MUC1 Initiates Src-CrkL-Rac1/Cdc42–Mediated Actin Cytoskeletal Protrusive Motility after Ligating Intercellular Adhesion Molecule-1

QiangShen, 1 Jennifer J. Rahn,1 JingZhang, 1 Nirosha Gunasekera,1 Xuejun Sun,2 Andrew R.E. Shaw,2 Michael J. Hendzel,2 Pat Hoffman,3 Ashlyn Bernier,1 and Judith C. Hugh1

1Department of Laboratory Medicine and Pathology, Heritage Medical Research Centre and 2Department of Oncology, Cross Cancer Institute, University of Alberta, Edmonton, Alberta, Canada; and 3ICOS Corporation, Bothell, Washington

Abstract motility, which is a tightly orchestrated process of numerous MUC1, a transmembrane glycoprotein of the mucin molecular events, characterized by directed cytoskeletal family, when aberrantly expressed on breast cancer cells rearrangements. Thus, clarifying the mechanism underlying is correlated with increased lymph node metastases. the hyperactivated cytoskeletal reorganization in cancer cells We have previously shown that MUC1 binds intercellular may suggest targets for effective therapeutic strategies to reduce adhesion molecule-1 (ICAM-1) on surrounding breast cancer mortality. accessory cells and facilitates transendothelial MUC1 (also called DF3, CA15-3, or episialin) is a migration of MUC1-bearing cells. Nevertheless, the transmembrane heterodimeric glycoprotein normally restricted underlyingmolecular mechanism is still obscure. to the apical surface of mammary epithelium (1, 2). However, In the present study, we used a novel assay of actin the apical polarization is frequently lost in breast cancers, cytoskeletal reorganization to show that by ligating substituted by highly overexpressed MUC1 either throughout ICAM-1, MUC1 triggers Rac1- and Cdc42-dependent the cytosol or circumferentially around the plasma membrane actin cytoskeletal protrusive activity preferentially at the (3, 4). Clinically, this aberrant expression of MUC1 in breast heterotypic cell-cell contact sites. Further, we show that cancer cells is associated with a poor prognosis and increased these MUC1/ICAM-1 interaction–initiated lamellipodial lymph node metastases (4). We have previously reported that and filopodial protrusions require Src family kinase and MUC1 binds intercellular adhesion molecule-1 (ICAM-1; CT10 regulator of kinase like (CrkL) accompanied by refs. 5, 6), an adhesion molecule normally involved in the firm the rapid formation of a Src-CrkL signaling complex at arrest and the subsequent extravasation of leukocytes through the MUC1 cytoplasmic domain. Through inhibition of the vascular endothelium during inflammation (7). Analogous Src kinase activity, we further revealed that Src is to the involvement of ICAM-1 in the extravasation of required for recruitingCrkL to the MUC1 cytoplasmic leukocytes, our recent transwell studies suggested that the domain as well as mediatingthe observed actin MUC1/ICAM-1 interaction also facilitates in vitro trans- cytoskeleton dynamics. These findings suggest a novel endothelial migration of MUC1-bearing cells through a MUC1-Src-CrkL-Rac1/Cdc42 signaling cascade monolayer of ICAM-1–expressing cells (8). This is significant, following ICAM-1 ligation, through which MUC1 as the adhesion of tumor cells to endothelium and the regulates cytoskeletal reorganization and directed subsequent exit from vasculatures of secondary organs are cell motility duringcell migration. thought to be critical steps in metastasis (9, 10). However, the (Mol Cancer Res 2008;6(4):555–67) fundamental mechanism(s) of MUC1/ICAM-1–potentiated cell transmigration is largely unknown. Introduction MUC1 contains a large extracellular domain composed of a variable number of tandem repeats of a 20-amino-acid se- The development of metastases in distant vital organs such quence, which has multiple O-glycosylation sites and represents as the lungs, liver, and brain is the principal cause of death in the binding site for the extracellular domain-1 of ICAM-1 on breast cancer patients. Metastasis requires increased cell adjacent cells (5, 11). The 72-amino-acid cytoplasmic domain of MUC1 (MUC1-CD) contains seven highly conserved Received 9/12/07; revised 12/18/07; accepted 1/6/08. tyrosine residues, four of which are confirmed phosphorylation Grant support: Canadian Breast Cancer Foundation-Prairie/NWT Chapter and sites and construct potential docking motifs for SH2 or SH3 the Canadian Breast Cancer Research Alliance. The costs of publication of this article were defrayed in part by the payment of domain–containing , such as Src family kinases and page charges. This article must therefore be hereby marked advertisement in growth factor receptor binding 2 (12, 13). Thus far, accordance with 18 U.S.C. Section 1734 solely to indicate this fact. epidermal growth factor receptor (14, 15), PKCy (16), and Src Requests for reprints: Judith C. Hugh, Department of Laboratory Medicine and Pathology, 5B4.21 Mackenzie Health Sciences Centre, University of Alberta (14, 17, 18) have been found to physically associate with Hospital, Edmonton, AB, Canada T6G 2R7. Phone: 780-407-2872; Fax: 780- and phosphorylate MUC1-CD. This indicates that MUC1, 407-3009. E-mail: [email protected] Copyright D 2008 American Association for Cancer Research. although without intrinsic kinase activity (19), could function as doi:10.1158/1541-7786.MCR-07-2033 a scaffold protein in signal transduction. Supporting this, we

Mol Cancer Res 2008;6(4). April 2008 555 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. 556 Shen et al.

