Unraveling the protein kinase C/NDRG1 signaling network in breast cancer

Abstract

N-myc downstream-regulated gene 1 (NDRG1) is a member of the NDRG family of intracellular proteins and plays a central role in a wide range of biological processes including stress response, differentiation, and metabolism. The overexpression of NDRG1 is an indicator of poor prognosis in various types of cancer. Here, we found that NDRG1 is an independent prognostic marker of poor outcome in breast cancer (BC). Analysis of the TCGA dataset showed a significant positive correlation between NDRG1 and PRKCA expression, suggesting a mechanistic role of protein kinase C (PKC) in the regulation of NDRG1. We then assessed the hypothesis that PKC might modulate the activity of NDRG1, and observed that different acute stress conditions converging on PKC activation lead to enhanced NDRG1 expression. This mechanism was found to be specific for NDRG1 as the expression of other NDRG members was not affected. Moreover, CRISPR-based inhibition of NDRG1 expression was obtained in a BC cell line, and showed that this protein is a key driver of BC cell invasion through the Rho-associated coiled-coil containing protein kinase 1 (ROCK1)/phosphorylated cofilin pathway that regulates stress fiber assembly, and the modulation of extracellular matrix reorganization related genes. Together, our findings highlight the potential of NDRG1 as a new BC biomarker and uncover a novel mechanism of regulation of NDRG1 expression that might lead to innovative therapeutic strategies.

Keywords: Breast cancer; NDRG1; PKC; TNBC.

Fig. 1

NDRG1 is highly expressed in breast cancer samples and TNBC subtypes. A Schematic representation of micro-proteomics workflow. Overview of trypsin deposition, protein extraction, and LC–MS/MS analysis. B The heat map based on Euclidean distance showed a significant separation between the healthy and tumor sections. The color scale ranges from red to green (highest to lowest relative expression). Each column of the heat map represents an independent sample and each row represents a specific protein. C Volcano plot of healthy vs tumor sections, from Perseus software. Significant proteins are determined using permutation-based FDR calculation with high confidence of 0.01 (solid line). Thresholds are displayed in the graph. D Spatial expression of NDRG1 in healthy and tumor sections. E The heat map based on Euclidean distance showed a significant separation between the Luminal and TNBC samples. The color scale ranges from red to green (highest to lowest relative expression). Each column of the heat map represents an independent sample and each row represents a specific protein. F Volcano plot of TNBCs vs Luminal, from Perseus software. The luminal group is chosen as a negative control. Significant proteins are determined using permutation-based FDR calculation with high confidence of 0.01 (solid line). Thresholds are displayed in the graph. G Differential expression of NDRG1 between Luminal A and TNBC samples analyzed by RNA seq

Fig. 2

NDRG1 is highly expressed in a TNBC model in vitro. A The heat map based on Euclidean distance showed a significant separation between the Luminal model MCF-7 and the TNBC model MDA-MB-231. The color scale ranges from red to green (highest to lowest relative expression). Each column of the heat map represents an independent sample and each row represents a specific protein. B Volcano plot of MDA-MB-231 vs MCF-7, from Perseus software. The differential expression of NDRG1 and NDRG3 is highlighted. C Western blotting analysis for NDRG1, phospho-NDRG1 (S330), NDRG2, NDRG3, NDRG4, E-cadherin, Vimentin, SGK1, p53 and YAP of lysates obtained from MDA-MB-231 and MCF-7 cells. D) RT-qPCR of NDRG1, NDRG2, and NDRG3 mRNAs in MDA-MB-231 and MCF-7 cells. The p-value was calculated using the Student’s t-test. The error bar represents ± SD. p-value ** < 0.01, *** < 0.001. E Posttranslational and transcriptional mechanisms of NDRG1 regulation. Left. The mTORC2/SGK1 pathway regulates the phosphorylation of NDRG1. Right. The red stars indicate differentially expressed proteins involved in the regulation of NDRG1 expression and identified by MS/MS. The lists of proteins detected are presented in Supplementary MS/MS Data 2

