Role of Krüppel-like Factor 4-p21^CIP1 Axis in Breast Cancer Stem-like Cell Inhibition by Benzyl Isothiocyanate

Novel interventions for chemoprevention of breast cancer, especially for estrogen receptor (ER)-negative disease, are urgently needed because of the side effects associated with the currently available clinical agents like selective ER-modulators (e.g., tamoxifen) and aromatase inhibitors such as exemestane, and their lack of activity against ER⁻ subtypes (1–3). Common edible plants as well as their bioactive phytochemical constituents continue to gain momentum for identification of safe and efficacious cancer chemopreventive approaches (4). Continued interest in the search for dietary cancer chemopreventive interventions for breast cancer is substantiated by the results of population-based epidemiologic studies suggesting a cancer protective role for certain edible plants including cruciferous vegetables (5, 6). Common edible cruciferous vegetables include broccoli, cabbage, watercress, garden cress, and so forth. Isothiocyanate (R-N=C=S) family of phytochemicals is partly responsible for the cancer-protective effect of cruciferous vegetables (4, 7, 8). Structurally dissimilar isothiocyanates are present in cruciferous vegetables differing by the nature of the side chain connected to the N=C=S moiety, including aromatic (e.g., phenethyl-N=C=S and benzyl-N=C=S) and thio-alkyl [e.g., 4-(methylsulfinyl)-butyl-N=C=S; commonly known as sulforaphane; ref. 8)]. In a structure–activity relationship study, we demonstrated previously that benzyl isothiocyanate (BITC) was a relatively more potent inhibitor of human breast cancer cell growth (MCF-7 and MDA-MB-231) and apoptosis inducer in vitro than phenethyl-N=C=S or sulforaphane (9).

Chemically induced as well as transgenic rodent models of breast cancers have been utilized to demonstrate chemopreventive efficacy of BITC (10, 11), whereas the growth inhibitory effect of this agent in a therapeutic setting has been shown in human and murine xenograft models (12, 13). Wattenberg (10) was the first to document BITC-mediated chemoprevention of breast neoplasm in rats in which cancer was induced by exposure to a chemical carcinogen [7,12-dimethylbenz(a)anthracene]. We showed subsequently that feeding of mouse mammary tumor virus-neu (MMTV-neu) transgenic mice, in which ER⁻ breast cancer is driven by mammary gland–specific overexpression of EGFR 2 oncogene, with a diet supplemented with BITC caused prevention of breast cancer in association with decreased cellular proliferation as well as apoptosis induction when compared with mice fed a basal diet (11). In a therapy setting, oral administration of BITC to female BALB/c mice orthotopically implanted with murine 4T1 breast cancer cells also caused a significant decrease of tumor growth in vivo (13). We have also shown recently that the multiplicity of skeletal metastasis induced by intracardiac injection of MDA-MB-231 cells in nude mice is greatly inhibited by five times/week oral treatment with BITC (14).

Inhibition of breast cancer stem-like cell population (bCSC) is another interesting characteristic of BITC (15). Mammosphere number as well as aldehyde dehydrogenase 1 (ALDH1)-positive bCSC fraction, which are well-accepted markers of bCSC, in MCF-7 and SUM159 human breast cancer cells were decreased markedly in the presence of BITC in vitro when compared with corresponding vehicle-treated control cells (15). Moreover, oral administration of BITC to MMTV-neu mice resulted in a significant decrease in ALDH1-positive bCSC fraction in vivo in the mammary tumors in comparison with controls (15). However, the mechanism underlying BITC-mediated inhibition of bCSC is still poorly understood. Studies have shown that Krüppel-like factor 4 (KLF4) is involved in maintenance of bCSC (16, 17). Interestingly, treatment of a colon cancer cell line with the BITC analogue sulforaphane resulted in induction of KLF4 expression (18). On the basis of these published findings (16–18), this study was designed to explore possible contribution of KLF4 in bCSC inhibition by BITC.

IHC and Western blot analyses for KLF4 protein expression were performed using mammary tumors/tumor sections of control and BITC-treated MMTV-neu mice. This study was approved by the Institutional Animal Care and Use Committee. Details of BITC treatment and tissue collection have been described previously (11, 15).