have shown that MUC1 can initiate an intracellular calcium cytoskeletal changes in monoculture (data not shown). How- signal in response to ICAM-1 ligation (20). Calcium-based ever, dramatic actin cytoskeletal reorganization and membrane signaling is frequently implicated in actin cytoskeletal ‘‘tread- ruffling were provoked in the MUC1-positive breast cancer cells milling’’ dynamics and cell migration (21). Also, the signal and 293T transfectants within 4 minutes after contact with mediators Src family kinase and PLCg involved in the MUC1/ ICAM-1–transfected NIH3T3 cells. These initial membrane ICAM-1–induced calcium oscillation (20) could play crucial dynamics rapidly developed into continually extending lamelli- roles in directing cell migration (22, 23). All these findings podial and filopodial protrusions, which preferentially occurred imply that, in addition to mediating cell adhesion, the MUC1/ at the heterotypic cell-cell contact sites (Fig. 1C, first and third ICAM-1 interaction may play a signaling role in cell motility. lanes). These protrusions were highly motile, as they frequently Rho family small GTPases (Rac1, Cdc42, and RhoA) are migrated out of the focus-plane and were accompanied by central regulators of actin-based cell motility (24, 25). Once forward motion of the cell body (Fig. 1C, third lane, far right). activated in the GTP-bound forms, Rac1 and Cdc42 can In marked contrast, if either the plated cells lacked MUC1 or the consequently interact with the downstream effectors, Arp2/3 NIH3T3 transfectants lacked ICAM-1, the breast cancer cells activators of WAVE/WASP family, to initiate membrane and 293T transfectants exhibited only a low degree of ruffling and formation of membrane lamellipodial/filopodial membrane ruffling and minor protrusions, and these membrane protrusions (26), whereas RhoA is frequently involved in events were always transient and ceased within minutes cellular contraction and focal adhesion dynamics (27). Acting (Fig. 1C, second and fourth lanes; Supplementary Fig. S1), upstream of Rho GTPases, the adaptor protein CrkL has been indicating that the presence of both MUC1 and ICAM-1 in this identified as a crucial mediator of cell migration through the system is essential for the augmented cytoskeletal rearrangements. association with guanine nucleotide exchange factors, such as To quantitatively analyze the protrusive dynamics of the Dock180, for catalyzing the GDP/GTP exchange and Rho actin cytoskeleton, a novel method was developed in this study GTPase activation (28, 29). In response to migratory cues, like for quantifying the proportion (%) of actin voxel intensity sum integrin engagement, CrkL can be recruited to focal contacts (AVIS) in membrane protrusions compared with the whole cell where Src forms a signaling complex with focal adhesion (Fig. 2A-F). Compared with conventional indices based on kinase (FAK) and phosphorylates FAK-associated p130CAS or either the number or area of protrusions, this technique is more on the conserved Y-x-x-P binding motifs for the CrkL- sensitive to cytoskeletal changes because it compensates for the SH2 domain (23, 30), thereby facilitating Rho GTPase– volume of each protrusion (Supplementary Fig. S2). Further, mediated cytoskeletal rearrangements and cell migration (30). by tracing the proportion of AVIS in protrusions over time, we In the present study, we investigated the potential role of found that the cytoskeletal protrusive activity seems to be a MUC1 in the regulation of actin-based cell motility after ligating wave-like dynamic process with MUC1-positive cells showing ICAM-1. Our data revealed that the MUC1/ICAM-1 interaction increased but varying amplitude in the presence of ICAM-1, triggers Rho GTPase Rac1- and Cdc42-dependent actin cyto- compared with their controls (Fig. 2G). Because there were skeletal reorganization and increased cell protrusive motility. We wide cell-to-cell asynchronization and variation in these also showed that these MUC1-induced cytoskeletal dynamics dynamic cytoskeletal responses, a quantifiable index–actin require Src and CrkL, both of which are recruited to MUC1-CD cytoskeletal reorganization factor (ACRF), which is defined as in response to ICAM-1 ligation. Further, using selective inhibi- the average of the AVIS proportion changes in protrusions tion of Src kinase activity, we show that Src functions upstream during the 90-minute assay (Fig. 2H), was established and used of CrkL in these Rho GTPase-mediated cytoskeletal rearrange- to represent the level of actin cytoskeletal rearrangements in ments. These results suggest a novel MUC1-initiated Src-CrkL- each experiment. For each experimental condition, ACRFs Rac1/Cdc42 promigratory signaling, which may represent a were obtained from four to nine independent trials and the crucial mechanism underlying the MUC1/ICAM-1 interaction average ACRF value was subsequently calculated to facilitate in promoting breast cancer metastasis. comparisons between different experimental conditions. It was shown that significantly higher levels of cytoskeletal rearrange- ments were initiated exclusively in the MUC1-positive breast Results cancer T47D and MCF-7 cells (Fig. 3A), as well as in 293T MUC1 Induces Dramatic Actin Cytoskeletal Rearrange- SYM25 and SYM33 cells (Fig. 3B) in response to NIH3T3- ments and Protrusive Motility in the Presence of ICAM-1 ICAM-1 stimulation, compared with NIH3T3-Mock stimula- To investigate if MUC1 initiates actin-based cell motility in tion or in the MUC1-negative Hs578T and 293T SYM3 cells response to ICAM-1 ligation, phosphorylated enhanced green (Fig. 3A and B). These findings suggest that MUC1 may fluorescent protein (pEGFP)-actin expression vector was trans- initiate actin cytoskeletal reorganization and protrusive motility fected into three breast cancer cell lines (T47D, MCF-7, and in the presence of ICAM-1. Hs578T) and three MUC1-transfected 293T cell sublines To exclude the possible bias due to the unrelated differences (SYM25, SYM33, and SYM3), which differed in their MUC1 in breast cancer cell lines or the 293T SYM subclone selection, expression levels (Fig. 1A). Then, using time-lapse confocal we generated a Flp-In T-REx 293 MUC1-inducible expression microscopy, actin cytoskeletal dynamics were examined in system (Fig. 3C). Time-lapse confocal microscopy showed these pEGFP-actin–transfected cells in the presence or absence results similar to those seen previously, in that the Flp-In T-REx of human ICAM-1 expressed on NIH3T3 transfectants 293 MUC1+ cells exhibited significantly (f3-fold) higher (Fig. 1B). We found that stable expression of exogenous levels of cytoskeletal protrusive motility in response to MUC1 on 293T cells did not elicit obvious cell membrane and NIH3T3-ICAM-1 stimulation compared with NIH3T3-Mock

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. MUC1/ICAM-1 Induces Actin Cytoskeleton Reorganization 557

stimulation (Fig. 3D). On the other hand, Flp-In T-REx 293 164B, targets the MUC1 ligation site on ICAM-1, thereby MUC1-cells showed only minor levels of cytoskeletal protru- blocking the MUC1/ICAM-1 interaction (6). By stimulating sive dynamics regardless of the presence or absence of ICAM-1 breast cancer cells (T47D and Hs578T) and MUC1-transfected on the contacting NIH3T3 transfectants (Fig. 3D). These results 293T cells (SYM25 and SYM3) with NIH3T3-ICAM-1 cells substantiate that MUC1 is the key molecule that induces these that had been preincubated with either 18E3D or 164B mAb, increased actin cytoskeletal rearrangements upon ICAM-1 we found that the anti–ICAM-1 mAb 18E3D significantly stimulation. (>70%) blocked MUC1-initiated actin cytoskeletal rearrange- ments in the T47D cells, but had no effect on the MUC1- MUC1/ICAM-1 Interaction Initiates the Increased Actin negative Hs578T cells (Fig. 4A). Conversely, the anti–ICAM-1 Cytoskeletal Reorganization mAb 164 B pretreatment did not show a significant abrogating To further confirm that this MUC1-initiated cytoskeletal effect on the MUC1-induced cytoskeletal reorganization when reorganization is specifically induced by the ligation of ICAM-1, compared with the T47D cells stimulated with nonpretreated anti–ICAM-1 monoclonal antibody (mAb; 18E3D or 164B) NIH3T3-ICAM-1 cells (Fig. 4A). Similar results were also blockade was introduced before the cytoskeletal reorganization obtained in 293T SYM25 and SYM3 cells where 18E3D, but assays. Our previous work has shown that mAb 18E3D, but not not 164B, abrogated the MUC1-induced actin cytoskeletal

FIGURE 1. Cytoskeletal reorganization is initiated in MUC1-positive cells in re- sponse to ICAM-1 stimulation. A. MUC1 expression levels were examined in breast can- cer cell lines (T47D, MCF-7, and Hs578T) and MUC1- transfected 293T sublines (SYM25, SYM33, and SYM3). The MUC1 heterodi- mer dissociates under reduc- ing conditions into a large extracellular domain (f200 kDa) and a smaller cytoplas- mic domain (15-30 kDa). Tubulin was used as a loading control. IB, immunoblotting. B. ICAM-1 expression levels were examined in NIH3T3- ICAM-1 and NIH3T3-Mock transfectants. C. All the cell lines, indicated in A,were transfected with pEGFP-actin. As representatives to be shown here, T47D (green cells in the first and second lanes) and SYM25 (green cells in the third and fourth lanes) cells were plated on MatTek micro- well dishes and then subjected to cytoskeletal reorganization assays upon stimulation by NIH3T3-ICAM-1 (gray cells in the first and third lanes) and NIH3T3-Mock (gray cells in the second and fourth lanes) transfectants. White arrows, membrane ruffling; red arrows, membrane lamel- lipodial-filopodial protrusions. The time ‘‘0 min’’ represents the moment of dropping NIH3T3 transfectants, and the images at each time point are three-dimensionally recon- structed Z-stacks, including all planes of these living cells. Bars, 5 Am.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. 558 Shen et al.

protrusive motility in the presence of ICAM-1 (Fig. 4B). and formation of motile lamellipodial or filopodial protrusions Collectively, we conclude that MUC1 initiates these increased (24, 25). As the data presented above show that the MUC1/ actin cytoskeletal rearrangements and motile membrane ICAM-1 interaction initiates actin-based cell protrusive motil- protrusions by ligating ICAM-1. ity, we speculated that these Rho GTPases may be involved. To test this hypothesis, T47D and 293T SYM25 cells were Rho GTPases Rac1 and Cdc42 Are Required for the microinjected with the Rac1-T17N, Cdc42-T17N, or RhoA- MUC1/ICAM-1 Interaction–Induced Cytoskeletal Rear- T19N expression plasmid, which confers the dominant negative rangements phenotype of the corresponding Rho GTPase, using Dextran- Rho family small GTPases Rac1, Cdc42, and RhoA are Alexa 568 as an injection tracer (Fig. 5A). Time-lapse confocal crucial regulators in directing actin cytoskeletal reorganization microscopy revealed that coexpression of the dominant

