Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2021 Jun 29;35(13):109293.
doi: 10.1016/j.celrep.2021.109293.

Breast tumor stiffness instructs bone metastasis via maintenance of mechanical conditioning

Adam W Watson  1 , Adam D Grant  2 , Sara S Parker  3 , Samantha Hill  4 , Michael B Whalen  3 , Jayati Chakrabarti  4 , Michael W Harman  5 , Mackenzie R Roman  2 , Brittany L Forte  2 , Cody C Gowan  3 , Raúl Castro-Portuguez  2 , Lindsey K Stolze  3 , Christian Franck  6 , Darren A Cusanovich  3 , Yana Zavros  4 , Megha Padi  7 , Casey E Romanoski  8 , Ghassan Mouneimne  9
Affiliations

Breast tumor stiffness instructs bone metastasis via maintenance of mechanical conditioning

Adam W Watson et al. Cell Rep. .

Abstract

While the immediate and transitory response of breast cancer cells to pathological stiffness in their native microenvironment has been well explored, it remains unclear how stiffness-induced phenotypes are maintained over time after cancer cell dissemination in vivo. Here, we show that fibrotic-like matrix stiffness promotes distinct metastatic phenotypes in cancer cells, which are preserved after transition to softer microenvironments, such as bone marrow. Using differential gene expression analysis of stiffness-responsive breast cancer cells, we establish a multigenic score of mechanical conditioning (MeCo) and find that it is associated with bone metastasis in patients with breast cancer. The maintenance of mechanical conditioning is regulated by RUNX2, an osteogenic transcription factor, established driver of bone metastasis, and mitotic bookmarker that preserves chromatin accessibility at target gene loci. Using genetic and functional approaches, we demonstrate that mechanical conditioning maintenance can be simulated, repressed, or extended, with corresponding changes in bone metastatic potential.

Keywords: ATACseq; RUNX2; biomechanics; bone metastasis; breast cancer; matrix stiffness; mechanical memory; osteolysis; phenotypic plasticity; tumor microenvironment.

PubMed Disclaimer

Conflict of interest statement

Declaration of interests A.W.W. and G.M. have equity in MeCo Diagnostics LLC, which is commercializing the MeCo score under license from the Arizona Board of Regents. G.M. has disclosed an outside interest in MeCo Diagnostics LLC to the University of Arizona. Conflicts of interest resulting from this are being managed by the University of Arizona in accordance with its policies. All of the other authors declare no competing interests.

