented (n=5). D, Progression-free survival is presented. Differences between groups were analyzed by Log-rank test. C+ D, One of two independent experiments is shown (n=5). E, BM was harvested on day 14 and analyzed for the presence of regulatory T cells (Tregs). Left panel: Treg (CD45⁺CD3⁺CD4⁺CD25⁺FoxP3⁺) frequency of CD3⁺CD4⁺ T cells; Middle panel: Treg frequency of CD45⁺ BM cells; right panel: CD25 MFI of CD3⁺CD4⁺FoxP3⁺ T cells. Mean and SEM are presented. Combined data from three independent experiments are shown (n=7–16). F, Schematic diagram of IT848-mediated effects on MM cells and microenvironment. Three major mechanisms contributing to anti-myeloma activity of IT848 are depicted: Inhibition of NF-KB DNA binding in MM cell nuclei, Oxidative stress induced apoptosis of MM cells and enhancement of immune checkpoint blockade by depletion of bone marrow Treg pool.

Discussion

Even though constitutive NF-κB activity is crucial to the survival and progression of MM cells and plays a role in the development of resistance to cancer therapeutics, targeting NF-κB safely remains a challenging balancing act. In addition, the crosstalk between myeloma cells and the immune components of the tumor microenvironment within the BM further complicates the therapeutic design (42). As a result, even though hundreds of NF-κB inhibitor compounds have been designed over the years, to date there are no approved agents that directly inhibit NF-κB transactivation (43). However, multiple indirect strategies to target NF-κB, including agents that were not originally conceived as NF-κB inhibitors, have yielded some significant advances. Examples include immunomodulatory drugs (IMiDs) such as lenalidomide that suppresses IκB kinase activity (via TNFα, IL-1β inhibition) (44), and proteasome inhibitors such as bortezomib and carfilzomib that target IκBα degradation (30). These classes of drugs have been used in MM and B cell lymphoma but lack specificity, have off-target effects, and dose limiting toxicities (31,45). Inhibitors of a central regulator of NF- κB activation, the IκB kinase (IKK) complex (46), on the other hand, are highly NF-κB specific, but their efficacy in clinical trials has been underwhelming (47). As far as pharmaceutical inhibition of NF-κB DNA binding is concerned, a limited number of molecules (DHMEQ, SEMBL, PBS-1086) derived from the natural compound epoxyquinomicin C have been researched in the past, but the poor bioavailability and PK restricted their use and potential for clinical translation (48–50). On the other hand, our synthetic compound IT848 is characterized by drug-like physicochemical properties. IT848 inhibits NF-κB transcriptional activity by reducing the ability of NF-κB subunits that migrated to the nucleus to bind to DNA, thereby interfering with both canonical and non-canonical NF-κB signaling. Importantly, during healthy steady state conditions, NF-κB is kept in an inactive state in the cytoplasm and as such, in the absence of NF-κB activating stimuli, healthy cells are not expected to be negatively affected by pharmaceuticals specifically preventing DNA binding of NF-κB. Although small molecules alone are unlikely to induce durable responses in aggressive cancers such as MM, we demonstrated significant in vivo efficacy of IT848 monotherapy. The above-mentioned adverse events following prolonged courses of clinically irrelevant intraperitoneal administration of IT848 in mice (e.g., diarrhea, ascites) do not occur with intravenous or oral administrations; both administration routes are viable options, given the favorable PK profile of IT848 and its efficacy in recently used both oral and intravenous regimens in mouse models of T cell lymphoma (Supplementary Figure 9) and rheumatoid arthritis (Supplementary Figure 10).

