The c-Myc Target Glycoprotein1bα Links Cytokinesis Failure to Oncogenic Signal Transduction Pathways in Cultured Human Cells

An increase in chromosome number, or polyploidization, is associated with a variety of biological changes including breeding of cereal crops and flowers, terminal differentiation of specialized cells such as megakaryocytes, cellular stress and oncogenic transformation. Yet it remains unclear how cells tolerate the major changes in gene expression, chromatin organization and chromosome segregation that invariably accompany polyploidization. We show here that cancer cells can initiate increases in chromosome number by inhibiting cell division through activation of glycoprotein1b alpha (GpIbα), a component of the c-Myc signaling pathway. We are able to recapitulate cytokinesis failure in primary cells by overexpression of GpIbα in a p53-deficient background. GpIbα was found to localize to the cleavage furrow by microscopy analysis and, when overexpressed, to interfere with assembly of the cellular cortical contraction apparatus and normal division. These results indicate that cytokinesis failure and tetraploidy in cancer cells are directly linked to cellular hyperproliferation via c-Myc induced overexpression of GpIbα.


Introduction
The transition from the restrained and controlled growth of normal cells to the accelerated and dysregulated growth of cancer cells requires multiple changes, including enhancement of the signaling pathways controlling division and survival. But additional changes not directly related to increased proliferation usually accompany these cellular alterations. These include genetic instability (GI), aneuploidy, and centrosome amplification, all of which are associated with a loss of genomic integrity [1,2,3,4]. The reason the two phenotypes of enhanced growth and GI so often appear together is currently unknown. It is commonly believed that GI imparts a ''mutator'' phenotype to the cancer cells, increasing the genetic diversity necessary for the selection of mutant clones with enhanced growth and survival [5]. But since GI is strongly associated with senescence and apoptosis [6,7,8], it is unclear how cells tolerate the deleterious effects of GI long enough for these cellular evolutionary steps to occur. It is also unclear whether the mechanisms that cause polyploidization are directly related to the signals that cause enhanced growth or whether they are an indirect consequence of elevated proliferation rates.
Two key, and related, genomic destabilizing events that are believed to contribute to cancer are tetraploidization, the doubling of the chromosome number, and centrosomal amplification, which increases the number of microtubule organizing centers in the cell. It has long been believed that tetraploidy is an important intermediate in cellular transformation, as cancer cells typically have increased chromosome numbers [1,9,10]. More recently, tetraploidy has been directly linked to tumorigenesis in mice [11,12], and centrosome amplification has been linked to tumor growth in flies [13]. But in both of these model systems, tetraploidy and centrosome amplification were artificially induced by mechanisms not directly associated with carcinogenesis. The root cause of tetraploidy and centrosome amplification in cancer cells therefore remain mostly uncharacterized.
One of the classic oncoproteins that enhance growth and proliferation of cancer cells is the transcription factor c-Myc. Highly overexpressed in malignant cells, c-Myc modifies a variety of processes including cell proliferation, differentiation, survival, GI and metabolism [14]. Overexpression of c-Myc is sufficient for acute transformation of immortalized rodent cell lines, allowing them to become tumorigenic in immunocompromised mice. One of the many targets of c-Myc transcriptional regulation is GpIba, a subunit of the von Wilebrand factor receptor (vWFR) that is responsible for the adhesion, aggregation and activation of platelets upon binding to damaged epithelium [15,16]. Recent data shows that GpIba has additional functions that are independent of the blood-clotting pathway but are linked to c-Myc mediated transformation and induction of GI. These include reducing the need for growth factors, inhibiting apoptosis, causing DNA and nuclear damage, promoting tetraploidy and transforming immortalized cells [12,17]. GpIba is also necessary to promote tetraploidy by c-Myc activation and is sufficient to do this in the absence of overt c-Myc deregulation [17].
To understand in more detail the role of GpIba in promoting GI, we have identified the genomic-destabilizing events associated with GpIba overexpression. We show here that GpIba localizes to the cleavage furrow of dividing primary cells and that overexpression of GpIba interferes with the correct localization of key divisional proteins at the cleavage furrow associated with failure of cytokinesis or cell division. These observations provide the first direct mechanistic link between stimulation of cell proliferation and transformation, via the c-Myc signaling pathway, and the genomic destabilizing events of polyploidization and centrosomal amplification.

