Lipid Rafts Remodeling in Estrogen Receptor–Negative Breast Cancer Is Reversed by Histone Deacetylase Inhibitor

238

Lipid rafts remodeling in estrogen receptor–negative breast cancer is reversed by histone deacetylase inhibitor

Anna Ostapkowicz,1 Kunihiro Inai,1,4 Leia Smith,5 that during transition to invasive breast cancer there is a Silvia Kreda,3 and Jozef Spychala1,2 significant structural reorganization of lipid rafts and underlying cytoskeleton that is reversed upon histone deace- 1Lineberger Comprehensive Cancer Center; 2Department of tylase inhibition. [Mol Cancer Ther 2006;5(2):238–45] Pharmacology; and 3Cystic Fibrosis/Pulmonary Research and Treatment Center, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina; 4Faculty of Medical Science, University of Fukui, Fukui, Japan; and 5Seattle Genetics, Inc., Introduction Bothell, Washington Establishment and clinical use of tumor markers that define invasive and metastatic breast carcinoma is critical for more Abstract individualized therapeutic strategies in the future. Because breast cancer is becoming an increasingly heterogeneous Recently, we have found dramatic overexpression of disease, the task of defining specific cancer cell phenotypes ecto-5V-nucleotidase (or CD73), a glycosylphosphatidyli- is especially challenging. Several individual breast cancer nositol-anchored component of lipid rafts, in estrogen markers have been established for target-specific pharma- receptor–negative [ER( )] breast cancer cell lines and À cologic intervention. The clinically proven therapies in- in clinical samples. To find out whether there is a more clude targeting estrogen receptor (ER) and progesterone general shift in expression profile of membrane proteins, receptor and more recently Erb2 and epidermal growth we undertook an investigation on the expression of factor receptor (EGFR). Major differences in expression selected membrane and cytoskeletal proteins in aggres- profiles of wide number of genes has been documented in sive and metastatic breast cancer cells. Our analysis ER(À) and ER(+) carcinomas (1). Both in in vitro studies revealed a remarkably uniform shift in expression of a and in the clinic, these differences were associated with broad range of membrane, cytoskeletal, and signaling either more motile and invasive phenotype or more proteins in ER( ) cells. A similar change was found in À aggressive course of the disease in the case of ER(À) breast two in vitro models of transition to ER( ) breast cancer: À carcinoma (2). Membrane proteins and associated cytoskel- drug-resistant Adr2 and c-Jun-transformed clones of MCF- eton mediate the communication with the extracellular 7 cells. Interestingly, similar expression pattern was milieu and the composition of membrane proteins is critical observed in normal fibroblasts, suggesting the common- for cell behavior in general and invasive and metastatic ality of membrane determinants of invasive cancer cells properties in particular. with normal mesenchymal phenotype. Because a number Among different membrane microdomains, lipid rafts are of investigated proteins are components of lipid rafts, our one of the least understood subcellular elements. Although results suggest that there is a major remodeling of lipid their lipid composition has been investigated in several rafts and underlying cytoskeleton in ER( ) breast cancer. À studies (3–7), protein components were not systematically To test whether this broadly defined ER( ) phenotype À compared between invasive and noninvasive cells. Because could be reversed by treatment with differentiating agent, several in vitro models of breast cancer exemplify transition we treated ER( ) cells with trichostatin A, an inhibitor of À to more invasive and metastatic state, we have chosen histone deacetylase, and observed reversal of mesenchy- this model to study changes in lipid rafts composition after mal and reappearance of epithelial markers. Changes in loss of ER expression. Our recent finding that ER(À) breast gene and protein expression also included increased cancer cells express high level of ecto-5V-nucleotidase, a capacity to generate adenosine and altered expression glycosylphosphatidylinositol-anchored protein and a profile of adenosine receptors. Thus, our results suggest marker of lipid rafts, provides an early argument for the lipid raft remodeling during transition to more aggressive breast carcinoma (8). In this study, we aimed to analyze Received 7/5/05; revised 11/29/05; accepted 12/16/05. whether there is a consistent alteration in expression of Grant support: RO1-CA34085 and Department of Defense grant cytoskeletal membrane and lipid raft protein components DAMD17-01-1-0351. across a wider population of ER(+) and ER(À) cells that The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked would suggest a coordinate expression consistent with advertisement in accordance with 18 U.S.C. Section 1734 solely to the motile and invasive phenotype. Within this phenotype, indicate this fact. we also investigated the expression of genes and proteins Requests form reprints: Jozef Spychala, Department of Pharmacology, involved in adenosine generation and signaling. Although University of North Carolina, MEJ Building, CB7365, Chapel Hill, NC 27599-7365. Phone: 919-493-8996; Fax: 919-493-3299. limited by the availability of suitable antibodies, our unique E-mail: [email protected] focus on expressed proteins, rather than mRNA profile, Copyright C 2006 American Association for Cancer Research. allowed us to directly relate protein expression with the doi:10.1158/1535-7163.MCT-05-0226 specific cell phenotype.