BITC was purchased from LKT Laboratories. Stock solution of BITC was prepared in DMSO. Control cells were treated with equal volume of DMSO, and its final concentration did not exceed 0.1%. Reagents for cell culture were purchased from Invitrogen-Life Technologies. An antibody against KLF4 was from Cell Signaling Technology; antibodies against p21^CIP1 and β-Catenin were from BD Biosciences; anti-KLF4 antibody for immunofluorescence and chromatin immunoprecipitation (ChIP) was from Santa Cruz Biotechnology; and anti–β-Actin antibody was from Sigma-Aldrich. A nonspecific control siRNA and KLF4-targeted siRNA were purchased from Qiagen, whereas p21^CIP1-targeted siRNA was from Cell Signaling Technology. The MCF-7 and MDA-MB-231 cells were purchased from the ATCC and more recently authenticated in March of 2017. The SUM159 cell line was purchased from Asterand and authenticated in March of 2017. Each cell line was maintained at 37°C in an atmosphere of 95% air and 5% CO₂ according to the recommendations of the providers. The MCF-7 cells stably transfected with pRetroSuper vector or the same vector with KLF4 small-hairpin RNA (hereafter abbreviated as control shRNA and KLF4 shRNA cells, respectively) were generously provided by Dr. Yong Wan (University of Pittsburgh, Pittsburgh, PA).

Whole-cell lysates from control and BITC-treated cells, and supernatants from mammary tumor tissues of control and BITC-treated MMTV-neu mice were prepared as described previously (19, 20). BITC concentrations used in this study for the cellular in vitro studies were within the plasma-achievable level based on a mouse pharmacokinetic study (21). Western blotting analysis was done as described previously (19).

Cells were plated on glass coverslips in 24-well plates (3 × 10⁴), allowed to attach for overnight, and treated with DMSO or 5 μmol/L BITC for 24 hours. After treatment, cells were fixed with 2% paraformaldehyde for 1 hour, and permeabilized with 0.5% Triton X-100 for 5 minutes. Cells were then incubated with buffer containing 0.5% BSA and 0.15% glycine in PBS for 1 hour at room temperature followed by overnight incubation with anti-KLF4 antibody (1:500 dilution) at 4°C. The next day, cells were incubated with Alexa Fluor 568–conjugated secondary antibody for 1 hour at room temperature and then treated with Hoechst 33342 for 5 minutes to stain nuclear DNA. After washing with PBS, the cells were mounted and observed using an Olympus FluoView FV1000 confocal microscope at ×60 objective magnification.

tRNA from cells was isolated using the RNeasy Kit (Qiagen). The cDNA was synthesized using Superscript Reverse Transcriptase (Invitrogen-Life Technologies) with oligo (dT)₂₀ primer. Primers were as follows: KLF4 forward, 5′-CGAACCCACACAGGTGAGAA-3′ and reverse, 5′-TACGGTAGTGCCTGGTCAGTTC-3′; glyceraldehyde-3-phosphate dehydrogenase (GAPDH) forward, 5′-GGACCTGACCTGCCGTCTAGAA-3′ and reverse, 5′-GGTGTCGCTGTTGAAGTCAGAG-3′. The qPCR was done using 2 × SYBR Green Master Mix (Thermo Fisher Scientific) with 95°C (15 seconds), 55°C annealing (60 seconds), and 72°C (30 seconds) for 40 cycles. Human GAPDH was used as a control. Relative gene expression was calculated using the method described previously (15).

Cells were seeded in 6-well plates and transiently transfected with a control nonspecific siRNA, a KLF4-targeted siRNA for 48 hours or a p21^CIP1-targeted siRNA for 24 hours. After transfection, the cells were treated with DMSO or BITC and used for ALDH1 activity determination or mammosphere assay.

The ALDH1 activity was determined by flow cytometry using the ALDEFLUOR kit from Stem Cell Technologies as recommended by the supplier.

Mammosphere assay was done as described previously (15). Briefly, 1,000 cells were plated in ultralow attachment plates in medium containing B27, insulin, hydrocortisone, EGF, basic FGF, 2-mercaptoethanol, and methylcellulose. Desired concentration of BITC was then added to the plates. Same volume of DMSO was added to control samples. After a 5-day (MCF-7) or 7-day (SUM159) incubation, the mammospheres were scored under an inverted microscope.