GpIba overexpression caused failure of cytokinesis
GpIba is widely overexpressed in a variety of tumors and tumor cell lines and GpIba overexpression gives rise to tetraploidy in primary human foreskin fibroblasts (HFF; [12,17]. To determine if GpIba overexpression was the cause of nuclear amplification in cancer cells, GpIba was stably knocked down by a short hairpin RNA in HeLa, OS osteosarcoma, and MCF7 breast cancer cell lines ( Figure 1A and Figure S1A and S1B). The frequency of multinucleates (an example is shown in Figure S1C, a common result of cytokinesis failure), was markedly (p,0.05) reduced in HeLa and OS cell lines, and moderately (p,0.1) reduced in the MCF7 cell line ( Figure 1B and C). In addition, the frequency of multipolar spindles (MPS, an example is shown in Figure S1D), a hallmark of centrosome amplification, was also significantly (p, 0.05) reduced in HeLa and OS cells, and moderately (p,0.1) reduced in MCF7 cells. Many other mitotic and cytokinesis defects including anaphase bridges, lagging chromosomes and micronuclei, demonstrated similar trends after GpIba knockdown in tested cancer cells ( Figure 1B and C), showing that overexpression of GpIba is a significant cause of cytokinesis failure and mitotic defects in malignant cells. Furthermore, these results were validated by expressing a murine shRNA-resistant GpIba (mGpIba) in HeLa-shGpIba cells and as expected, we observed increases in mitotic and cytokinesis defects compared with control shGpIba cells (vector alone) showing the specificity of the knockdown phenotype ( Figure 1B).
We next examined whether overexpression of GpIba was sufficient to impair cytokinesis in noncancer primary cells. A series of HFF cells stably immortalized with human telomerase (hTERT) were used for this study, including HFF-vector (stably transfected with empty vector), HFF-shp53 (p53 stably knocked down by a short hairpin RNA), HFF+GpIba (stably overexpressing GpIba), and HFF-shp53+GpIba (stably overexpressing GpIba with p53 knockdown) [12]. The frequency of binucleates in interphase cells increased markedly when GpIba was overexpressed, but only in a p53 knockdown background ( Figure 1D), consistent with previous findings [12]. To confirm that the binucleation was due to cytokinesis failure, we observed the division of .300 cells by livecell differential interference contrast (DIC) microscopy. Cytokinesis failure was seen in approximately 2% of the vector-alone cells, shp53 or GpIba overexpressing cells. However, failure of division increased by .4-fold in HFF-shp53+GpIba cells ( Figure 1E and Movies S1 and S2). These results show that overexpression of GpIba is sufficient to lead to cytokinesis failure in immortalized primary cells lacking p53 and provide an explanation for the increased multinucleation and ploidy of cancer cells.