MolCancerTher2006;5(2).February2006

Materials and Methods bodies against cytoskeletal proteins a-smooth muscle actin Cell Lines (clone 1A4), cytokeratin (clone K8.13 detecting forms 1, 5, 6, Breast cancer ER(+) cell lines MDA-MB-474, ZR-75-1, 7, 8, 10, 11, and 18; clone K8.12 detecting forms 13, 15, and MCF-7 and negative SK-BR-3, MDA-MB-468, MDA-MB- 16) were from Sigma (St. Louis, MO). 435s, MDA-MB-231, BT-549, Hs578t, nontransformed Lipid Raft Isolation MCF-10A, c-Jun-transformed MCF-7/c-Jun clone 2-33 Cell lysates were prepared by mixing equal volumes of and control MCF-7/neo clone 7-1, and human fibroblasts cell pellets with 2% Triton X-100 on ice for 1 minute and WI-38 were obtained from either Tissue Culture Facility subsequent dilution twice with PBS and twice further with at Lineberger Comprehensive Cancer Center/University 35% Nycodenz [5V-(N-2,3-dihydroxypropylacetamido)- of North Carolina, American Type Culture Collection 2,4,6-triiodo-N,N-bis(2,3-dihydroxypropyl)-isophtalamide] (Manassas, VA), or developed as described before (9). in PBS. At each step, mixing was achieved by pipetting the The Adr2 and AdrR MCF-7 Adriamycin-resistant cell lysate up and down several times with Eppendorf pipettor. sublines were from Dr. Y.M. Rustum (Roswell Park A modified procedure for density gradient centrifugation Cancer Institute, Buffalo, NY). Cells were maintained in using Nycodenz from Sigma-Aldrich (St. Louis, MO) MEM supplemented with Eagle salts, NaPyr, nonessential was used to fractionate Triton X-100–soluble and Triton amino acids, and 10% fetal bovine serum (most cell X-100–insoluble membrane and cytoskeletal subdomains lines); in McCoy’s supplemented with 15% fetal bovine and complexes (11). For the purpose of centrifugation, cell serum (SK-BR-3 cells); in Leibowitz L-15 supplemented lysates containing 3 to 4 mg total protein were diluted with 10% fetal bovine serum (MDA-MB-468 cells); and in 2-fold with 35% Nycodenz. Density step gradient was mammary epithelial growth medium supplemented with generated by applying 0.5 mL aliquots of increasing bovine pituitary extract, human epidermal growth factor, concentration of Nycodenz (35%, 25%, 22.5%, 20%, lysate insulin, hydrocortisone, and 10% fetal bovine serum in 17.5%, 15%, 12%, 8%, and 4%) sequentially into Beckman (BioWhittaker medium for MCF-10A cells) in CO2/O2 (Palo Alto, CA) 13 Â 51 mm polyallomer tubes. Lysates atmosphere at 37jC, except for MDA-MB-468 cells that were placed in the middle of Nycodenz gradient premixed were grown at ambient atmosphere at 37jC. All media in 17.5% Nycodenz. Tubes were centrifuged at 46,000 Âg contained penicillin and streptomycin. Original cell stocks for 4 hours in a Beckman 55 Ti rotor at 4jC. Following were stored in liquid nitrogen and each sample was kept centrifugation, 0.5 mL fractions were carefully withdrawn in culture for no more than 15 passages. and small pellet was resuspended in PBS containing 0.5% Reagents SDS and 1% Triton X-100 (fraction 10). Total of 10 fractions All general reagents were ACS or the highest purity and control input lysate were analyzed for the distribution commercially available. The following antibodies were of proteins by Western blot. Typically, components of light used. Rabbit polyclonal anti-ecto-5V-nucleotidase antibodies lipid rafts and caveolae distributed into first four fractions: (for Western blot) were generated as described before (10) Soluble cell components, including cytosolic proteins, or purchased from BD Biosciences (PharMingen, San Jose, remained in fractions 5 and 6 and cytoskeleton-associated CA; for immunofluorescence). Anti-Gas/olf sc-383, Gai-2 high-density fractions were distributed in fractions 7 to 9. sc-7276, Ga12 sc-409, Ga13 sc-410, thy-1 sc-9163, CD24 Western Blotting sc-11406, Gh sc-378, Gh2 sc-380, ankyrin B (clone 2.20), Cell extracts, obtained by scraping cells in PBS in the cyclin D1 sc-8396, integrin h1 sc-8978, integrin h2 sc-6624, presence of protein phosphatase and protease inhibitors integrin h3 sc-6627, ERa sc-8005, fyn sc-434, lyn sc-15, lck and lysing with ice-cold 1% Triton X-100/PBS, were loaded sc-433, c-fgr sc-130, hck sc-72, MDR1 sc-8313, N-cadherin on the SDS-PAGE at 30 Ag per lane. Separated proteins sc-8424, OB-cadherin sc-9997, caveolin-1 sc-894, PKCa sc were transferred onto Immobilon-P 0.45 Amol/L (Millipore, 8393, fascin, sc-16579, lamin B1, sc-6216, and secondary Bedford MA) polyvinylidene difluoride membrane and horseradish peroxidase conjugated against goat and rat used for probing with specific antibodies. Two buffer IgG were from Santa Cruz (Santa Cruz, CA). E-cadherin systems were used during incubations with antibodies: PBS C20820, fak F15020, integrin h1 I41720, integrin a5 I55220, supplemented with 5% Carnation fat-free dry milk and PKCh P17720, PKCy P36520, PKCu P15120, moesin M36820, 0.2% Tween 20 or 25 mmol/L Tris (pH 8.4) supplemented EBP50 E83020, ezrin M36820, gelsolin G37820, and flotillin-1 with 130 mmol/L NaCl, 5 mmol/L potassium phosphate, F65020 antibodies were from Transduction Laboratories 5% fat-free dry milk, and 0.2% Tween 20. Blots were reused (Lexington, KY). Anti-CD44s 13-5500 antibody was from several times after mild stripping by air drying overnight at Zymed (San Francisco, CA). Anti-c-Yes antibody 06-514 room temperature when necessary. Secondary antibodies and c-Src GD11 were from Upstate (Lake Placid, NY); rabbit conjugated to horseradish peroxidase and BM Chemilumi- anti-uPAR (399R) were from American Diagnostica, Inc. nescence Western Blotting kit (Roche, Indianapolis IN) (Greenwich, CT); anti-h-actin was from Oncogene (Boston, were used to develop images on Kodak (Rochester, NY) MA); antifilamin antibody MAB1678, integrin av AB1930, X-Mat Blue XB-1 film. talin MAB1676, and vimentin AB1620 were from Chemicon Reverse Transcription-PCR (Temecula, CA); and anti-intestinal alkaline phosphatase Extraction of total RNA was done using TRI reagent from antibodies were from Biogenesis (Kingston, NH). Anti- Molecular Research Center (Cincinnati, OH). Reverse