FIGURE 2. Quantitative analysis of cytoskeletal protrusive motility. A. pEGFP-actin cytoskeleton was three-dimensionally reconstructed from Z-stack images at each time point in the surpass mode of Imaris. B. The three-dimensional actin cytoskeleton image was surface-rendered by setting a background threshold, allowing the quantitation of the AVIS of the whole cell. C. The cell body was defined at each plane of the cell image in the Z-series. D. A contoured surface of the cell body was generated. E. The voxels inside the contoured surface were set to zero. F. AVIS in protrusions was obtained. G. The cytoskeletal dynamics were tracked by tracing the proportion of AVIS in protrusions relative to whole-cell AVIS every 10 min over the trial period. H. Using a 293T SYM25 cell stimulated with NIH3T3-ICAM-1 as an example, ACRF represents the average of the changes of the AVIS in protrusions during the cytoskeletal reorganization assays. Bars, 5 Am.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. MUC1/ICAM-1 Induces Actin Cytoskeleton Reorganization 559

FIGURE 3. MUC1 initiates increased cytoskeletal protrusive motility in the presence of ICAM-1 stimulation. Cytoskeletal reorganization assays were done on (A) breast cancer cells (T47D, MCF-7, and Hs578T) and (B) MUC1-transfected 293T sublines (SYM25, SYM33, and SYM3) in the presence or absence of ICAM-1 stimulation. The resultant actin cytoskeleton dynamics (columns, average ACRF from at least four independent experiments; bars, SE) revealed that the MUC1-positive breast cancer cells (T47D and MCF-7) and 293T transfectants (SYM25 and SYM33) displayed statistically significant increased cytoskeletal protrusive motility following ICAM-1 stimulation, compared with the mock stimulation. C. In the Flp-In T-REx 293 system, the MUC1 expression levels were examined in Flp-In T-REx 293 MUC1+ and Flp-In T-REx 293 MUC1À cells. Tubulin was used as a loading control. D. Based on the average ACRF from at least four independent experiments (columns), the MUC1-positive Flp-In T-REx 293 cells exhibited significantly increased cytoskeletal rearrangements in the presence of ICAM-1 stimulation, compared with the controls; bars, SE. *, P < 0.05; **, P < 0.01. negative Rac1-T17N, as well as Cdc42-T17N, led to complete spatial-temporal regulation of Rho GTPase Rac1-Cdc42 inhibition of the MUC1/ICAM-1 interaction–induced cyto- activation and cell migration (30). Our previous work has skeletal protrusive motility (Fig. 5B and C). Moreover, the shown that Src family kinase and PLC are involved in the initial membrane ruffling and cell movement were also MUC1/ICAM-1 interaction–initiated calcium signal (20). considerably abrogated in both of these cell lines. In contrast, These findings suggest that CrkL may cooperate with Src in coexpression of dominant negative RhoA-T19N had no this MUC1/ICAM-1 interaction–induced actin cytoskeletal obvious inhibitory effect on the MUC1/ICAM-1–induced protrusive motility. In addition, PLC-mediated cleavage of cytoskeletal activity compared with the control cells injected PtdIns(4,5)P2 regulates the actin cytoskeleton modulating only with Dextran-Alexa 568 (Fig. 5B and C). Further, to proteins, profilin, gelsolin, and cofilin, which together with confirm these live imaging results, Rac1, Cdc42, and Rho Ins(1,4,5)P3-mediated Ca2+ signaling could result in direc- activation assays were carried out. As shown in Fig. 5D, in tional actin cytoskeletal rearrangements at the migrating contrast to Rho, both Rac1 and Cdc42 were significantly leading edge (21). activated in T47D and 293T SYM25 cells following NIH3T3- To investigate the role of Src family kinase and PLC in the ICAM-1 stimulation, compared with the control cells that were MUC1/ICAM-1 interaction–induced cytoskeletal dynamics, serum starved and stimulated with NIH3T3-Mock transfectants. T47D cells, as well as 293T SYM25 cells, were pretreated Therefore, we conclude that the Rho family small GTPases with the pharmaceutical inhibitors PP2 (Src family kinase inhi- Rac1 and Cdc42, but not RhoA, are required for mediating the bitor), U-73122 (PLC inhibitor), or U-73343 (inactive analogue MUC1/ICAM-1 ligation–initiated actin cytoskeletal rearrange- of U-73122) and then compared with untreated control cells in ments and protrusive motility. the cytoskeletal reorganization assays. To characterize the role of CrkL in this MUC1/ICAM-1–induced cytoskeletal protru- Role of Src Family Kinase, CrkL, and PLC in MUC1/ICAM- sive motility, T47D cells and 293T SYM25 cells were trans- 1 Interaction–Initiated Cytoskeletal Rearrangements fected with Cy3-labeled CrkL small interfering RNA (siRNA; We then investigated the potential mediators downstream 100 nmol/L) before the cytoskeletal reorganization assays of MUC1 in this Rac1- and Cdc42-dependent cytoskeletal (Fig. 6A). As shown in Fig. 6B, compared with the mock promigratory responses. The adaptor protein CrkL, as a transfections, the CrkL protein levels were efficiently reduced downstream mediator of Src, plays a crucial role in the by f50% in both T47D and 293T SYM25 cells at 24 hours

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. 560 Shen et al.