IHC for KLF4 protein expression in 4–5 μm mammary tumor sections of control and BITC-treated MMTV-neu mice was performed as described previously for other proteins (20). At least six nonoverlapping fields on each slide were captured and analyzed using Nuclear v9.1 algorithm of the Aperio ImageScope software.

RNA-Seq data from The Cancer Genome Atlas (TCGA) were analyzed to examine the association between expression of KLF4 and that of p21^CIP1 (CDKN1A) or β-Catenin (CTNNB1). TCGA data set were analyzed using University of California Santa Cruz Xena Browser (https://xena.ucsc.edu/public-hubs/).

MCF-7 and SUM159 cells were prepared for ChIP assay using a kit from Millipore. Other details were essentially the same as described previously (22). Briefly, DNA and associated proteins in cells were cross-linked with formaldehyde by treatment for 20 minutes. The cells were sonicated in SDS lysis buffer to shear DNA. Subsequently, immunoprecipitation was done using an anti-KLF4 antibody or control IgG antibody for overnight. After reversal of cross-linking, DNA fragments were purified on spin column. The KLF4-binding sites in the p21^CIP1 promoter were amplified (60°C, 1 minute, 40 cycles) with the following region-specific primers. Site 1, forward, 5′-TCTGGGGTTTAGCCACAATC-3′ and reverse, 5′-CATGCCCAGTCTTCTTCCTC-3′ site 2, forward, 5′-GCAAATGTTTCAGGCACAGA-3′ and reverse, 5′-CCCTCATTTGCAGATGGTTT-3′. Fold enrichment was normalized to the input.

One-way ANOVA followed by Dunnett or Newman–Keuls test or Student t test was performed for statistical comparisons. A significance level was set at 0.05. Statistical analyses were performed using a GraphPad Prism (v 7.02).

BITC treatment decreases bCSC population in preclinical models in vitro and in vivo (15), but the molecular regulators of this effect largely remain elusive. Initially, we tested the possibility of whether bCSC inhibition by BITC was associated with suppression of KLF4 because studies have implicated this transcription factor in maintenance of bCSC (16, 17). Surprisingly, exposure of human breast cancer cell lines (MCF-7, MDA-MB-231, and SUM159) to BITC resulted in a concentration-dependent increase in expression of KLF4 protein that was evident 6 hours posttreatment (Fig. 1A). Confocal microscopy confirmed BITC-mediated induction of KLF4 protein in each cell line (Fig. 1B). Moreover, BITC treatment resulted in nuclear accumulation of KLF4 protein in MDA-MB-231 and SUM159 cells, which are triple-negative breast cancer cell lines. To the contrary, BITC treatment caused an increase in both nuclear and cytosolic levels of KLF4 in the MCF-7 cell line, which is a luminal-type breast cancer cell line (Fig. 1B). KLF4 mRNA level was also increased with BITC treatment in each cell line as revealed by qPCR analysis. This effect was statistically significant at 5 μmol/L in each cell line at each time point BITC (Fig. 1C). Together, these results indicated induction of KLF4 upon BITC treatment in a panel of human breast cancer cell lines.

We performed knockdown studies to determine the functional significance of KLF4 induction in BITC-mediated inhibition of bCSC population. The level of KLF4 protein was decreased by 70% to 90% upon transient transfection of MCF-7 and SUM159 cells with a KLF4-targeted siRNA when compared with cells transfected with a nonspecific control siRNA (Fig. 2A). Figure 2B shows flow histograms for ALDH1 activity in MCF-7 and SUM159 cells transfected with the specified siRNA and treated for 48 hours with DMSO (control) or BITC in the absence or presence of 20 μmol/L N,N-diethylaminobenzaldehyde (DEAB), which is an inhibitor of ALDH1 activity (23). The percentage of ALDH1-positive bCSC fraction was low in both cell lines consistent with the published literature (22, 24, 25). For example, in the absence of DEAB treatment, the percentage of ALDH1-positive bCSC fraction was about 1.4% in control MCF-7 cells that was reduced to 0.1% in the presence of DEAB (24). In this study, the percentage of ALDH1-positive bCSC population was about 1.2% in vehicle-treated control MCF-7 cells in the absence of DEAB. As expected, the ALDH1-positive bCSC fraction was decreased in MCF-7 and SUM159 cells in the presence of DEAB (Fig. 2B). The KLF4 knockdown alone significantly decreased ALDH1 activity in both cells but this effect was relatively more pronounced in the MCF-7 cell line than in the SUM159 cells (Fig. 2C). BITC-mediated inhibition of ALDH1 activity was augmented by knockdown of KLF4 protein in both MCF-7 and SUM159 cells (Fig. 2C). In agreement with these results, a combination of BITC treatment and KLF4 knockdown was more effective for suppression of mammosphere formation when compared with BITC treatment or KLF4 knockdown alone (Fig. 2D).