Mol Cancer Ther 2006;5(2). February 2006

transcription-PCR assays were done using Advantage RT defined in previous studies (12, 13), thus enabling us to and Titanium Taq PCR kits from Clontech BD Biosciences correlate protein expression with specific cell phenotypes. (Mountain View, CA). The following PCR oligonucleotides In this cell panel, MCF-10A cells is an example of for adenosine receptor subtypes were used: A1 forward: nontumorigenic mammary epithelial control cells and SK- AATTGCTGTGGACCGCTACCTC, reverse: CGACACCTT- Br-3 and MDA-MB-468 represent ER(À) cells that are less CTTGTTGAGCTG; A2a forward: TTGACCGCTACATTGC- tumorigenic than MDA-MB-435s, MDA-MB-231, BT-549, CATCCG, reverse: GAAGATCCGCAAATAGACACC; and Hs578t cells (12). Recently, we showed dramatic up- A2b forward: ACCAACTACTTCCTGGTGTCC, reverse: regulation of ecto-5V-nucleotidase, a glycosylphosphatidyli- GCAGCTTTCATTCGTGGTTCC: A3 forward: ATCGCTGT- nositol-anchored membrane ecto-protein, in ER(À) breast GGACCGATACTTG, reverse: AATGCACCTGTCTCTTTG- cancer cells (8). Here, we compared the expression of ecto- GAG. Amplification reactions were run at the following 5V-nucleotidase with other lipid raft components, such as conditions: 1 minute each at 94jC, 62jC, and 72jC, total CD24 and alkaline phosphatase. The expression of pro- 40 cycles. The size of PCR products were as follows: AR1 344 posed breast cancer marker CD24 concurred with the ER bp, AR2a 305 bp, AR2b 374 bp, AR3 352 bp, glyceraldehyde- receptor status and was also found in SK-Br-3 and MDA- 3-phosphate dehydrogenase 983 bp, and trypsin inhibitor MB-468 cells. A reciprocal pattern was found for intestinal 407 bp. alkaline phosphatase (Fig. 1A). The broader membrane and cytoskeletal protein survey revealed an alteration in expression of specific proteins in ER(+) and ER(À) cell Results lines with somewhat variable expression in SK-Br-3 and We have done a broad survey of the expression of MDA-MB-468 cells (Fig. 1A). Although E-cadherin was membrane, cytoskeletal, and signaling proteins in breast expressed mostly in ER(+) and control cells, other adhesion cancer cell lines that represent ER(+) and ER(À) pheno- receptor integrins h1, a5, and aV; CD44; and OB-cadherin types. The cell panel consists of well-characterized cell and N-cadherin tended to coexpress with ecto-5V-nucleo- lines, which invasive and metastatic potential have been tidase in ER(À) cells (Fig. 1B). Similarly, cytoskeletal

Figure 1. Differential expression of membrane and cytoskeletal proteins in hyperplastic MCF-10A cells and in a panel of breast cancer cell lines. A, representative cytoskeletal and lipid raft – associated proteins. eN, ecto-5’-nucleotidase; ALP, alkaline phosphatase. B, representative membrane and signaling proteins with roles in adhesion. C, representative cytoskeleton-associated proteins. Note that moesin (bottom) and ezrin (top band on the bottom panel) are detected simultaneously using M36820 antibodies from Transduction Laboratories. Cell lysates were prepared in enough quantity so different proteins could be compared using the same lysate. They were loaded onto each lane (30 Ag) and processed for SDS-PAGE and Western blotting procedure as described in Materials and Methods.