GpIba colocalizes with F-actin at the cleavage furrow during cytokinesis
To determine if GpIba plays a role in cell division, its localization was evaluated in dividing HFF-hTERT cells by immunofluorescence. In late mitosis, endogenous GpIba concentrated at the contractile ring in the midzone of the dividing cell, co-localizing with F-actin, filamin A, and myosin heavy chain (MHC, Figure 2A). This is notably different from the ER localization of GpIba described previously in interphase cells where GpIba is distributed diffusely throughout the cytoplasm and in association with the ER ( [18] and Figure S2A). As a control, another membrane-associated marker, CD44, did not concentrate at the cleavage furrow ( Figure  S2B), showing the cleavage furrow enrichment is specific to a subset of membrane-associated proteins. To examine the changing dynamics of GpIba positioning, GFP was fused to the C-terminus of GpIba and the fusion protein was transiently expressed in HeLa cells, which tolerated the expression better than primary cells. As observed with immunolocalization in primary cells, GpIba-GFP concentrated at the cleavage furrow in mitotic HeLa cells ( Figure 2B and Movie 3), thus confirming that GpIba associated with contractile structures of the cell during division. However, unlike primary cells GpIba-GFP staining was only observed in a fraction of the dividing HeLa cells (discussed further below). More diffuse cytoplasmic staining was seen in interphase cells, consistent with the previously described ER localization [12]. GpIba-GFP also partially colocalized with F-actin fibers near the cell cortex of interphase cells ( Figure 2C) indicating an association with the actin cytoskeletal in nondividing cells.
GpIba overexpression causes mislocalization of filamin A, F-actin, MHC and RhoA from the contractile ring As we documented real-time divisional failure in GpIbaoverexpressing cells ( Figure 1E), we also observed defects of the cortical structure of the cells consistent with abortive contraction. One such defect was membrane blebbing seen by immunofluorescence with antibodies to MHC or F-actin ( Figure 3A, arrows). Similar blebbing structures have been seen with failed cytokinesis from other sources [19,20]. We also noted polar contraction, defined as cortical contraction and F-actin and myosin accumulation outside of the cleavage furrow during division ( Figure 3A). Both blebbing and polar contraction are abnormal features and were only rarely observed in HFF-vector cells, but were found ,30% of the HFF-shp53+GpIba cells ( Figure 3B). When these aberrant divisional structures and processes formed, they typically contained both F-actin and GpIba, as shown in Figure 3C for blebbing in the top panel and polar contraction in the bottom panel. Small molecule inhibition (ML-7) of the signaling protein myosin light chain kinase blocked cytokinesis but did not lead to blebbing or polar contraction confirming that these are symptoms of contractile or abscission defects and are not found in all cases of cytokinesis failure. These observations suggest that the actomyosin contractile cytoskeletal organization and function in dividing cells is defective following overexpression of GpIba.
To examine in more detail the molecular nature of the divisional defect, the localization of a variety of divisional proteins were examined in GpIba-overexpressing cells. Several key cytokinesis proteins were missing in a subset of the dividing cells. Surprisingly, GpIba itself was missing from the contractile rings in about 60% of anaphase HFF-shp53+GpIba cells ( Figure 4A). (Comparable results were observed by live cell imaging of HeLa cells stably ). (D) HFF-hTERT cells with stable knockdown of p53 and/or overexpression of GpIba were stained with DAPI and the frequency of binucleated cells were determined by fluorescent microscopy (n = 300-500 cells per sample). Note that cancer cells occasionally have more than two nuclei and cells with two or more nuclei were categorized in Figure 1B and 1C as ''multinucleated''. HFF-hTERT cells very rarely had more than two nuclei and cells with two nuclei were categorized in Figure 1D  transfected with the GpIba-GFP described above.) Similarly, filamin A, F-actin and MHC were also often absent from the cleavage furrow of dividing HFF-shp53+GpIba cells, while the interphase localizations of filamin A and F-actin were not affected ( Figure 4B, C, D and G). We believe that these cytokinesis protein mislocalizations were related to the abnormal divisional structures described above. Fully 80.8% of HFF-shp53+GpIba cells with abnormal filamin A localization showed blebbing during division. Filamin A deficiencies have been previously shown to cause blebbing during cell locomotion [21]. The cytokinesis activator, RhoA, was often asymmetrically localized in HFF-shp53+GpIba cells, with stronger staining at one edge of the furrow ( Figure 4E). In contrast, the mitotic signaling kinase Aurora B was normally positioned at the cleavage furrow in dividing HFF-shp53+GpIba cells ( Figure 4F), showing that the mislocalization was specific to a subset of cytokinesis proteins.
The above studies showed that GpIba overexpression resulted in the mislocalization of key divisional proteins from the cleavage furrow of dividing primary cells. We next determined if the localization of the same proteins was compromised in cancer cells. The distribution of GpIba, F-actin and filamin A during cytokinesis in four cancer cell lines including HeLa, liver adenocarcinoma SK-HEP-1 and oral squamous cell carcinoma derived UPCI:SCC40 and UPCI:SCC103 was examined by immunofluorescence. All of the tested cancer cell lines showed frequent mislocalization of these divisional markers, similar to HFF-shp53+GpIba cells in Figure 4, and we observed a correlation between the frequency of binucleated/multinucleated cells and the frequency of GpIba, F-actin and filamin A mislocalization ( Figure 5A). These observations show that the mislocalization of cytokinesis proteins seen with GpIba-overexpression in primary cells can also be seen in malignant cells and is associated with failure of division. To determine if a reduction of GpIba was able to reverse the marker mislocalization in cancer cells, we compared localization of filmain A and F-actin in HeLa cells before and after shRNA knockdown of GpIba ( Figure S3). In both cases a small decrease was observed, but this was significant only for Filamin A (p = 0.016). We interpret these results to indicate that GpIba overexpression does play a role in cytokinesis failure and divisional protein mislocalization in cancer cells, but that additional unknown factors may be acting to interfer with cytokinesis protein localization.

