Cytokeratin 5 positive cells represent a steroid receptor negative and therapy resistant subpopulation in luminal breast cancers

Abstract

A majority of breast cancers are estrogen receptor (ER) positive and have a luminal epithelial phenotype. However, these ER⁺ tumors often contain heterogeneous subpopulations of ER⁻ tumor cells. We previously identified a population of cytokeratin 5 (CK5) positive cells within ER⁺ and progesterone receptor positive (PR⁺) tumors that is both ER⁻PR⁻ and CD44⁺, a marker of breast tumor-initiating cells (TICs). These CK5⁺ cells have properties of TICs in luminal tumor xenografts, and we speculated that they are more resistant to chemo- and anti-ER-targeted therapies than their ER⁺ neighbors. To test this, we used ER⁺PR⁺ T47D and MCF7 breast cancer cells. CK5⁺ cells had lower proliferative indices than CK5⁻ cells, were less sensitive to 5-fluorouracil and docetaxel, and cultures became enriched for CK5⁺ cells after treatments. CK5⁺ cells were less prone to drug-induced apoptosis than CK5⁻ cells. In cells treated with 17β-estradiol (E) plus anti-estrogens tamoxifen or fulvestrant, ER protein levels decreased, and CK5 protein levels increased, compared to controls treated with E alone. In ER⁺ tumors from patients treated with neoadjuvant endocrine therapies ER gene expression decreased, and CK5 gene expression increased in post compared to pre-treatment tumors. The number of CK5⁺ cells in tumors also increased in post- compared to pre-treatment tumors. We conclude that an ER⁻PR⁻CK5⁺ subpopulation found in many luminal tumors is resistant to standard endocrine and chemotherapies, relative to the majority ER⁺PR⁺CK5⁻ cells. Compounds that effectively target these cells are needed to improve outcome in luminal breast cancers.

Fig. 1

Luminal ER⁺ breast cancers contain a subpopulation of ER⁻PR⁻CK5⁺ cells. a Paraffin sections of primary ER⁺ breast tumors and T47D and MCF7 breast cancer xenografts were stained by dual IHC for ER (brown)/CK5 (pink) or PR (brown)/CK5 (pink). Lower and higher magnification images are shown. Arrows indicate ER⁻CK5⁺ or PR⁻CK5⁺ cells in each higher magnification image. b Graph illustrates the percent of CK5⁺ cells that were positive or negative for each steroid receptor plus/minus SEM (n = 10 tumors T47D (PR), MCF7 (ER); n = 5 tumors T47D (ER), MCF7 (PR); n = 9 primary tumors (ER, PR)). ^* P < 0.0001 t test comparing the percent of CK5⁺ cells that were positive versus negative for ER or PR in each tumor type. (Color figure online)

Fig. 2

Relative quiescence of CK5⁺ compared to CK5⁻ cells in luminal breast cancer cell lines. a T47D_T cells were stained by dual fluorescent ICC for PR (red)/CK5 (green) or ER (red)/CK5 (green). Arrows indicate PR⁺(CK5⁻) and ER⁺(CK5⁻) cells and CK5⁺(ER⁻PR⁻) cells. b T47D_T and MCF7 cell lines were placed in 2D (top) or 3D Matrigel (bottom) culture conditions and allowed to grow into colonies for 3 days. Cultures were spiked with 0.25 mg/mL BrdU for 1 h prior to fixation. Representative fields of T47D_T and MCF7 cells stained by dual ICC (2D) or IHC (3D) for CK5 (green) and BrdU (red) are shown. Sections were counterstained for DAPI (blue). Nuclear BrdU staining and cytoplasmic CK5 staining are indicated in panel one. An asterisk marks a dual CK5⁺/BrdU⁺ cell (T47D_T, 2D). c Graph of BrdU incorporation in T47D_T and MCF7 cells in 2D and 3D culture. Multiple fields (10–12, 2D), or colonies (20–22, 3D) were scored for the number of DAPI, BrdU, CK5, and PR (T47D_T) positive cells. The percent of cells negative or positive for CK5, or PR (T47D_T) that were also BrdU⁺ were calculated and plotted (average plus/minus SEM). ^* P < 0.05 compared to CK5⁻ or PR⁺, one-way ANOVA/Tukey; ^** P < 0.01 and ^*** P < 0.001 t test compared to CK5⁻ cells. (Color figure online)