We found an additional mechanism of action of IT848 that exploits the heightened ROS production in cancer cells, a consequence of redox homeostasis imbalance, oncogenic stimulation, increased metabolic activity, and mitochondrial malfunction (51,52). We observed a dose-dependent increase in superoxide levels and a decrease of GSH in MM cell lines but not in healthy control cells where basal ROS levels are minimal. A previous study found a direct correlation between GSH depletion and ROS-mediated apoptosis of cancer cells (53). Notably, NF-κB activity is required for cellular GSH production (54), and GSH depletion mediated specifically by NF-κB inhibition has been proposed as a therapeutic strategy for cancer treatment (54). Our findings were further reiterated by RNA-seq and qPCR data supporting a transcriptional upregulation of oxidative stress signatures and genes involved in oxidative stress-induced apoptosis. Collectively, our data indicate that therapeutic manipulation by IT848 can preferentially kill myeloma cells via NF-κB/ROS-mediated mechanisms utilizing both caspase-dependent and independent apoptosis pathways. Induction of cancer cell-selective oxidative stress by IT848 is closely related to GSH depletion and inadequate upregulation of oxidative stress response genes, which can both be attributed to the NF-κB inhibitory effects of the molecule. This suggests that IT848 has the potential to be effective against a wide range of cancer cells and its clinical application may not have to be limited to cancer cell types that rely on constitutive NF-κB activity for their survival.

We showed that IT848 induces prominent anti-myeloma activity against cell lines, primary cells, and in a xenograft model, where combining IT848 with bortezomib enhanced antitumor efficacy. However, since tumor modeling in the absence of a functional immune system limits clinical relevance (especially when studying approaches that have the potential to alter immune responses), we additionally explored the effect of IT848 in combination with PD-1 blockade in an immunocompetent mouse model of MM. It has been shown that malignant plasma cells are able to induce T cell anergy via PD-L1/PD-1 interaction (55), and PD-L1 expression in MM cells correlates with BCMA signaling, which activates the canonical NF-κB pathway (56). We found that single agent treatment with either anti-PD-1 checkpoint blockade or IT848 was not effective, but combination therapy significantly decreased infiltration of the BM by Tregs and MM cells, and it improved survival. In vitro testing of this combination therapy confirmed that combination of direct NF-κB inhibition with immune checkpoint blockade induces a shift of the MM microenvironment toward enhanced cytotoxicity. MDSC populations were not quantitatively affected by IT848 or anti-PD-1 treatment, and there was no evidence of a role of MDSCs in promoting Treg activity through cytokines release. However, we found that IT848/anti-PD-1 combination treatment resulted in significant downregulation of the interleukin 2 (IL-2) receptor alpha chain (CD25), which is an NF-κB target gene (57), in Tregs, suggesting that decreased responsiveness to the Treg growth factor IL-2 may be a contributing factor for the depletion of BM resident Tregs (58). In addition, IT848 likely suppresses thymic Treg development, given that NF-κB regulates FoxP3 expression (59). Lastly, it is well-established that immune checkpoint blockade can be associated with autoreactive T cell mediated adverse events such as colitis or hepatitis. Our previous findings in GVHD (35) and rheumatoid arthritis models suggest that IT848 treatment might ameliorate autoimmune-related adverse events associated with immune checkpoint blockade (60) without compromising anti-tumor activity of T cells (21,35), providing an additional rationale for the development of IT848 combination therapies with immune checkpoint blockade. Taken together, these cellular and molecular interactions suggest that while high-dose IT848 displays anti-tumor activity in the absence of a functional immune system, harnessing the immunomodulary effects of IT848 for cancer immunotherapy promises synergistic therapeutic effects at lower doses in the absence of toxicities. Studies designed to further validate the latter concept are currently underway.

In conclusion, our study shows that the small molecule IT848 is a novel and highly specific type of NF-κB inhibitor and cancer cell-selective oxidative stress inducer with significant anti-myeloma activity against both cell lines and primary cells. IT848 demonstrated in vivo efficacy as single agent and in combination therapies designed to overcome immunosuppression and drug resistance in the tumor microenvironment. IT848 has a favorable PK profile and we found no toxicities associated with clinically relevant administration routes. Our data therefore provide a strong preclinical rationale for the clinical development of IT848 in B cell malignancies.