MolCancerTher2006;5(2).February2006

protein vimentin and partially smooth muscle actin, in contrast to several cytokeratins, coexpressed with ecto-5V- nucleotidase in ER(À) cells. Lamin B1 expression was independent of cell type; however, a-tubulin tended to express at higher levels in more aggressive ER(À) cells (Fig. 1A). Analysis of proteins associated with cytoskeleton show that although EBP50 and gelsolin were associated with less invasive cells, fimbrin, talin, filamin, and especially fascin and moesin tended to express at higher level in more invasive breast cancer cells (not shown). Interestingly, caveolin-1 expression strongly coincided with ecto-5V-nucleotidase, further suggesting that, in addition to ecto-5V-nucleotidase lipid rafts, caveolae may have specific function in more invasive cells. Other membrane or membrane-associated signaling molecules, such as EGFR (not shown), lyn, and PKCa, also showed similar expression pattern. On the other hand, FAK, PKCy, PKC~, and integrin a3 did not display cell-specific expression (not shown). This comprehensive analysis helped us subcategorize our breast cancer cell panel into three distinct profiles. BT474, ZR-75-1, and MCF-7 cells having epithelial and MDA-MB- 435s, MDA-MB-231, BT-549, and Hs578t having more mesenchymal features. Significantly, cell lines SK-Br-3 and MDA-MB-468 fall in between: Although losing many epithelial markers, such as E-cadherin and certain cytoker- Figure 2. Differential expression of membrane, adhesion, cytoskeletal, atins, they did not yet acquire full set of mesenchymal and regulatory proteins in in vitro models of breast cancer progression: features, and, as have been shown previously, are less drug-resistant (Adr2) and c-Jun-transformed MCF-7 cells. Comparison of tumorigenic (12). Interestingly, during long cell culture expression profiles with human normal fibroblasts WI38. Experimental conditions as in Fig. 1. (>20 cell passages), we occasionally observed temporal expression of ecto-5V-nucleotidase in SK-Br or lower expression in MDA-MB-468 cells, but no change in expression up-regulation of Src, lyn, and EGFR and down-regulation in other cell lines, suggesting some intrinsic phenotypic of Lck in ER(À) cells and in fibroblasts. The expression of plasticity in these particular cells. Despite their less fimbrin and flotillin-1 did not show, however, significant aggressive phenotype, this could explain metastatic behav- variation in these cells (not shown). ior of MDA-MB-468 cells in mouse xenograft model (14). To determine which membrane, cytoskeletal, and signal- The MCF-10A cell line, on the other hand, expresses more ing proteins were associated with lipid rafts, we used the complex mixture of epithelial and mesenchymal markers, modified density gradient centrifugation with Nycodenz and thus may likely have myoepithelial origin. (11) and used MDA-MB-231 and MCF-7 cell lysates as To further test whether transition to ER(À) status and representative for ER(À) and ER(+) phenotypes. These two more invasive phenotype will show similar shift in cell lines are very well characterized in previous studies expression profile, we used two independent in vitro and therefore are better suited for comparative analysis. models of breast cancer progression. Development of drug Furthermore, MDA-MB-231 cells were shown to reexpress resistance and overexpression of c-Jun in MCF-7 cells were ERa and E-cadherin upon trichostatin A treatment (16), both shown to correlate with the loss of ER expression and suggesting an intrinsic plasticity important for the purpose transition to more invasive and tumorigenic phenotype of this work. Other cell lines shown in Fig. 1, both ER(À) (9, 15). Results presented in Fig. 2 show that in these and ER(+), were also tested for lipid raft distribution of two models, there is a very similar shift in expression selected membrane proteins and the results were similar to profile to that in ER(À) cells shown in Fig. 1A-C. To further MCF-7 and MDA-MB-231 cells (not shown). Preliminary analyze whether the expression of membrane or cytoskel- experiments also determined that cholesterol, an important etal proteins may differentiate between ‘‘cancer metastatic’’ component of lipid rafts, distributed mostly with low- and normal motile phenotypes, we included in this survey density fractions 2 to 4 (here defined as lipid rafts). lysates from normal human fibroblasts WI-38. Normal As shown in Fig. 3A, ecto-5V-nucleotidase and flotillin-1, fibroblasts are commonly used to represent mesenchymal two independent markers of lipid rafts, were found phenotype. This comparison revealed that drug-resistant, predominantly in low-density fractions. On the other c-Jun-transformed, and all other invasive cells shown in hand, a-tubulin (disassembled to monomers under low- Fig. 1A-C exhibit the expression pattern of membrane and temperature conditions) and adenosine kinase were dis- cytoskeletal proteins that is remarkably similar to fibro- tributed predominantly to fractions 5 to 7, which represent blasts. In addition to data presented in Fig. 2, we observed Triton X-100 soluble cell extracts (not shown). Vimentin,