Signal peptide and filamin A binding domains of GpIba are indispensible for GpIba-overexpression mediated cytokinesis failure
We further explored which domains of GpIba were important for inhibition of cytokinesis. One region of interest was the filamin A binding domain to test the significance of interactions of GpIba with this actin modifying protein. A second region of interest was the signal peptide domain to determine if transit through the secretory pathway was required for overexpressed GpIba to inhibit cytokinesis. Cellular fractionation was used to verify the mutant lacking the signal peptide was unable to localize to the ER ( Figure S4). When overexpressed, these mutants were much less effective at increasing the binucleation frequency observed in DAPI-stained HFF-hTERT cells ( Figure 5B), or the frequency of cytokinesis failure viewed by live-cell DIC microscopy ( Figure 5C). These results indicated that interference with cytokinesis required that the overexpressed GpIba be capable of entering the ER secretory pathway and binding to filamin A, thus further supporting the conclusion that GpIba overexpression inhibits cytokinesis by interfering with the cortical F-actin filament network. These findings are consistent with previous observations showing that GpIba-induced GI was abrogated by loss of either the filamin A-binding domain or signal peptide of GpIba [17].

GpIba overexpression is responsible for transformationrelated features of cancer cells
It is conventionally believed that tetraploidy resulting from cytokinesis failure is an intermediate step towards tumorigenesis [11,12]. As our data have demonstrated that GpIba overexpression led to cytokinesis failure and tetraploidization, we next investigated whether GpIba overexpression was required for the elevated growth rates and tumorigenic properties of cancer cells.
When endogenous GpIba in tumor cells was knocked down, we observed markedly reduced clonogenicity in soft agar compared with controls, even when high serum concentrations were maintained or the periods of culture were extended to compensate for possible reduced rates of proliferation ( Figure 6A and data not shown). Moreover, the colonies that did arise from shRNA cells were invariably of much smaller overall size ( Figure 6B). These results were validated by expression of a shRNA resistant murine mGpIba which restored enhanced growth showing the specificity of the shRNA knockdown ( Figure S5). We therefore conclude that anchorage-independent growth of the tested cancer cells lines was profoundly influenced by endogenous GpIba levels.
Furthermore, when we tested each shRNA cell line and its control counterpart by inoculating immunocompromised nu/nu mice with equivalent numbers of cells, we found that in all three cases, endogenous GpIba knockdown resulted in a significant impairment of tumor growth ( Figure 6C). Collectively, we conclude that GpIba overexpression is responsible for hyperproliferation and tumorigenesis of the tested cancer cells.

