Promyelocytic Leukemia (PML) Protein Plays Important Roles in Regulating Cell Adhesion, Morphology, Proliferation and Migration

PML protein plays important roles in regulating cellular homeostasis. It forms PML nuclear bodies (PML-NBs) that act like nuclear relay stations and participate in many cellular functions. In this study, we have examined the proteome of mouse embryonic fibroblasts (MEFs) derived from normal (PML+/+) and PML knockout (PML−/−) mice. The aim was to identify proteins that were differentially expressed when MEFs were incapable of producing PML. Using comparative proteomics, total protein were extracted from PML−/− and PML+/+ MEFs, resolved by two dimensional electrophoresis (2-DE) gels and the differentially expressed proteins identified by LC-ESI-MS/MS. Nine proteins (PML, NDRG1, CACYBP, CFL1, RSU1, TRIO, CTRO, ANXA4 and UBE2M) were determined to be down-regulated in PML−/− MEFs. In contrast, ten proteins (CIAPIN1, FAM50A, SUMO2 HSPB1 NSFL1C, PCBP2, YWHAG, STMN1, TPD52L2 and PDAP1) were found up-regulated. Many of these differentially expressed proteins play crucial roles in cell adhesion, migration, morphology and cytokinesis. The protein profiles explain why PML−/− and PML+/+ MEFs were morphologically different. In addition, we demonstrated PML−/− MEFs were less adhesive, proliferated more extensively and migrated significantly slower than PML+/+ MEFs. NDRG1, a protein that was down-regulated in PML−/− MEFs, was selected for further investigation. We determined that silencing NDRG1expression in PML+/+ MEFs increased cell proliferation and inhibited PML expression. Since NDRG expression was suppressed in PML−/− MEFs, this may explain why these cells proliferate more extensively than PML+/+ MEFs. Furthermore, silencing NDRG1expression also impaired TGF-β1 signaling by inhibiting SMAD3 phosphorylation.


Introduction
PML protein has been extensively studied because of its involvement in the etiology of Acute Promyelocytic Leukemia (APL) [1]. Approximately 95% of human APL contains the homologous recombination between chromosome 15 and 17 [2]. It involves a translocation between the PML gene in the breakpoint region on chromosome 15q22 and the partner gene Retinoic Acid Receptor alpha (RARa) located on chromosome 17q21. The chimeric proteins formed, PML-RARa and/or RARa-PML, are thought to be responsible for activating the oncogenic events associated with the transformation to the APL phenotype [3]. PML is one of the key components involved in the regulation of numerous important biological functions through its ability to inhibit the ubiquitination and proteasomal degradation processes. PML monitors the activation of p53/ TP53 via phosphorylation in the nucleolus following DNA damage and participates in neoangiogenesis and tumor vascularization. It also plays an important role in regulating gene transcription, and other nuclear events, including repression of cell-cycle progression, regulation of cellular senescence, cell death and neurodevelopment [4, 5, 6 & 7]. It has been reported that pRB, p53, mTOR, cJun, Akt, Daxx and EIf4e are all capable of interacting with PML to regulate homeostasis in immune response, to repress cell proliferation and to regulate protein synthesis [8 & 9]. Nevertheless, it is still unclear whether PML is capable of interacting with other proteins during cellular processes.
PML is a Ringer-finger protein that belongs to the tripartite motif family of proteins [10]. In human, there are seven splice variants of PML and all of them perform a variety of distinct functions [7]. Recently, it has been demonstrated that PML variant II is one of the most abundant and plays a key role in regulating the function and structure of PML-NBs [11]. In this study, we have used comparative proteomics to elucidate all the proteins that are differentially expressed in MEFs that are incapable of expressing PML [12]. In addition, we have also examined how the absence of PML affected cell adhesion, morphology, proliferation, migration and gene expression.