posttransfection and f70% after 48 hours posttransfection, observed in 293T SYM25 cells (Fig. 6D). These results show using ImageJ as the quantitative analyses software with tubulin that Src and CrkL mediate the Rac1- and Cdc42-dependent as a loading control (Fig. 6B; Supplementary Fig. S3). In cytoskeletal rearrangements in response to MUC1/ICAM-1 contrast to CrkL, the CrkI, CrkII, and Src expression levels ligation. remained similar at all time points (Fig. 6B; Supplementary Fig. S3), indicating that the down-regulation by CrkL siRNA Src Functions Upstream of CrkL in the MUC1/ICAM-1 was specific for CrkL. In the ensuing time-lapse confocal Interaction–Induced Promigratory Signaling microscopy experiments, we found that the MUC1/ICAM-1 To further clarify the signaling mechanism underlying the interaction–induced cytoskeletal protrusive motility was sub- MUC1/ICAM-1 ligation–induced cytoskeletal rearrangements, stantially (>75%) abrogated in breast cancer T47D cells under we investigated the possible interaction(s) between MUC1 and the conditions with selective inhibition of either Src or CrkL these signaling mediators after ICAM-1 ligation. In this study, (Fig. 6C). However, inhibition of PLC activity using U-73122 breast cancer T47D cells and 293T SYM25 cells were serum exhibited only a modest decrease in the MUC1/ICAM-1– starved and then stimulated with either NIH3T3-ICAM-1 or induced cytoskeletal dynamics, which was not statistically NIH3T3-Mock transfectants for 10 seconds or 1 minute, fol- significant when compared with the inactive analogue U- lowed by MUC1 immunoprecipitation using anti-MUC1 mAb 73343–pretreated cells (Fig. 6C). Similar results were also B27.29. The immunoprecipitates were then analyzed by SDS- PAGE and sequential immunoblotting for tyrosine-phosphory- lated proteins, Src, CrkL, and MUC1. Using the whole-cell lysates as positive controls, we found that there was always a basal level of association of Src with MUC1 under all con- ditions (Fig. 7A). However, in both T47D and 293T SYM25 cells that were stimulated with NIH3T3-ICAM-1 transfectants, the anti-MUC1 mAb B27.29 coimmunoprecipitate increased amounts of associated Src compared with the NIH3T3-Mock– stimulated and serum-starved controls (Fig. 7A; Supplementary Fig. S4A). Similar results were also obtained when probing the membranes for CrkL. In contrast to the negligible levels observed in controls, CrkL was significantly increased in MUC1 immunoprecipitates within 10 seconds of stimulation by ICAM-1 (Fig. 7A; Supplementary Fig. S4A). These ICAM-1– induced increased associations of MUC1 with Src and CrkL were further confirmed in the reciprocal immunoprecipitations using anti-Src antibody (Fig. 7B; Supplementary Fig. S4B) and anti-CrkL antibody (Fig. 7C; Supplementary Fig. S4C). Taken together, these results suggest that MUC1/ICAM-1 ligation recruits Src and CrkL to MUC1-CD where they form a signaling complex crucial for actin-based cell motility. The localization and activation of CrkL, the SH2-SH3-SH3 adaptor protein, is controlled by protein tyrosine kinase– mediated phosphorylation, which creates the SH2 domain binding motifs for recruiting CrkL (31). The MUC1-CD has seven tyrosine residues, four of which are established phosphorylation sites of protein tyrosine kinases, including Src (12, 18, 32). Therefore, we postulate that Src kinase activity may be required for recruiting CrkL to the MUC1-CD in the ICAM-1 ligation–induced promigratory signaling. To address this, coimmunoprecipitation experiments were done on the Src kinase inhibitor PP2-pretreated T47D and 293T SYM25 cells using anti-MUC1 mAb B27.29. The resultant immunoprecipi- tates were then examined for Src, CrkL, MUC1, and tyrosine- phosphorylated proteins. We found that, compared with the non–PP2-pretreated controls, which showed significantly FIGURE 4. MUC1/ICAM-1 ligation is required for this increased cytoskeletal reorganization. NIH3T3-ICAM-1 cells were pretreated with increased tyrosine phosphorylation of the MUC1-CD following anti – ICAM-1 mAb (18E3D or 164B) before stimulating breast cancer cells MUC1/ICAM-1 interaction, no obvious MUC1-CD tyrosine and MUC1-transfected 293T cells in the cytoskeletal reorganization phosphorylation was observed in the PP2-pretreated T47D and assays using the non-pretreated NIH3T3-ICAM-1 or NIH3T3-Mock cells as controls. Columns, average ACRF for (A) breast cancer T47D, Hs578T 293T SYM25 cells under all experimental conditions (Fig. 8A). cells and (B) MUC1-transfected 293T SYM25, SYM3 cells from at least Correspondingly, there was no increase in Src and CrkL in the three independent experiments; bars, SE. 18E3D, but not 164B, significantly abrogated ICAM-1 – induced cytoskeletal reorganization in MUC1 immunoprecipitates of the PP2-pretreated cells follow- the MUC1-positiveT47D and 293T SYM25 cells. *, P < 0.05; **, P < 0.01. ing ICAM-1 stimulation (Fig. 8B). Further, PP2 pretreatment

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. MUC1/ICAM-1 Induces Actin Cytoskeleton Reorganization 561

FIGURE 5. Rho family GTPase Rac1 and Cdc42 mediate the MUC1/ICAM-1 interaction – induced cyto- skeletal rearrangements. A. An example of a pEGFP- actin – transfected 293T SYM25 cell microinjected with dominant negative Rac1-T17N, using Dextran Alexa 568 as a tracing dye. Bars, 5 Am. The subsequent cytoskel- etal reorganization assays on both T47D cells (B) and 293T SYM25 cells (C). Columns, average ACRF from at least three independent experiments; bars, SE. *, P < 0.05. D. Activation assays showing increased activation of Rac and Cdc42 with no change in Rho in T47D and 293T Sym25 cells, demonstrating that Rho GTPases Rac1 and Cdc42 (but not RhoA) are required for the MUC1/ICAM-1 interaction – induced actin cytoskeletal protrusive motility.

significantly decreased transmigration of ICAM-1–stimulated transiting cell and is essential in mediating cell migration in T47D cells in a transwell assay (Supplementary Fig. S5). Based leukocytes (7, 36), a crucial role is suggested for MUC1/ICAM- on all the evidence above, we conclude that Src kinase activity 1 interaction during breast cancer metastasis. However, the is required for recruiting CrkL to MUC1-CD, where Src molecular mechanism(s) underlying the promigratory property functions upstream of CrkL in the Rho GTPases Rac1- and of MUC1 remained elusive. Cdc42-mediated cytoskeletal protrusive motility. In this study using a novel assay of actin cytoskeletal reorganization, we show that MUC1 initiates dramatic cytoskeletal rearrangements and increased protrusive activity Discussion in response to ICAM-1 ligation. Human breast cancer is not a MUC1, as the fundamental member of the mucin family homogeneous disease. There are at least four genetic subtypes glycoproteins, has long been associated with mammary defined by hierarchical analysis of gene expression (37). T47D tumorigenesis and malignant progression. In addition to the and MCF-7 differ from Hs578T in that they are positive for the potential roles that have been suggested for MUC1 in tumor cell estrogen receptor and are representative of one of the two proliferation, scattering, and survival (15, 33-35), we were the luminal subtypes of breast cancer, which together account for first to report that MUC1 was a proadhesive molecule capable 70% of breast cancers. Luminal A is very indolent whereas of mediating heterotypic cell-cell adhesion by binding ICAM-1 Luminal B tumors show a progressive decline in survival, on accessory cells (5, 6). Further, we showed that increased eventually being equivalent to the estrogen receptor–negative, transendothelial migration could be obtained under conditions HER2, and basal subtypes (38). Because T47D and MCF-7 of increased MUC1/ICAM-1 interaction (8). As ICAM-1 is were isolated from metastatic disease, they most likely present throughout the entire expected migratory track of a represent Luminal B tumors, the single most common genetic

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. 562 Shen et al.

variant. Therefore, our results using cells that have an gratory signaling into cells and inducing the dynamic cyto- inherently slower motility and clinical progression may be skeletal rearrangements. These findings based on our previous particularly relevant to a large proportion of human breast studies showing that the MUC1/ICAM-1 interaction mediates cancers in which the ability of these cells to metastasize is low cell adhesion (6) provide the first evidence that MUC1, by and perhaps dependent on a recognition phenomenon. The ligating ICAM-1, initiates actin-based cell protrusive motility. rapid adverse clinical course of the estrogen receptor–negative Because cancer metastatic spread is characterized by increased tumors, as exemplified in vitro by the Hs578Ts, probably tumor cell motility, this result represents a crucial step for fully indicates a different mechanism of interaction with the host understanding the potential role of MUC1 in breast cancer vasculature that is not ICAM-1 dependent. In support of this, metastasis. During metastasis, tumor cells lose their original we did not find a difference in protrusions when the Hs578Ts tissue contacts, invade the extracellular matrix, transit via the were exposed to ICAM-1–expressing cells. lymphatic/blood system, and adhere to and then extravasate Because the enhanced cytoskeletal promigratory responses at the secondary metastatic site. Most recently, consistent with are only induced in the MUC1-expressing cells and exclusively our previous transendothelial migration study (8), we found abrogated by the anti–ICAM-1 antibody that specifically that MUC1, by ligating ICAM-1 on stromal fibroblasts, also blocks the MUC1/ICAM-1 ligation, an initiating role of the potentiates cell invadopodial protrusions and stromal invasion MUC1/ICAM-1 interaction is shown for relaying the promi- (data not shown). This implies that tumor cells may use MUC1

