Matrix stiffness-induced IKBKE and MAPK8 signaling drives a phenotypic switch from DCIS to invasive breast cancer

Abstract

Ductal carcinoma in situ (DCIS) is not life threatening unless it transitions into invasive breast cancer (IBC). However, although breast cancer cell exposure to matrix stiffening in vitro phenotypically mimics the DCIS to IBC switch, the molecular changes driving this switch remains unclear. Here, breast cancer cell kinome activity profiling suggested matrix stiffness-upregulation of 53 kinases, among which 16 kinases were also regulated by integrin β1. Functional validation identified matrix stiffness-activation of inhibitor of nuclear factor kappa-B kinase subunit epsilon (IKBKE) and mitogen-activated protein kinase 8 (MAPK8) signaling as critical for the stiffness-driven IBC phenotype, including for cell proliferation. The IKBKE-inhibitor Amlexanox, clinically utilized for aphthous ulcers, as well as the MAPK8 inhibitor JNK-IN-8, reinstalled the DCIS-like phenotype of breast cancer cells on high matrix stiffness. This suggests that IKBKE and/or MAPK8 inhibitors could enhance the arsenal of treatments to prevent or treat breast cancer.

Keywords: Cellular signaling; Integrin; Kinase; Kinome; Mechanotransduction.

Fig. 1

Matrix stiffness induced switch in MCF10CA1a (CA1a) cancer cell phenotype reversed by integrin β1 inhibition. a, Representative phenotype of CA1a cells on rBM-conjugated hydrogels of different stiffness mimicking normal breast tissue stiffness (0.5 kPa) or breast cancer stiffness (8 kPa). Scale bar = 100 μm. b, Representative spinning disc confocal immunofluorescence images of laminin (pink), integrin β4 (green) and nuclei (blue) in CA1a cells on low (top row) and high (bottom row) stiffness hydrogels. Scale bar = 20 μm. c, CA1a cells on high stiffness hydrogels treated with anti-integrin b1 monoclonal AIIB2 antibody or vehicle alone. Scale bar = 100 μm

Fig. 2

Kinome activity profiling reveals matrix stiffness-induced changes in tyrosine and serine/threonine peptide phosphorylation partially dependent on integrin β1. a, Experimental setup for peptide array-based kinome profiling of CA1a cells on hydrogels. b, Principal component analysis (PCA) of low and high stiffness samples using all QC peptide phosphorylation data from PTK (left) and STK (right) chip arrays. Colour indicate condition and number indicate replicate c, d, Volcano plots of peptide phosphorylation data from high stiffness vs. low stiffness PTK (c) and STK (d) chip arrays assessed by unpaired t-test. Red spots are peptides with p < 0.05. e, PCA analysis of high stiffness and high + AIIB2 treated samples using only the most differentially phosphorylated peptides (p < 0.01) from the low versus high stiffness comparison. f, g, Volcano plots of peptide phosphorylation data from high stiffness AIIB2 treated vs. high stiffness control treated PTK (f) and STK (g) chip arrays assessed by unpaired t-test. Red spots are peptides with p < 0.05. h, i, Kinase score plots for high stiffness versus low stiffness (h) or for AIIB2 versus control (i) from PTK (left) and STK (right) peptide chip array data. Upstream kinases are predicted from differentially phosphorylated peptides using the proprietary Bionavigator software, with a mean specificity score > 1

Fig. 3

MAPK8 and IKBKE mediate the stiffness-induced breast cancer phenotype. a, Venn diagram (left) and table (right) showing overlap between kinases predicted to be upregulated by stiffness and downregulated by AIIB2 treatment. b, Workflow for the siRNA-based screen used to test the function of predicted kinases, and all conditions were run in duplicates. c, Representative phenotype of CA1a cells on high stiffness hydrogels transfected with siRNA as specified. FAK siRNA serves as positive control. Images are single tiles from 6 × 6 montages. Scale bar = 50 μm. d, Area fraction of segmented cell clusters expressed relative to MOCK (Gene of interest/MOCK). The area fraction is calculated as the ratio between the total area and the area of the thresholded objects. Positive control (FAK siRNA) is shown in red. Data represents the average of two technical replicates. Genes whose knockdown results in a similar or lower relative area fraction as the positive control are considered hits, including MAPK8 and IKBKE

Fig. 4

Mechano-regulated IKBKE signaling controls breast cancer cell growth. a, Immunoblot of IKBKE in CA1a cells (left) cultured on 0.4 kPa or 5 kPa hydrogels as indicated. Densitometric analysis shows IKBKE levels normalized to loading controls (ponceau S staining, Supplementary Fig. 2a) and expressed relative to the low stiffness level (right). Data are presented as mean ± S.E.M., with p-values according to an unpaired t-test. b, Representative immunofluorescent images of Edu staining in CA1a cells grown on 5 kPa transfected with IKBKE or control siRNA pool (left) and quantification of three biological repeats (right), each with more than one thousand cells. Scale bar = 50 μm. Data are mean ± S.E.M and p-values according to an unpaired t-test. c, Representative images of actin (top row) and Edu staining (bottom row) in HCC1143 cells treated with IKBKE or control siRNA (left). Scale bar = 50 μm. Area fraction of segmented cell clusters expressed relative to control siRNA. Quantification of three or more biological repeats (right). Data are mean ± S.E.M and p-values derived by one-way Anova followed by Dunnett’s multiple comparison test. d, e, Representative spinning disc confocal images of CA1a (d) and HCC1143 (e)cells on 5 kPa treated with 180 µM Amlexanox or vehicle for 48 h (left). Quantification of actin staining (top row) and Edu staining (bottom row) from three or more independent experiments (right). Scale bar = 50 μm. Relative area fraction as in (c). Data and statistics as in (b)

Fig. 5

Matrix stiffness-induced MAPK8 activity is critical for breast cancer cell proliferation in vitro. a, Immunoblot analysis of MAPK8 (left), and phospho-MAPK8 (right) of CA1a cells cultured on 0.4 kPa or 5 kPa hydrogels. Densitometric analysis shows MAPK8 and phospho-MAPK8 levels normalized to loading controls (ponceau S staining, Supplementary Fig. 2a and 2c) and expressed relative to the low stiffness levels. Data are presented as mean ± S.E.M., with p-values according to an unpaired t-test. b, Representative Edu staining images of CA1a cells grown on 5 kPa transfected with MAPK8 or control siRNA pool (left) and quantification of three biological repeats (> 1000 cells each; right). Scale bar = 50 μm. Data are mean ± S.E.M and p-values according to an unpaired t-test. c, Representative images of actin (top row) and Edu staining (bottom row) in HCC1143 cells treated with MAPK8 or control siRNAs (left). Quantification of area fraction of segmented cell clusters relative to control siRNA of three biological repeats (right). Scale bar = 50 μm. Data are mean ± S.E.M and p-values derived by one-way Anova with Dunnett’s multiple comparison test. d, e, Spinning disc confocal images of breast cancer cells on 5 kPa treated with 5 µM (CA1a, d) or 10 µM ( HCC1143, e) JNK-IN-8 or vehicle for 24 h(left). Quantification of actin staining (top row) and Edu staining (bottom row) of three or more independent experiments(right). Scale bar = 50 μm. Relative area fraction as in (c). Data and statistics as in (b)