Mol Cancer Ther 2006;5(2). February 2006

h-actin, and lamin B1 were distributed mostly to fractions PKCa, caveolin-1, integrins h1 and a5, EGFR, OB-cadherin, 5 to 10, which represent heavier Triton X-100–soluble and and moesin, were all strongly down-regulated by trichos- Triton X-100–insoluble (cytoskeletal) fractions of cell lysate tatin A. On the other hand, epithelial markers, such as (not shown). cytokeratins (detectable with K8-13 antibody), E-cadherin, Among signaling membrane-associated proteins, src- ERa, EBP50, and gelsolin, were up-regulated. The expres- family members, with the exception of c-yes, distributed sion of CD24, another marker of lipid rafts in ER(+) cells, almost exclusively to lipid rafts (Fig. 3A). In contrast, protein was also significantly increased. The expression of several kinase Ca was found exclusively in soluble cell compart- other proteins, including lyn (not shown) and vimentin, ment and neither associated with lipid rafts nor with was only slightly down-regulated; however, at 48 hours of insoluble cytoskeleton (Fig. 3A). CD44 and Gai2, known drug treatment, there was increased incidence of apoptosis lipid raft components, also distributed almost exclusively to that may have masked further changes in gene expression. lipid rafts. On the other hand, Gas and Gh2 distributed in Along with the complete disappearance of CD44 standard significant part to soluble and cytoskeletal compartments form, we have noticed an appearance of a higher molecular (Fig. 3B). Integrin h1, EGFR, and OB-cadherin proteins form, probably a splice variant (18), that was also seen in highly expressed in MDA-MB-231 cells were distributed in untreated MDA-MB-468 cells (Fig. 1A). all three cell compartments and FAK was found only in Because ecto-5V-nucleotidase, the major adenosine- soluble and cytoskeletal fractions. Analysis of cell lysates producing enzyme in epithelial cells (19), was dramatically from ER(+) breast cancer MCF-7 cells that express very low down-regulated in ER(À) cells exposed to trichostatin A levels of ecto-5V-nucleotidase also showed lipid raft distri- (Figs. 1A and 4A), we tested the expression of adenosine bution of this protein, along with glycosylphosphatidylino- receptors in MDA-MB-231 cells treated with trichostatin A. sitol-linked CD24, src, lyn, and Gas, suggesting that this As shown in Fig. 4B, expression of adenosine receptors was distribution is cell type independent (not shown). unevenly affected by trichostatin A: Although A2a and A2b Previous studies showed that histone deacetylase inhib- were unchanged, the expression of A1 was decreased and itors caused shift in expression pattern of selective protein A3 was increased during this trichostatin A–induced groups. In breast cancer cells, trichostatin A has been transition to epithelial phenotype (Fig. 4B). Thus, increased shown to induce the expression of ERa and E-cadherin in capacity to produce adenosine seem to correlate with MDA-MB-231 cells (16, 17). Based on these data, we asked increased signaling through A1 receptors in aggressive whether trichostatin A may reverse the expression of a breast cancer cells and this feature is reversed by histone broader range of mesenchymal proteins that define the deacetylase inhibitors. invasive and metastatic phenotype in breast cancer cells. Our preliminary studies revealed that 0.5 Amol/L trichos- Discussion tatin A concentration was most effective without triggering Analysis of lipid components of membranes in breast cancer substantial apoptosis (see also ref. 16). Treatment of MDA- cells was done previously and revealed significantly altered MB-231 cells with trichostatin A for 48 hours caused down- ratio of phospholipids (3), increased level of gangliosides, regulation of ecto-5V-nucleotidase protein (Fig. 4A). Similar and components of lipid rafts (4–6) in ER(À)cells. result was seen with ecto-5V-nucleotidase mRNA (not Furthermore, increased circulating levels of gangliosides shown). Other proteins that have been shown to contribute were found in breast cancer patients when compared with to invasive cell behavior, such as CD44 (standard form), healthy individuals (7). Although association of few

Figure 3. Distribution of proteins from MDA-MB-231 cells after density gradient centrifugation. A and B, analysis of proteins associated with lipid rafts and cytoskeleton in MDA-MB-231 cells. Other conditions as in Fig. 1.