FIGURE 6. Src family kinase and CrkL are required for the MUC1/ICAM-1 interaction – induced cytoskeletal protrusive motility. A. Following CrkL siRNA transfection, the successfully transfected cells were identified by the red fluorescence of the siRNA-tagged Cy3. Bars, 5 Am. B. Using mock transfection (no siRNA) as a control, the expression levels of CrkL protein were significantly knocked down by CrkL siRNA (f50%) 24 h posttransfection in both T47D and 293T SYM25 cells, and the levels of CrkL were further reduced (f70%) after 48 h of transfection. The knockdown was specific for CrkL as protein levels of CrkII and CrkI were not significantly altered (see Supplementary Fig. S3 for quantification). Tubulin and Src were used as loading controls. The following cytoskeletal reorganization assays on T47D (C) and 293T SYM25 (D) cells subsequent to selectively inhibiting Src family kinase (PP2), CrkL (siRNA), and PLC (U-73122) using the inactive analogue U-73343 as a control revealed that Src family kinase and CrkL are required for the MUC1/ICAM-1 – induced cytoskeletal protrusive motility. Columns, average ACRF from at least three independent experiments; bars, SE. *, P < 0.05.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. MUC1/ICAM-1 Induces Actin Cytoskeleton Reorganization 563

FIGURE 7. MUC1/ICAM-1 interaction induces formation of a Src-CrkL signaling com- plex at MUC1-CD. A. Coimmu- noprecipitation experiments weredoneonbothT47D and 293T SYM25 cells that were grown in culture medium, serum starved, and then stim- ulated with NIH3T3-Mock or NIH3T3-ICAM-1 transfectants for 10 s and 1 min. Using MUC1 itself as the loading control, increased Src (red arrows) and CrkL were detected in the MUC1 immu- noprecipitates of both T47D and 293T SYM25 cells in response to ICAM-1 stimula- tion. The nonspecific bands, which were detected just be- low the Src bands, are anti- body heavy chains. Whole-cell lysates were used as positive controls. The ‘‘acellular’’ sam- ples consisting of only protein G – agarose (PAG) and/or antibody were used as nega- tive controls. The reciprocal immunoprecipitation proce- dures were also conducted using anti-Src mAb (B) and anti-CrkL mAb (C), with similar results showing that in re- sponse to the stimulation by ICAM-1, Src, and CrkL are recruited to MUC1-CD.

as an aberrantly up-regulated adhesion receptor to facilitate nuclear factor-nB and nuclear factor-AT (39, 40). Considering cell–extracellular matrix interactions, thereby initiating actin- that nuclear factor-nB and nuclear factor-AT are efficiently based cell invasive motility. activated by intercellular calcium oscillations (41) and fre- Using selective inhibition of Src family kinase, CrkL, and quently implicated in malignant cell survival (42), the MUC1/ PLC, we found that Src family kinase and CrkL are required ICAM-1–initiated calcium signaling may have an antiapoptotic for this MUC1/ICAM-1 ligation–initiated cytoskeletal rear- effect that is parallel to the MUC1/ICAM-1–induced invasive rangements and cell protrusive motility. In contrast to the motility. MUC1/ICAM-1–induced calcium oscillatory signaling, inhibi- This study revealed that Src is a critical regulator of this tion of PLC activity caused only a minor decrease in the actin-based cell motility. In response to the MUC1/ICAM-1 MUC1-induced cytoskeletal promigratory activities in response interaction, Src is shown to be recruited to the MUC1-CD to ICAM-1 ligation. This result indicates that, although the where it forms a signaling complex with CrkL. Because the PLC-mediated activation of actin cytoskeleton-regulatory MUC1-CD lacks intrinsic kinase activity, its association with proteins (e.g., profilin, gelsolin, cofilin) and calcium-based the Src non–receptor tyrosine kinase is crucial for understand- signaling have been implicated in cytoskeleton reorganization ing the initial signaling events downstream of ICAM-1 ligation. as well as cell migration (21, 22), the regulatory role of PLC in There is evidence suggesting that the intracellular targeting of the MUC1/ICAM-1–induced cytoskeletal dynamics may be Src is a kinase-independent mechanism that is determined merely auxiliary or redundant for Src and CrkL. Therefore, we principally by interactions involving the SH3 domain with speculate that this MUC1/ICAM-1–induced cytoskeletal pro- modulation by the SH2 domain (43). Supporting this, previous trusive activity does not share the same molecular mechanism studies by Li et al. (18) showed that Src interacts with MUC1 underlying the previously reported calcium signaling following via its NH2-terminal SH3 domain and is able to phosphorylate MUC1/ICAM-1 interaction. There are several lines of evidence the tyrosine residues of MUC1-CD to generate conserved suggesting that MUC1 modulates transcriptional activities via binding sites for SH2-containing proteins. In this report, we

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. 564 Shen et al.

found that there was always a minor level of association and Cdc42, but not RhoA, mediate the MUC1/ICAM-1–induced between Src and MUC1 in both breast cancer cells and MUC1- cytoskeletal protrusive motility. This is significant, as the regu- transfected 293T cells. Subsequent to MUC1/ICAM-1 interac- latory role of CrkL in actin-based cell motility is mediated by tion, the MUC1-Src association was significantly increased and Rho GTPase Rac1 and Cdc42 (30, 45). accompanied by increased MUC1-CD phosphorylation and In summary, this study presents evidence showing that CrkL recruitment. Further, using inhibition of Src kinase MUC1, by interacting with ICAM-1 on the accessory cells, activity, we revealed that Src functions upstream of CrkL in the initiates actin-based cell invasive motility, which is mediated by MUC1/ICAM-1–induced promigratory signaling. Src, CrkL, as well as Rho GTPases Rac1 and Cdc42. In The adaptor protein CrkL is usually localized and activated addition, we showed that Src functions upstream of CrkL in this by recruitment to a tyrosine-phosphorylated ‘‘Y-x-x-P’’ motif Rho GTPase Rac1– and Cdc42–mediated cytoskeletal rear- via its SH2 domain (23, 30). For example, in the integrin- rangements, suggesting a novel MUC1-initiated Src-CrkL- mediated cell migration, Src interacts directly with FAK to form Rac1/Cdc42 promigratory signaling pathway in response to a ‘‘Src-FAK signaling complex’’ at the newly formed focal ICAM-1 ligation. This work builds on our previous studies and contacts, where Src can subsequently phosphorylate the FAK- provides not only functional evidence that implicates MUC1/ associated p130CAS/paxillin at the conserved binding sites for ICAM-1 ligation as a key molecular event in the aggressive CrkL thereby initiating the Rho GTPase Rac1 and Cdc42- tumor phenotype, but also an insight into the underlying mediated cell motility (23, 44). In this study, FAK was not molecular mechanisms that tumor cells may use to promote associated with the MUC1-CD in the MUC1 immunoprecipi- tumor metastasis. tates, regardless of the presence or absence of ICAM-1 stimu- lation (data not shown). The MUC1-CD contains two highly conserved Y-x-x-P sequences (Y35VPP38 and Y60TNP63), Materials and Methods which represent potential docking sites for the SH2 domain Antibodies and Reagents of the adaptor protein CrkL. Therefore, we propose that in B27.29 mouse mAb against the MUC1-ECD was a gift from response to MUC1/ICAM-1 ligation, Src-mediated phosphor- Biomira, Inc. CT2 Armenian hamster mAb against MUC1-CD ylation of MUC1-CD serves to directly recruit CrkL and induce was generously provided by Dr. Sandra Gendler (Mayo Clinic, the actin-based cell motility in a FAK-independent manner. Scottsdale, AZ). Mouse anti-human ICAM-1 mAbs, 18E3D It is well established that the Rho GTPases (Rac1, Cdc42, and 164B, were from ICOS Corporation. Mouse mAbs against and RhoA) are key effectors of actin cytoskeleton protrusion Src (clone GD11) and CrkL were purchased from Upstate Cell machinery (25). In this study, using coexpression of the respec- Signaling Solutions. Mouse anti-Crk mAb was from BD tive dominant negative Rho GTPase, as well as Rho GTPases Transduction Laboratories. Mouse anti-CrkII mAb and anti- (Rac1, Cdc42, and Rho) activation assays, we revealed that Rac1 phosphotyrosine mAb PY99 were purchased from Santa Cruz

