Identification of Genes Regulating Breast Cancer Dormancy in 3D Bone Endosteal Niche Cultures

Abstract

Tumor cell dormancy is a significant clinical problem in breast cancer. We used a three-dimensional (3D) in vitro model of the endosteal bone niche (EN), consisting of endothelial, bone marrow stromal cells, and fetal osteoblasts in a 3D collagen matrix (GELFOAM), to identify genes required for dormancy. Human triple-negative MDA-MB-231 breast cancer cells, but not the bone-tropic metastatic variant, BoM1833, established dormancy in 3D-EN cultures in a p38-MAPK-dependent manner, whereas both cell types proliferated on two-dimensional (2D) plastic or in 3D collagen alone. "Dormancy-reactivation suppressor genes" (DRSG) were identified using a genomic short hairpin RNA (shRNA) screen in MDA-MB-231 cells for gene knockdowns that induced proliferation in the 3D-EN. DRSG candidates enriched for genes controlling stem cell biology, neurogenesis, MYC targets, ribosomal structure, and translational control. Several potential DRSG were confirmed using independent shRNAs, including BHLHE41, HBP1, and WNT3. Overexpression of the WNT3/a antagonists secreted frizzled-related protein 2 or 4 (SFRP2/4) and induced MDA-MB-231 proliferation in the EN. In contrast, overexpression of SFRP3, known not to antagonize WNT3/a, did not induce proliferation. Decreased WNT3 or BHLHE41 expression was found in clinical breast cancer metastases compared with primary-site lesions, and the loss of WNT3 or BHLHE41 or gain of SFRP1, 2, and 4 in the context of TP53 loss/mutation correlated with decreased progression-free and overall survival. IMPLICATIONS: These data describe several novel, potentially targetable pathways controlling breast cancer dormancy in the EN.

Figure 1.. Dormancy induction in 3D-EN is p38-MAPK-dependent.

Relative cell numbers of MDA-MB-231 (A), MDA-MB-231[BoM1833] (B) or MDA-MB-231 cells with p38 knockdown (vs. shCont.) (C) grown for either 1 or 7 d in 3D-EN or 3D, or in 2D (control) conditions. N = independent replicates; error bars, SEM; **, p <0.001.

Figure 2.. Analysis of HBP1 and WNT3 as potential DSRG.

(A) Gene Set Enrichment Analysis of module-1 DSRG candidates identified 18 of 65 hits as being MYC target genes. (B) qRT-PCR showing knockdown of HBP1 in MDA-MB-231 cells. (C and D) Knockdown of HBP1 (C) or WNT3 (D) induces proliferation in 3D-EN vs. 3D (“C”) or 2D cultures. Error bars, SEM of three independent replicates; *, p < 0.01; **, p <0.001. (E) Confirmation of WNT3 knockdown by qRT-PCR. Error bars, SEM of three independent replicates. (F) Immunoblot of lysates of MDA-MB-231 or BoM1833 transduced with scramble shRNA (“shCont”), or WNT3-knockdown MDA-MB-231 cells probed for total or activated (poT202/Y204) ERK1/2, total or activated (poT180/Y182) p38-MAPK or α-tubulin (as a loading control). Digital quantifications are shown as normalized to the shControl. This blot is typical of three independent experiments. (G) Overexpression of SFRP2 or 4, but not SFRP3, in MDA-MB-231 induces proliferation in 3D-EN cultures, whereas the overexpression of WNT3 in BoM1833 suppresses 3D-EN proliferation. Error bars, SEM of three independent replicates; *, p < 0.01; **, p <0.001. (H) Immunoblot of MDA-MB-231 lysates transduced with lentivirus expressing V5-tagged SFRP2, 3, or 4 (or empty vector), or BoM1833 (“1833”) cells transduced with WNT3 (or empty vector), probed for V5 or GAPDH. Molecular weight markers are at right.

Figure 3.. WNT3 expression in clinical BrCa datasets and correlation with survival.

(A) Oncomine TCGA Breast and Radvanyi datasets showing relative WNT3 expression in primary (1°) vs. metastatic BrCa cases. N = number of cases. (B) Copy number variations, mutations and expression changes of TP53, WNT3, ERBB2, PGR and ESR1 in the TCGA Breast dataset produced through cBioPortal, with vertical bars representing a single patient, and the percentages representing the total changes for a given gene. (C and D) Progression-free (C) and overall survival (D) for TGCA Breast and METABRIC datasets, respectively, based on combined TP53, WNT3, ERBB2, PGR and ESR1 losses in the TGCA data, and TP53 and WNT3 losses in the METABRIC dataset. The number of cases with or without these gene changes, as well as the median number of disease-free months, are shown below. (E) Copy number variations, mutations and expression changes of TP53, SFRP1, 2 and 4 in the TCGA Breast dataset produced through cBioPortal. (F and G) Progression-free (F) and overall survival (G) for TGCA Breast and METABRIC datasets, respectively, based on combined TP53, SFRP1, 2 and 4 losses, with numbers of cases with or without changes (below).

Figure 4.. BHLHE41 expression in clinical BrCa datasets and correlation with survival.

(A) Oncomine Bittner, TCGA Breast and Radvanyi datasets showing relative BHLHE41 expression in primary (1°) vs. metastatic BrCa cases. N = number of cases. (B) Copy number variations, mutations and expression changes of TP53 and BHLHE41 in the TCGA Breast dataset produced through cBioPortal, with vertical bars representing a single patient, and the percentages representing the total changes for a given gene. (C and D) Progression-free (C) and overall survival (D) for TGCA Breast and METABRIC datasets, respectively, based on combined TP53 and BHLHE41 losses in the METABRIC dataset. The number of cases with or without these gene changes, as well as the median number of disease-free months, are shown below.