MolCancerTher2006;5(2).February2006

that there is a consistent shift in expression pattern between noninvasive and invasive cell phenotypes. The use of widely characterized cell lines allowed us to directly correlate the expression pattern with well-defined specific cellular phenotypes. Remarkably, the expression profile, common to all six more aggressive cell lines (MDA-MB-435s, MDA- MB-231, BT549, Hs578t, MCF-7/Adr2, and MCF-7/c-Jun), was to large extent recapitulated in normal fibroblasts. The extent of similarities, including increased expression of signaling molecules, such as EGFR, PKCa, lyn, and Gai-2, suggests that both membrane structural proteins and a number of proteins representing regulatory circuitry has been adopted from normal mesenchymal cells. The down- regulation of EBP50 and gelsolin, two proteins indepen- dently associating with actin cytoskeleton, correlate with up- regulation of moesin, fascin, and, to a lesser extent, talin and filamin. EBP50 may associate with ERM proteins (ezrin, radixin, and moesin) and thereby regulate anchoring of lipid rafts to cytoskeleton (27). Overall, our results may provide a direct evidence for epithelial to mesenchymal transdifferentiation in breast cancer. We found that several membrane proteins residing in lipids rafts are differentially expressed in invasive breast cancer cells, suggesting that there is a major remodeling of this membrane microdomain during breast cancer progression. Glycosylphosphatidylinositol-linked proteins are typically considered markers of lipid rafts. In our Figure 4. A, effect of trichostatin A on the expression of selected membrane, cytoskeletal, and associated proteins in MDA-MB-231 cells. study, they show most dramatic shift in expression Cells were treated with 0.5 Amol/L trichostatin A for 48 h, harvested, and profile. The complete down-regulation of CD24 in ER(À) processed as described in Materials and Methods. Note that some Western cells and the emergence of ecto-5V-nucleotidase and blots were exposed much longer than shown in Fig. 1, which caused some proteins, such as cytokeratins, EBP50, or gelsolin, to be detected in alkaline phosphatase most likely have physiologic con- control MDA-MB-231 cells. Each experiment was repeated at least twice sequences. CD24 is a new marker for breast and ovarian with similar result. B, effect of trichostatin A on the expression of carcinoma (28, 29) that coincides with ER(+) status (30). adenosine receptor mRNAs determined by reverse transcription-PCR (40 This mucin-like heavily glycosylated protein was shown cycles). The sizes of PCR products are as follows: AR1 344 bp, AR2a 305 bp, AR2b 374 bp, AR3 352 bp, glyceraldehyde-3-phosphate dehydroge- to be a ligand for P-selectin and proposed to mediate nase 983 bp, and trypsin inhibitor 407 bp. rolling in endothelium (31). On the other hand, ecto-5V- nucleotidase and alkaline phosphatase, which seem to membrane and cytoskeletal proteins with the ER status replace CD24 during breast cancer progression, both have were reported before, no comprehensive analysis of phosphohydrolase activity and participate in dephosphory- membrane and cytoskeletal proteins in a broad cell panel lation of extracellular nucleotides (mostly AMPs) and of breast cancer cells has been done thus far. Among these generation of signaling adenosine (32). Adenosine, the proteins, vimentin, EGFR, CD44, fascin, and E-cadherin product of both ecto-5V-nucleotidase and alkaline phospha- were shown to be highly correlated with ER status both in tase enzymatic activities (19), acting through a family of breast cancer cell lines and in clinical samples (17). EGFR receptors, has well-established roles in adhesion, growth and vimentin were shown to coexpress in specific subset of regulation, vasodilation, and angiogenesis. These activities advanced breast carcinoma distinct from both erbB2(+) are likely to be relevant for breast cancer progression and ER(+) cases (17). CD44, moesin, and fascin were also (reviewed in refs. 32, 33). Although ecto-5V-nucleotidase found to express at higher level in advanced breast cancer was also reported to participate in cell adhesion (34, 35), (20–22). A number of functional associations between it is at present unclear how ecto-5V-nucleotidase would proteins that overexpress in mesenchymal cells have also specifically contribute to invasive cell behavior. Nonethe- been established. CD44/moesin and CD44/lyn interactions less, increased capacity to generate adenosine on one hand, collaborate in triggering invasiveness and chemoresistance and increased expression of A1 and decreased expression (23, 24). Vimentin and fimbrin form functional complexes in of A3 on the other hand, support the view that adenosine macrophages (25). Recent study found a striking correlation signaling may have distinct roles in epithelial and of CD44+ and CD24À cells derived from breast carcinoma mesenchymal cells. Recent demonstration of increased with their tumorigenic potential (26). ecto-5V-nucleotidase expression and adenosine formation Our analysis of the expression of membrane, cytoskeletal, upon wnt-1 pathway activation further collaborate this and associated proteins in 12 breast cancer cell lines show notion (36).