MCF-7 cells stably transfected with a control shRNA or a KLF4-targeted shRNA were also used for confirmation of the results with siRNA. The level of KLF4 protein was lower by about 90% in MCF-7 cells upon stable transfection with KLF4-targeted shRNA when compared with control shRNA transfected cells (Fig. 3A). Figure 3B shows mammospheres resulting from MCF-7 cells transfected with each shRNA and with or without treatment with BITC. The BITC-mediated suppression of mammosphere number (Fig. 3C) as well as ALDH1 activity (Fig. 3D) was augmented by KLF4 knockdown. Collectively, these results indicated that bCSC maintenance was partly regulated by KLF4, and that KLF4 induction attenuated bCSC inhibition by BITC treatment.

Western blotting using mammary tumor supernatants from BITC-treated MMTV-neu mice revealed an approximately 3.6-fold higher levels of KLF4 protein when compared with that of controls, but the difference was not significant because of large variability especially in the experimental group (Fig. 4A). IHC also indicated a nearly 4-fold higher H-score for KLF4 protein expression in the BITC treatment group in comparison with the control group (Fig. 4B). The p21^CIP1 and β-Catenin are known downstream targets of KLF4 (26, 27), and thus their protein expression was also compared in mammary tumors of control and BITC-treated MMTV-neu mice. Expression of both p21^CIP1 and β-Catenin proteins was relatively higher in mammary tumors from BITC-treated mice in comparison with controls (Fig. 4A).

Knockdown of KLF4 resulted in suppression of p21^CIP1 protein level in both MCF-7 and SUM159 cells (Fig. 4C and D). BITC treatment caused induction of KLF4 and p21^CIP1 proteins in both cell lines that was markedly attenuated by knockdown of KLF4. However, a similar association was not discernible for β-Catenin (Fig. 4C and D). A marked induction of p21^CIP1 protein upon BITC treatment was still visible even after about 70% to 90% knockdown of KLF4 suggesting alternate mechanisms in BITC-mediated induction of p21^CIP1.

Next, we examined the association between expression of KLF4 with that of p21^CIP1 (CDKN1A) and β-Catenin (CTNNB1) in TCGA breast cancer database. Analyses of the RNA-Seq data for breast cancer revealed a significant positive association between expression of KLF4 with that of p21^CIP1 (CDKN1A) and β-Catenin (CTNNB1) in TCGA breast cancer database (Fig. 5). Because of inconsistent data for β-Catenin (Fig. 4C, D, and Fig. 5), only p21^CIP1 was studied further. Possible KLF4-binding sites (CACCC and GGGGTGT; 24, 25) were identified at the promoter of p21^CIP1 (Fig. 6A). ChIP assay revealed recruitment of KLF4 at site 1 and site 2 of the p21^CIP1 promoter in both MCF-7 and SUM159 cells (Fig. 6B) but site 3 was not tested. Collectively, these results indicated that KLF4 was a direct regulator of p21^CIP1 expression in breast cancer cells.

Consistent with the results for KLF4, knockdown of p21^CIP1 protein alone caused a significant decrease in ALDH1 activity in MCF-7 and SUM159 cells (Fig. 6C and D). Inhibition of ALDH1 activity resulting from BITC treatment was augmented by knockdown of the p21^CIP1 protein (Fig. 6D).

