SEARCH

SEARCH BY CITATION

Keywords:

  • Breast cancer;
  • Cancer stem cell;
  • Epithelial-mesenchymal transition;
  • Keratin;
  • ROCK1

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The differentiation of stem-like tumor cells may contribute to the cellular heterogeneity of breast cancers. We report the propagation of highly enriched mouse mammary cancer stem cells that retain the potential to differentiate both in vivo and in culture and their use to identify chemical compounds that influence both self-renewal and differentiation. We identify epithelial tumor-initiating cells (ETICs) that express lineage markers of both basal and luminal mammary cell lineages and retain the potential, from even single cells, to generate heterogeneous tumors similar to the tumor of origin. ETICs can progress through a Rho-associated coiled-coil containing protein kinase 1 dependent, epithelial to mesenchymal transition to generate mesenchymal tumor-initiating cells capable of initiating tumors of limited heterogeneity. The propagation of ETICs may allow for the identification of new therapeutic compounds that may inhibit or prevent progression of some types of breast cancer. Stem Cells2013;31:12–22


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Cancer stem cells (CSCs) have been defined by their capacity to self-renew and to contribute to the heterogeneous cellular composition of tumors [1]. Cell surface markers such as CD44, CD61, CD29, CD24, CD90, CD49f, and epithelial cell adhesion molecule (EpCAM) have been used to enrich for breast CSCs that have been tested in vivo by transplantation [2–5]. If CSCs differentiate along a normal developmental lineage to a nontumorigenic, benign state, one of the unique opportunities of CSCs is differentiation therapy as an alternative to cytotoxicity [6]. The search for compounds and genes regulating the differentiation of breast CSCs is limited by the number of cell lines with CSC properties and growth conditions for screening in cell culture. Normal human and mouse mammary stem or progenitor cells have been propagated in culture as heterogeneous spheroids containing both differentiated cells and stem cells with self-renewal capacity. Serial propagation is commonly limited by spontaneous differentiation [7–9] that may be moderated by Wnt ligands [10]. Application of mammosphere culture methods has been used successfully to generate and propagate cells from breast cancer specimens [9, 11]. However, the generally low efficiency of serial propagation and the heterogeneity of the resulting cell population complicate cell culture screening strategies and the interpretation of gene expression profiles. Widely used, established, human cancer cell lines have been developed and adapted to growth in monolayer culture in serum containing medium [12]. However, the strong selection for growth in these cell lines may restrict the potential to differentiate.

We combined the methods of spheroid culture [7], the stimulatory effects of extracellular matrix [13–15], and the differentiation inhibitory conditions of low oxygen atmosphere [16, 17] to identify conditions that permit mouse mammary epithelial tumor-initiating cells (ETICs) to be amplified in culture in a highly enriched state while retaining the potential for differentiation both in culture and in vivo. We have used these cultured cells to identify chemical inhibitors of Rho-associated coiled-coil containing protein kinase 1 (ROCK1) that stimulates self-renewal and inhibits an epithelial-mesenchymal transition (EMT) that spontaneously generates a mesenchymal tumor-initiating cell (MTIC) with limited differentiation potential.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Cell Lines, Cell Culture, and Lentiviral Transduction

WMG300, WMG316, WMG49, and WMG307 tumor-initiating cells were isolated from mammary tumors of MMTV-Wnt1/MELK-GFP bitransgenic mice [18]. Lines maintained only by transplantation are designated by the appended number T and a number indicating the transplant generation. All cell lines were derived from single-cell-initiated spheroids in Matrigel. Cells were cultivated either embedded in a three-dimensional (3D) Matrix (Matrigel/MSCM mixed 1:1) or in monolayer on gelatin-coated plates (Nunc, Rochester, NY, #150288) in the presence of serum-free mammary stem cell medium (MSCM). MSCM consists of mammary epithelial basal medium (Lonza, Walkersville, MD), supplemented with penicillin and streptomycin, Fungizone (Omega Scientific, Tarzana, CA) 1×, heparin (Sigma, St. Louis, MO) 4 μg/ml, epidermal growth factor (Sigma) 20 ng/ml, basic fibroblast growth factor (BD Bioscience, Bedford, MA) 20 ng/ml, B27 Supplement (Invitrogen-GIBCO, Grand Island, NY) 1×, and Gentamycin (Sigma) 35 μg/ml. For routine culture, cells were mixed with the Matrigel/MSCM mixture in the cold and plated onto low attach 24-well plates (Costar, Corning, NY, #3473). The mix solidified in 30 minutes at 37°C. Each gel was overlaid with 500 μl of MSCM. Five hundred microliters MSCM were added to each well on a weekly basis. Cells were incubated in an atmosphere of 3% O2, 5% CO2, and 92% N2 atmosphere for 3 weeks before subcultivation. Single-cell suspensions were prepared by treatment with equal volume of Dispase (BD Biosciences, San Jose, CA) for 20 minutes at 37°C, the digest was stopped with addition of EDTA (pH 8.0) to 5 μM. Spheroids were recovered by centrifugation, washed twice in phosphate buffer saline (PBS), and then treated with Accutase (MP Biomedicals, Solon, OH) at 37°C for 20 minutes to get a single-cell suspension upon repeated pipetting. ETICs were maintained in MSCM supplemented with 1 μM of a ROCK inhibitor (EMD Biochemical, San Diego, CA, catalog number 555550, H-1152). For MTICs, cells were maintained in MSCM supplemented with 300 nM a glycogen synthase kinase 3β (GSK3β) inhibitor (EMD Biochemical, catalog number 402086, indirubin-3′-monoxime). Knockdown cell strains were generated by transduction with lentivirus encoding ROCK I or ROCKII shRNAs (MISSION, Sigma) upon selection in 2 μg/ml puromycin. In each case, multiple shRNA vectors were tested and at least two for each gene gave similar results.