Mol Cancer Ther 2006;5(2). February 2006

Inhibitors of histone deacetylases, including trichostatin 10. Yegutkin GG, Henttinen T, Samburski SS, Spychala J, Jalkanen S. The evidence of two opposite, ATP-generating and ATP-consuming, A, have potent effects on global gene expression, inhibit cell extracellular pathways on endothelial and lymphoid cells. Biochem J 2002; proliferation, and induce differentiation and apoptosis 367:121 – 8. (37–39). Although histone deacetylase inhibitors caused 11. Hostager BS, Catlett IM, Bisghop GA. Recruitment of CD40 and tumor necrosis factor receptor-associated factors 2 and 3 to membrane changes in many different cell types, mesenchymal cells microdomains during CD40 signaling. J Biol Chem 2000;275:15392 – 8. were proposed to be particularly sensitive (40). In 12. Sommers CL, Byers SW, Thompson EW, Torri JA, Gelmann EP. particular, previous studies have shown that MDA-MB- Differentiation state and invasiveness of human breast cancer cell lines. 231 cells induced the expression of E-cadherin and ERa Breast Cancer Res Treat 1994;31:325 – 35. after treatment with histone deacetylase inhibitor trichos- 13. Zajchowski DA, Bartholdi MF, Gong Y, et al. Identification of gene expression profiles that predict the aggressive behavior of breast cancer tatin A (16). Given the known genome-wide effects of cells. Cancer Res 2001;61:5168 – 78. histone deacetylase inhibitors, such response suggested 14. Walsh MD, Luckie SM, Cummings MC, Antalis TM, McGuckin MA. that trichostatin A may have induced differentiation, which Heterogeneity of MUC1 expression by human breast carcinoma cell lines in the context of mesenchymal cells may be regarded as in vivo and in vitro. Breast Cancer Res Treat 2000;58:255 – 66. mesenchymal to epithelial transdifferentiation. Indeed, the 15. Vickers PJ, Dickson RB, Shoemaker R, Cowan KHA. Multidrug- À resistant MCF-7 human breast cancer cell line which exhibits cross- expression of most markers that differentiate ER( ) and resistance to antiestrogens and hormone-independent tumor growth ER(+) cells were affected in a reciprocal manner. Thus, in vivo. Mol Endocrinol 1988;2:886 – 92. our expression profiling of a broader set of membrane and 16. Yang X, Ferguson AT, Nass SJ, et al. Transcriptional activation of cytoskeletal proteins provide a strong evidence for the estrogen receptor a in human breast cancer cells by histone deacetylase inhibition. Cancer Res 2000;60:6890 – 4. mesenchymal to epithelial transdifferentiation after treat- 17. Wilson KS, Roberts H, Leek R, Harris AL, Geradts J. Differential gene ment of breast cancer cells of mesenchymal phenotype with expression patterns in HER2/neu-positive and -negative breast cancer cell trichostatin A. Previous studies have shown that histone lines and tissues. Am J Pathol 2002;161:1171 – 85. deacetylase inhibitors increased the expression of metasta- 18. Oliferenko S, Paiha K, Harder T, et al. Analysis of CD44-containing sis suppressors, such as breast metastasis suppressor 1 (41), lipid rafts: recruitment of Annexin II and stabilization by the actin cytoskeleton. J Cell Biol 1999;146:843 – 54. tissue inhibitor of metalloproteinase 3 (42), and nm23 (43) 19. Picher M, Burch LH, Hirsch AJ, Spychala J, Boucher RC. Ecto 5V- and thereby may have therapeutic activity in metastatic nucleotidase and nonspecific alkaline phosphatase. Two AMP-hydrolyzing phase of breast and other carcinomas. In this context, our ectoenzymes with distinct roles in human airways. J Biol Chem 2003;278: results suggest that such antimetastatic activity, at least in 13468 – 79. À 20. Kaufmann M, Heider KH, Sinn HP, von Minckwitz G, Ponta H, Herrlich the subset of ER( ) breast cancer cells, may be due to the P. CD44 variant exon epitopes in primary breast cancer and length of reversion of intrinsically invasive and metastatic mesen- survival. Lancet 1995;345:615 – 9. chymal phenotype. 21. Bankfalvi A, Terpe HJ, Breukelmann D, et al. Gains and losses of CD44 expression during breast carcinogenesis and tumour progression. Histopathology 1998;33:107 – 16. Acknowledgments 22. Grothey A, Hashizume R, Sahin AA, McCrea PD. Fascin, an actin- We thank Dr. Elzbieta Kulig for performing reverse transcription-PCR for bundling protein associated with cell motility, is upregulated in hormone this study. receptor negative breast cancer. Br J Cancer 2000;83:870 – 3. 23. Herrlic P, Morrison H, Sleeman J, et al. CD44 acts both as a growth- References and invasiveness-promoting molecule and as a tumor-suppressing cofac- tor. Ann N Y Acad Sci 2000;910:106 – 18. 1. Martin KJ, Kritzman BM, Price M, et al. Linking gene expression patterns to therapeutic groups in breast cancer. Cancer Res 2000;60: 24. Bates RC, Edwards NS, Burns GF, Fisher DEA. CD44 survival 2232 – 8. pathway triggers chemoresistance via lyn kinase and phosphoinositide 3-kinase/Akt in colon carcinoma cells. Cancer Res 2001;61:5275 – 83. 2. Daidone MG, Coradini D, Martelli G, Veneroni S. Primary breast cancer in elderly woman: biological profile and relation with clinical outcome. Crit 25. Correia I, Chu D, Chou Y-H, Goldman RD, Matsudaira PT. Integrating Rev Oncol Hematol 2003;45:313 – 25. the actin and vimentin cytoskeletons: adhesion-dependent formation of fimbrin-vimentin complexes in macrophages. J Cell Biol 1999;146:831– 42. 3. Le Moyec L, Tatoud R, Eugene M, et al. Cell and membrane lipid analysis by proton magnetic resonance spectroscopy in five breast cancer 26. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. cell lines. Br J Cancer 1992;66:623 – 8. Prospective identification of tumorigenic breast cancer cells. Proc Natl Acad Sci U S A 2003;100:3983 – 8. 4. Lavie Y, Cao H, Bursten SL, Giuliano AE, Cabot MC. Accumulation of glucosylceramides in multidrug-resistant cancer cells. J Biol Chem 1996; 27. Itoh K, Sakakibara M, Yamasaki S, et al. Cutting edge: negative 271:19530 – 6. regulation of immune synapse formation by anchoring lipid raft to cytoskeleton through Cbp-EBP50-ERM assembly. J Immunol 2002;168: 5. Nohara K, Wang F, Spiegel S. Glycosphingolipid composition of MDA- 541 – 4. MB-231 and MCF-7 human breast cancer cell lines. Breast Cancer Res Treat 1998;48:149 – 57. 28. Fogel M, Friederichs J, Zeller Y, et al. CD24 is a marker for human breast carcinoma. Cancer Lett 1999;143:87 – 94. 6. Lucci A, Cho WI, Han TY, Giuliano AE, Morton DL, Cabot MC. Glucosylceramide: a marker for multiple-drug resistant cancers. Anticancer 29. Kristiansen G, Denkert C, Schluns K, Dahl E, Pilarsky C, Hoauptmann Res 1998;18:475 – 80. S. CD24 is expressed in ovarian cancer and is a new independent prognostic marker of patient survival. Am J Pathol 2002;161:1215 – 21. 7. Wiesner DA, Sweeley CC. Circulating gangliosides of breast-cancer patients. Int J Cancer 1995;60:294 – 9. 30. Schindelmann S, Windisch J, Grundmann R, Kreienberg R, Zeillinger R, Deissler H. Expression profiling of mammary carcinoma cell lines: 8. Spychala J, Lazarowski E, Ostapkowicz A, Ayscue LH, Jin A, Mitchell correlation of in vitro invasiveness with expression of CD24. Tumour Biol BS. Role of estrogen receptor in the regulation of ecto-5V-nucleotidase (eN) 2002;23:139 – 45. expression and extracellular adenosine generation in breast cancer. Clin Cancer Res 2004;10:708 – 17. 31. Aigner S, Ramos CL, Hafezi-Moghadam A, et al. CD24 mediates rolling of breast carcinoma cells on P-selectin. FASEB J 1998;12:1241 – 51. 9. Smith LM, Wise SC, Hendricks DT, et al. cJun overexpression in MCF-7 breast cancer cells produces a tumorigenic, invasive and hormone 32. Spychala J. Tumor-promoting functions of adenosine. Pharmacol resistant phenotype. Oncogene 1999;18:6063 – 70. Ther 2000;87:161 – 73.