Discussion
This study makes two advances towards understanding GI in tumor cells. The first is to establish a mechanism for cytokinesis failure in cancer cells and the second is to link cytokinesis failure mechanistically with enhanced growth and proliferation via c-Myc.
Increased GpIba expression, a common feature of tumor cells [18], was shown to contribute to cytokinesis failure in the tested cancer cell lines. Furthermore, we were able to establish a working model for how tetraploidy originates in cancer cells by overexpression of GpIba and p53 inhibition in immortalized primary cells. In these cells, cytokinesis failure was accompanied by the appearance of abnormal contractile structures and mislocalization of essential divisional proteins, including F-actin, filamin A and RhoA. Similar mislocalization of F-actin, filamin A and GpIba could be seen in tumor-derived cells and in each case was correlated with the appearance of multinucleation, a feature of cytokinesis failure. These results show for the first time that the genomic destabilizing event of cytokinesis failure in cancer cells can be defined at the molecular level and reproduced with similar phenotypes in primary cells.
The phenotypes from GpIba overexpression in primary cells were markedly more severe in the absence of p53. It has been observed previously that loss of this genomic checkpoint protein facilitates c-Myc induced tetraploidy [22] and promotes survival of the cells following genomic damage from GpIba overexpression [12]. Similarly, the loss of p53 may be required here to bypass cellular checkpoints that otherwise inhibit abnormal cytokinesis in primary cells, although this explanation alone is insufficient to explain the protein mislocalization we see from p53 knockdown. Additionally, while a reduction of GpIba led to some normalization of cytokinesis protein localization in cancer cells, GpIba, F-actin and filamin A remained mislocalized in many HeLa cells after knockdown of GpIba demonstrating that other factors also interfere with cytokinesis protein positioning in these cells.
Inhibition of cytokinesis by GpIba overexpression requires filamin A binding and we have shown that filamin A localizes to the mammalian cleavage furrow in a GpIba-dependent manner. Previously, filamin A was found in chick embryonic cells at the cleavage furrow [23]. Filamin A is known to bind to GpIba as part of the vWFR signaling pathway [24,25], and we propose that GpIba-filamin A interactions are also important for cell division. Filamin A homodimerizes at a flexible hinge and crosslinks polymerized actin into a 3-dimensional gel, promoting F-actin networks rather than the anti-parallel arrays associated with contractile fibers in skeletal sarcomeres [21,26]. It may therefore seem surprising that filamin A function would be important for contractile mechanisms in cytokinesis. However, the contractile forces at the cleavage furrow have also been proposed to result from disordered actin arrays [27,28], and we propose that filamin A crosslinking of F-actin may be an important part of that process. Filamin A binds RhoA [29] that was also mislocalized in GpIba- overexpressing cells. It is controversial whether RhoA activity is essential for formation of the cleavage furrow [19,30], but it is important to activate formin leading to actin polymerization during division [31] and is thought to be required for cortical contraction [32]. Deficiency in RhoA also leads to blebbing as we observe in GpIba-overexpressing cells [19]. Thus, GpIba-induced interference with filamin A and RhoA positioning/function may explain the actomyosin and contractile deficiencies we observe in GpIba-overexpressing cells.
The observation that a source of cytokinesis failure in cancer cells is a target of the c-Myc pathway that stimulates growth and division, could help explain the linkage between oncogenic transformation and tetraploidy. Since both pathways are activated concurrently, by the same molecular changes, it is logical that they would be found together in the same cells. Furthermore, this may help explain how the cell tolerates the mitotic disruption of centrosome and chromosome amplification. We propose that a tight phenotypic linkage between the cause of cytokinesis failure and stimulated cell growth and proliferation offsets the intrinsic cost of abnormal division and polyploidy on cell survival. This may allow the cells to thrive despite the selective disadvantage of GI. In this model, we interpret cytokinesis failure and tetraploidy to be linked to enhanced cellular proliferation, not as a direct consequence of proliferation, but because both processes are induced by c-Myc activation and GpIba overexpression.
We have demonstrated a role for GpIba in cytokinesis and the increases in ploidy common in cancer cells. Previously, GpIba was known for its role as a platelet-and megakaryocyte-specific cell surface receptor [15,16]. Its presence on other cell lineages, where there is little or no expression of the other vWFR components, raises interesting questions regarding the origin and functionality of this subunit. It is possible that the original function of GpIba was related to cytokinesis, and only later did it evolve into other specialized roles in platelets and megakaryocytes. It is also possible that some of its known functions in platelets and megakaryocytes could be related to a role in cytokinesis. During maturation megakaryocytes undergo endomitosis, several rounds of mitosis without cell division [33]. Recent evidence shows that endomitosis in megakaryocytes involves a failure of the contractile ring [34]. Both endomitosis and aborted division in cancer cells may utilize similar pathways involving GpIba, although further investigations will be required to test this hypothesis.

GpIba shRNA knockdown
Retroviral vectors (pHUSH) encoding human GpIba shRNA 29-mers and a puromycin-selectable cassette (Cat. Numbers TR12692) were obtained from Origene, Inc. (Rockville, MD) and 100 units/ml Penicillin G + 100 mg/ml Streptomycin as previous described [35]. Retroviral transfections were performed in Phoenix-A cells as previously described using Superfect (Qiagen, Chatsworth, CA; [36]. Phoenix A supernatants were then harvested daily beginning 48 hr after transfection with retroviral vectors, filtered by passage through 0.45 mM filters (Millipore, Bedford, NY) and applied to cancer cell line monolayers for 24 hr in the presence of 8 mg/ml Polybrene (Sigma-Aldrich, St. Louis, MO). After 2-3 applications, cells were cultured in fresh, virus-free medium for 48 hr followed by selection in puromycin-containing medium (1 mg/ml; Sigma-Aldrich). Puromycin-resistant colonies were pooled for all subsequent studies and were intermittently maintained in puromycin-containing medium.
Plasmid and DNA transfections 2610 5 cells were seeded on 22622 mm glass coverslips (VWR) in 6-well plates and incubated with pre-warmed OPTI-MEM (Invitrogen) medium. After six hours, cells were transfected with 2 mg of plasmid using 6 ml of the FuGENE6 transfection reagent (Roche Diagnostics) following the manufacture's protocol. Fresh medium was added 12 hours later. Cells were examined 24-48 hours after transfection.