Figures

Figure 1.
Figure 1.. Mechanical conditioning manifests distinctively in cellular dynamics and invasion
(A) Schematics showing multifunctional analysis workflow for testing mechanical memory. (B) Mechanoresponse readouts for a panel of breast cancer cell lines and patient-derived xenografts (denoted by asterisks). (C) Dynamics (overlaid cell traces) of iRFP-Lifeact-expressing SUM159 cells preconditioned on stiff and/or soft hydrogels as indicated, showing 1-h intervals starting 10 h after plating on glass. Scale bars, 10 μm. See Video S1. (D) Quantification of (C) (n = 36 cells in each condition from n = 3 biological replicates). (E) Multidimensional traction force microscopy of SUM159 cells on bead-embedded 8.5 kPa gels, preconditioned on regular stiff and/or soft hydrogels, as indicated. Panels show vector maps of displacement magnitude. Heat scales (μm) show bead displacement. (F) Quantification of (E) (n = 17–28 cells in each condition from n = 3 biological replicates). (G and H) Analysis of actin cytoskeletal structures in 7-day soft and stiff preconditioned cells before and after plating on glass for 1 day. Quantification (G) of cells in situ on gels and on glass, and corresponding representative images (H). Scale bar, 10 μm. (I) SUM159 cell position along invasion fronts after 16 h of live-cell tracking in 3D collagen. Cells were preconditioned on stiff and/or soft hydrogels as indicated. Gray dots, non-invasive cells; black dots, invasive cells; white triangles, invasion front at start of imaging; black triangles, invasion front at end of imaging. See Figure S1F and Video S2. (J and K) Quantification of (I) (n = 3 biological replicates with n = 3 technical replicates). Data are means ± SEMs. *p < 0.05, **p < 0.01, ***p < 0.001; 1-way ANOVA with Tukey’s multiple comparisons test.
Figure 2.
Figure 2.. Primary tumor mechanical conditioning is associated with bone metastasis
(A) Heatmap of differentially regulated genes (R2-fold) from RNA-seq of 2-week soft- and stiff-preconditioned SUM159 cells, which constitute the raw mechanical conditioning genes (n = 3 biological replicates). (B) Kaplan-Meier curve of patients in the METABRIC 2019 study (molecular dataset cohort), assigning each patient a raw mechanical conditioning (MeCo) score derived from the differential gene expression in (A) comparing upper and lower quartiles of the MeCo score (n = 476 high MeCo, 476 low MeCo). See Method details for MeCo score derivation. (C) Metascape ontology of stiffness-induced genes (>4-fold). Bold text indicates skeletal ontologies. (D) Kaplan-Meier curve of bone metastasis-free survival in the combined cohort, split at median MeCo score (n = 280 high MeCo, 280 low MeCo). (E) Time to bone metastasis for patients in (D) split at median MeCo score (n = 93 high MeCo, 92 low MeCo). (F) Kaplan-Meier curve of bone metastasis-free survival in the combined cohort, split at median MeCorefined score (n = 281 high MeCorefined, 279 low MeCorefined). (G) Time to bone metastasis for patients in (F), split at median MeCorefined score (n = 93 high MeCorefined, 92 low MeCorefined). (H and I) Large validation cohort for the MeCorefined score showing Kaplan-Meier curve of bone metastasis-free survival (H) and time to bone metastasis (I) in METABRIC 2019 (using all patients with distant relapse annotation), comparing upper and lower quartiles of MeCorefined score. n = 422 high MeCorefined, 421 low MeCorefined in (H), and n = 65 high MeCorefined, 64 low MeCorefined in (I). (J) Kaplan-Meier curve of bone metastasis-free survival in the NKI cohort, split at median MeCorefined score (n = 147 high MeCorefined, 148 low MeCorefined). (K) Time to bone metastasis for patients in (J), split at median MeCorefined score (n = 27 high MeCorefined, 26 low MeCorefined). The median time to metastasis for patients with high MeCo scores was 27 months, compared to 39 months for those with low MeCo scores. (L) Radiograms of tibia from mice injected with 7-day soft-, stiff-, and plastic-preconditioned SUM159 cells, imaged 4 weeks after intracardiac injection. (M) Quantification of (L) (n = 4 mice per group; TC, tissue culture). Data are means ± SEMs. *p < 0.05, ***p < 0.001, ****p < 0.0001; 2-way ANOVA with Tukey’s multiple comparisons test. Kaplan-Meier p values calculated with Wilcoxon test.