Fig. 3

NDRG1 is over-expressed in BC samples. A Differential NDRG1 expression in BC. Compared to normal breast tissue (left), invasive ductal carcinoma (right) shows a marked increase in NDRG1 expression (original magnification, × 200). Scale bar = 20 µm. Images were obtained on an Axion Image 2 upright microscope (Zeiss, Oberkochen, Germany) with an Axiocam 512 color camera. B On the left, representative images of immunohistochemical staining in BC tissues. The iamge displays the representative expression of NDRG1 in TNBC and luminal phenotype. (original magnification, × 200). Scale bar = 50 µm. Images were obtained on an Axion Image 2 upright microscope (Zeiss, Oberkochen, Germany) with an Axiocam 512 color camera. On the right, expression levels of NDRG1 in TNBCs compared with Luminal phenotype. Values are expressed as the median (horizontal red line in each box). Dots indicate outliers. ***p < 0.001. C Representative images of immunohistochemical staining in the TNBC subgroups. On the left, a case with high NDRG1 expression; on the right, a case with low NDRG1 expression. Scale bar = 50 µm. Kaplan–Maier curve analysis and log-rank test. D Kaplan–Maier curve for disease-free survival (DFS) according to high NDRG1 versus low expression in BC patients (p = 0.0028). E DFS in the subgroups of Luminal (Luminal A and B) and TNBC tumors, according to high and low NDRG1 expression (p = 0.0099). F DFS of TILs/NDRG1 co-expression showed that the patients with low TILs/high NDRG1 tumors had a worse DFS with respect to the other phenotypes considered (p = 0.0096). G The Kaplan‒Meier Plotter website was used to investigate the relationship between DRG1 expression and survival probability in TNBC human samples (n(high) = 288, n(low) = 104)

Fig. 4

NDRG1 is a stress-responsive protein. A Western blotting analysis for NDRG1 of lysates obtained from MDA-MB-231 cells exposed to 1% serum or maintained in a medium without lipids for 24 h. Cofilin was used as a loading control. B RT-qPCR of NDRG1 and NDRG3 mRNAs in control and MDA-MB-231 cells exposed to 1% serum for 24 h. The p-value was calculated using the Student’s t-test. The error bar represents ± SD. p-value *** < 0.001. C Western blotting analysis for NDRG1 of lysates obtained from MCF-7 and MDA-MB-231 cells exposed to 1% serum for 5 h or 24 h. Cofilin was used as a loading control. D Western blotting analysis for NDRG1, GRP78 and p62 of lysates obtained from MDA-MB-231 cells exposed to thapsigargin (1 μM) or LY294002 (10 μM) for 24 h. Cofilin was used as a loading control. E Representative images of MDA-MB-231 cells cultured in a normal condition medium or treated palimitic acid (200 μM), or oleic acid (200 μM) for 24 h. Images were acquired using an inverted wide-field microscope (EVOS FLoid Cell Imaging Station, Thermo). Scale bar 100 μm. F Western blotting analysis for NDRG1, GRP78 and p62 of lysates obtained from MDA-MB-231 cells exposed to palmitic acid or oleic acid at the concentration of 200 μM for 24 h. G Representative images of MDA-MB-231 cells cultured in a normal condition medium or treated with Vandetanib (10 μM), or Crizotinib (10 μM) for 24 h. Images were acquired using an inverted wide-field microscope (EVOS FLoid Cell Imaging Station, Thermo). Scale bar 100 μm. H Western blotting analysis for NDRG1, GRP78 and p62 of lysates obtained from MDA-MB-231 cells exposed to Vandetanib and Crizotinib treatment at the concentration of 10 μM for 24 h

Fig. 5

Results of TCGA data analysis. Correlation of PRKCA, NDRG1, NDRG2, NDRG3 and NDRG4 mRNAs expression in BC samples of the TCGA database

Fig. 6

Different stress conditions activate PKC. Western blotting analysis for Phospho-PKC Substrate Motif [(R/K)XpSX(R/K)] of lysates obtained from MDA-MB-231 cells exposed to PMA, thapsigargin, Crizotinib, palmitic acid, and Vandetanib for 30 min

Fig. 7

Mass spectrometry analysis of PMA-treated MDA-MB-231 cells. A The heat map based on Euclidean distance showed a significant separation between the control and PMA-treated MDA-MB-231 cells. The color scale ranges from red to green (highest to lowest relative expression). Each column of the heat map represents an independent sample and each row represents a specific protein. The window contains the expression profiles of NDRG1. B Western blotting analysis for Phospho-PKC Substrate Motif [(R/K)XpSX(R/K)] of lysates obtained from MDA-MB-231 cells exposed to PMA (100 nM), Ro318220 (1 μM) alone or in combination for 30 min. C Bright-field images of MDA-MD-231 cells treated with PMA (100 nM), Ro318220 (1 μM) alone or in combination for 24 h. Images were acquired using an inverted wide-field microscope (EVOS FLoid Cell Imaging Station, Thermo). Scale bar 100 μM. D Western blotting analysis for NDRG1 and phospho-NDRG1 of lysates obtained from MDA-MD-231 cells treated with PMA (100 nM), Ro318220 (1 μM) alone or in combination for 24 h. E RT-qPCR in MDA-MB-231 cells subjected to exposed to PMA (100 nM), Ro318220 (1 μM) alone or in combination for 24 h. The p-value was calculated using the ANOVA test. The error bar represents ± SD. p-value **** < 0.0001