Immunofluorescence
Cells on coverslips were fixed in 4% paraformaldehyde at room temperature and washed in PBS. 0.1% Triton X-100 was used to permeabilize the cells and 1.5% BSA/PBS was used as blocking solution. Various primary antibodies were used including rAb- MHC (Sigma, 1:500), mAb-actin (Cytoskeleton, 1:100), mAbfilamin (a gift from Dr. Nakamura, Translational Medicine Division, Brigham and Women's Hospital, Boston, MA, 1:500), ratAb-Gp1ba (Emfret, 1:100), mAb-RhoA (Santa Cruz Biotechnology, 1:100), and mAb-CD44 (BD Pharmingen, 1:1000). All primary antibodies were diluted in the blocking solution and incubated for 30 minutes at room temperature. Fluorescent labeled goat anti-rabbit or anti-mouse or anti-rat IgG (Invitrogen, 1:500) were diluted in the blocking solution as secondary antibodies. After PBS wash, cells were incubated with the desired secondary antibody for 30 minutes at room temperature followed by staining with 4,6-diamidino-2-phenylindole (DAPI) at 1 mg/ml (Sigma) for 5 minutes. The coverslips were mounted and examined by Olympus BX60 epifluorescence microscope with 1006 oil immersion objectives. Hamamatsu Argus-20 CCD camera was used to capture the images. Confocal microscopy was performed using Nikon Eclipse E800 (Nikon) with BioRad Radiance 2000 system.

Live microscopy analysis
2610 5 cells were seeded on 35 mm glass-bottom Petri dishes (MatTek Corporation) and subject to live cell imaging either after transfection with desired DNA plasmids as described above or without any treatment. Cells were videoed while being maintained at 37uC with a moisturized-warm air microscope chamber (Life Imaging Services, Reinach, Switzerland). DIC microscopy and epifluorescence microscopy were performed on Nikon Eclipse TE2000-U inverted microscope with Coolsnap HQ digital camera (Roper Scientific Photometrics). Images were taken and analyzed using MetaMorph software (Molecular Devices).

In vivo tumorigenesis studies
All studies were reviewed and approved by The University of Pittsburgh's Institutional Animal Use and Care Committee. 6-8 wk old nu/nu mice were purchased from Harland Laboratories (Indianapolis, IN). They were housed under sterile, germ-free conditions with 12 hr day-night cycles and were allowed access to feed and water ad libitum. Animals were inoculated subcutaneously in the flank with 10 7 tumor cells that had been trypsinized, washed, and immediately resuspended in PBS. They were monitored at least twice weekly and tumor volumes were calculated as previously described [12].

Real time qRT-PCR-RNA
Extraction was performed as previously described followed by treatment with TurboDNAse as recommended by the supplier (Ambion, Austin TX) [37]. qRT-PCR was performed using a QuantiTect SYBR Green RT-PCR kit according to the directions of the supplier (Qiagen) and as previously described. Primers for the detection of human GpIba were identified using the Primer3 program (www.frodo.wi.mit.edu/). They consisted of the sequences between nt 781 (forward) and 91 (reverse) of the transcript (GenBank Accession no. NM_000173) and were synthesized by IDT (Coralville, IA). Cycling was performed on triplicate samples on a Roche LightCycler 2.0 apparatus (Roche Diagnostics, Indianapolis, IN) and values were adjusted to those obtained for GAPDH qRT-PCR reactions performed in parallel. showing reduced expression of GpIba in shRNA lines versus control lines. As a control, cells were also stained with calnexin (green) and with DAPI (blue) as previously described [21].  Figure S3 Changes in filamin A, and F-actin localization after GpIba knockdown in HeLa cells. The frequency of protein mislocalization following stable shGpIba transfection in HeLa cells was determined by immunofluorescence. A modest, but statistically significant, restoration of filamin A localization was observed indicating that GpIba overexpression contributes to mislocalization in these cancer cells. But other unknown factors are apparently also controlling cytokinesis protein mislocalization in malignant cells. Standard error about the means is shown.

Supporting Information
Found at: doi:10.1371/journal.pone.0010819.s003 (0.07 MB TIF) Figure S4 GpIba mutants lacking signal peptide is truly defective in localizing to ER. Top two panels: Western blotting shows successful cellular fractionation to separate ER proteins and non-ER proteins. Calnexin: an ER protein marker. Middle two panels: wild-type GpIba was found in both ER and non-ER fractions, while signal peptide-deleted GpIba was only found in non-ER fraction. Bottom panel: loading control beta-tubulin. Movie S1 An example of cytokinesis failure. DIC live-cell imaging microscopy was used to visualize the cell division. Note: a binucleated (tetraploid) cell was generated after cytokinesis failure.