Figure 3.
Figure 3.. RUNX2 maintains mechanical conditioning in soft microenvironments
(A) Schematics of discovery approach identifying candidate drivers of mechanical memory-mediated metastasis. Gray dots, mechanically sensitive upstream regulators from RNA-seq analysis (141 genes); blue dots, Human Cancer Metastasis Database metastasis-associated genes (1,811 genes); black/red dots, intersecting genes (123 genes); red dots, intersecting genes that are known gene bookmarkers (5 genes); inset shows further characterization. See Method details for in-depth description. (B) Heatmap showing clustering of ATAC-seq samples; scale shows Jaccard index. See Figure S4A for sample annotation. (C) Graphic representation of change in chromatin accessibility over time after changing the mechanical environment; inset shows RUNX and BACH family motifs as representative of the significantly enriched motifs in the delayed-closing and quick-closing sites, respectively (motifs analysis performed by HOMER). (D) The number of pairwise differentially decreased accessibility sites are shown consequent of RUNX2-knockdown (vertical arrows) and longitudinally upon transition to soft matrix (horizontal arrows). Data represent 3 biological replicates per condition. (E) Top enriched motifs are shown for the vertical comparisons in (D), where the first percentage is the number of differential peaks with the motif and the second is the number of GC-matched random genomic sequences with motif.
Figure 4.
Figure 4.. RUNX2 is activated by matrix stiffness via mechanotransduction
(A) Immunoblot of RUNX2 in patient-derived xenograft (PDX) primary cells and breast cancer cell lines, preconditioned on soft and stiff hydrogels for 7 days (representative of n = 2 biological replicates). (B) Immunoblot of RUNX2 in SUM159 and T47D cells cultured for 7 days in 3D matrix consisting of soft 1.0 mg/mL rat-tail collagen-I, or stiff 1.0 mg/mL rat-tail collagen-I crosslinked with PEG-di(NHS) to stiffen the collagen lattice without changing ligand density (representative of n = 3 biological replicates). (C) qRT-PCR of 4 RUNX2 target genes in SUM159 cells preconditioned for 7 days on soft and stiff hydrogels with non-targeting small hairpin RNA (shRNA) (GIPZ), or on stiff hydrogels with 2 shRNAs targeting RUNX2 (n = 3 biological replicates). Data are means ± SEMs. *p < 0.05, **p < 0.01, ***p < 0.001; 1-way ANOVA with Tukey’s multiple comparisons test. (D) qRT-PCR of RUNX2 in SUM159 cells preconditioned for 7 days on soft and stiff hydrogels with non-targeting shRNA (GIPZ), or on stiff hydrogels with 2 shRNAs targeting RUNX2 (n = 3 biological replicates). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; 1-way ANOVA with Sidak’s multiple comparisons test. (E) qRT-PCR of RUNX2 and 4 target genes, plus CTGF (YAP target) and PLIN1 (adipogenic biomarker) in SUM159 cells preconditioned, as indicated (n = 3 biological replicates). Data are means ± SEMs. *p < 0.05, **p < 0.01, ***p < 0.001; 1-way ANOVA with Tukey’s multiple comparisons test. (F) qRT-PCR of the RUNX2 gene target OPN, and 3 YAP targets—CTGF, CYR61, and ANKRD1—in SUM159 cells preconditioned as indicated, and without media change for 48 h before sample collection (n = 3 biological replicates). Data are means ± SEMs. *p < 0.05, **p < 0.01, ***p < 0.001; 1-way ANOVA with Tukey’s multiple comparisons test. (G) Immunoblot of RUNX2, ERK, and pERK in SUM159 cells stably expressing lentiviral shRUNX2 or GIPZ (non-targeting control), preconditioned for 7 days on soft or stiff hydrogels (representative of n = 3 biological replicates). (H) Immunoblot of pERK and ERK in SUM159 cells cultured on stiff hydrogels with 20 μM PD98059, 30 μM blebbistatin, 100 nM dasatinib, 1 μM Faki14, or DMSO for 1 h before lysis (representative of n = 3 biological replicates). (I) Immunofluorescence of pFAK, paxillin, and F-actin in SUM159 cells cultured on stiff or soft hydrogels for 7 days. Scale bar, 10 μm. (J) Immunoblot of OPN in SUM159 cells preconditioned for 7 days on