The biguanide polyamine analog verlindamycin promotes differentiation in neuroblastoma via induction of antizyme

Abstract

Deregulated polyamine biosynthesis is emerging as a common feature of neuroblastoma and drugs targeting this metabolic pathway such as DFMO are in clinical and preclinical development. The polyamine analog verlindamycin inhibits the polyamine biosynthesis pathway enzymes SMOX and PAOX, as well as the histone demethylase LSD1. Based on our previous research in acute myeloid leukemia (AML), we reasoned verlindamycin may also unblock neuroblastoma differentiation when combined with all-trans-retinoic acid (ATRA). Indeed, co-treatment with verlindamycin and ATRA strongly induced differentiation regardless of MYCN status, but in MYCN-expressing cells, protein levels were strongly diminished. This process was not transcriptionally regulated but was due to increased degradation of MYCN protein, at least in part via ubiquitin-independent, proteasome-dependent destruction. Here we report that verlindamycin effectively induces the expression of functional tumor suppressor-antizyme via ribosomal frameshifting. Consistent with previous results describing the function of antizyme, we found that verlindamycin treatment led to the selective targeting of ornithine decarboxylase (the rate-limiting enzyme for polyamine biosynthesis) as well as key oncoproteins, such as cyclin D and Aurora A kinase. Retinoid-based multimodal differentiation therapy is one of the few interventions that extends relapse-free survival in MYCN-associated high-risk neuroblastoma and these results point toward the potential use of verlindamycin in this regimen.

Trial registration: ClinicalTrials.gov NCT02395666 NCT01586260.

Conflict of interest statement

The authors declare no competing interests.

© 2021. The Author(s).

Figures

Fig. 1. Verlindamycin inhibits neuroblastoma cell proliferation and enhances growth-inhibitory effect of ATRA.

A Structure of verlindamycin. B Dose response by SRB assay for a panel of non MYCN-amplified (blue) and MYCN-amplified (red) neuroblastoma cell lines treated with verlindamycin for 72 h. 50% growth-inhibitory (GI50) values are shown for individual cell lines (left panel) and non-MYCN-amplified versus MYCN-amplified cell lines (right panel). C SK-N-BE(2)-C and Kelly cells cultured as tumor spheroids were treated with verlindamycin for 72 h and changes in spheroid diameter were assessed using a Celigo S Imaging Cell Cytometer. D Viability assessment as measured by CellTiter Glo (Promega) in SK-N-BE(2)-C and Kelly cells grown in monolayer. Metabolic activity of cells treated with 0.5× GI50 verlindamycin in combination with 1 μM ATRA was measured by CellTiter Glo after 6 days of treatment. E Cells treated with 0.5× GI50 verlindamycin and 1 μM ATRA for 10 days were tested for their ability to form colonies from single cells within 14 days. All the experiments were performed in triplicates and a representative result or mean is shown. Student’s t test was performed to calculate statistical significance, *p < 0.05, **p < 0.005, ***p < 0.0005, error bars show standard deviation.

Fig. 2. Verlindamycin enhances ATRA-induced differentiation in neuroblastoma.

SK-N-BE(2)-C and SK-N-AS were treated with 0.5× GI50 verlindamycin and 1 μM ATRA for 6 days. A Brightfield microscopic pictures show morphological differences upon treatment. B Cells were fixed and stained with neurofilament light chain (NFL, red) and DAPI (blue). C mRNA expression of ATRA-target gene (CRABP2) and neural markers (RARB, RET) was measured by RT-qPCR and normalized to GAPDH. All the experiments were performed in triplicates and a representative result or mean is shown; error bars show standard deviation.

Fig. 3. Gene Set Enrichment Analysis (GSEA) of differential gene expression in SK-N-BE(2)-C cells following co-treatment with verlindamycin (2d) and ATRA.

Gene expression data generated by expression microarray analysis following 6 days of co-treatment with 1 μM ATRA and 0.5× Gi50 verlindamycin versus untreated (DMSO) control were analyzed using GSEA to extract biological knowledge and highly significantly enriched gene sets are shown. The most upregulated genes in vehicle control are shown on the left side (red), while the most upregulated genes following ATRA + 2d treatment are shown on the right side (blue). Black bars represent the positions of the vehicle control versus ATRA + 2d upregulated signature genes in the ranked list. Green curves represent the evolution gene density. Normalized enrichment scores (NES) reflect the degree to which genes are overrepresented. When the distribution is random, the enrichment score is zero. Enrichment of signature genes at the top of the ranked list results in a large positive deviation of the NES from zero. q-value false discovery rate (FDR)-adjusted q-value.

