Targeting S6K/NFκB/SQSTM1/Polθ signaling to suppress radiation resistance in prostate cancer

Abstract

In this study we have identified POLθ-S6K-p62 as a novel druggable regulator of radiation response in prostate cancer. Despite significant advances in delivery, radiotherapy continues to negatively affect treatment outcomes and quality of life due to resistance and late toxic effects to the surrounding normal tissues such as bladder and rectum. It is essential to develop new and effective strategies to achieve better control of tumor. We found that ribosomal protein S6K (RPS6KB1) is elevated in human prostate tumors, and contributes to resistance to radiation. As a downstream effector of mTOR signaling, S6K is known to be involved in growth regulation. However, the impact of S6K signaling on radiation response has not been fully explored. Here we show that loss of S6K led to formation of smaller tumors with less metastatic ability in mice. Mechanistically we found that S6K depletion reduced NFκB and SQSTM1 (p62) reporter activity and DNA polymerase θ (POLθ) that is involved in alternate end-joining repair. We further show that the natural compound berberine interacts with S6K in a in a hitherto unreported novel mode and that pharmacological inhibition of S6K with berberine reduces Polθ and downregulates p62 transcriptional activity via NFκB. Loss of S6K or pre-treatment with berberine improved response to radiation in prostate cancer cells and prevented radiation-mediated resurgence of PSA in animals implanted with prostate cancer cells. Notably, silencing POLQ in S6K overexpressing cells enhanced response to radiation suggesting S6K sensitizes prostate cancer cells to radiation via POLQ. Additionally, inhibition of autophagy with CQ potentiated growth inhibition induced by berberine plus radiation. These observations suggest that pharmacological inhibition of S6K with berberine not only downregulates NFκB/p62 signaling to disrupt autophagic flux but also decreases Polθ. Therefore, combination treatment with radiation and berberine inhibits autophagy and alternate end-joining DNA repair, two processes associated with radioresistance leading to increased radiation sensitivity.

Fig. 1.

RPS6KB1 and response to radiation in prostate tumors. A. Log-rank test shows significant correlation between survival of patients with prostate cancer and RPS6KB1 expression. B. Parental, RPS6KB1-knockout (S6K-KO), and RPS6KB1-overexpressing (S6K-OE) LNCaP cells were exposed to escalating doses of ionizing radiation (0, 0.25, 0.5, and 1 Gy) and assessed for colony formation. Colonies were stained with crystal violet after 18–20 days (parental and S6K-KO), or 8–10 days (S6K-OE). Colonies with >50 cells were counted. Surviving fraction in response to radiation was calculated by normalizing to mock radiation control. Data shown are mean ± SD of three independent experiments with triplicate technical repeats. ****P < 0.0001 relative to control (Student’s t-test). C. Principal component analysis (PCA) of mechanical properties of tumor-associated circulating cells: EpCAM⁺ CTCs (canonical EpCAM+, CD45⁻, CD11c−), EMT-CTCs (EpCAM−, CD45⁻, CD11c−), M1-like macrophages (EpCAM−, CD45⁺, CD11c−), M2-like macrophages (EpCAM−, CD45⁺, CD11c+). D. PCA of stiffness, adhesion and deformation of circulating tumor cells isolated from mice implanted with parental and S6K-KO cells following exposure to radiation.

Fig. 2.

Effects of pharmacological inhibition of RPS6KB1 with berberine on cancer cell outcomes. A–B. Immunoblotting of whole-cell lysates from LNCaP and C4–2B cells treated with increasing doses of berberine (0, 1, 2.5, 5, 10, or 25 μg/ml) for 48 h to determine total and phosphorylated S6K1 protein levels. β-actin was used as a loading control. A representative blot from three independent experiments is shown. C–D. Logarithmically growing LNCaP (C) or C4–2B (D) cells were treated with escalating concentrations of berberine for 48 h. Colonies were allowed to form for 18–20 days. Surviving fraction was determined by counting colonies containing >50 cells normalized to vehicle control. Data shown are mean ± SD of three independent experiments with triplicate technical repeats. ****P < 0.0001 relative to control (Student’s t-test). E-F. Berberine inhibits migratory ability of LNCaP (E) and C4–2B (F) cells. Logarithmically growing LNCaP or C4–2B cells were seeded at a density of 3 × 10⁵ and 2 × 10⁵ cells/well, respectively. After a wound was created, cells were treated with increasing doses of berberine (0, 5, and 10 μg/ml). Images were used to document wound healing at 0, 12, 24, and 48 h after wound scratch using a Zeiss PrimoVert Inverted microscope. Migration distance was measured with ImageJ software. Wound area is displayed as mean ± SD of three independent experiments. ****P < 0.0001 relative to vehicle control (two-way ANOVA). G-I. Violin plots display cell adhesion (nN; G), cell stiffness (Young’s modulus, kPa; H), and deformation (μM; I) in parental LNCaP (P) cells treated with 5 or 10 μg/ml berberine for 48 h. Data presented is mean ± SD of two independent experiments. P values shown are relative to vehicle control (t-test).

