Targeting the ribosome to treat multiple myeloma

Abstract

The high rates of protein synthesis and processing render multiple myeloma (MM) cells vulnerable to perturbations in protein homeostasis. The induction of proteotoxic stress by targeting protein degradation with proteasome inhibitors (PIs) has revolutionized the treatment of MM. However, resistance to PIs is inevitable and represents an ongoing clinical challenge. Our first-in-human study of the selective inhibitor of RNA polymerase I transcription of ribosomal RNA genes, CX-5461, has demonstrated a potential signal for anti-tumor activity in three of six heavily pre-treated MM patients. Here, we show that CX-5461 has potent anti-myeloma activity in PI-resistant MM preclinical models in vitro and in vivo. In addition to inhibiting ribosome biogenesis, CX-5461 causes topoisomerase II trapping and replication-dependent DNA damage, leading to G2/M cell-cycle arrest and apoptotic cell death. Combining CX-5461 with PI does not further enhance the anti-myeloma activity of CX-5461 in vivo. In contrast, CX-5461 shows synergistic interaction with the histone deacetylase inhibitor panobinostat in both the Vk∗MYC and the 5T33-KaLwRij mouse models of MM by targeting ribosome biogenesis and protein synthesis through distinct mechanisms. Our findings thus provide strong evidence to facilitate the clinical development of targeting the ribosome to treat relapsed and refractory MM.

Keywords: CX-5461; MT: Regular Issue; RNA polymerase I; multiple myeloma; panobinostat; ribosome biogenesis.

Graphical abstract

Figure 1

CX-5461 impairs ribosome biogenesis and induces the DDR in MM cells (A) CX-5461 EC50 doses that induce 50% cell death at 24 h in HMCLs. Cell death was measured by the propidium iodide exclusion assay and the values of EC50 are outlined in Figure S1A. Error bars represent mean ± SEM. Statistical analysis was performed using two-tailed one-way ANOVA Mann-Whitney test. ∗p < 0.05. (B) CX-5461 reduces Pol I transcription rates in MM cells. Cells were treated with 1 μM CX-5461 for 1 h and the abundance of 47S pre-rRNA 5′ETS transcript was measured by quantitative real-time PCR and normalized to β-2-microglobulin and expressed as fold change relative to corresponding vehicle control. Primer sequenced are listed in Table S5. The cell lines are ordered based on EC50 values of CX-5461 doses from low to high values. (C) AMO-1 cells were treated with either vehicle, 0.5 μM CX-5461, or cycloheximide (CHX) 100 μg/mL (a control for inhibiting protein synthesis) for 6 h, and labeled with 50 μM L-azidohomoalanine for 1 h before fixing. The newly synthesized proteins were labeled with Alexa Fluor 488-azide by Click chemistry and analyzed by flow cytometry. Error bars represent mean ± SEM (n = 3). Statistical analysis was performed using one sample t test. ∗∗p <0.01. (D) Co-IF assays of fibrillarin (FBL) and nucleophosmin (NPM) in AMO-1cells treated with 0.5 μM CX-5461 or vehicle for 3 h. Representative images of two biologically independent experiments. Scale bar, 10 μm. Signal intensities were analyzed using CellProfiler. Raw mean intensity values were normalized to the median value of the vehicle control. Error bars represent mean ± SD. Statistical analysis was performed using a two-tailed nonparametric Mann-Whitney t test. ∗∗∗∗p < 0.0001. (E) Western blotting of cells treated with 0.5 μM CX-5461. Vinculin was probed as a loading control. Representative images of n = 3. (F) Co-IF analysis of pRPA S33 and the nucleolar protein UBF in cells treated with 0.5 μM CX-5461 or vehicle for 3 h. Cells were pulse labeled with EdU 30 min prior to drug treatment. (G) Co-IF analysis of γH2AX with UBF in cells treated as (F). (F and G) Scale bar, 10 μm. Quantification of nucleolar (signal overlap with UBF) and nuclear (signal overlap with DAPI) of pRPA S33 (F) or γH2AX (G) signal intensity was performed using CellProfiler. n = 300 EdU+ and n = 300 EdU-negative cells per treatment condition were analyzed over three or four biologically independent experiments. Raw mean intensity values were normalized to the median value of vehicle control. Error bars represent mean ± SD. Statistical analysis was performed using one-way ANOVA, Kruskal-Wallis multiple comparisons test. ∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01; ns, non-significant with p > 0.05. (H) RADAR assay. DNA-protein covalent complexes were isolated and TOP2α probed. Representative images of n = 3.