Animal Studies

All procedures for treatment, housing, and euthanasia of mice have been approved by the Sanford-Burnham Medical Research Institute (SBMRI) Animal Care and Use Committee. Serial dilution transplantation of ETICs and MTICs was performed as previously described [19]. The indicated numbers of ETIC, MTIC, sh902, or shCntrl cells were suspended in 20 μl of Matrigel diluted 1:1 in MSCM and injected into cleared mammary fat pads of 19-day-old FVB/N females. Tumor development was monitored by palpitation three times a week. The frequency of tumor-initiating cells was estimated using the Extreme Limiting Dilution Analysis software provide by the Walter and Eliza Hall Institute of Medical Research (http://bioinf.wehi.edu.au/software/elda/index.html).

Immunofluorescence Staining and Microscopy

Depending on the combination of antibodies and reagents, either 100% MeOH or 4% formaldehyde was used for fixation (10–15 minutes at room temperature). Following fixation, cells grown on gelatin-coated plates or coverslips were washed twice with PBS and permeabilized with 0.05% Tween 20/Tris-HCl buffered saline (1× Tris-HCL buffered saline Tween-20) prior to blocking with 1 mg/ml ovalbumin (Sigma) in PBS. Primary antibodies were incubated in blocking buffer at recommended dilutions overnight at 4°C and consisted of Keratin 14 (PRB-155P, Covance, Dedham, MA), Keratin 6 (PRB-169P, Covance), Keratin 5 (PRB-160P, Covance), Keratin 8 (Troma-1, DTSHB, Iowa City, IO), Keratin 18 (Troma-II, kindly provided by Rolf Kemler of the Max-Planck Institute, Freiburg, Germany), Keratin 19 (Troma-III, DTSHB), phosphorylated myosin light chain 2 (MLC2; Cell Signaling, Danvers, MA), vimentin (GTX62264, GeneTex, Irvine, CA), and E-Cadherin (612130, BD Biosciences, San Jose, CA). After washing three times with PBS, secondary antibodies (anti-rabbit or anti-rat coupled to Alexa-488 or -566 from Invitrogen-Molecular Probes, Grand Island, NY) were incubated at 37°C for 30 minutes and subsequently washed and held in buffer containing 4,6′-diamidino-2-phenylindole dihydrochloride (DAPI; Sigma). Primary tumor sections were deparaffinized and incubated with antibodies using the same protocol. Fluorescence images were obtained with a spinning disc confocal microscope (AI Observer Zeiss Axio, Thornwood, NY) equipped with a CCD camera (Coolsnap HQ2, Photometrics, Tucson, AZ) and a Nikon EC-Plan NeoFluor 10x/0.5 N.A objective using MetaMorph 7.6.3.0 software (MDS Analytical Tech, Sunnyvale, CA). All images were taken at room temperature and further processed with Image J software.

Differentiation Assay/Annotated Kinase Inhibitor Screen

Spheroid colonies were dissociated into single cells and seeded onto gelatin-coated coverslips or 384-well flat bottom, μClear plates (Greiner Bio-One, Monroe, NC) for the kinase screen. Three hundred and eighty-four-well kinase assay plates contained 242 annotated kinase inhibitors (10 mM) that were delivered by the Labcyte Echo 550 (Labcyte Inc., Sunnyvale, CA) acoustic pipetter to reach a final concentration of 1 μM upon cell seeding, 50 μl final total volume. ETICs were seeded in 30 μl of media containing 1,500 cells per well and three replicate wells. Dimethyl sulfoxide (DMSO)-only control wells were present on each assay plate screen. After 3 days of growth, in a 3% O2, 5% CO2, and 92% N2 atmosphere, cells were fixed in 100% methanol and used for immunohistochemistry (IHC).

Fixed cells were stained with DAPI and cell numbers were determined by evaluating nine images acquired by the IC100 (Beckman Coulter Inc., Danvers, MA) automatic focusing imaging system with a Hamamatsu Orca-ER scientific CCD camera, a Nikon S Fluor ×10/0.5 N.A objective, and 2 × 2 camera pixel binning. The algorithm to identify individual cells based on keratin content was designed/implemented with CyteSeer Software (Vala Sciences Inc., San Diego, CA). A Z-score was established for each individual kinase inhibitor using the formula Z = (|X − μ|)/SD where μ is the mean of the negative controls and SD is the standard deviation of the whole population. X is the sample value calculated based on the average DAPI count per well.

Colony-Forming Assays

Assessment of 3D colony-forming ability was carried out as described previously [7]. One thousand cells per well were seeded in 24-well ultra low attachment plates (Costar) in MSCM. For studies incorporating Matrigel, 200 cells per well were suspended in 150 μl Matrigel diluted 1:1 in MSCM and seeded on 24-well ultra low attachment plates. Monolayer colony-forming assays were carried out on 12-well flat bottom tissue culture plates (Costar) coated with gelatin. Four hundred cells per well were seeded in MSCM, grown for 5–6 days, fixed in 100% methanol, and visualized by staining with 0.4% Crystal violet (Sigma). Influence of ROCK and GSK3β inhibitors on colony forming units (CFU) was quantified by generation of dose-response curves for each cell matrix under the same cell density and culture conditions.

Fluorescence-Activated Cell Sorting and Flow Cytometry

Cells in Matrigel were sequentially treated with Dispase (to break down the Matrigel) and Accutase (to achieve a single-cell suspension). Single cells were resuspended in 1 ml MSCM and viability was assessed by staining with trypan blue. Up to 4 × 106 cells were resuspended in fluorescence-activated cell sorting buffer (5% fetal calf serum in PBS) and incubated with primary coupled antibodies; CD29 (102213, Biolegend, San Diego, CA), CD24 (101803, Biolegend), CD49f (551129, BD Biosciences; 313617, Biolegend), CD61 (553345, BD Pharmingen), EpCAM (118211, Biolegend), and Lineage cocktail (Ter-119, CD31, CD45, EBioscience, San Diego, CA). Cells were sorted using a BD Bioscience FACSort. Sorted fractions of cells were plated in Matrigel as described above or inoculated into cleared mammary fat pads for tumor-initiation assay.