Fig. 3.

Berberine interacts with RPS6KB1 and potentiates radiation effects. A. Logarithmically growing parental, RPS6KB1-knockout (S6K-KO), and RPS6KB1-overexpressing (S6K-OE) LNCaP cells were treated with 0, 5, 10, 15, 25, or 30 μg/ml berberine or vehicle control (methanol) for 72 h. CellTiter 96 Aqueous One solution assay was used to determine the effect on growth ability. Data presented are mean ± SD from three independent experiments. B. A ribbon representation of bilobal S6K1-berberine model highlights mode 1 of berberine (red stick) binding to ATP pocket (light blue, N-lobe; light green, C-lobe; dark blue, c-clasp; orange, P-loop; magenta; HM or hydrophobic motif; cyan sticks, interacting amino acids). C. The mode 2 of berberine (dark green stick, berberine) reveals its extensive rotation away from the canonical ATP-binding mode (mode 1) resulting in its close proximity to c-clasp region. D. An overlay of two modes of berberine-S6K1 interaction displays relative orientation of berberine in the ATP pocket and amino acid residues (cyan sticks) that stabilize the two modes. Logarithmically growing LNCaP (E) and C4–2B (F) cells at ~60–70 % confluency were pretreated with berberine (5 μg/ml) for 48 h and irradiated with 0, 0.25, or 0.5 Gy radiation. Cells were then plated (5000 cells/well) for colony formation over 18–20 days. Stained colonies with >50 cells were counted, and the surviving fraction was determined. Relative surviving fraction was calculated by normalizing to mock radiation control. Data presented are mean ± SD from two or three independent experiments with three technical replicates for each experiment. **P < 0.01, ****P < 0.0001, compared with vehicle control (two-way ANOVA). G. Two weeks after orthotopic implantation of C4–2B cells, athymic nude mice received no treatment (control), berberine alone (5 mg/kg in DMSO), radiation alone, berberine for three weeks followed by a single dose of 9 Gy radiation (CONC group), or radiation followed by berberine treatment (SEQ group; n = 6–8 per group). The experiment was terminated seven weeks after tumor cell implantation. Blood was collected to measure serum levels of PSA at the indicated time points.

Fig. 4.

The role of NFκB-SQSTM1 in berberine’s effect on autophagy. A. Volcano plot of top upregulated and downregulated genes in S6KO vs. parental LNCaP cells (n = 3). Total RNA extracted from parental and S6K-KO cells was subject to transcriptome analysis. Differentially expressed genes (DEGs) were filtered with FDR-adjusted P-value <0.05 and fold change >2. B. Berberine inhibits SQSTM1 reporter activity. SQSTM1-luciferase reporter activity in LNCaP cells treated with berberine (5 μg/ml) 48 h post transfection. To demonstrate the role of NFκB in berberine-associated reduction of SQSTM1 reporter activity, cells were pretreated with the NFκB inhibitor BAY 11–7082 (10 μM) for 24 h following transfection, followed by treatment with berberine (5 μg/ml) for an additional 48 h. Data are mean ± SD of SQSTM1-luciferase activity normalized to Renilla-luciferase activity from at least 3–5 (3 for NFκB inhibitor studies; 5 for berberine studies) independent experiments. ****P < 0.001, relative to vehicle control for berberine treatment and relative to berberine for NFκB inhibition (Student’s t-test). C. Immunoblotting of extracts from LNCaP cells that were pretreated with berberine (0, 1, 5, or 10 μg/ml) for 48 h and exposed to radiation (1 Gy) or no radiation (0 Gy). Immunoblotting was carried out with antibodies against total and phosphorylated S6K1 and total and phosphorylated p65. β-actin was used as a loading control. A representative blot from three independent experiments is shown. D. Whole-cell extracts from C4–2B cells treated with berberine alone (5 μg/ml), radiation alone (2 Gy), or a combination of berberine and radiation were immunoblotted with antibodies against total and phosphorylated p65, SQSTM1, and pIκBα. β-actin was used as a loading control. A representative blot from three independent experiments is shown. E. Autophagy induction increases radiosensitivity. Logarithmically growing LNCaP cells were treated with rapamycin (100 nM), chloroquine (10 μM), or solvent control for 48 h. After exposure to radiation (0, 0.25, or 0.5 Gy), cells were re-plated cells and allowed to form colonies for 18–20 days. Colonies containing >50 cells were counted. Data presented are the mean number of colonies ± SD from three independent experiments, each with three technical replicates. **P < 0.01, ****P < 0.0001, compared with vehicle control (two-way ANOVA). F-G. Berberine treatment disrupts autophagic flux in C4–2B (F) and LNCaP (G) cells. Immunoblotting of extracts from C4–2B or LNCaP cells treated with berberine (5 μg/ml; BER), chloroquine (10 μM; CQ), combination of berberine plus chloroquine (BC), or vehicle control (VCM, methanol; VCD, DMSO) at base line (extracts prepared immediately following radiation exposure) and following 12 h exposure (0.25Gy). Immunoblotting was performed with antibodies against total and phosphorylated RPS6KB1, SQSTM1, and LC3. β-actin was used as a loading control. A representative blot from three independent experiments is presented. H. LNCaP cells treated with berberine (5 μg/ml) and chloroquine (10 μM) alone or in combination for 48 h were exposed to 0.25 Gy radiation. Colony formation determined 18–20 days after re-plating is shown. Data are the mean number of colonies ± SD from three independent experiments with three technical replicates for each experiment. **P < 0.01, ****P < 0.0001, compared with vehicle control (two-way ANOVA). VCM: vehicle control; BER: berberine; BC: berberine + chloroquine; IR: radiation; BIR: berberine + radiation; BCIR: berberine + chloroquine + radiation.