Fig. 8

Mechanisms for NDRG1 regulation by PKC. A Western blotting analysis for p53, NDRG1 of lysates obtained from MDA-MB-231 cells exposed to PMA (100 nM) for 24 h. Cofilin was used as a loading control. B and C Western blotting analysis for phospho-PKC Substrate Motif [(R/K)XpSX(R/K)], PKCα, PKCε, phospho-MARCKS (Ser167/170), MARCKS, phospho-Erk1/2, Erk1/2, phospho-Cofilin, and Cofilin of lysates obtained from MDA-MB-231 cells exposed to PMA (100 nM) for 30 min. Cofilin was used as a loading control. ROCK inhibition induces NDRG1 expression. D Representative images of MDA-MB-231 cells cultured in a normal condition medium or treated with 100 nM Y-27632 for 24 h. Images were acquired using an inverted wide-field microscope (EVOS FLoid Cell Imaging Station, Thermo). Scale bar 100 μm. E) Western blotting analysis for phospho-Cofilin, and Cofilin of lysates obtained from MDA-MB-231 cells exposed to Y-27632 (10 μM), for 30 min. F Western blotting analysis for NDRG1 of lysates obtained from MDA-MB-231 cells exposed to PMA (100 nM), Y-27632 (10 μM) alone or in combination for 24 h. G RT-qPCR of NDRG1, NDRG3, Zeb1 and c-Myc mRNAs in control and MDA-MB-231 cells exposed to Y-27632 (10 μM) for 24 h. The p-value was calculated using the Student’s t-test. The error bar represents ± SD. p-value ** < 0.01, *** < 0.001. H Representative images of Hs 578 T cells cultured in a normal condition medium or treated with 100 nM Y-27632 for 24 h. Images were acquired using an inverted wide-field microscope (EVOS FLoid Cell Imaging Station, Thermo). Scale bar 100 μm. I RT-qPCR of NDRG1 mRNA in control and Hs 578 T cells exposed to Y-27632 (10 μM) for 24 h. The p-value was calculated using the Student’s t-test. The error bar represents ± SD. p-value *** < 0.001

Fig. 9

CRISPR/Cas9 knockdown of NDRG1. A Western blotting analysis for NDRG1, pNDRG1 (S330), NDRG3 and NDRG4 of lysates obtained from MDA-MB-231 Empty and NDRG1-CRISPR cells. Cofilin was used as a loading control. B RT-qPCR of NDRG3 and NDRG4 mRNAs in MDA-MB-231 Empty and NDRG1-CRISPR cells. CRISPR/Cas9 knockdown of NDRG1 modulates the proteome of MDA-MB-231 cells. C Heat map based on Euclidean distance showing a significant separation between the MDA-MB-231 Empty and NDRG1-CRISPR cells. Each row of the heat map represents a protein, and each column represents an independent sample. Two main clusters were identified from the hierarchical clustering, and their pattern is reported. D Enrichment maps of deregulated genes in Empty and NDRG1-CRISPR cells

Fig. 10

NDRG1 is required for breast cancer cell invasion in vitro. A Schematic representation of inverse Matrigel invasion test. B Inverse Matrigel invasion of MDA-MB-231 Empty and NDRG1-CRISPR cells. Green fluorescent cells result from live-cell staining with calcein-AM. The grey arrow indicates the direction of cell migration during the process of Matrigel invasion. C In vitro growth rates of MDA-MB-231 Empty and NDRG1-CRISPR cells. Empty CRISPR cells are in black; NDRG1 CRISPR cells are in red. Bars, SEM. D Western blotting analysis for Vinculin, Vimentin, phospho-Cofilin, and Cofilin of lysates obtained from MDA-MB-231 Empty and NDRG1-CRISPR cells. E The windows contain the expression profiles of NDRG1 and ROCK obtained after LC–MS/MS analysis. F Actin localization visualized using phalloidin staining of MDA-MB-231 Empty and NDRG1-CRISPR cells. G The Kaplan‒Meier Plotter website was used to investigate the relationship between NDRG1 low/ROCK high and NDRG1 high/ROCK low expression and survival probability in BC human samples

Fig. 11

A schematic representation of the prognostic and molecular impact of NDRG1 in TNBCs. Tumor cell expression was found to be significantly increased compared to normal tissues and correlated with poor patient outcomes and molecular characteristics. Stress signals including chemotherapy or hostile environmental conditions converge to PKC to regulate NDRG1 expression through enhanced transcription which in turn regulates multiple pathways associated with endoplasmic reticulum and invasion