Poised chromatin at the ZEB1 promoter enables breast cancer cell plasticity and enhances tumorigenicity

Abstract

The recent discovery that normal and neoplastic epithelial cells re-enter the stem cell state raised the intriguing possibility that the aggressiveness of carcinomas derives not from their existing content of cancer stem cells (CSCs) but from their proclivity to generate new CSCs from non-CSC populations. Here, we demonstrate that non-CSCs of human basal breast cancers are plastic cell populations that readily switch from a non-CSC to CSC state. The observed cell plasticity is dependent on ZEB1, a key regulator of the epithelial-mesenchymal transition. We find that plastic non-CSCs maintain the ZEB1 promoter in a bivalent chromatin configuration, enabling them to respond readily to microenvironmental signals, such as TGFβ. In response, the ZEB1 promoter converts from a bivalent to active chromatin configuration, ZEB1 transcription increases, and non-CSCs subsequently enter the CSC state. Our findings support a dynamic model in which interconversions between low and high tumorigenic states occur frequently, thereby increasing tumorigenic and malignant potential.

Figure 1. Basal breast cancer CD44^lo non-CSC cell populations spontaneously switch to a CD44^hi CSC state in vivo

(A) Tumorigenicity of FACS-purified luminal BrCa CD44^lo cells or basal BrCa CD44^lo and CD44^hi cell populations following orthotopic injection into NOD/SCID mice (n ≥ 6/group). (B) Representative FACS plots for CD44 expression and quantification of CD44^hi cells generated from luminal or basal CD44^lo-derived tumors generated in (A). (C) Basal CD44^lo digested tumors from (B) were cultured in vitro to generate ex vivo cell lines (ExV). ExV lines were purified by FACS into CD44^lo and CD44^hi components and injected orthotopically into NOD/SCID mice (n ≥ 8/group). Tumor incidence displayed as percentages on the graph. Data represented as mean ± SEM. See also Figure S1.

Figure 2. ZEB1 is an essential mediator of CD44^lo-to-CD44^hi cell transitions

(A) Western blot for markers of the epithelial (CDH1) or mesenchymal (CDH2, VIM) phenotype in immortalized human mammary epithelial cells (HME), HME-flopc and single cell clones derived from HME-flopc population enriched for the CD44^lo phenotype (clones F1 and F2) or CD44^hi phenotype (clones F3 and F4). (B) qPCR for EMT transcription factors and MIR200B/C in non-transformed CD44^lo (HME and HME-flopc-CD44^lo) or HME-flopc-CD44^hi cells. (C) FACS analysis for the ability of HME-flopc-CD44^lo cells to switch to the CD44^hi cell state. Cells express a doxycycline (dox) inducible control shRNA (control) or shRNA targeting ZEB1 (sh1 and sh2). -/- no dox, +/- dox on for 8 days then removed for the remaining 8 days, +/+ dox on for the duration of the experiment. Inhibition (%) at day 16 is also shown (*p < 0.0001, **p < 0.0008, compared to (-/-). (D) Purified HME-flopc-CD44^lo cells treated with MIR200B/C inhibitors (I) or mimetics (M) to determine effects on switching from CD44^lo to CD44^hi cell state. Data are mean ± SEM. (E) Purified HME-flopc-CD44^lo cells expressing a dox-inducible control shRNA (control) or shRNA targeting ZEB1 (sh1, sh2 or sh3) were analyzed for their ability to switch to the CD44^hi state in the presence (+) or absence (-) of dox, and in response to a MIR200C inhibitor (I) or mimetic (M). (F) Transformed HME-flopc-CD44^lo cells (with SV40-Early Region and RAS oncoprotein) expressing a dox-inducible shRNA targeting ZEB1 (sh1) were analyzed for conversion to the CD44^hi state in the presence (+) or absence (-) of ZEB1-knockdown. Cells were monitored by FACS for 8 days. Data represented as mean ± SEM. See also Figure S2.

Figure 3. Inhibition of CD44^lo-to-CD44^hi conversions by blocking ZEB1 decreases tumorigenicity

(A) Western blot comparing the expression of ZEB1 in basal BrCa cell lines (HMLER and HCC38) purified for CD44^lo or CD44^hi subpopulations, and luminal BrCa cell lines (ZR-75-1, T47D, MCF7 and MCF7R). Quantification of differential ZEB1 expression in basal cell lines (n = 4). (B) qPCR assessing ZEB1, MIR200B and MIR200C mRNA expression in BrCa cell lines. (C) Schematic illustrating expression of MIR200 family members and ZEB1 protein expression in basal (CD44^lo and CD44^hi subpopulations) and luminal BrCa cell lines. (D) Purified CD44^lo cells from HMLER or HCC38 and SUM159 basal BrCa cell lines expressing dox-inducible control shRNA (control) or shRNA targeting ZEB1 (sh1 and sh2) were analyzed for tumorigenic potential. Final tumor mass and incidence are represented (n ≥ 5/group). Data represented as mean ± SEM. See also Figure S3