FIGURE 8. Src kinase ac- tivity is required for recruiting CrkL to MUC1-CD following MUC1/ICAM-1 interaction. Subsequent to the inhibition of Src kinase activity using PP2, coimmunoprecipitation (IP) experiments were done on T47D and 293T SYM25 cells following stimulation with NIH3T3-Mock or NIH3T3- ICAM-1 cells for 10 s and 1 min. A. By reprobing MUC1 itself as the loading control, the phosphorylation of MUC1-CD (red arrows) was significantly reduced in PP2-pretreated T47D cells and 293T SYM25 cells following MUC1/ICAM-1 interaction, compared with the non-PP2 – pretreated controls. B. The recruitment of CrkL to MUC1-CD was also signi- ficantly abrogated in PP2- pretreated T47D cells and 293T SYM25 cells following MUC1/ICAM-1 interaction.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. MUC1/ICAM-1 Induces Actin Cytoskeleton Reorganization 565

Biotechnology, Inc. The goat anti-mouse and anti-Armenian [152 mmol/L NaCl, 5.4 mmol/L KCl, 0.8 mmol/L MgCl2, hamster horseradish peroxidase–conjugated secondary anti- 1.8 mmol/L CaCl2, 10 mmol/L HEPES, 5.6 mmol/L glucose bodies were from Jackson ImmunoResearch Laboratories, Inc. (pH 7.2); refs. 20, 46], and left in imaging buffer at 37jCfor MOPC 31C mouse IgG1 and B-5-1-2 anti-tubulin mAb were 30 min until imaged. Meanwhile, NIH3T3-ICAM-1 or purchased from Sigma-Aldrich. Protein G–agarose was NIH3T3-Mock transfectants were trypsinized, rinsed once with purchased from Roche Diagnostics. ECL Plus Western Blotting PBS, and resuspended in imaging buffer at f1 Â 107/mL. detection reagents were purchased from GE Healthcare Then, the MatTek dish was placed in a 37jC microscope stage (Amersham Biosciences). U-73122, U-73343, protease inhib- heater on a Zeiss 2-photon confocal microscope (Zeiss NLO itor cocktail, phosphatase inhibitor cocktail 2, and gelatin were 501), and a representative cell was focused with a Plan Neofluar from Sigma-Aldrich. PP2 was obtained from Tocris. Dextran- Â40/1.3 oil immersion lens. The EGFP was excited using a 488 Alexa Fluor 568 was purchased from Molecular Probes. nm laser line of a 25 mW Argon excitation source (for the DMEM, fetal bovine serum (FBS), G418, blasticidin S, hygro- dominant negative Rho GTPase–microinjected cells, a 543 nm mycin B, and tetracycline were purchased from Invitrogen, Inc. laser line of a 1 mW HeNe laser was used for excitation of the Alexa Fluor 568 signal) with the intensity output set to 25%. A Expression Plasmids Eight-bit images were collected with a 1.60- s dwell time per pEGFP-actin plasmid was obtained from Clontech Labo- pixel. The pinhole was set to 1 Airy unit and images were A ratories, Inc. The pcDNA5/FRT/TO-MUC1 plasmid was collected at 0.11 m/pixel resolution. The gain and offset were constructed by releasing the MUC1 insert from the set to expand the signal from the cell being imaged over the pcDNA1-hMUC1-TR+ plasmid (kindly provided by Dr. 255 gray value scale of the 8-bit data range. Then, Z-stacks f A Sandra Gendler) with NotI, and ligated into a dephosphory- ( 0.45- m intervals from cell bottom to top) were acquired lated, NotI digested pcDNA5/FRT/TO plasmid (Invitrogen). every 2 min for a total of 45 min, recorded over the 1.5-h Digestion with KpnI was used to check correct directionality experiment under both differential interference contrast micros- of the MUC1 insert, which has a KpnI restriction site close to copy and a 505-nm-long pass filter (EGFP). Six hundred the 3¶ end. The pOG44 Flp recombinase expression plasmid microliters of NIH3T3-ICAM-1 or NIH3T3-Mock cell suspen- was purchased from Invitrogen. The dominant negative Rac1- sion were gently added onto the plated cells immediately T17N, Cdc42-T17N, and RhoA-T19N plasmids were kindly after the first image was taken. For the ICAM-1 blockade provided by cDNA Resource Center, University of Missouri- experiments, NIH3T3-ICAM-1 transfectants were incubated A Rolla (Rolla, MO). with anti–ICAM-1 mAb 164-B (20 g/mL) or mAb 18E3D (20 Ag/mL) for 30 min at 37jC before adding onto the plated cells and commencing cell imaging. In the pharmaceutical inhi- Cell Culture and Transfections bitory experiments, the plated cells were preincubated with PP2 Human breast cancer cells T47D, MCF-7, and Hs578T were (10 Amol/L), U-73122 (10 Amol/L), or U-73343 (10 Amol/L) from the American Type Culture Collection and were in imaging buffer for 30 min at 37jC, followed by NIH3T3 cells maintained in DMEM containing 10% FBS. The MUC1- stimulation and cell imaging. For the selective CrkL inhibitory transfected 293T SYM3, SYM25, and SYM33 sublines, as experiments, the plated cells were transfected with Cy3-labeled previously described (20), were maintained in DMEM CrkL siRNA duplexes for 48 h before the stimulation by containing 10% FBS and 200 Ag/mL G418. Human ICAM- NIH3T3-ICAM-1 or NIH3T3-Mock transfectants. 1–transfected NIH3T3 cells and the mock-transfected counter- parts were a generous gift of Dr. Ken Dimock (University of Quantitative Analysis of Cytoskeletal Reorganization Ottawa, Ottawa, ON, Canada) and were maintained in DMEM Under the Expert mode of LSM software, the image stacks containing 10% FBS and 2 Ag/mL blasticidin S. To generate the of Z-series were retrieved at every fifth time point (i.e., every Flp-In T-REx 293 MUC1-inducible expression system, the Flp- 10 min) from time zero (i.e., the first image Z-stack). The In T-REx 293 parental cells (Invitrogen) were cotransfected analyses of actin cytoskeletal rearrangements were done using with pcDNA5/FRT/TO-MUC1 plasmids and pOG44 plasmids Imaris software (Bitplane AG, version 4.2). First, the three- using Lipofectamine 2000 (Invitrogen), according to the dimensional time-lapse images were reconstructed. Then, in the manufacturer’s protocol. The generated Flp-In T-REx 293 Surpass mode of Imaris software, the differential interference MUC1 cells were maintained in DMEM containing 10% FBS, contrast microscopic channel was deleted and the three- 15 Ag/mL blasticidin S, and 150 Ag/mL hygromycin B. As a dimensional EGFP images at each time point were surface result, MUC1 expression can be induced or inhibited in this rendered by setting a threshold value, which subtracted the system by tetracycline-responsive regulation. nonspecific intensity values of background and generated surface objects enclosing the voxels with intensity values Cytoskeletal Reorganization Assays above the threshold. The isosurface was then used as a mask to Thirty-five–millimeter glass-bottomed microwell dishes measure the whole-cell pEGFP-AVIS at each time point and the (MatTek Corp.) were coated with 100 AL of FBS or 0.1% resultant data were exported to MS Excel with the built-in (w/v) gelatin as described previously (20). One hundred micro- statistical function of the software. Then, the cell body was liters of pEGFP-actin transfected breast cancer cells, 293T manually contoured in each slice of the imaging stacks. Further, SYM sublines, or Flp-In T-REx 293 MUC1 cells at 5 Â 104/mL by masking the channel with the contoured surface (i.e., cell were plated on the precoated dishes and equilibrated overnight. body) and setting the voxels inside the contoured surface to The cells were then washed once with 37jC imaging buffer zero, the AVIS in protrusions was obtained in the same way

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. 566 Shen et al.