MolCancerTher2006;5(2).February2006

33. Merighi S, Mirandola P, Varani K, et al. A glance at adenosine 39. Fuino L, Bali P, Wittmann S, et al. Histone deacetylase inhibitor receptors: novel target for antitumor therapy. Pharmacol Ther 2003;100: LAQ824 down-regulates Her-2 and sensitizes human breast cancer cells to 31 – 48. trastuzumab, taxotere, gemcitabine and epothilone B. Mol Cancer Res 34. Airas L, Hellman J, Salmi M, et al. CD73 is involved in lymphocyte 2003;2:971 – 84. binding to the endothelium: characterization of lymphocyte-vascular 40. Ailenberg M, Silverman M. Differential effects of trichostatin A on adhesion protein 2 identifies it as CD73. J Exp Med 1995;182:1603 – 8. gelatinase A expression in 3T3 fibroblasts and HT-1080 fibrosarcoma 35. Airas L, Jalkanen S. CD73 mediates adhesion of B cells to follicular cells: implications for use of TSA in cancer therapy. Biochem Biophys Res dendritic cells. Blood 1996;88:1755 – 64. Commun 2003;302:181 – 5. 36. Spychala J, Kitajewski J. Wnt and h-catenin signaling target the 41. Meehan WJ, Welch DR. Breast cancer metastasis suppressor 1: expression of ecto-5V-nucleotidase and increase extracellular adenosine update. Clin Exp Metastasis 2003;20:45 – 50. generation. Exp Cell Res 2004;296:99 – 108. 42. Gagnon J, Shaker S, Primeau M, Hurtubise A, Momparler RL. 37. Sambucetti LC, Fischer DD, Zabludoff S, et al. Histone deacetylase Interaction of 5-aza-2V-deoxycytidine and depsipeptide on antineoplastic inhibition selectively alters the activity and expression of cell cycle activity and activation of 14-3-3j, E-cadherin and tissue inhibitor of proteins leading to specific chromatin acetylation and antiproliferative metalloproteinase 3 expression in human breast carcinoma cells. Antican- effects. J Biol Chem 1999;274:34940 – 7. cer Drugs 2003;14:193 – 202. 38. Marks P, Rifkind RA, Richon VM, Breslow R, Miller T, Kelly WK. 43. Yasui W, Oue N, Ono S, Mitani Y, Ito R, Nakayama H. Histone Histone deacetylases and cancer: causes and therapies. Nat Rev Cancer acetylation and gastrointestinal carcinogenesis. Ann N Y Acad Sci 2003; 2001;1:194 – 202. 983:220 – 31.

Mol Cancer Ther 2006;5(2). February 2006

Downloaded from mct.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Lipid rafts remodeling in estrogen receptor−negative breast cancer is reversed by histone deacetylase inhibitor

Anna Ostapkowicz, Kunihiro Inai, Leia Smith, et al.

Mol Cancer Ther 2006;5:238-245.

Updated version Access the most recent version of this article at: http://mct.aacrjournals.org/content/5/2/238

Cited articles This article cites 42 articles, 15 of which you can access for free at: http://mct.aacrjournals.org/content/5/2/238.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/5/2/238.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/5/2/238. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.