Immunoblots

Protein was extracted with 1× RIPA buffer (Cell Signaling) supplemented with 1 mM phenylsulfonyl fluoride (Pierce, Rockford, IL). Protein content was quantified using the Bradford protein assay reagent kit (Bio-Rad, Hercules, CA). Protein lysates were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis sample buffer on 4%–20% Tris-Glycine Gels (Invitrogen). Gels were transferred to polyvinylidene fluoride membranes (Millipore, Billerica, MA), blocked with 5% non-fat milk/1× TBST and probed with the following primary antibodies; ROCK 1 (sc-5560; Santa Cruz, Santa Cruz, CA), ROCK 2 (sc-5561; Santa Cruz), and actin (A1978; Sigma). The primary antibodies were revealed with horse radish peroxidase-linked IgG secondary antibodies (Millipore) and visualized with enhanced chemiluminescent reagent (Denville Scientific, Metuchen, NJ).

RNA Preparation and Microarray Analysis

RNA was isolated with the Nucleospin RNA II kit (Machery-Nagel, Bethlehem, PA). Labeled cRNA is generated from the whole RNA sample using an RNA amplification kit where first-strand cDNA is made after priming with a T7 promoter-oligo (dT) primer. Following second strand cDNA synthesis, T7-mediated transcription in the presence of biotin-16-UTP is used to generate biotinylated cRNA. This amplified, biotinylated cRNA is labeled by incubation with streptavidin-Cy3. We analyzed RNA with Illumina (San Diego, CA) MouseRef-8 BeadChips using the manufacturer's BeadArray Reader and collected primary data using the supplied Scanner software. All the arrays were normalized using the normalize quantiles routine from the Affymetrics package in Bioconductor. This procedure accounted for any variation in hybridization intensity between the individual arrays. Finally, these normalized data were imported into GeneSpring. The groups of biological replicates were described to the software, and significantly differentially expressed genes were determined based on t tests and fold difference changes in expression level. Gene expression data were analyzed with Bioconductor and GeneGo software. The array data have been deposited at the NCBI GEO under GSE33433.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Enrichment for Tumorsphere-Initiating Cells from MMTV-Wnt1 Tumors

Solid heterogeneous tumors that arise sporadically from hyperplastic mammary glands in the MMTV-Wnt1 transgenic mouse line are associated with secondary activating mutations in genes that activate mitogen-activated protein kinases and cooperate with the activated Wnt pathway [20, 21]. MMTV-Wnt1 tumors are composed of cells of both luminal and basal epithelial cell lineages that express keratin 8 (K8) and keratin 14 (K14), respectively (Fig. 1A, 1C, 1E, 1G). Lineage tracing experiments have confirmed the fidelity of K8 and K14 as lineage markers of luminal and myoepithelial lineages after birth [22]. Vimentin-positive cells were found in patterns generally distinct from the epithelial organization (Fig. 1B, 1D, 1F, 1H). Serial transplantation of all lines, except WMG300, maintained a differentiated tumor phenotype containing basal and luminal epithelia. Histological analysis of multiple independent and serially derived tumors revealed biased contributions of K8- and K14-positive cells characteristic of each line (Supporting Information Fig. S1A and S1B) independent of the sampling throughout an entire tumor (Supporting Information Fig. S1C). A variable, minority proportion of tumor cells express both K8 and K14 (Fig. 1I). Similar cells that express both K8 and K14 or K5 and K18 have been found during development of the mammary gland and have been suggested to be progenitors to luminal and basal epithelial lineages [22–24].

thumbnail image

Figure 1. Keratin expression in primary and secondary MMTV-Wnt1 tumors. (A--H): Immunoflourescence images of sections of two independent MMTV-Wnt1 primary tumors labeled WMG316 (A, B) and WMG300 (E, F) and their secondary tumors derived from cultured epithelial tumor-initiating cell lines WMG316 (C, D) and WMG300 (G, H) are shown. Ethanol-acetic acid-fixed tumor sections were stained for K8, K14 colocalization (A, C, E, G), and vimentin (B, D, F, H) as indicated. The nuclei in all images are visualized by DAPI staining. (I): Cells from dissociated tumors were prepared by cytospin and stained with both K8 and K14 antibodies. Reactivity was assessed on captured images taken at the same exposure time and without digital manipulation. The bars represent the percent of total cells stained with the indicated antibodies from one tumor. Error bars indicate the standard deviation of scoring multiple microscopic fields. (J): Cytospin preparations of the two cultured cell lines were stained and scored as in M. Note the small fraction of K8+, K14+ cells in tumors and the increased population in cell cultures. Scale bars represent 60 μm. Abbreviation: DAPI, 4,6′-diamidino-2-phenylindole dihydrochloride.

Download figure to PowerPoint

Mammary tumor-initiating cells have been successfully cultivated in suspension culture as cellular spheroids in serum-free medium [7]. To overcome the low colony-forming efficiency of this method (commonly approximately 1% or less), we grew MMTV-wnt1 tumor cells in reconstituted extracellular matrix (Matrigel). The recovery of colony-forming activity from dissociated spheroids decreased with serial cultivation, presumably due to continued differentiation [18]. Because a low oxygen atmosphere inhibits the spontaneous differentiation of multiple other stem cell types [16, 25], we introduced a low oxygen atmosphere (3% O2, 5% CO2, and 92% N2) into our cultures resulting in an increase in the cloning efficiency of MMTV-Wnt1 tumor cells compared to normal air (5% CO2, air) (Fig. 2A). The use of both Matrigel and low O2 atmosphere permitted the serial propagation of cell lines from multiple, but not all (six out of eight) independent MMTV-Wnt1 tumors. Evaluation of K8 and K14 expression in cells of spheroids revealed that most cells expressed both keratins (Figs. 1J, 2I). The four different lines of ETICs (WMG49, WMG300, WMG307, and WMG316) all shared the characteristics of coexpression of K8, K14, and K18 (Supporting Information Fig. S1D). However, they differ with respect to expression of K5, K19, and vimentin (Supporting Information Fig. S1D, S1E).

thumbnail image

Figure 2. Cultivation and characterization of epithelial tumor-initiating cells (ETICs). (A): Spheroid-forming activity of WMG300 in Matrigel in two different O2 concentrations. (B, C): Original tumors (B) and cultured ETICs (C) were dissociated to single cells and analyzed by fluorescence-activated cell sorting (FACS) for the expression of surface markers CD29 and CD24. Values for percent double-positive cells are indicated in the upper right corner. (D, E): WMG316 cultured cells (E) and secondary tumors (D) were dissociated and analyzed by FACS for surface markers EpCAM and CD49f. (F--H): The four fractions indicated in panel (D) were recovered and analyzed for expression of K8 and K14 on cytospin preparations (F) and quantified for individual cell types (G). Individual fractions were also examined for colony-forming ability in Matrigel (H). (I): Confocal section through a typical spheroid from the WMG300 cell line under hypoxic conditions. Spheroid was stained for K8, K14, and DAPI. Magnification is ×100. Abbreviations: CFU, colony forming units; DAPI, 4,6′-diamidino-2-phenylindole dihydrochloride.

