KRT13 promotes stemness and drives metastasis in breast cancer through a plakoglobin/c-Myc signaling pathway

Human breast cancer cell lines MCF7 (indolent, low endogenous KRT13 expression) and HCC1954 cells (aggressive, high endogenous KRT13 expression) were provided by Dr. X. J. Cui at Cedars-Sinai Medical Center (CSMC). MCF7 cells were maintained in DMEM (ThermoFisher Scientific, Waltham, MA) and HCC1954 was maintained in RPMI1640 (ThermoFisher Scientific). Both culture media were supplemented with 10% fetal bovine serum (Atlanta Biologicals, Flowery Branch, GA), 100 IU/ml penicillin, and 100 μg/ml streptomycin. All cells were maintained at 37 °C in a humidified atmosphere with 5% CO₂.

For KRT13 overexpression, a cDNA containing the full-length open reading frame of human KRT13 (NM_153490) was subcloned into pLVX-AcGFP1-N1 (pLV) (Clontech, CA) by introducing EcoR I and BamH I sites. The construct of pLVX-AcGFP1-N1-KRT13 (pLV-KRT13) was confirmed by DNA sequencing. The plasmid DNA was transfected to the 293 T cells to produce lentiviral particles, following the manufacturer’s instructions (System Biosciences, Mountain View, CA). MCF7 cells were transduced with the lentivirus and selected with puromycin (2 µg/ml for 2 weeks). For shRNA-mediated knockdown, non-targeting control or KRT13 shRNA lentiviral particles (Santa Cruz Biotechnology, Dallas, Texas) were used to infect the HCC1954 cells. The plasmid pcDNA3-c-Myc (Addgene, Cambridge, MA) was used to generate c-Myc-overexpressing cells. To silence c-Myc, cells were transfected with c-Myc siRNA (Santa Cruz Biotechnology) using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA).

To determine growth rates, breast cancer cells were seeded on 24-well plates for 5 days, and cell numbers from triplicate wells were counted daily with a TC20 automatic cell counter (Bio-Rad, Hercules, CA). Cancer cell migration was examined in triplicate transwells (8 μm pore size) coated with collagen I. For invasion assays, transwells were coated with collagen I, and each well was overlaid with growth factor reduced Matrigel (BD Biosciences, San Jose, CA) as previously described [19]. After incubation at 37 °C for 24–48 h, cells on the upper surface of the filters were removed by cotton swab, and cells that had invaded and migrated to the lower surface were fixed with 10% formaldehyde and stained with 0.5% crystal violet. After washing, stain was eluted with Sorensen’s solution. The absorbance of each well was measured at 590 nm.

Cytoplasmic and nuclear extracts or whole cell lysates were prepared with NE-PER nuclear and cytoplasmic extraction reagents (ThermoFisher Scientific). Western blot analysis was performed as previously described [19]. The primary antibodies used were to KRT13 (EPR3671, Abcam, Cambridge, UK or B-2, Santa Cruz Biotechnology), CD44 (156-3C11, Cell Signaling, Danvers, MA), c-Myc (D84C12, Cell Signaling), ALDH1A1 (B-5, Santa Cruz Biotechnology), Nanog (5A10, Santa Cruz Biotechnology), PG (A-6, Santa Cruz Biotechnology), DSP [EPR4383(2), Abcam], Actin (AC-15, Santa Cruz Biotechnology) and Lamin B (C-20, Santa Cruz Biotechnology). Immunoblots were subjected to morphometric analysis by Image Studio Software (LI-COR, Lincoln, NE).

Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany). For RNA-seq analyses, duplicate samples were submitted to University of California Los Angeles (UCLA) Clinical Microarray Core for RNA-seq analysis. The RNA-seq data were first subjected to differentially gene expression with the DESeq2 program. Gene set enrichment analysis was then performed. For correlation study, RNA-seq gene expression pattern was compared to breast cancer subtypes of The Cancer Genome Atlas (http://​cancergenome.​nih.​gov). To assess correlation between breast cancer subtypes and KRT13 overexpression cells, we compared expression centroids of the five defined breast cancer subtypes [20, 21] with the KRT13 overexpressing MCF7 cells. Unsupervised clustering with Pearson’s correlation was performed to display the similarity in the heatmap with dendrogram. To analyze specific gene expression, qRT-PCR was conducted with experimental settings as we previously reported. Sequences of the primer pairs are listed in Additional file 1: Table S1.