Figure 2

CX-5461 has potent activity in PI-resistant MM models (A) Culture-adapted bortezomib-resistant human JJN-3 and murine 5T33 myeloma cells and parental controls were treated with bortezomib (upper) and CX-5461 (bottom) for 72 h and cell death was measured using propidium iodide exclusion assay. Error bars represent mean ± SEM of n = 3. (B) C57BL-KaLwRij mice were transplanted with bortezomib-resistant 5T33 cells and treated with either 5 mg/kg bortezomib via i.p. injection weekly, 35 mg/kg CX-5461 via oral gavage twice weekly, or the vehicle control (n = 6 per group). Kaplan-Meier plot is shown. Statistical analysis was performed using log rank test. ∗∗p < 0.01. (C) Label-free quantitative proteomic analyses of parental and bortezomib-resistant 5T33 cells treated with 0.5 μM CX-5461 or vehicle for 24 h. Gene Ontology enrichment analysis identified the top-ranked biological processes regulated by CX-5461. (D) Gene Ontology enrichment analyses of differentially expressed proteins shared by both 5T33_S and 5T33_R cells identified the top-ranked biological processes regulated by CX-5461. (E) Quantitative phosphoproteomic analyses of parental and bortezomib-resistant 5T33 cells treated with 0.5 μM CX-5461 or vehicle for 24 h. Gene Ontology enrichment analysis identified the top-ranked biological processes regulated by CX-5461. (F) The most significantly upregulated and downregulated kinases (p < 0.05, substrates ≥ 3) identified from kinase-substrate enrichment analysis using Phosphomatics (version 2.0) are shown and ranked by the kinase Z scores. (G) Cells were treated with 0.5 μM CX-5461 or vehicle for 24 h and protein lysates collected for western blotting. Vinculin was probed as a loading control. Representative images of n = 3. (H and I) Gene Ontology enrichment analyses of differentially expressed proteins (H) and differently phosphorylated proteins (I) in bortezomib-resistant cells relative to parental cells identified the top-ranked biological processes associated with bortezomib resistance.

Figure 3

The combination with PIs does not further enhance the anti-myeloma activity of CX-5461 (A) Cells were treated with CX-5461 and bortezomib for 72 h and the number of viable cells were counted by Coulter Counter. The mean of n = 3 is shown in the dose-response matrix (left). The synergy scores were calculated using Bliss (middle) and Loewe (right) models on SynergyFinder Application website (SynergyFinder.com). (B) C57BL/KaLwRij mice were transplanted with 2 × 10⁶ luciferase-expressing 5T33 MM cells by intravenous tail injection. The treatment started on day 12 post transplantation with carfilzomib 5 mg/kg i.p. injection weekly, CX-5461 25 mg/kg oral gavage three times per week, the drug combination, or the vehicle control (n = 8 per condition). A Kaplan-Meier plot is shown. Statistical analysis was performed by log rank test. ∗∗p < 0.01; ns, non-significant. (C) Western blotting of HMCLs treated with 5 nM bortezomib or 500 nM CX-5461 for 24 h. Actin was probed as a loading control. Representative images of n = 3.

Figure 4

CX-5461 is synergistic with the histone deacetylase inhibitor panobinostat (A) Cells were treated with increasing concentrations of CX-5461 in the presence or absence of 50 nM panobinostat for 72 h. Cell death was assessed by propidium iodide exclusion assay. Error bars represent mean ± SEM, n = 3. (B) Combination indices (CIs) calculated by CalcuSyn software. CI < 1 indicates synergistic cell death (blue), CI > 1 indicates antagonism (red), CI = 1 indicates additive effect (black). Error bars represent mean ± SEM, n = 3. (C and D) 3.5 × 10⁵ spleen cells from Vκ∗MYC mice (clone 4929) were transplanted into C57BL/6 mice following sub-lethal irradiation. After 7 weeks, mice were randomized based on the paraprotein levels into four groups and started treatment with 35 mg/kg CX-5461 twice weekly via oral gavage (n = 6 mice), 7.5 mg/kg panobinostat via i.p. injection three times weekly (n = 6 mice), the combination of CX-5461 and panobinostat (n = 6 mice), or vehicle control (n = 5 mice). The paraprotein level is presented as the percentage of total proteins, as measured by serum protein electrophoresis at baseline, 2-week, 6-week, and 10-week time points (C). Error bars represent mean ± SEM. Statistical analysis is performed using one-way ANOVA, Tukey’s multiple comparations test. ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Kaplan-Meier plot (D). Statistical analysis is performed using log rank test. ∗p < 0.05, ∗∗p < 0.01. (E–G) 2 × 10⁶ luciferase-expressing 5T33 cells were transplanted into C57BL-KaLwRij mice. Mice were randomized based on bioluminescent imaging on day 12 post transplantation into four groups and started treatment with 25 mg/kg CX-5461 via oral gavage three times weekly (n = 6 mice), 5 mg/kg panobinostat via i.p. injection three times weekly (n = 6 mice), the combination of CX-5461 and panobinostat (n = 5 mice), or vehicle control (n = 5 mice). Bioluminescent images are shown (E). Disease burden measured by bioluminescent signal intensity (photons per second) (F). Kaplan-Meier plot (G). Statistical analysis was performed using log rank test, ∗∗p < 0.01.

Figure 5

The combination of CX-5461 and panobinostat enhances multiple myeloma cell death through inhibition of protein synthesis via independent pathways (A) MM.1S cells were treated with vehicle, 500 nM CX-5461, 50 nM panobinostat, or the combination for 24 h and cellular RNA was extracted for 3′ RNA-seq analysis (n = 3). Changes of activities of hallmarks upon treatment were determined by single-sample GSEA using MSigDB hallmark gene sets. (B and C) AMO-1 (B) and JJN3 (C) cells were treated as in (A) and analyzed by western blotting. Vinculin was probed as a loading control. Representative images of n = 3. (D and E) AMO-1(D) and JJN-3 (C) cells were treated as in (A) for 6 h and labeled with 50 μM L-azidohomoalanine for 1 h before fixing followed by labeling with 1 μM Alexa Fluor 488-azide and flow cytometry analysis. The median fluorescence intensity was normalized to vehicle control. Error bars represent mean ± SEM, n = 3. Statistical analysis was performed using one-way ANOVA Tukey’s multiple comparisons test. ∗p < 0.1, ∗∗p < 0.01.