Download figure to PowerPoint

The tumorigenicity of the cell lines was assessed by transplantation of cells into cleared fat pads of syngeneic immune competent host animals (Table 1). The estimated frequency of tumor-initiating cells for two cell lines (WMG300, WMG316) was 100-fold greater than dissociated whole tumors. The efficiency of tumor formation from single transplanted cells (44%) was comparable to the colony formation efficiency of single cells in Matrigel culture that averaged 36.5% ± 11% over passages 4–7 (Fig. 2A). The proportion of cells of primary or secondary tumors that simultaneously expressed both K8 and K14 varied from 4% to 20%. In contrast, the cells of the cultured lines predominantly express both K8 and K14 (Fig. 1J). A clonal subline (WMG300A, Table 1) formed tumors with the same efficiency and phenotype as the parental line. Furthermore, tumors derived from single transplanted cultured cells had similar epithelial heterogeneity as the parental tumor (Table 1; Fig. 1C, 1G). The heterogeneous tumors derived from single cultured cells demonstrate both their potent tumorigenicity and differentiation potential.

Table 1. Tumorigenicity of MMTV-Wnt1 tumor cells
Thumbnail image of

The analysis of the cell surface markers previously used to enrich for tumor-initiating cells revealed that WMG300 and WMG316 cells express more CD29 and CD24 than most cells of the tumors (Fig. 2B, 2C). Consistent with previous findings [26, 27], the candidate basal and luminal populations of ETIC cultures and transplanted tumors were positive for CD49f and EpCAM antigens, respectively (Fig. 2D, 2E). This provided another basis for enrichment of ETICs from the differentiated luminal and basal cells of ETIC-initiated tumors. EpCAM+CD49f cells were primarily K8+ luminal cells (Fig. 2D--2G, Quadrant IV), while EpCamCD49f+ were predominately K14+ basal cells (Fig. 2D--2G, Quadrant I). Cells expressing neither EpCam nor CD49f were predominately negative for both keratin markers (host hematopoietic and endothelial cells were depleted by reaction with a combination of Ter-119, CD45, CD31, antibodies). K8+K14+ double-positive cells were reactive with both CD49f and EpCam antibodies (Fig. 2D--2G, Quadrant II). However, a smaller number of K8+ K14+ cells (5%–7%) were also found in the EpCam+CD49f and EpCamCD49f+ populations (Fig. 2F, 2G, Quadrant I and IV). Limiting dilution transplantation confirmed that the CD49f+EpCam+ cells were tumorigenic while the CD49fEpCam+ luminal cell population had greatly diminished tumor-initiating activity (Table 1). Similarly, increased spheroid-forming activity in culture was correlated with EpCam+CD49f+ (Fig. 2H). In summary, WMG300 and WMG316 cells serially cultivated as spheroids in low oxygen contain a high proportion of ETICs that retain the ability to efficiently initiate tumors and generate both basal and luminal epithelial lineages within tumors. Cultivated ETICs are relatively uniform as judged by cell surface markers EpCam, CD49f, CD29, CD24 and by coexpression of K8 and K14 (Fig. 2B, 2C, 2E, 2I).

Differentiation in Culture

The successful propagation of ETICs provided the opportunity to develop a cell-based assay for screening compounds and genes that might influence self-renewal or differentiation behavior. To determine the differentiation potential of single ETICs in culture, we identified the immediate derivates of individual cells in monolayer by antibody staining. Dissociated single cells were found either as single cells or colonies of only a few cells 1 day after plating in monolayer culture (Fig. 3A). The total numbers of colonies (composed of one or more cells) were comparable in 3% O2 and atmospheric O2 atmospheres (Fig. 3D). Thus, differential effects of the low oxygen atmosphere are not due to selective survival of plated cells. Antibody staining of colonies for K8 and K14 detected single candidate differentiated cells within clones. Cells that expressed only K8 or K14 but not both were scored as differentiated. The fraction of differentiated colonies increased with each day in atmospheric O2, reaching 80% of the scored colonies, whereas a low oxygen environment restricted the number of differentiated colonies (Fig. 3B, 3C, 3E, 3F). This cell lineage analysis shows that the progeny of ETICs are capable of differentiating to both candidate luminal (K8) and basal (K14) cell lineages in culture as observed in vivo.

thumbnail image

Figure 3. Epithelial tumor-initiating cell (ETIC) differentiation in culture. (A-C): ETIC single cells (A) day 1 were grown for up to 3 days in 3% O2 (B) or air (C) atmospheres supplemented with 5% CO2. Cells were fixed and stained with antibodies to K8 (red) and K14 (green). Size bar in (A--C) indicates 50 μm. The number of colonies (one or more cells) and reactivity of individual cells of colonies were determined from eight digital images of each condition taken with same exposure time. (D): Total colony number decreases with time but is similar in the two conditions. (E): The number of colonies with differentiated cells (K8 or K14 single-positive cells) increased in normoxia. (F): The percentage of colonies with differentiated cells increased in normoxic atmosphere. Abbreviation: DAPI, 4,6′-diamidino-2-phenylindole dihydrochloride.