Formalin-fixed and paraffin-embedded tissue specimens including 41 primary, 11 bone and 10 brain metastatic breast cancer tissues were obtained from West China Hospital, Chengdu, China. Handling of tissue specimens conformed to the policies and practices of the institutions. The use of the specimens was approved by the Institutional Research Board of the CSMC (IRB: Pro00054328).

Our immunohistochemical (IHC) staining protocol is previously published [22]. Primary antibodies to KRT13 (ERP3671, Abcam), PG (A-6, Santa Cruz Biotechnology), and c-Myc (D84C12, Cell Signaling) were used. For this study, images of stained sections were scanned with a Keyence BZ-X710 microscope (Itasca, IL). The stains were scored with combined percentage of positive cells and intensity as reported [23].

For immunofluorescence staining, sections were stained with Alexa Fluor 594‐conjugated goat anti‐rabbit IgG (H + L) or Alexa Fluor 488‐conjugated goat anti‐mouse IgG (H + L) secondary antibodies (ThermFisher Scientific). Sections were further counterstained with 4′,6-diamidino-2-phenylindole (DAPI) in mounting medium (Vector Laboratories, Burlingame, CA) and examined by Zeiss Axio Observer Z1 fluorescence microscope.

Multiplexed quantum dot labeling (mQDL) analysis was performed as we previously reported [24]. Tissue sections were sequentially incubated with antibodies to KRT13, PG and c-Myc. Areas of interest were defined with manual segmentation by a pathologist (M. S. Lewis).

Flow cytometric analyses were conducted as described previously [23]. Cultured cells at 80% confluence were prepared in single cell suspension and fixed in 4% formaldehyde, washed in phosphate buffered saline (PBS), resuspended in PBS containing 1% BSA, and incubated with APC-conjugated anti-CD44 (IM7, Biolegend) and PE-conjugated anti-CD24 (ML5, Biolegend) antibodies at 4 °C for 30 min. Matched APC- or PE- conjugated isotypes were used as negative controls.

Cells were plated in ultra-low attachment 96-well plates (Corning Inc., Corning, NY) in serum-free DMEM/F12 medium (ThermoFisher Scientific) supplemented with 20 ng/ml epidermal growth factor (ThermoFisher Scientific), 10 ng/ml basic fibroblast growth factor (Pepro Tech, Rocky Hill, NJ), 5 µg/ml insulin (VWR, Radnor, PA), 1 × B-27 supplement (ThermoFisher Scientific) and 0.4% bovine serum albumin (Sigma, St. Louis, MO). For each experimental group, 500 cells in 100 µl medium were plated to each of 10 wells. After cultured for a week [25], spheres were counted and documented with phase contrast microscopy. Each study was repeated for three times.

Cell lysates (50 µg) were precleared by incubating with 1 µg rabbit/mouse control IgG in 20 µl Protein A/G PLUS-Agarose suspension (Santa Cruz Biotechnology) at 4 °C for 1 h. After centrifugation, the supernatant was incubated with primary antibodies at 4 °C overnight, and then with 20 µl of Protein A/G PLUS-Agarose at 4 °C for 2 h. Immunoprecipitates were collected and rinsed four times with cell lysis buffer. Half of each sample was submitted to the Mass Spectrometry and Biomarker Discovery core at CSMC for interactome analysis, and the remaining half was used to verify the interactome results by western blot.

All animal procedures were performed according to a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of the CSMC as previously described [18, 19]. MCF7-KRT13 or HCC1954-shKRT13 cells were tagged for luciferase reporter expression and inoculated orthotopically to the fat pads of mammary glands (3 × 10⁶ cells/100 µl PBS containing 50% Matrigel) or intracardially to the left ventricle (1 × 10⁶ cells/50 µl PBS) of 6- to 8-week-old female Fox Chase SCID Beige mice (Charles River, Wilmington, MA). Tumor volume and metastasis were monitored and assessed weekly by bioluminescence imaging (BLI) using an IVIS® Spectrum or Lumina XR Optical Imaging System. A Scanco viva CT 40 system was used to examine skeletal lesions. At the end of the studies, mice were euthanized, and tumor tissues and organs were harvested for histopathological analysis.

All assays were done in triplicates for each of the independent experiments. Differences between groups were analyzed using Student’s t test (two groups) or one-way ANOVA (three or more groups). For Kaplan–Meier survival analysis, statistical significance was determined by the log-rank test. Other data were analyzed using GraphPad software (GraphPad Prism version 5.01 for Windows). Results are expressed as the mean ± SD. The p value < 0.05 was considered statistically significant.