Studies have demonstrated context/cell line–dependent oncogenic or tumor-suppressive effects of KLF4 in breast cancer (28–33). In one such study, KLF4-mediated inhibition of CD44 affected pancreatic cancer stem cell–like characteristics and metastasis (28). A tumor-suppressive function of KLF4 was shown in ER⁻ and EGFR 2-enriched SK-BR-3 human breast cancer cell line in which its overexpression caused suppression of tumor growth in vivo (30). This study found lower level of KLF4 expression in breast tumors and breast cancer cell lines in comparison with nonmalignant mammary tissues and a nontransformed mammary epithelial cell line (30). In another study with ER⁺ MCF-7 cells, KLF4 suppressed estrogen-dependent cell growth by inhibiting transcriptional activity of ER-α by binding to its DNA-binding domain (33). Expression of KLF4 mRNA was observed in 68% of the breast cancer but its level was low in adjacent morphologically normal breast epithelial cells (31). Nuclear KLF4 expression was shown to be associated with an aggressive phenotype in early-stage primary human-infiltrating ductal breast cancer (32). This study shows that KLF4 knockdown suppresses bCSC population, and these results are consistent with the data reported by Yu and colleagues (16). Although the mechanism(s) underlying inconsistent role of KLF4 in pathogenesis of breast cancer is still not fully understood, the p21^CIP1 has been suggested to function as a switch to affect outcome of KLF4 activity (26). This study shows that KLF4 expression is positively associated with that of p21^CIP1 in breast cancer TCGA (Fig. 5). Moreover, KLF4 knockdown markedly suppresses expression of p21^CIP1 protein in both MCF-7 and SUM159 cells. We also found recruitment of KLF4 at the p21^CIP1 promoter by ChIP assay. These results indicate that KLF4 is a direct regulator of p21^CIP1 and this axis contributes to bCSC maintenance irrespective of the ER-α status.

Data presented herein demonstrate that induction of KLF4-p21^CIP1 proteins by BITC attenuates its inhibitory effect on bCSC population. This conclusion is based on the following observations: (i) BITC treatment at plasma-achievable doses causes robust induction of KLF4 and p21^CIP1 proteins in breast cancer cells in vitro and mammary tumors of MMTV-neu mice in vivo albeit insignificantly, and (ii) inhibition of ALDH1 activity and/or mammosphere formation by BITC is boosted by knockdown of both KLF4 and p21^CIP1 proteins. The translational implication of these findings is that bCSC inhibition by BITC, and consequently its chemopreventive activity is likely enhanced if administered in combination with a KLF4 inhibitor. Although further studies are needed to determine the mechanism by which BITC treatment increases expression of KLF4 protein in breast cancer cells, the p21^CIP1 induction resulting from BITC exposure cannot be fully explained by KLF4 status. Both proteins are induced upon BITC treatment in MCF-7 and SUM159 cells transfected with a control siRNA, but BITC-mediated induction of p21^CIP1 protein is partially maintained even after 70% to 90% knockdown of KLF4.

Induction of p21^CIP1 protein by BITC treatment has been reported in other types of cancer cells. In a human non–small-cell lung carcinoma cell line (A549), BITC-induced apoptosis was associated with induction of p21^CIP1 protein (34). BITC-mediated G₂–M phase cell-cycle arrest in pancreatic cancer cells was accompanied by induction of p21^CIP1 protein (35). On the other hand, BITC treatment decreased p21^CIP1 expression in a glioma cell line (36). Functional experiments in cell lines with gain- or loss-of-function of p21^CIP1 were not performed in any of these studies to determine its role in cell-cycle arrest or apoptosis induction from BITC treatment. This study clearly indicates that p21^CIP1 induction is protective against bCSC inhibition by BITC.

In conclusion, this study reveals that bCSC inhibition by BITC is partially attenuated by induction of KLF4 and p21^CIP1 proteins. We also show that KLF4 is a direct regulator of p21^CIP1 in breast cancer cells.

No potential conflicts of interest were disclosed.

Conception and design: S.-H. Kim, S.V. Singh

Development of methodology: S.-H. Kim

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.-H. Kim

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.-H. Kim, S.V. Singh

Writing, review, and/or revision of the manuscript: S.-H. Kim, S.V. Singh

Study supervision: S.V. Singh

This work was supported by the NCI (grant no. RO1 CA129347, to S.V. Singh). This research used the Flow Cytometry Facility and the Tissue and Research Pathology Facility supported in part by Cancer Center Support Grant from the NCI (P30 CA047904, to principal investigator, Robert L. Ferris).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.