Download figure to PowerPoint

High-Content Screen of Kinase Inhibitors

The identification of four types of cells (K8+K14+, K8+K14, K8K14+, and K8K14) by the combination of DNA and keratin antibody staining was adapted to a multiwell format and analyzed using an automated high-content imaging system with accompanying CyteSeer software. Comparison of cell number identified by K8 and K14 antibody staining to cells measured by DAPI staining confirmed the routine successful identification and enumeration of greater than 95% of the keratin staining cells (Supporting Information Fig. S2A). A library of 242 chemical kinase inhibitors was screened under hypoxic conditions in triplicate with final concentrations of 1 μM (Z' score of 0.635). The results of this screen identified candidate inhibitors that modulated the growth of ETICs while also influencing the fraction of K8 or K14 single-positive cells (Supporting Information Fig. S2B, S2C). Twenty-nine inhibitors increased cell number by greater than 50% as judged by either DAPI-stained nuclei or total K8, K14 double-positive cells (Supporting Information Table S1). Kinase inhibitors that increased the proportion of candidate differentiated cells had annotated activity on multiple kinase targets and will require additional validation and genetic deconvolution. However, four of the six most active growth promoting compounds, as evaluated by DAPI staining, were directed against ROCK1 (Fig. 4A, R1-R4). All four ROCK inhibitors and one inhibitor of GSK3β (Fig. 4A, G1) were confirmed in follow-up dose-response experiments in monolayer culture (Fig. 4B, 4C). The most potent of the ROCK inhibitors (R1) and the putative GSK3β inhibitor (G1) stimulated colony formation at limiting dilution in standard tumorsphere suspension culture (Fig. 4D), monolayer culture (Fig. 4E), and Matrigel (Fig. 4F). Moreover, combination of the two inhibitors was additive suggesting distinct pathways of growth stimulation for the two compounds (Supporting Information Fig. S3A).

thumbnail image

Figure 4. Chemical kinase inhibitor screen reveals compounds that stimulate proliferation. (A): Influence of 242 kinase inhibitors on cell number after 3 days of exposure to 1 μM in a 3% O2 atmospheric condition. The y-axis represents the average Z-scores for the change in cell growth upon treatment with an individual kinase inhibitor. The Z score = |average value − mean of controls|/SD of all tests. Controls are represented by black squares. Six candidate inhibitors had a Z-score >2, four of which targeted rho-associated coiled-coil containing protein kinase 1 (R1, R2, R3, and R4), one inhibitor of glycogen synthase kinase 3β (G1), and one inhibitor of checkpoint kinase one (C1). (B, C): Dose-response of fold change in cell number as function of the R1 and G1 inhibitors (mean ± SD). (D--F): Stimulation of WMG300 colony-forming activity by R1 and G1 in suspension (D), monolayer (E), and Matrigel (F) (mean ± SD). Note differences in vertical scale of the three panels. Abbreviations: CFU, colony forming units; DMSO, dimethylsulfoxide.

Download figure to PowerPoint

Inhibition of ROCK1 Promotes Growth of ETICs

The two ROCK kinases, ROCK1 and ROCK2, were targeted with multiple lentiviral expressed shRNAs in our WMG300 ETIC culture. Transduction with ROCK1 shRNA (sh902 and sh903) increased colony formation while ROCK2 shRNA did not (sh919 and sh922) (Fig. 5A). A subsequent ETIC polyclonal population was selected by resistance to puromycin for sh902 and sh919. Western blot analysis confirmed that these cells had significantly decreased full length ROCK1 and ROCK2, respectively (Fig. 5B). Suppression of ROCK1 resulted in an increased ability to form colonies in the absence of the R1 inhibitors (Fig. 5C), whereas the ROCK2 knockdown line remained dependent on the R1 inhibitor (Supporting Information Fig. S3B). The increase in colony-forming ability could not be strictly attributed to reduced anoikis as the number of cells containing cleaved caspase-3 did not significantly change upon suppression of ROCK1 (Supporting Information Fig. S3C). However, both the R1 inhibitor and the suppression of ROCK1 by shRNA resulted in increased proliferation of K8+K14+ cells in monolayer culture without altering the proportion of K8 or K14 single-positive cells (Supporting Information Fig. S3D). We conclude that the stimulatory effect of R1 is on preferential proliferation of ETICs and not on decreased cell death.

thumbnail image

Figure 5. ROCK1-dependent stimulation of colony-forming efficiency and stabilization of ETIC. (A): Stimulation of WMG300 monolayer colony formation by ROCK1 shRNA (902 and 903) but not ROCK2 shRNA (919 and 922) upon viral transduction. Each column represents the average of three replicates. Cells were plated in the absence of the R1 inhibitor. (B): Immunoblots of whole cell lysates from WMG316 (316), WMG300 (300), WMG300sh902 (sh902), and WMG300sh919 (sh919; separate blot) with ROCK1 and ROCK2 antibodies. Actin serves as the loading control. (C): Monolayer colony formation dependence on the ROCK inhibitor. ROCK1 knockdown cells are insensitive to additional suppression with the ROCK1 inhibitor. (D, E): WMG300 ETIC (spheroid passage 4) were serially cultivated in monolayer in the presence or absence of R1 or G1. Portions of the cells were evaluated by keratin (D) or E-cadherin staining (E) at the indicated monolayer passage. (F): Representative images of WMG300 monolayer colonies showing the immunofluorescence profile of most ETIC colonies in the presence of the ROCK inhibitor (F, 300R1 panel): K8, K14, ECAD, and VIM. MTICs isolated from G1 treatment in (D, E) have little K8, K14, and ECAD antibody reaction but stain strongly for vimentin (F, 300G1 panel). Scale bars represent 60 μm. (G): Colony morphologies of ETIC and MTIC grown in Matrigel. Abbreviations: CFU, colony forming units; DAPI, 4,6′-diamidino-2-phenylindole dihydrochloride; ROCK1, rho-associated coiled-coil containing protein kinase 1.