Comparison of PML +/+ and PML 2/2 MEFs Morphology
We first confirmed the status of PML +/+ and PML 2/2 MEFs by Western blotting and RT-PCR. The results confirmed that the PML 2/2 MEFs did not express PML transcripts and protein (Figs. 1A & B). We then estimated the average size of PML +/+ MEFs, 8 and 32 hrs after culture, to be 1.160.03 and 1.660.043 mm 2 , respectively. For PML 2/2 MEFs, the average size after 8 and 32 hr incubations were 0.4860.023 and 0.9860.036 mm 2 , respectively (Fig. 2C). The results indicated that the PML 2/2 MEFs were significantly smaller than PML +/+ MEFs. Hematoxylin and eosin staining showed that there was no obvious morphological differences between PML +/+ ( Fig. 2A) and PML 2/2 (Fig. 2B) MEFs under the light microscope. However, under the SEM, the cell surface of PML +/+ MEFs were found to contain distinct elongated membranous projections. These projections irradiated from the edges of the cells and were attached on the culture plate (Fig. 2D). Similar membrane projections were not found on the cell surface of the PML 2/2 MEFs; instead, the cell surface was covered by extracellular matrix proteins (Fig. 2E). We also cultured both MEFs on gelatin-coated culture dishes and then stained the MEFs with Phalloidin dye to show the arrangement of the cytoskeleton. Phalloidin staining revealed that filamentous actin was highly organized in both types of MEFs. However, the staining intensity that we produced from PML 2/2 MEFs (Fig. 2F) were always weaker than PML +/+ MEFs (Fig. 2G). This may be attributed to PML 2/2 MEFs containing fewer filamentous actin.
Differences in Cellular Adhesions between PML +/+ and PML 2/2 MEFs We measured the efficacy of PML +/+ and PML 2/2 MEFs to adhere to plastic culture dishes 30, 60, 90 and 120 mins after plating. The aim was to establish whether there were any differences in the cell adhesion properties between PML +/+ and PML 2/2 MEFs. We established that there were significantly fewer PML 2/2 MEFs adhering to the culture dish than PML +/ + MEFs, when examined, 30 and 60 mins after plating. The PML 2/2 MEFs required at least 90 min for cells to fully coat the entire surface of the plastic dish compared with 30 min for PML +/+ MEFs (Figs. 3A-D). We also stained the MEFs, after different time intervals of plating, with 0.25% crystal violet dye and then extracted the dye for spectrophotometery to quantify the extent of cell adhesion. The results revealed that at all time points examined, there were significantly fewer PML 2/2 MEFs adhering to the culture wells than PML +/+ MEFs (Fig. 3E).

Effects of PML Null Mutation on Cell Proliferation
We investigated whether there were any differences between the cell cycle profile of PML 2/2 and PML +/+ MEFs. All cells were cultured for 36 hours and then dissociated from the culture flasks. They were stained with PI dye, sorted in a FACSAria II Flow Cytometer and analyzed using a ModFit LT TM software. We established that there were approximately 1263.3% more PML 2/ 2 MEFs distributed at S-phase than the control PML +/+ MEFs (Figs. 4A & B). However, there were 10.363.6% and 1.960.58% less PML 2/2 MEFs distributed at G0/G1 and G2/M phases respectively than PML +/+ MEFs (Fig. 4C). The results suggest that PML 2/2 MEFs proliferate significantly faster that PML +/+ MEFs.