Fig. 5.

A. Survival probability of patients with prostate cancer from the TCGA_PRAD dataset correlates with expression levels of a 39-gene signature identified from integration of transcriptomic data from S6K-KO and berberine treated LNCaP cells. Expression analyses of the upregulated and downregulated genes in the TCGA_PRAD project showed seven upregulated and 39 downregulated DEGs. Higher expression of the 39-gene signature is associated with poor survival. B. Transcriptomic data integration of LNCaP cells with RPS6KB1 genetic knockout or suppression with berberine to identify common differentially expressed genes (DEGs). Log2 fold-change values of 39 genes from XRT_vs_CON (radiation vs. untreated) and RTBER_vs_CON (combination vs. untreated) were used to generate the plots presented. We calculated the difference between the two comparisons for each gene and labeled the genes with |difference value |>1. C. Log rank test of survival of human prostate cancer patients from TCGA_PRAD correlated with H2AFX expression level. D. RPS6KB1 loss differentially affects DNA repair pathways. RPS6KB1 knockout (S6K-KO) reduced alternative end-joining (Alt-EJ) and increased non-homologous end-joining (NHEJ) repair events in S6K-KO cells compared with parental LNCaP cells. E. Significantly downregulated expression of selected DNA repair and autophagy genes in S6K-KO cells (n = 2) compared with parental LNCaP (n = 3) or RPS6KB1-overexpressing (n = 3) cells based on RNA-seq data. F. Scatter plot (Pearson’s correlation test) illustrating the correlation between the gene expression levels of POLQ and H2AFX in patients with prostate cancer.

Fig. 6.

A. Combination of berberine and radiation decreases POLθ protein and POLQ mRNA expression in LNCaP cells. A representative blot of three independent experiments is shown. A representative blot of three independent experiments is shown. B. Changes in p70S6K, pp65, and POLθ protein levels in mouse prostate tumors (n = 5 per group) following treatment with berberine alone, radiation alone, or berberine plus radiation. β-actin was used as a loading control for immunoblotting. C. Logarithmically growing S6K-OE LNCaP cellswere transfected with si-control or si-POLQ using Lipofectamine 2000 and Opti-MEM media.72 hposttransfection, cells were exposed to radiation(0.25, 0.5, 0.75, or 1Gy) from a CellRad faxitron. Immediately after radiation, cells were harvested and re-plated in triplicates(500 cells/well) for colony formation. Colonies with >50 cells were counted 8–9 days post-radiation and visualized using crystal violet staining. D. Hypothetical model of radioresistance in prostate cancer. Inhibition of S6K decreased POLQ and inhibiting POLQ in S6K expressing prostate cancer cells decreased S6K showing the existence of a feedback loop between S6K and POLQ. Biological outcome experiments showed restoration of radiosensitivity. These results show that POLQ is downstream of S6K, a key DNA repair enzyme involved in radioresistance. Thus the ability of berberine to inhibit POLQ and the feedback loop is important to overcoming radioresistance and the downstream effects of S6K on NF-κB-p62 mediated effects on autophagy and ultimately growth promotion following radiation.

Fig. 7.

A-C. Elevated expression of 4-gene signature RPS6KB1/NFκB/SQSTM1/POLQ is associated with increased (A) aneuploidy, (B) genomic alterations, and (C) tumor aggressiveness of prostate cancer patients. The median value was used as the cutoff between low and high expression levels of each gene in prostate tumor samples in TCGA_PRAD cohort. The samples showing expression of 4 genes above the median (n = 38) were selected to compare with the samples showing expression of 4 genes below the median (n = 31). Data presented is based on cBioPortal. Wilcoxon test was used for aneuploidy score and fraction genome altered, while Chi-squared test was used for tumor stage.