Download figure to PowerPoint

To determine whether the stimulatory effects of the R1 and G1 inhibitors might be sufficient to propagate ETICs in the absence of extracellular matrix, WMG300 cells were serially subcultivated in monolayer culture at moderate density in the presence or absence of the two inhibitors. Sufficient numbers of cells were recovered under all conditions to permit serial propagation. However, analysis of epithelial cell markers revealed that while cells cultivated in the presence of R1 maintained expression of K8, K14, and E-cadherin, cells propagated in G1 or vehicle soon failed to express keratin and E-cadherin (Fig. 5D, 5E). Single-cell-initiated colonies of WMG300 cells in monolayer have a compact epithelial morphology (Fig. 5F), while a minority of colonies (15.9%) appeared heterogeneous with central epithelial cells surrounded by migratory cells (Fig. 5F; 300R1 bottom panel). The nonepithelial cells at the periphery of epithelial colonies or those selectively cultivated in monolayer without R1 both expressed elevated vimentin (Fig. 5F; 300 G1 bottom panel). Vimentin-positive, keratin-negative, spindle-like cells are designated MTIC based on additional data described below. These cells grew as disperse, vimentin-positive motile cells that initially retained some K8 reactivity (Fig. 5F; 300 G1 panel, Supporting Information Fig. S4C, S4D). The dispersed single cells formed irregular, invasive-like colonies in Matrigel unlike the relatively uniform spheroids formed by ETICs (Fig. 5G). Treatment of isolated MTICs with the R1 inhibitor did not stimulate proliferative capacity or cause a reversal of EMT.

ROCK1-Dependent EMT

Gene expression profiling revealed that RNAs previously associated with EMT were increased in MTICs (Supporting Information Tables S2, S4). RNAs for epithelial markers E-cadherin, K5, K14, K8, K18, and epidermal growth factor receptor were all diminished in MTICs while EMT markers AP2, Snail 1, Twist2, Zeb1, and Foxc2 were all increased (Fig. 6A). MTICs also had decreased expression of CD29 and CD24 (Supporting Information Fig. S4D). Limiting dilution transplantation studies revealed MTICs were still very tumorigenic (Table 1) but generated tumors deficient in epithelial cells (Supporting Information Fig. S4). Interestingly, serial transplantation of the WMG300 line resulted in accelerated tumor growth after three to four passages with the concomitant loss of epithelial contribution and predominance of vimentin-positive spindle-like cells (Supporting Information Fig. S4A, S4B). Even WMG300 ETICs cultivated in the presence of the ROCK inhibitor ultimately generated tumors dominated by vimentin-positive cells when transplanted (Fig. 6B). This progression through an EMT could be attributed to the lack of ROCK1 inhibition in vivo because WMG300 ETICs with ROCK1 suppressed by shRNA (WMG300sh902) generated tumors that maintained both luminal and epithelial components like the originating MMTV-Wnt 1 tumor (Fig. 6B). To test whether ROCK1 kinase activity was also associated with EMT, the phosphorylation status of MLC2, a well-known target of ROCK1 activity, was determined by use of an antibody specific for the phosphorylated form of MLC2 (pMLC2). Treatment with the R1 inhibitor reduced the number of cells expressing pMLC2 as well as the number of vimentin-expressing cells (Supporting Information Fig. S5A). Strikingly, all cells found to contain pMLC2 also expressed vimentin (Supporting Information Fig. S5B, S5C). These results support a role of ROCK1 activity in the transition of ETIC to MTIC.

thumbnail image

Figure 6. EMT-associated RNA signature, tumor formation, and the evolution of tumor-initiating cells. (A): Comparison of ETIC and MTIC RNA levels of selected genes implicated in an epithelial to mesenchymal transition. (B): Histological analysis of tumors from stably transduced ETIC lines, 300shCntrl (cultivated in the presence of the R1 inhibitor) and 300sh902. “p” indicates the number of passages since isolation of the line, K8, K14, and vimentin staining are shown. (C): Schematic representation of a proposed cellular hierarchy responsible for the generation of HER2, luminal A,B, and basal-like tumors from a K8+,K14+ progenitor. Scale bars represent 60 μm. Abbreviations: DAPI, 4,6′-diamidino-2-phenylindole dihydrochloride; EMT, epithelial-mesenchymal transition; ETIC, epithelial tumor-initiating cell; MTIC, mesenchymal tumor-initiating cell.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We have serially propagated MMTV-Wnt1 tumor cells highly enriched in tumor-initiating capability and with persistent differentiation potential. These ETICs can be considered CSCs because of their potent tumorigenicity and ability to differentiate to luminal and basal mammary cell lineages both in vivo and in culture. The ability of these cells to differentiate distinguishes them from most cancer cell lines, capable of forming tumors of limited heterogeneity. The self-renewal capability of an ETIC is reflected by the efficient colony-forming activity of single-cell-initiated spheroids for successive passages. Similarly, a single ETIC is capable of initiating a heterogeneous tumor that can be serially propagated. The success of transplanting single ETICs was similar (44%) to that of colony formation in culture under optimal conditions (25%–40%). Thus, considering the stress of enzymatic dissociation and handling, it is likely that at least half of WMG300 and WMG316 cells are likely functional ETICs. It follows that these ETICs express both K8 and K14 and surface markers (CD24, CD29, and CD49f) previously associated with the tumor-initiating population of MMTV-Wnt1 tumors.

The coexpression of K8 and K14 in ETICs is unusual because these proteins are normally characteristic of luminal or basal epithelial lineages, respectively [22, 23]. Normal mammary epithelial cells that express both luminal and basal epithelial keratins simultaneously are rare but are found in presumptive progenitor cells during normal mammary gland development [28, 29] (Fig. 6C). Both WMG300 and WMG316 ETIC lines coexpress K8, K14, and K18 (Supporting Information Fig. S1), consistent with the suggestion that they are mammary multipotent progenitor cells [27]. However, they differ with respect to K5 and K6 expression even though they are clonally derived. While variable K6 expression may reflect sensitivity to microenvironmental cues as observed during wound healing of epidermis [30], K5 is more consistently expressed in mammary basal epithelia. While K5 and K14 are normally expressed coordinately in mature epithelia, K5 may be expressed at an earlier stage than K14 during mammary gland development [7, 24, 31].