Figure 4. ZEB1 is essential for the stem cell/CSC activity of CD44^hi cells

(A) FACS analysis for CD44 expression in purified HME-flopc-CD44^hi cells. Cells express a doxycycline (dox) inducible control shRNA (control) or shRNA targeting ZEB1 (sh1 and sh2). -/- no dox, +/- dox on for 8 days then removed for the remaining 8 days, +/+ dox on for the duration of the experiment. The percentage of spontaneously arising CD44^lo cells was determined by FACS over a 16 day time period. (B) Purified HME-flopc-CD44^hi cells expressing control or shRNA targeting ZEB1 (sh1, sh2 and sh3) were assessed for mammosphere-forming ability with or without dox-induction. Cells were treated with a MIR200C inhibitor (I) or mimetic (M). p<0.001, two-way ANOVA followed by Tukey's multiple comparisons test, *- different to miR-control and miR-200c-M, **- different to miR-control and miR-200c-I (C) Transformed HME-flopc-CD44^hi cells expressing control, sh1 or sh2 were assessed for mammosphere formation with or without dox-induction (p<0.0001, one-way ANOVA followed by Tukey's multiple comparisons test, *-different to sh(-)). (D) Transformed HME-flopc-CD44^hi cells (control, sh1 and sh2) were purified by FACS and implanted into the fat pad of NOD/SCID mice (n = 8/group). Tumor weight and incidence are shown. Data represented as mean ± SEM. See also Figure S4

Figure 5. The ZEB1 promoter is maintained in a bivalent chromatin state in basal CD44^lo non-CSCs

(A) Schematic showing the location of primer sets used for ChIP-qPCR. (B) and (C) ChIP-qPCR for the H3K4me3, H3K27me3 and H3K79me2 histone modifications at the ZEB1 promoter in (B) non-transformed CD44^lo or CD44^hi cells and (C) luminal CD44^lo cells and basal CD44^lo and CD44^hi sorted populations. Data are mean ± SEM of biological duplicates performed as technical replicates.

Figure 6. TGFbeta can induce CD44^lo-to-CD44^hi switching and modulates the chromatin at the ZEB1 promoter

(A) Purified HME-flopc-CD44^lo cells expressing dox-inducible shRNA targeting ZEB1 (sh1 and sh2) were monitored by FACS for their ability to switch to the CD44^hi state following TGFbeta treatment in vitro. *p<0.0001 - different to control, **p<0.001- different to control (-dox). (B) Transformed HME-flopc-CD44^lo cells expressing sh1 targeting ZEB1 were treated with TGFbeta and monitored by FACS for switching to the CD44^hi state. (C) Purified CD44^lo cells from luminal (MCF7R and ZR-75-1) and basal (HMLER and HCC38) BrCa cell lines monitored by FACS for switching to the CD44^hi state following TGFbeta or SB431542 treatment in vitro *p<0.0001 – two-way ANOVA followed by Tukey's multiple comparisons test. (D) Representative FACS plots of HME-flopc-CD44^lo cells expressing sh1 or sh2 targeting ZEB1 treated with control (PBS), TGFbeta (2ng/ml) or SB431542 (10μM). (E) ChIP-qPCR for the H3K4me3, H3K27me3 and H3K79me2 histone modifications at the ZEB1 promoter in cells from (D) (*p<0.0001, n = 4, two-way ANOVA followed by Tukey's multiple comparison test, different to control and SB431542). (F-H) MCF7R cells expressing a dox-inducible empty vector (control) or ZEB1 overexpression construct were treated with dox and monitored by FACS for their ability to switch to the CD44^hi state (F), for the ability to form tumorspheres in vitro, *p<0.0001, one-way ANOVA and Tukey's multiple comparisons test, different to control (G) and for tumorigenicity in vivo (tumor initiating ability marked as percentages on each bar), *p = 0.03, one-way ANOVA and Tukey's multiple comparisons test, different to Control and ZEB1-lo (H). Data represented as mean ± SEM. See also Figure S6.

Figure 7. ZEB1 in clinical cases of breast cancer

(A) MIR200C accounts for 93% (sd=5%) of mature miRNA in the MIR200BC family, which also includes MIR200B (6%) and MIR429 (1%), so subsequent analysis uses MIR200C to represent the MIR200BC family. (B) Levels of ZEB1 and MIR200C are inversely correlated in basal (p=9.4e-4; r2=0.13), luminal A (p=2.8e-4; r2=0.07) and luminal B (p=6.8e-4; r2=0.12) subtypes (but not Her2). ZEB1 is shown as median-centered values and MIR200C by log2-transformed reads per million mapped reads (RPM). (C) mRNA abundance of ZEB1, ZEB2, and MIR200C by subtype. Asterisks indicate a difference compared to the basal subtype (p<0.05; ANOVA with Dunnett post-hoc): *=p<0.05; **=p<0.01; *=p<0.05. (D) Human BrCa tissue array stained with an antibody targeting ZEB1 (100x and 400x images provided). ** p=0.017, TN compared to Luminal A, Fisher's Exact Test followed by Bonferroni correction for multiple-hypothesis testing. See also Figure S7. (E) Schematic depicting: 1) an alternative CSC model for basal-type BrCa that includes bidirectional conversions between CSCs and non-CSCs, and 2) a model of non-CSC-to-CSC conversion that includes a microenvironmental stimulus acting on non-CSCs harboring bivalent chromatin marks at the ZEB1 promoter enabling a rapid activation of ZEB1 and switch to a CSC state.