We previously reported that KRT13 induced prostate cancer progression and metastasis [18]. In this study, we investigated the potential role of KRT13 in breast cancer using cell lines that expressed low (MCF7) or high (HCC1954) endogenous KRT13. We established KRT13-overexpressing MCF7 cells (MCF7-KRT13) and knocked down KTR13 levels in HCC1954 cells (HCC1954-shKRT13) (Fig. 1a). MCF7-KRT13 cells displayed increased proliferation (Fig. 1b), migration (Fig. 1c) and invasion (Fig. 1d); while HCC1954-shKRT13 cells showed reduced aggressive behaviors. Moreover, MCF7-KRT13 cells inoculated orthotopically to mouse mammary glands showed accelerated tumor growth, forming hemorrhagic tumors quite different from the control MCF7-con tumors (Fig. 2a), with significantly increased tumor weight (Fig. 2b) and volume (Fig. 2c). When inoculated intracardially, MCF7-KRT13 cells colonized multiple sites (Fig. 2d) leading to tumor formation in lung, liver, adrenal gland, mesenteric lymph nodes, jaw, and other bones (Fig. 2e and Additional file 1: Table S2), which were confirmed by IHC analysis (Fig. 2f). Interestingly, when MCF7-KRT13 cells were cultured in vitro, high levels of KRT13 could be readily detected by western blotting (Fig. 1a); whereas in xenograft tumors, KRT13 overexpression was often seen unevenly by IHC analysis, probably suggesting temporal or spatial modification of the protein in the tumor microenvironment. Bone µCT scans revealed severe osteolytic lesions (Fig. 2g), and the mice had significant weight loss (Fig. 2h) and shortened survival (Fig. 2i).

Similar correlations between KRT13 levels and xenograft tumor formation were observed with HCC1954 cells. HCC1954-shcon xenograft tumors caused prevalent metastases to the liver, adrenal glands and lung as early as 8 weeks post-inoculation. Conversely, HCC1954-shKRT13 cells formed smaller tumors (Fig. 2j) with lower incidence (6/10) (Fig. 2k). KRT13 knockdown delayed HCC1954 tumor metastasis, which was only detected by week 19 (Fig. 2l). These results indicate strongly that KRT13 is associated with elevated aggressive behaviors in breast cancer cells.

We established single cell colonies using MCF7-KRT13 cells and detected KRT13 by immunostaining. Interestingly, the staining was stronger at the extending edge than at the center of the colony (Additional file 1: Fig. S1). The edge-positive phenomenon is in accord with our finding that KRT13 is associated with the tumor invasive front in breast cancer tissue specimens (see below).

To investigate the mechanism of KRT13-induced morphologic change and aggressive behavior, we conducted RNA-Seq gene expression profiling with the MCF7-KRT13 cells. Bioinformatic data analysis suggested that KRT13 overexpression had switched the cells from a typical epithelial type to more invasive basal cell-like and Her2-positive subtypes (Additional file 1: Fig. S2) [26]. Gene set enrichment analysis (GSEA) suggested that KRT13-overexpression induces EMT (Fig. 3a) and stemness (Fig. 3b) properties. This explains the marked morphologic changes from connected cobble-stone epithelial MCF7 cells to a less cohesive mesenchymal-like arrangement of the MCF7-KRT13 cells (Fig. 3c and Additional file 1: Fig. S1). MCF7-KRT13 cells were also found with decreased epithelial markers E-cadherin and claudin-7 but with increased mesenchymal markers vimentin and twist. Conversely, claudin-7 were up-regulated and twist was down-regulated in HCC1954-shKRT13 cells (Fig. 3d).

We explored the relationship of KRT13 to stem cell properties. A panel of stemness-related marker and regulator proteins were determined in MCF7-KRT13 and HCC1954-shKRT13 cells. Overexpression of KRT13 increased the expression of CD44, c-Myc, ALDH1A1 and Nanog, whereas knockdown of KRT13 was seen with decreased expression of these proteins (Fig. 3e). MCF7-KRT13 cells displayed a more robust self-renewal capability as reflected by accelerated mammosphere formation and increased mammosphere counts (Fig. 3f). In addition, MCF7-KRT13 cells expressed significantly elevated CD44⁺CD24⁻ stem cell features (Fig. 3g), while HCC1954-shKRT13 cells showed slightly decreased CD44⁺ levels (Fig. 3h). These results suggest that a high KRT13 level promotes stem cell properties.