Although largely dependent upon the specific antibodies used, basal epithelial keratin-positive cells are detected in as many as 30% of human breast cancer [32, 33]. Mammary tumors containing basal epithelial components, commonly detected as K5-, K14-, and or K17-positive cells also contain luminal derivatives that express K8 and K18. The cell of origin of mutant BRCA1 basal tumors in humans has been suggested to be a luminal progenitor cell [34]. In mice, conditional inactivation of Brca1 within either K14 basal stem cells or candidate luminal progenitor cells combined with p53 loss causes heterogeneous tumors with both basal and luminal components. Despite different cells of origin and tumor phenotypes, both resultant tumors have gene expression profiles similar to human basal-like tumors [35]. The presence of cells of apparent basal and luminal lineages in MMTV-Wnt1 tumors may be due to either tumor progenitor cells still capable of differentiating to both mammary lineages [3] or due to an aberrant or altered differentiation program [34]. We consider progenitor cell differentiation the likely origin of basal and luminal-like cells within MMTV-Wnt1 tumors because of the well-differentiated epithelial organization of the primary tumors, the concordance of basal and luminal keratins with CD49f and EpCAM cell surface marker expression in differentiated cells, and the decreased tumorigenicity of luminal-like derivates from the tumors. If ETICs maintain the characteristics of the cell of origin of MMTV-Wnt1 tumors then that cell appears to be a bipotential progenitor cell that expresses characteristics of both lineages [3] (Fig. 6C). A bipotential progenitor may account for the detection of basal keratin-positive cells within luminal epithelial positions [36].

ETICs are the cell of origin of MTIC. The generation of a mesenchymal-like cell population, also capable of initiating tumors of limited heterogeneity, demonstrates that at least in this system, tumor initiation is not confined to one cell type (Fig. 6C). In addition to the changes in epithelial morphology and protein expression (keratin, E-cadherin, and vimentin), changes in RNA expression confirm the epithelial to mesenchymal transition. The progression from ETIC to an MTIC occurs spontaneously both in serial transplanted WMG300 tumors and in serially cultivated cells in some but not all ETIC lines. The primary tumor of the WMG300 line had significant contributions of cells that did not stain with keratin antibodies (Fig. 1). After a few transplantation generations, the tumors were dominated by vimentin-positive, spindle-like cells that also lost expression of CD29 (Supporting Information Fig. S4). The inhibition of vimentin-positive cells arising in WMG300 ETIC cultures by R1 treatment and the localization of pMLC2 only within vimentin-positive cells give further indication of an active role for ROCK1 in an EMT (Supporting Information Fig. S5). The distinguishing characteristic of MTIC tumors is the uniformity and absence of epithelial components. Thus, MTICs appear to have a low frequency of reversal of the EMT in vivo. The variation in frequency of EMT from different ETIC tumor isolates may be due in part to the variation in type of secondary genetic event necessary for MMTV-Wnt1 hyperplastic epithelia to form solid tumors [21]. While MTICs lose their epithelial phenotype, they retain anchorage-independent growth and grow more rapidly than ETICs, providing a competitive advantage both in culture and in vivo. EMT of mammary epithelial cells has been reported to result in tumor-initiating cells with characteristics of stem cells [37–39]. We found ETICs are the cells of origin of MTICs and the differentiated epithelial components of the tumor. If ETIC EMT occurs in human tumors, the model of CSC-driven tumors may need to be revised and the identification of cellular targets reconsidered. A dispersed population of MTICs might be difficult to distinguish from host stromal cells. A differential sensitivity of ETICs and MTICs to therapeutic agents would also complicate treatment strategies. However, the key opportunity of differentiation therapy for ETICs would also prevent the emergence of MTICs but would more likely succeed at early stages of tumor progression.

The successful expansion of ETICs in culture provided the opportunity of developing a cell-based, high-content assay capable of distinguishing four cell types by nuclear staining and K8 and K14 content. We identified a ROCK inhibitor, R1 (H-1152), which has greater potency than the widely used inhibitor Y-27632 [40]. Y-27632 has been previously recognized to facilitate human embryonic stem cell and neural stem cell culture by inhibiting the apoptosis of dissociated cells [41]. However, in this study, both the ROCK inhibitor and the suppression of ROCK1 by shRNA had only moderate effects on apoptosis of ETICs in suspension culture but dramatically stimulated growth in monolayer. A function for ROCK in neural crest EMT has been previously suggested [42]. Here, ROCK1 inhibition both promotes self-renewal of ETICs (Fig. 4) and restricts EMT (Figs. 5, 6A, 6B and Supporting Information Fig. S5). Both of these effects would contribute to the selective advantage of MTIC both in vivo and in culture when ROCK1 is not inhibited. These effects appear to be independent because treatment with the ROCK inhibitor stimulated growth with all ETIC cell lines, irrespective of their ability to undergo an EMT.