Fig. 3

The CK5⁺ population is enriched in T47D_T cells following chemotherapy. T47D_T cells were plated at 10⁵ cells/well and treated with vehicle, 150 μM 5-FU, or 10 nM Dx for 3 days. a Representative fields of cells treated with vehicle (control), 5-FU, or Dx and immunostained for CK5 (green) and DAPI (blue). b Multiple fields (n = 10–12) were counted for DAPI⁺ and CK5⁺ cells and plotted as the percent of CK5⁺ cells/field ± SEM. ^* P < 0.05 control versus 5-FU, ^** P < 0.01 control versus Dx, one-way ANOVA/Dunnett’s post test. (Color figure online)

Fig. 4

CK5⁺ cells resist chemotherapy induced apoptosis compared to CK5⁻ cells in luminal breast cancer cell lines. T47D_T, T47D (not derived from xenograft tumors), and MCF7 cells were plated into 3D Matrigel culture. T47D and MCF7 cells were treated with 100 nM MPA (+P) or 10 nM E + 100 nM MPA (E + P), respectively, for 24 h prior to drug treatments to induce CK5⁺ cells. Cells were treated with vehicle (control), 150 μM 5-FU, or 10 nM Dx for 3 days. a Representative immunostains of control or 5-FU-treated T47D_T colonies immunostained for CK5 (green), cleaved caspase 3 (red), and DAPI (blue). Arrows mark CK5⁺/cleaved caspase 3⁻ cells. b Graphs indicate the total percent of CK5⁺ and CK5⁻ cells that stained negative or positive for cleaved caspase 3 in control, 5-FU, and Dx-treated T47D_T cultures (n = 20), and in 5-FU-treated T47D/(+P) cultures (n = 10). The far graph indicates the total percent of CK5⁺ and CK5⁻ cells that stained negative or positive for cleaved PARP in MCF7/(E + P) cells treated with vehicle (cont) or 5-FU (n = 10). ^* P < 0.0001, t tests comparing CK5⁺ versus CK5⁻ for each treatment. (Color figure online)

Fig. 5

Loss of ER and gain of CK5 expression in T47D_T cells following endocrine therapy treatments. a T47D_T cells were maintained for 3 weeks in estrogen depleted media supplemented with 10 nM 17β-estradiol (E), no hormone (EWD), 10 nM E + 100 nM Tam (E + Tam), or 10 nM E + 100 nM ICI-182780 (E + ICI). Cells extracts were separated by SDS-PAGE and immunoblots probed with antibodies to ER, CK5, and β-actin (actin). b T47D_T cells were plated in 2D culture and treated with control or 150 μM 5-FU for 3 days. Immunoblots of extracts from control (cont) or 5-FU-treated cultures were probed for ER (top) and β-actin (actin, bottom)

Fig. 6

ER⁺ breast tumors from patients treated with neoadjuvant endocrine therapy have increased CK5 and decreased ER expression in post- versus pre-treatment samples. Patients with ER⁺ tumors from a phase II clinical trial were treated with neoadjuvant tamoxifen plus/minus exemestane for 4 months. Tumor samples were collected pre-and post-treatment. a mRNA levels for the genes encoding ER (ESR1) and CK5 (KRT5) were compared from published microarray data sets from these tumors (n = 6 pairs) [24]. Normalized expression levels plus/minus SEM are depicted. ^* P < 0.05, paired t test. b Sections from pre- and post-treatment tumors were immunostained for CK5 (n = 6 pairs). Two representative sections each of pre- and post- treatment tumors are shown. c Sections were scored for CK5 staining and plotted as number of cells per field plus/minus SEM (n = 6 pairs). ^* P < 0.05, paired t test