Effects of PML Null Mutation on Cell Migration
The scratch assay was used to assess whether the migration efficiency of PML +/+ and PML 2/2 MEFs were different because of the null mutation. PML +/+ and PML 2/2 MEFs were allowed to grow until confluent and then prevented from further proliferation by Mitomycin C treatment. A scratched (2.5 mm wide) was then created longitudinally in the monolayer culture. Micrographs were taken of these cultures at different time intervals. Two dotted lines were drawn on the plastic dish to define the original width of the gap created in the cultures. The total numbers of MEFs that have migrated into the gap (i.e. in front of the dotted lines) were counted at different time intervals for both PML +/+ and PML 2/2 MEFs (Fig. 5A). The results showed that there were significantly more PML +/+ MEFs that have migrated into the gap than PML 2/2 MEFs at alltime intervals examined. The percentage of migrant cells in the gap at different time intervals was calculated as shown in Figure 5B. After 24 hr culture, 87.766.9% area of the gap was filled up with PML +/+ MEFs versus 48.3615% for PML 2/2 MEFs. The results imply that PML +/+ MEFs were significantly more mobile than PML 2/2 MEFs in vitro.

NDRG1 Expressions in PML +/+ and PML 2/2 MEFs
We have selected one of the proteins (called NDRG1) that were differentially expressed in the proteomic analysis for further investigation. In the proteomics, NDRG1 was expressed in PML +/ + MEFs but was barely detectable in PML 2/2 MEFs. Semiquantitative RT-PCR was performed and the results correlated with the proteomic findings that there was significantly less NDRG1 transcripts in PML 2/2 than PML +/+ MEFs (Fig. 7). We also examined NDRG1 expression in MEFs using immunofluorescent staining. In PML +/+ MEFs, NDRG1 was distributed in the cytoplasm as a small oval-shaped aggregate and as speckles when viewed under the confocal microscope (Fig. 8B). In contrast, there were only small speckles of weakly stained NDRG1 distributed in the cytoplasm of PML 2/2 MEFs (Fig. 8C). We also examined NDRG1 expression in PML +/+ and PML 2/2 MEFs under the transmission electron microscope (Figs 8D-I). The MEFs were stained with mouse NDRG1 antibody and anti-mouse QDot conjugated secondary antibody. In PML +/+ MEFs, there were numerous electron dense Qdot labels present in the rough endoplasmic reticulum (RER) and Golgi complex (Figs. 8D-F). In Golgi complex the QDots were distributed as an aggregate (Fig. 8F) reminiscent of the oval-shaped NDRG1 fluorescent staining detected under the confocal microscopy ( Fig. 8B). For PML 2/2 MEFs, there was a significant reduction in QDot labeling in the RER and Golgi complex (Figs. 8G-I). We also examined the distribution of PML using PML/QDot staining (Fig. 8A). In PML +/+ MEFs, PML was distributed mainly in the RER and as numerous small aggregates in the nucleus. Hence, it appears that the distribution of PML and NDRG1 proteins overlap in the RER.

Differences in Chemotactic Response between PML +/+ and PML 2/2 MEFs
The proteomic results revealed that CTRO, TRIO and CLF1 expression were suppressed in PML 2/2 MEFs. These proteins play an important regulatory role in cell adhesion, spreading and migration. Therefore, we want to establish whether the ability of PML 2/2 MEFs to chemotactically respond to factors present in fetal bovine serum was affected, when compared with PML +/+ MEFs. For this study a Neuro multi-chemotaxis chamber was deployed, where 0, 1% and 5% fetal bovine sera were added to the  Proteins that are differentially expressed may not be correspondingly reflected at the transcriptional level. (A) LC-ESI-MS/ MS identified proteins that were differentially expressed in PML 2/2 and PML +/+ MEFs. Semi-quantitative RT-PCR was performed to establish whether the proteins that we identified were differentially expressed were also correspondingly expressed at the mRNA level. (B) The bar chart shows the intensity of the PCR bands for each gene in (A). The measurements were normalized against b-actin internal control. The data is presented as Mean 6 SD by t-test and *p,0.05. The experiment was repeated in triplicates. doi:10.1371/journal.pone.0059477.g007