as for the whole-cell pEGFP-actin. Thus, the cytoskeletal done using Lipofectamine 2000 (Invitrogen) according to the rearrangements can be tracked by tracing the proportion (%) of manufacturer’s instructions. AVIS in the protrusions compared with that in the whole cell over time. The ACRF was calculated as the average of the AVIS Coimmunoprecipitation and Western Blot Analysis changes in protrusions, during the course of the entire Following treatment under each experimental condition, cytoskeletal reorganization assay, thereby numerically repre- cells were harvested in lysis buffer [50 mmol/L Tris (pH 7.6), senting the level of cytoskeletal dynamics under each 0.5% NP40, 100 mmol/L NaCl, 0.5 mmol/L EDTA, freshly experimental condition (Fig. 2). supplemented with 0.5% protease inhibitor cocktail and phosphatase inhibitor cocktail 2] and the insoluble components Microinjection of Single Cells were pelleted at 14,000 Â g for 1 min. The protein Microinjection was done on a Zeiss Axiovert 100M concentrations of the supernatants were then determined using fluorescent microscope by using an Eppendorf microinjection a Bio-Rad detergent-compatible protein assay (Bio-Rad) and system. The pEGFP-actin transfected T47D or SYM25 cells equal amounts (1.8 mg) of proteins were incubated with were plated as described previously for the actin cytoskeletal specific antibodies for 2 h at 4jC with gentle agitation. The reorganization assays. Rac1-T17N, Cdc42-T17N, or RhoA- incubation continued with addition of protein G-agarose, rinsed T19N plasmid (100 Ag/mL) was mixed with Dextran-Alexa once with lysis buffer, for 6 h at 4jC with end-to-end agitation. 568 in sterilized PBS. Then, the mixture was spun down at The immunocomplexes were then pelleted and washed 14,000 Âg for 30 min at 4jC, and the supernatant was taken for extensively with lysis buffer. Subsequently, the immunocom- microinjection. Following microinjection, the cells were rinsed plexes were separated by SDS-PAGE, transferred to an once with PBS and then incubated with fresh medium for Immobilon-P membrane (Millipore Corp.), and immunoblotted another 8 to 12 h before the actin cytoskeleton reorganization with appropriate primary antibodies. After washing and assay. The successfully injected cells were identified by the incubating with horseradish peroxidase–conjugated secondary fluorescence of Alexa Fluor 568 using a Cy3 filter set. antibodies, the proteins of interest were visualized using ECL Plus detection system and Kodak BioMax MR Film. The films were then scanned and processed with ImageJ software (NIH) Rho Family Small GTPase Activation Assay for quantification. Where indicated, the membrane was stripped The intracellular activities of Rho family GTPases Rac1, and reprobed with other antibodies. Cdc42, and RhoA were examined using Rac1, Cdc42, and Rho activation assay kits, respectively (Upstate Biotechnology) according to the manufacturer’s protocols. In brief, breast Acknowledgments We thank Drs. Sandra Gendler and Ken Dimock, ICOS Corporation, and UMR cancer T47D cells and 293T SYM25 cells were grown in 10-cm cDNA Resource Center for sharing their plasmids, antibodies, and cell lines; and cell culture dishes to f75% confluence. Then, the cells were Dr. Bo Sun, Jeffrey Chow, Mandy Mah, and Gerry Baron for the valuable advice serum starved and stimulated with NIH3T3-Mock or ICAM-1 and technical support concerning this project. transfectants for 4 min. The cells were then rinsed with ice-cold PBS and lysed with Mg2+ lysis/wash buffer. After clarifying the References cell lysates with glutathione agarose and quantifying the protein 1. Braga VM, Pemberton LF, Duhig T, Gendler SJ. Spatial and temporal expression of an epithelial mucin, Muc-1, during mouse development. concentrations, aliquots with equal amounts of proteins were Development 1992;115:427 – 37. incubated with Rac/Cdc42 assay reagent (PAK-1 PBD, agarose) 2. Pemberton LF, Rughetti A, Taylor-Papadimitriou J, Gendler SJ. The epithelial or Rho assay reagent (Rhotekin RBD, agarose) at 4jC for 45 to mucin MUC1 contains at least two discrete signals specifying membrane 60 min, using the GTPgS-pretreated lysates as positive localization in cells. J Biol Chem 1996;271:2332 – 40. 3. Kufe D, Inghirami G, Abe M, Hayes D, Justi-Wheeler H, Schlom J. controls. The precipitated GTP-bound Rac1, Cdc42, or Rho Differential reactivity of a novel monoclonal antibody (DF3) with human were then eluted in Laemmli reducing sample buffer, resolved malignant versus benign breast tumors. Hybridoma 1984;3:223 – 32. in a 4% to 20% gradient SDS-PAGE, and immunoblotted using 4. Rahn JJ, Dabbagh L, Pasdar M, Hugh JC. The importance of MUC1 cellular specific antibodies against Rac1, Cdc42, or Rho (Upstate localization in patients with breast carcinoma: an immunohistologic study of 71 patients and review of the literature. Cancer 2001;91:1973 – 82. Biotechnology). 5. Kam JL, Regimbald LH, Hilgers JH, et al. MUC1 synthetic peptide inhibition of intercellular adhesion molecule-1 and MUC1 binding requires six tandem CrkL siRNA and Transfections repeats. Cancer Res 1998;58:5577 – 81. CrkL siRNA duplexes were obtained from Dharmacon, 6. Regimbald LH, Pilarski LM, Longenecker BM, Reddish MA, Zimmermann G, Hugh JC. The breast mucin MUCI as a novel adhesion ligand for endothelial Inc. The sequences targeting the CrkL mRNA are sense 5¶- intercellular adhesion molecule 1 in breast cancer. Cancer Res 1996;56:4244 – 9. CCGAAGACCUGCCCUUUAAUU-3¶ and antisense 5¶- 7. Kuby J. Leukocyte migration and inflammation. Immunology. 3rd ed. New PUUAAAGGGCAGGUCUUCGGUU-3¶ for duplex 1, sense York: WH. Freeman; 1997. p. 361 – 77. 5¶-GAAGAUAACCUGGAAUAUGUU-3¶ and antisense 5¶- 8. Rahn JJ, Chow JW, Horne GJ, et al. MUC1 mediates transendothelial PCAUAUUCCAGGUUAUCUUCUU-3¶ for duplex 2, sense migration in vitro by ligating endothelial cell ICAM-1. Clin Exp Metastasis 2005; 22:475 – 83. 5¶-GUCACAAGGAUGAAUAUAAUU and antisense 5¶- 9. Glinskii OV, Huxley VH, Turk JR, et al. Continuous real time ex vivo PUUAUAUUCAUCCUUGUGACUU-3¶ for duplex 3, and epifluorescent video microscopy for the study of metastatic cancer cell sense 5¶-AAUAGGAAUUCCAACAGUUUU-3¶ and antisense interactions with microvascular endothelium. Clin Exp Metastasis 2003;20: 5¶-PAACUGUUGGAAUUCCUAUUUU-3¶ for duplex 4. Cy3 451 – 8. 10. Glinsky VV, Glinsky GV, Glinskii OV, et al. Intravascular metastatic cancer labeling of CrkL siRNA duplexes was done using a Cy3-siRNA cell homotypic aggregation at the sites of primary attachment to the endothelium. labeling kit (Ambion), and the transfection of siRNA was then Cancer Res 2003;63:3805 – 11.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. MUC1/ICAM-1 Induces Actin Cytoskeleton Reorganization 567