soft and stiff hydrogels with non-targeting shRNA (GIPZ), on stiff hydrogels with 2 shRNAs targeting RUNX2, on soft hydrogels with constitutively active MEK-DD expression, or on stiff hydrogels with MEK inhibitor PD98059 (20 μM) (representative of n = 3 biological replicates). (K and L) qRT-PCR of OPN (K) and GM-CSF (L) in SUM159 cells preconditioned for 7 days on stiff hydrogels conjugated with either poly D-lysine (PDL) to reduce integrin binding, or collagen conjugated with DMSO (control), 30 μM blebbistatin, 100 nM dasatinib, or 1 μM Faki14 in media changed every other day (n = 3 biological replicates). Data are means ± SEMs and normalized to 7-day stiff controls. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; 1-way ANOVA with Sidak’s multiple comparisons test. (M) Immunoblot showing phospho-AKT levels in SUM159 cells treated with DMSO (control) or AKT inhibitor (1 μM MK-2206) (representative of n = 2 biological replicates). (N) qRT-PCR of RUNX2 target genes in SUM159 cells preconditioned for 7 days on stiff hydrogels and treated with DMSO or 1 μM AKT inhibitor MK-2206 (n = 3 biological replicates). Data are means ± SEMs. Multiple t test with Holm-Sidak multiple comparisons; adjusted p values are not significant (n.s.).
Figure 5.
Figure 5.. RUNX2 nuclear localization is regulated by matrix stiffness
(A) Immunofluorescence staining of RUNX2 in SUM159 cells on soft and stiff hydrogels. Scale bars, 10 μm. (B) Immunofluorescence corresponding to Figure 4G. Cell boundaries are delineated in blue. Scale bars, 10 μm. (C) Quantification of (B) (n = 40 cells in each condition from n = 3 biological replicates). ****p < 0.0001; 2-tailed unpaired Student’s t test. (D and E) Immunofluorescence staining (D) and quantification of nuclear localization (E) of RUNX2 in PC3 prostate cancer cells (n = 50 cells each from n = 3 biological replicates). (F and G) Immunofluorescence staining of RUNX2 (F) and quantification of nuclear intensity of RUNX2 in CK+ cells (G) in supraphysiological stiffness-naive patient-derived xenograft HCI-005 tumor cells (n = 60 cells each from n = 3 biological replicates). (H) Immunofluorescence staining in SUM159 cells, preconditioned on soft and stiff hydrogels for 7 days before transferring to collagen-coated glass for 3 h. (I and J) Quantification of nuclear RUNX2 (I) and cell area (J) from cells in (H). Data are means ± SEMs. ****p < 0.0001; 2-tailed unpaired Student’s t test. (K) Correlation analysis of (I) and (J), showing no positive intracellular correlation between cell spreading and nuclear RUNX2 in either soft- or stiff-preconditioned cells spreading on glass (n = 40 cells each from n = 3 biological replicates). Soft-to-glass Pearson’s r = −0.31 (not significant [n.s.]); stiff-to-glass Pearson’s r = −0.30 (n.s.). (L) Immunofluorescence staining of RUNX2 in SUM159 cells preconditioned for 7 days on stiff hydrogels, treated with DMSO (control), 1 μM taxol, 50 nM jas-plakinolide, 30 μM blebbistatin, or 20 μM Y27632, added 3 h before fixation. Scale bars, 10 μm. (M) Immunofluorescence staining of RUNX2 in SUM159 cells preconditioned for 7 days on soft hydrogels, with DMSO (control), 1 μg/mL lysophosphatidic acid (LPA), 10 μg/mL Rho Activator II, or 10 μM nocodazole, added 3 h before fixation. Scale bars, 10 μm. (N) Quantification of (L) (n R 40 cells each condition from n = 3 biological replicates). (O) Quantification of (M) (n R 40 cells each condition from n = 3 biological replicates). Data are means ± SEMs. ****p < 0.0001; 1-way ANOVA with Dunnett’s multiple comparisons test.
Figure 6.
Figure 6.. RUNX2 is associated with proliferation-sensitive mechanical conditioning, which is maintained in bone marrow (BM) resident cancer cells