Elevated ROCK1 expression has been correlated with poorer survival of breast cancer patients and inhibition of ROCK is associated with decreased cell motility and invasion [43]. The roles of ROCK1 and matrix metalloproteases in invasive movement within collagen matrices may be indirectly regulated by Notch1 [44]. EMT can be initiated by inhibiting E-cadherin expression [39] and is dependent on the Twist [45] and Snail transcription factors [46]. Evidence of EMT in myc but not Neu or PyVmT-driven mouse transgenic mammary tumors was obtained by the analysis of the progeny of fibroblast and epithelial-specific Cre recombinase [47] and by subjecting Neu-driven tumors to recognition by the mouse host immune system [48]. While MTICs generated in our model system are potently tumorigenic, we did not find metastases, even in animals from which primary tumors were excised. It remains to be determined whether additional genetic alterations can cause further progression to a metastatic state in this model.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The serial cultivation of mammary CSCs in a highly enriched state and their spontaneous differentiation provides future opportunities of applying both chemical and genetic interrogation of self-renewal, differentiation, progression, and differential resistance. MMTV-Wnt1 tumors have similarities to basal-like human breast cancers. In the future, it will be important to determine whether ETICs from human basal-like breast cancers share characteristics with the mouse ETICs identified here.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We thank Roy Williams, Yoav Altman, Kang Liu, Ed Monosov, Robbin Newlin, and Animal Resources personnel of the SBMRI Shared Resources for expert technical assistance. This work was supported by fellowship BC093655 to J.M. from the DOD Breast Cancer Research Program and by a pilot grant from NCI Cancer Center Support Grant 5 P30 CA030199. D.J.C. was supported by NCI Postdoctoral Training Grant T32 CA121949.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
sc-12-0511_sm_SupplFigure1.pdf656KSupplemental Fig. 1. Expression of keratin and vimentin in ETIC tumors and cell lines. (A) Representative immunofluorescence images of four different MMTV-Wnt1 tumors analyzed for K14 (green) and K8 (red). Scale bars represent 50um (B) Quantitative analysis of K8 and K14 positive areas in individual MMTV-Wnt1 tumors. Data represent the average percent of the tumor area positive for the indicated antibody. Each bar represents the mean and SD of 9 separate images. (C) Variation of keratin staining within a single tumor. Sections of a WMG316 tumor at 100 μm intervals were analyzed for K14 (green bar) and K8 (red bar) expression as quantified by the positively stained areas. Increased representation of K8 positive cells relative to K14 is found through out the tumor. (D) WMG300 (left column) and WMG316 (right column) cells grown in monolayer, stained with the indicated antibodies against intermediate filament proteins. Antibodies are indicated as follows from top to bottom: keratin 8 (K8, red), keratin14 (K14, green), keratin 6 (K6, green), keratin 19 (K19, red), keratin 5 (K5, green) and keratin 18 (K18, red). Scale bars represent 40um. (E) Quantitative cell identification determined with use of CyteSeer image analysis is represented as the average percent of positive cells for each individual marker.
sc-12-0511_sm_SupplFigure2.tif518KSupplemental Fig. 2 Verification of keratin content assay and results of kinase inhibitor screen (A) Each spot represents the sum of the three detectable keratin populations (K8+K14+, K8+ K14- , K8- K14+) for each of the 242 kinase inhibitors. Each inhibitor was tested in three wells and the corresponding average cell count is indicated by DAPI stained nuclei. Correspondence between number of nuclei and the sum of keratin stained cells indicates efficient identification of keratin stained cells by antibody staining and image analysis. (B) Log2 fold change of cell number identified by DAPI staining relative to control wells for individual kinase inhibitors. (C) Relative fractions of K8 single positive (top panel) or K14 single positive (bottom panel) cells relative to total cell count (DAPI stain) for individual kinase inhibitor. Data represent the average of 3 replicate wells for wells with at least 200 scored cells. Inhibitor identification number is indicated on the X-axis.
sc-12-0511_sm_SupplFigure3.tif950KSupplemental Fig. 3 Investigation of ROCK and GSK3β pathways on growth and CFU. (A) Dose response relationship of R1 was investigated in the absence (solid squares) or presence of the annotated GSK3β inhibitor (inverted triangles) at the near optimal concentration of 0.3 μM. Data represent the average of 3 replicates ± SD. (B) Assessment of colony forming ability in monolayer culture with a stable ROCK 2 cell line (sh919) with increasing levels of R1 inhibitor. Data represent average of 3 replicates ± SD. ROCK2 knockdown does not alleviate stimulatory effect of R1. (C) Cytospin preparations were generated from suspension cultures of WMG300 and WMG300sh902 24hrs post-plating. Cytospins were stained for cleaved caspase-3, imaged and subsequently scored with CyteSeer analysis software. Data represent the average percent of scored cells of four separate image fields ± SD, *p-value=0.19, **p-value=0.11. (D) Cellular proliferation assay of WMG300 (left panel) and WMG300sh902 (right panel) in the presence or absence of the ROCK inhibitor (R1). Cells were plated in 96 well gelatin coated wells in triplicate in the absence of R1 (2000 cells/well) or in the presence of R1 (1 μM, 1000/cells per well). At the corresponding time point, cells were fixed, stained for K8 and K14, imaged and analyzed for DAPI positive cells via CyteSeer analysis software. Values for total cell number (DAPI) are very similar to the K8+K14+ double positive population shown.
sc-12-0511_sm_SupplFigure4.tif2306KSupplemental Fig. 4 Tumor progression of WMG300 to MTIC. (A) Tumor growth rate is indicated by the number of days necessary for tumors to reach maximum allowable size upon successive passages of WMG300T (solid triangles) and WMG49T (solid squares). Arrow indicates the tumors used for analysis in (B). (B) Representative H&E and vimentin (brown) IHC on ETIC (WMG49) and MTIC tumor (WMG300). Scale bar represents 100 μm. (C) K8 and K14 immunofluorescence of WMG300 MTIC culture. (D) FACS profile of WMG300 ETICs and WMG300 MTICs. Ethanol fixed cells were stained for K8 and K14. Live cells were stained for CD24 and CD29.
sc-12-0511_sm_SupplFigure5.tif2486KSupplemental Fig. 5 Evaluation of MLC2 phosphorylation by ROCK1. (A) Immunofluorescent analysis for vimentin (VIM) and phosphorylated MLC2 (pMLC2) of WMG300. CyteSeer software scored cells positive for VIM (red bar) and pMLC2 (green bar). Data represent the average fraction of cells positive for each antibody. Note decreased fraction of cells positive for vimentin and pMLC2 in cells treated with R1. (B) Preferential co-localization of pMLC2 in cell expressing vimentin. Cells expressing vimentin were scored for simultaneous expression of pMLC2. In the presence or absence of the R1, pMLC2 is exclusively associated with cells expressing vimentin. In the absence of R1, most vimentin positive cells also express pMLC2. In the presence of R1, a smaller fraction of the vimentin cells express pMLC2. (C) Representative images of VIM and pMLC2 stains of WMG300. The nuclei are visualized by DAPI.
sc-12-0511_sm_SupplTable1.pdf399KSupplemental Table 1 Cell count and Keratin analysis of Kinase screen.
sc-12-0511_sm_SupplTable2.pdf79KSupplemental Table 2 GeneGo Pathway analysis.
sc-12-0511_sm_SupplTable3.pdf372KSupplemental Table 3 Full details of the analysis of RNAs increased in ETICs.
sc-12-0511_sm_SupplTable4.pdf338KSupplemental Table 4 Full details of the analysis of RNAs increased in MTICs.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.