11. Hayashi T, Takahashi T, Motoya S, et al. MUC1 mucin core protein binds to 30. Li L, Guris DL, Okura M, Imamoto A. Translocation of CrkL to focal the domain 1 of ICAM-1. Digestion 2001;63 Suppl 1:87 – 92. adhesions mediates integrin-induced migration downstream of Src family kinases. 12. Wang H, Lillehoj EP, Kim KC. Identification of four sites of stimulated Mol Cell Biol 2003;23:2883 – 92. tyrosine phosphorylation in the MUC1 cytoplasmic tail. Biochem Biophys Res 31. Salgia R, Pisick E, Sattler M, et al. p130CAS forms a signaling complex with Commun 2003;310:341 – 6. the adapter protein CRKL in hematopoietic cells transformed by the BCR/ABL 13. Zrihan-Licht S, Baruch A, Elroy-Stein O, Keydar I, Wreschner DH. Tyrosine oncogene. J Biol Chem 1996;271:25198 – 203. phosphorylation of the MUC1 breast cancer membrane proteins. Cytokine 32. Singh PK, Wen Y, Swanson BJ, et al. Platelet-derived growth factor receptor receptor-like molecules. FEBS Lett 1994;356:130 – 6. h-mediated phosphorylation of MUC1 enhances invasiveness in pancreatic 14. Li Y, Ren J, Yu W, et al. The epidermal growth factor receptor regulates adenocarcinoma cells. Cancer Res 2007;67:5201 – 10. interaction of the human DF3/MUC1 carcinoma antigen with c-Src and h-catenin. 33. Pandey P, Kharbanda S, Kufe D. Association of the DF3/MUC1 breast J Biol Chem 2001;276:35239 – 42. cancer antigen with Grb2 and the Sos/Ras exchange protein. Cancer Res 1995;55: 4000 – 3. 15. Schroeder JA, Thompson MC, Gardner MM, Gendler SJ. Transgenic MUC1 interacts with epidermal growth factor receptor and correlates with mitogen- 34. Schroeder JA, Adriance MC, Thompson MC, Camenisch TD, Gendler SJ. activated protein kinase activation in the mouse mammary gland. J Biol Chem MUC1 alters h-catenin-dependent tumor formation and promotes cellular 2001;276:13057 – 64. invasion. Oncogene 2003;22:1324 – 32. 16. Ren J, Li Y, Kufe D. Protein kinase C y regulates function of the DF3/MUC1 35. Raina D, Kharbanda S, Kufe D. The MUC1 oncoprotein activates the anti- carcinoma antigen in h-catenin signaling. J Biol Chem 2002;277:17616 – 22. apoptotic phosphoinositide 3-kinase/Akt and Bcl-xL pathways in rat 3Y1 fibroblasts. J Biol Chem 2004;279:20607 – 12. 17. Gonzalez-Guerrico AM, Cafferata EG, Radrizzani M, et al. Tyrosine kinase c-Src constitutes a bridge between cystic fibrosis transmembrane regulator 36. Luscinskas FW, Gimbrone MA, Jr. Endothelial-dependent mechanisms channel failure and MUC1 overexpression in cystic fibrosis. J Biol Chem 2002; in chronic inflammatory leukocyte recruitment. Annu Rev Med 1996;47: 277:17239 – 47. 413 – 21. 18. Li Y, Kuwahara H, Ren J, Wen G, Kufe D. The c-Src tyrosine kinase 37. Sorlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of breast regulates signaling of the human DF3/MUC1 carcinoma-associated antigen with carcinomas distinguish tumor subclasses with clinical implications. Proc Natl GSK3 h and h-catenin. J Biol Chem 2001;276:6061 – 4. Acad Sci U S A 2001;98:10869 – 74. 19. Hilkens J, Wesseling J, Vos HL, et al. Involvement of the cell surface-bound 38. Fan C, Oh DS, Wessels L, et al. Concordance among gene-expression-based mucin, episialin/MUC1, in progression of human carcinomas. Biochem Soc Trans predictors for breast cancer. N Engl J Med 2006;355:560 – 9. 1995;23:822 – 6. 39.Magarian-BlanderJ,CiborowskiP,HsiaS,WatkinsSC,FinnOJ. 20. Rahn JJ, Shen Q, Mah BK, Hugh JC. MUC1 initiates a calcium signal after Intercellular and intracellular events following the MHC-unrestricted TCR ligation by intercellular adhesion molecule-1. J Biol Chem 2004;279:29386 – 90. recognition of a tumor-specific peptide epitope on the epithelial antigen MUC1. J Immunol 1998;160:3111 – 20. 21. Feldner JC, Brandt BH. Cancer cell motility—on the road from c-erbB-2 receptor steered signaling to actin reorganization. Exp Cell Res 2002;272:93 – 108. 40. Thompson EJ, Shanmugam K, Hattrup CL, et al. Tyrosines in the MUC1 cytoplasmic tail modulate transcription via the extracellular signal- 22. Piccolo E, Innominato PF, Mariggio MA, Maffucci T, Iacobelli S, Falasca M. regulated kinase 1/2 and nuclear factor-nB pathways. Mol Cancer Res 2006; The mechanism involved in the regulation of phospholipase Cg1 activity in cell 4:489 – 97. migration. Oncogene 2002;21:6520 – 9. 41. Dolmetsch RE, Xu K, Lewis RS. Calcium oscillations increase the efficiency 23. Playford MP, Schaller MD. The interplay between Src and integrins in and specificity of gene expression. Nature 1998;392:933 – 6. normal and tumor biology. Oncogene 2004;23:7928 – 46. 42. Fu L, Lin-Lee YC, Pham LV, Tamayo A, Yoshimura L, Ford RJ. Constitutive 24. Hall A. Rho GTPases and the actin cytoskeleton. Science 1998;279:509 – 14. NF-nB and NFAT activation leads to stimulation of the BLyS survival pathway in 25. Van Aelst L, D’Souza-Schorey C. Rho GTPases and signaling networks. aggressive B-cell lymphomas. Blood 2006;107:4540 – 8. Genes Dev 1997;11:2295 – 322. 43. Kaplan KB, Bibbins KB, Swedlow JR, Arnaud M, Morgan DO, Varmus HE. 26. Nobes CD, Hall A. Rho GTPases control polarity, protrusion, and adhesion Association of the amino-terminal half of c-Src with focal adhesions alters their during cell movement. J Cell Biol 1999;144:1235 – 44. properties and is regulated by phosphorylation of tyrosine 527. EMBO J 1994;13: 4745 – 56. 27. Rottner K, Hall A, Small JV. Interplay between Rac and Rho in the control of substrate contact dynamics. Curr Biol 1999;9:640 – 8. 44. Schaller MD, Parsons JT. pp125FAK-dependent tyrosine phosphorylation of paxillin creates a high-affinity binding site for Crk. Mol Cell Biol 1995;15: 28. Kiyokawa E, Hashimoto Y, Kobayashi S, Sugimura H, Kurata T, Matsuda M. 2635 – 45. Activation of Rac1 by a Crk SH3-binding protein, DOCK180. Genes Dev 1998; 12:3331 – 6. 45. Nishihara H, Maeda M, Oda A, et al. associates with CrkL and 29. Posern G, Zheng J, Knudsen BS, et al. Development of highly selective SH3 regulates Rac1 in human leukemia cell lines. Blood 2002;100:3968 – 74. binding peptides for Crk and CRKL which disrupt Crk-complexes with 46. Oosawa Y, Imada C, Furuya K. Temperature dependency of calcium DOCK180, SoS and C3G. Oncogene 1998;16:1903 – 12. responses in mammary tumour cells. Cell Biochem Funct 1997;15:113 – 7.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research. MUC1 Initiates Src-CrkL-Rac1/Cdc42−Mediated Actin Cytoskeletal Protrusive Motility after Ligating Intercellular Adhesion Molecule-1

Qiang Shen, Jennifer J. Rahn, Jing Zhang, et al.

Mol Cancer Res 2008;6:555-567.

Updated version Access the most recent version of this article at: http://mcr.aacrjournals.org/content/6/4/555

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2021/03/10/6.4.555.DC1

Cited articles This article cites 45 articles, 27 of which you can access for free at: http://mcr.aacrjournals.org/content/6/4/555.full#ref-list-1

Citing articles This article has been cited by 7 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/6/4/555.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://mcr.aacrjournals.org/content/6/4/555. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from mcr.aacrjournals.org on September 28, 2021. © 2008 American Association for Cancer Research.