We previously reported that KRT13 overexpression in prostate cancer cells was associated with stemness gene expression, including c-Myc [18]. Here, we manipulated c-Myc expression in KRT13-overexpressing and KRT13-knockdown breast cancer cells. Specific siRNAs effectively repressed c-Myc expression in MCF7-KRT13 cells (Fig. 4a). The c-Myc repression was accompanied by decreased expression of CD44 and Nanog (Fig. 4a), reduced number and size of mammosphere formation (Fig. 4b), cell growth (Fig. 4c), invasion and migration (Fig. 4d). In HCC1954-shKRT13 cells, overexpression of c-Myc induced the expression of CD44 and Nanog (Fig. 4a) and promoted mammosphere formation (Fig. 4b), cell growth (Fig. 4e), invasion and migration (Fig. 4f). Results from these analyses suggest that c-Myc is a key KRT13 downstream mediator promoting aggressive behavior and stem cell-like properties in breast cancer cells.

To explore how KRT13 elicits oncogenic signaling in cancer progression and metastasis, we carried out studies to identify intracellular proteins that were interactive to KRT13. In co-IP experiments, antibodies to KRT13 were used in MCF7-KRT13 cells to pull down protein complexes, which were characterized with liquid chromatography-tandem mass spectrometry (LC–MS/MS) and spectral counting methods. Bioinformatic analyses suggested 37 pulled-down proteins that were significantly higher in MCF7-KRT13 cells than MCF7-con cells (Additional file 1: Table S3). We found abundant DSP and PG co-precipitated with KRT13. In co-IP and western blotting analyses with the MCF7-KRT13 and HCC1954-shKRT13 cells, KRT13 interacted directly with both PG and DSP, because either anti-KRT13, anti-PG or anti-DSP antibody was able to pull down all three proteins (Fig. 5a).

Throughout these studies, MCF7-KRT13 cells were found with decreased PG and DSP levels, while HCC1954-shKRT13 cells expressed increased PG and DSP (Fig. 5a). When subcellular distribution was examined, MCF7-KRT13 cells showed significantly reduced cytoplasmic and nuclear PG. In reverse, HCC1954-shKRT13 cells showed significantly increased nuclear PG with a slight increment of PG in the cytoplasm (Fig. 5b). Similar results were detected by immunofluorescence assay (Fig. 5c). KRT13 overexpression seemed to suppress PG and DSP protein levels although the mechanism has yet to be elucidated. Among these cytoskeleton proteins, PG is known to be a potential signaling mediator, translocating to the cell nucleus to exert its tumor suppressor function [27]. Lowered PG levels upon KRT13 overexpression may contribute to the aggressive behavior observed in MCF7-KRT13 cells. These results suggest that KRT13 is capable of binding with the DSP/PG complex, sequestering and lowering the expression and nuclear translocation of PG to promote oncogenic signaling.

We determined KRT13 expression in 41 primary and 21 metastatic breast cancer tissue specimens by IHC (Fig. 6a). KRT13 expression in metastatic specimens (12/21) was significantly higher than in the primary breast cancer (13/41) (Fig. 6b). In addition, the Oncomine gene expression database (www.​oncomine.​org) showed that KRT13 expression in primary breast cancer was associated with decreased patient overall survival (Fig. 6c), similar to our report on prostate cancer [18]. KRT13-positive cancer cells were detected at the invasive front of breast cancer (Fig. 6d) as well as prostate cancer (Additional file 1: Fig. S3).

There was a positive correlation between KRT13 and c-Myc expression in this study and in the Oncomine database (Fig. 6e–g). Using three pairs of clinical specimens, we further confirmed the association between KRT13, PG, and c-Myc by mQDL (Fig. 6h), as both KRT13 and c-Myc levels were elevated, whereas both cytosolic and nuclear PG levels were lowered in metastatic tumors than in the primary tumor specimens. A high expression and positive nuclear PG was associated with low KRT13 and c-Myc expression, while low cytoplasmic PG expression was associated with moderate-high KRT13 and c-Myc expression (Fig. 6h and Additional file 1: Fig. S4). These data suggest that KRT13 can downregulate PG expression and reduce its nuclear translocation, abrogate PG-mediated c-Myc suppression, and promote cancer progression.