Effects of Silencing NDRG1 Expression on MEFs Proliferation
We have transfected PML +/+ MEFs with CTL-siRNAs (control) and NDRG1-siRNAs for 24 hours and then harvested them for analysis. Semi-quantitative RT-PCR revealed that our NDRG1-siRNA was able to silence NDRG1 expression by approximately 98% (Fig. 10). We also established that PML and p53 expression were also correspondingly inhibited by approximately 9561.8% and 8063.6% respectively. Some of these cells were also stained with PI dye and their cell cycle profile analyzed by flow cytometry. We established that for PML +/+ MEFs transfected with CTL-siRNA, 32.663.1% of the cells were distributed at G0/G1 phase, 42.063.0% at S phase and 25.362.5% at G2/M. For PML +/+ MEFs transfected with NDRG1-siRNA, 25.762.1% of cells were distributed at G0/G1 phase, 50.763.2% at S phase and 23.664.7% at G2/M phase. The results show that there were significantly more NDRG1-silenced MEFs distributed at S-phase than MEFs transfected with CTL-siRNAs (Fig. 11). Furthermore, there were significantly less NDRG1-silenced MEFs distributed in the G0/G1 phase. The results suggest that silencing NDRG1 expression increases cell proliferation. Effects of Silencing NDRG1 Expression on TGF-b1 Signaling It has been reported that the TGF-b1 signaling pathway was impaired in PML 2/2 such that Smad2 and Smad3 phosphorylation was reduced in MEFs [13]. Consequently, this inhibited the nuclear translocation of Smad3 -as the process is phosphorylationdependent. Presently, our RT-PCR results demonstrated that silencing NDRG1 expression also inhibited PML expression in PML +/+ MEFs. Hence, we wanted to establish whether Smad3 phosphorylation was also correspondingly reduced in NDRG1silenced PML +/+ MEFs. We transfected PML +/+ MEFs with CTL-siRNAs or NDRG1-siRNAs for 24 h and then treated the cells with 100 ng/ml TGF-b1. All of the cells were harvested for immunofluorescent staining after 5 hr of TGF-b1 induction. For MEFs transfected with CTL-siRNA, TGF-b1 induction resulted in an intense nuclear staining for phosphorylated SMAD3 in all of the cells (Figs. 12A & B). In contrast, for NDRG1-silenced MEFs, TGF-b1 treatment only induced weak nuclear phosphorylated SMAD3 staining in approximately 70% of the cells (Figs. 12C & D). The results suggest that TGF-b1 signaling is impaired in NDRG1silenced MEFs.
Discussions PML-NBs are sub-nuclear multi-protein structures that have been implicated in many diverse biological functions such as apoptosis, cell proliferation and senescence [14]. In this study, we have used MEFs generated from PML knockout mice to further investigate PML's biological functions [12]. These MEFs do not express all seven known variants of PML. It has been reported that PML knockout mice react to staphylococcus and listeria infection by hyperproliferative response resulting in the development of splenomegaly [15]. These mice were also very prone to developing spontaneous batryomycosis. Immunohistopathological examinations of the mutant's lymph nodes, extracted near infection sites, revealed that they were enlarged and hyperplastic. Furthermore, the plasma cells have proliferated dramatically and infiltrated the lesion sites.