(A) Schematics showing strategy to enrich high-proliferative cells (CVLOW) versus low-proliferative cells (CVHIGH). (B) Time course of mechanical memory loss, showing flow cytometry of SUM159 cells preconditioned as indicated, sorted, and stained for OPN (n = 3 biological replicates). See Figure S6A. (C) Enhanced depth-of-focus differential interference contrast (DIC) images of high-proliferative cells (CVLOW) versus low-proliferative cells (CVHIGH) before and after 16 h of invasion in 3D collagen. (D–F) SUM159 cell position (D) and quantification of single-cell invasion (E) and translocation of invasion front (F) after 16 h of live-cell tracking in 3D collagen. Gray dots, non-invasive cells; black dots, invasive cells; white triangles, invasion front at start of imaging; black triangles, invasion front at end of imaging. See Video S3 (n = 3 biological replicates with n = 3 technical replicates). Data are means ± SEMs. **p < 0.01; 2-tailed unpaired Student’s t test. (G) Immunoblot of OPN showing memory extension in SUM159 cells treated with DMSO, palbociclib (2.5 μM; CDK4/6 inhibitor) or decitabine (7 μM; DNMT1 inhibitor) on soft hydrogels for 7 days in phase 2, after stiff- or soft-preconditioning for 7 days in phase 1. Drugs were added only in phase 2 (representative n = 3 biological replicates). (H) Schematic of experiment for injection of stiff preconditioned SUM159/GFP+ control or shRUNX2 cells into mice. BM was collected after 7 days, and then GFP+ cancer cells were sorted and assayed by ATAC-seq (n = 3 biological replicates). (I) Differentially accessible sites between control and shRUNX2 are shown by scatterplot (fold change > 1.7; p < 10−6), and motifs enriched in corresponding subsets are shown in the inset. (J) Proportions of quick and delayed closing sites (defined in Figure 3C) are shown for each set from (I).
Figure 7.
Figure 7.. RUNX2-mediated mechanical conditioning instructs bone metastasis
(A) qRT-PCR of RUNX2 target genes in SUM159 cells overexpressing RUNX2-WT, RUNX2-SE, or RUNX2-SA, preconditioned for 7 days on soft or stiff hydrogels (n = 3 biological replicates). (B) Quantification of invasion of SUM159 cells preconditioned as indicated in (A) (n = 3 biological replicates with n = 3 technical replicates). See Figures S6L–S6N and Video S6. (C) Micro-CT 3D reconstructions of proximal tibia from mice 4 weeks after intracardiac injection of SUM159 cells preconditioned as in (A) or no cancer cells (control). (D) Micro-CT analysis of bone volume from mice in (C) (n, mice; soft RUNX2-WT 5; soft RUNX2-SE 6; stiff RUNX2-WT 5; stiff RUNX2-SA 5; control 3). (E) Time course of SUM159 cells spreading on synthetic bone matrix, preconditioned as in (A) (n = 36 cells in each condition from n = 3 biological replicates). See Video S7. Shaded regions are means ± SEMs. ****p < 0.0001; 2-way ANOVA with Tukey’s multiple comparisons test. (F) Tartrate-resistant acid phosphatase (TRAP) staining of RAW264.7 cells after 7 days’ incubation: 4 days with 50 ng/mL RANKL in growth media, and then 3 days with 50% SUM159-conditioned media (CM) + 50% growth media. CM was collected 24 h after addition to hydrogels with equal SUM159 cell counts in each experimental group. Scale bar, 100 μm. (G and H) Quantification of (F) (n = 3 biological replicates with n = 3 technical replicates). Two-tailed unpaired Student’s t test. Data are means ± SEMs. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; 1-way ANOVA with Holm-Sidak’s multiple comparisons test, except for (E) and (G).
See this image and copyright information in PMC

References

    1. Altenbuchinger M, Schwarzfischer P, Rehberg T, Reinders J, Kohler CW, Gronwald W, Richter J, Szczepanowski M, Masqué-Soler N, Klapper W, et al. (2017). Molecular signatures that can be transferred across different omics platforms. Bioinformatics 33, i333–i340. - PMC - PubMed
    1. Anders S, and Huber W (2010). Differential expression analysis for sequence count data. Genome Biol. 11, R106. - PMC - PubMed
    1. Bar-Kochba E, Toyjanova J, Andrews E, Kim K-S, and Franck C (2015). A Fast Iterative Digital Volume Correlation Algorithm for Large Deformations. Exp. Mech 55, 261–274.
    1. Bernards MT, Qin C, and Jiang S (2008). MC3T3-E1 cell adhesion to hydroxyapatite with adsorbed bone sialoprotein, bone osteopontin, and bovine serum albumin. Colloids Surf. B Biointerfaces 64, 236–247. - PMC - PubMed
    1. Blake JA, Bult CJ, Eppig JT, Kadin JA, and Richardson JE; Mouse Genome Database Group (2009). The Mouse Genome Database genotypes:phenotypes. Nucleic Acids Res. 37, D712–D719. - PMC - PubMed

Publication types

Substances