Fig. 4. Verlindamycin downregulates the expression of MYCN protein in MYCN-amplified neuroblastoma.

A MYCN protein levels were assessed in SK-N-BE(2)-C cells treated for 6 days: with different concentrations of verlindamycin (upper panel; GI50 = 3.5 μM) and with 0.5× GI50 verlindamycin combined with 1 μM ATRA (lower panel). BMYCN mRNA expression was measured by RT-qPCR (relative to GAPDH) in SK-N-BE(2)-C cells treated for 6 days with 0.5× GI50 verlindamycin alone or combined with 1 μM ATRA. C SK-N-BE(2)-C cells pretreated with 0.5× GI50 verlindamycin for 4 days were exposed to 25 µg/ml cycloheximide for up to 2 h. D SK-N-BE(2)-C cells pretreated with verlindamycin for 4 days were exposed to 10 µM MG-132 for 16 h. E PLA was performed on SK-N-BE(2)-C to assess MYCN-FBXW7 interaction; dots were quantified (left panel). All the experiments were performed in triplicates and a representative result or mean is shown; error bars show standard deviation. Student’s t test was performed to calculate statistical significance, *p < 0.05.

Fig. 5. Inhibition of KDM1A/LSD1 is not the mechanism of action of verlindamycin.

A SK-N-BE(2)-C cells were treated with a range of GSK-LSD1 concentrations combined with 1 μM ATRA for 6 days after which cell viability was assessed by CellTiter Glo. B MYCN expression level was tested in cells treated with GSK-LSD1 with or without addition of 1 μM ATRA for 6 days. C SK-N-BE(2)-C and Kelly cells with siRNA-mediated KDM1A knock-down (alongside non-targeting control) and treated with ATRA (for 6 days) were harvested to assess the expression of MYCN and LSD1. D mRNA expression of KDM1A was also tested in SK-N-BE(2)-C and Kelly cells treated with ATRA and KDM1A-targeting siRNA; expression measured by RT-qPCR relative to GAPDH. All the experiments were performed in triplicates and a representative result or mean is shown; error bars show standard deviation.

Fig. 6. Verlindamycin treatment induces frameshifting of antizymes 1 and 2.

A Schematic showing polyamine-induced frameshifting of antizyme. B Verlindamycin-induced frameshifting of antizymes 1 and 2 relative to 25 µM spermidine. Briefly, 293T cells were transfected with antizyme 1 or 2 frameshifting reporters or in-frame controls with the addition of 25 µM spermidine or verlindamycin as indicated. Percentage frameshifting (%FS) activity was determined with Dual-Glo (Promega) by obtaining firefly:renilla luciferase ratios, then dividing reporter values by in-frame control values. The background-corrected %FS activity of each compound concentration was then divided by the background-corrected %FS activity induced by 25 mM spermidine and multiplied by 100. All the experiments were performed in triplicates and a representative result is shown.

Fig. 7. Functional analysis of verlindamycin-induced frameshifting of antizyme.

A SK-N-BE(2)-C, Kelly, and SK-N-AS cells were treated with 0.5× GI50 verlindamycin for 4 days, after which the expression of Cyclin D1, Aurora A, and MYCN was measured. GAPDH expression is shown as a loading control. At the same time, the expression of BODC1 and COAZ1 mRNA was measured by qPCR, shown relative to GAPDH. D, E After 96 h of siRNA knock-down of OAZ1 combined with verlindamycin treatment, D the expression of OAZ1 targets was measured by western blotting. E Expression of OAZ1 mRNA was assessed by RT-qPCR (relative to GAPDH) in cells transfected with siRNA. All the experiments were performed in triplicates and a representative result or mean is shown; error bars show standard deviation. Student’s t test was performed to calculate statistical significance, *p < 0.05.

Fig. 8. DFMO affects neuroblastoma viability and MYCN expression but does not enhance differentiation-inducing potential of ATRA.

SK-N-BE(2)-C, Kelly, and SK-N-AS cells were treated with 5 mM DFMO, a well-studied ODC1 inhibitor, combined with 1 μM ATRA for 6 days. A Relative viability of those cells was measured by CellTiter Glo. B MYCN protein level was assessed in SK-N-BE(2)-C. C Cell morphology was observed via brightfield microscope. All the experiments were performed in triplicates and a representative result or mean is shown; error bars show standard deviation.