We used comparative proteomics to identify proteins that were differentially expressed in PML 2/2 MEFs. We established that NDRG1 expression was suppressed in the absence of PML. Hence, we decided to examine the relationship between these two proteins. Using immunofluorescent staining, we showed that NDRG1 was distributed as a small oval-shaped aggregate and also as small speckles in the cytoplasm of PML +/+ MEFs but only as small speckles in PML 2/2 MEFs. We further investigated the distribution of NDRG1 using immune TEM. The MEFs were stained with NDRG1and QDot antibodies. Under TEM, we established that the NDRG1-Qdots mainly labeled components distributed around RER and Golgi complex in PML +/+ MEFs. Hence, we speculated that NDRG1 was likely involved in the trafficking of endosomes between the RER and Golgi complex for modification and maturation. Furthermore, we also established that PML and NDRG1 were co-located at RER. It has been reported that NDRG1 was normally ubiquitously expressed either in the cytoplasm, nucleus, mitochondrion or cell membrane depending on the tissue type [16]. It appears that NDRG1 is a potential metastatic-related cancer suicide gene because when the gene is over-expressed, it induces apoptosis in human colonic cancer cell lines [17 & 18]. Cellular senescence is regards as an irreversible cell cycle arrest that is associated with the tumorsuppressive mechanism [19 & 20]. Importantly, when we silenced NDRG1 expression using NDRG1-siRNA in PML +/+ MEFs, there was a significant increase in cell proliferation. This was also accompanied by an inhibition of PML and p53 expression. It has been reported that the promoter region of NDRG1 contained a p53 binding site and consequently in human colon cancer (DLD-1-p53) cells, NDRG1 is expressed in a p53-dependent fashion following treatment with DNA-damaging chemotherapeutic drug doxorubicin [21]. However, NDRG1 alone was not sufficient for p53-mediated caspase activation and apoptosis. Lin et al. 2004 has demonstrated that cytoplasmic PML was an essential modulator of TGF-b1signaling [13]. Furthermore, PML 2/2 MEFs were resistant to TGF-b1 -dependent growth arrest, cellular senescence and apoptosis. TGF-b1 signaling is mediated by the phosphorylation and then nuclear translation of Smad2/3. Cytoplasmic PML can physically interact with Smad2/ 3 and in its absence Smad2/3 phosphorylation is impaired. In our study, we have shown that silencing NDRG1 expression in PML +/+ MEFs resulted in PML expression being suppressed. Hence, we investigated whether if TGF-b1 signaling was also correspondingly impaired in NDRG1-silenced MEFs. Our immunofluorescent staining revealed that phosphorylated Smad3 accumulated in all the nuclei of PML +/+ MEFs, 5 hours after TGF-b1 induction. However in NDRG1-silenced MEFs, only approximately 70% of the nuclei were weakly stained for phosphorylated Smad3 -while there was no staining in the rest of the nuclei. This suggests that silencing NDRG1 expression in MEFs impairs the TGF-b1 signaling pathway and this may be mediated by PML because our immune TEM results demonstrated that NDRG1and PML were co-localized in the RER.
Besides NDGR1, our proteomic experiments also identified CTRO, TRIO, ANXA4 and CFL1 peptides were also downregulated in PML 2/2 MEFs. These proteins normally play important roles in cell adhesion, spreading and migration. CTRO and TRIO regulate Rho-GTPase to stimulate the cellular responses to changes in cell morphology and directed migration [22 & 23]. Furthermore, ANXA4 has the ability to bind to phospholipids found on membrane surfaces and mediate in the p53-apoptotic pathway [24], enhancement of cancer cell chemoresistance [25], regulates membrane protein mobility [26], membrane trafficking [27] and Ca 2+ homeostasis [28]. In the developing embryo, CFL1 is involved in regulating cell migration during gastrulation and promoting actin filament assembly [29 & 30]. Bamburg et al. (1999) [31] reported that CFL1 can function as an actin-depolymerizing-factor to destabilize the actin filaments at the leading edge of a migrating cell. In our study, we observed that in the absence of PML the morphology of MEFs was dramatically affected. The PML 2/2 MEFs were significantly smaller than PML +/+ MEFs. Furthermore, PML 2/2 MEFs did not produce the highly elongated cytoplasmic projections that were so prominent around the cellular edges of PML +/+ MEFs as seen under the scanning electron microscope. We believe that these differences in PML 2/2 MEF morphologies may be attributed to them being less adhesive than PML +/+ MEFs (i.e. when cells are highly adhesive they give the appearance of being flatter and therefore larger on culture dishes than less adhesive cell types). Indeed, we have experimentally demonstrated that our PML 2/2 MEFs were less adhesive than PML +/+ MEFs. Besides changes in cell morphology, we have also established that PML 2/ 2 MEFs were less mobile and did not respond to chemotactic signals as efficiently as PML +/+ MEFs. This again may be attributed to PML 2/2 MEFs being less adhesive (less traction and therefore less motile) than PML +/+ MEFs. These observations raise the possibility that the PML mutation may affect development. Schreck and Gaiano (2009) [32] reported that PML was essential for neural progenitor cells to migrate and develop normally in the neocortex. The brain of PML mutants were shown to be smaller and the neocortical wall thinner than normal animals.
To date, the function of PML is still not fully understood. PML-NBs may act as a nuclear depot for regulating the nucleoplasmic levels of proteins, such as HSF2, SENP-1, PA28, pRb [33, 34, 35 & 36]. They also mediate in nuclear activities (such as repression of p53, Daxx, Mdm2 and Sp100) and act as a nuclear platform for post-translational modification (such as phosphorylation, acetylation, SUMOylation and ubiquitination) [37, 38, 39 & 40]. Interestingly, HSPB1, SUMO2, NSFL1C and STMN1 are proteins that we found up-regulated in our PML 2/2 MEFs. Most of these proteins are related to the PML-NBs associated proteins, especially to SUMO2 which is involved in SUMOylation [41]. HSPB1 is also known as HSP27 that acts as a chaperone by holding unfolded polypeptides during stress conditions [42]. Inhibition of HSP27 has been reported to induce the degradation of the histone deactylase HDAC6, transcription factor STAT2 and procapase-3 in human cancer cells. PCBP2 expression is upregulated in PML 2/2 MEFs. You et al., 2009 [43] reported that PCBP2 was an adaptor of AIP4 which degraded MAVS via the ubiquitin-proteasome pathway. Moreover, PML monitors PCBP2 expression to target proteins for degradation. NSFL1C, another protein that is up-regulated in PML 2/2 MEFs, is a p47 co-factor that regulates the ATPase activity of membrane fusion protein p97 and leads to Golgi cisternal regrowth [44]. NSFL1C is a substrate  of EGF signaling during breast cancer development [45]. It has been demonstrated that antisense-EGFR treatment can increase PML expression in gliobalstoma cells [46]. STMN1 is a 19 kDa cytosolic protein that participates in microtubule-destabilizing in the construction of the mitotic spindle [47]. STMN1 is a biomarker for gastric [48] and non-small cell lung [49] cancers. This protein is normally expressed at low levels but up-regulated in PML 2/2 MEFs. YWHAG is strongly expressed in PML 2/2 MEFs and that can interact with RAF1 or miRNA involved in lipid metabolismhence may inhibit various signal transduction pathways [50 & 51]. Recently, it has been reported that YWHAG is upregulated during osteogenic differentiation [52]. It is not surprising that YWHAG is up-regulated in PML 2/2 MEFs since PML normally inhibits YWHAG protein to modulate cell proliferation and differentiation [53]. In summary, we have demonstrated that PML plays an important role in cell adhesion, morphology, proliferation, migration and TGF-b1 signaling.

Cell Cultures
Normal (PML +/+ ) and mutant PML (PML 2/2 ) mouse embryonic fibroblasts (MEF) were generously provided by Professor PP Pandolfi [12]. The MEFs were cultured in Dulbecco's Modified Eagle Medium (DMEM, Life Technology, USA) supplemented with 10% FBS (Life technologies, USA) plus 100 units penicillin and 100 mg streptomycin. The MEFs were maintained at 37uC and 5% CO 2 in a humidified cell incubator. After the MEFs became confluent, they were trypsinized in 0.25% trypsin solution (Invitrogen, USA) and seeded onto new culture dishes at 1610 4 cells/ml.

Cell Adhesion Analysis
PML +/+ and PML 2/2 (3.5610 4 cells/ml) MEFs were plated onto uncoated 96-well plastic culture dish. All un-adhered cells were removed 30, 60, 90 and 120 mins after plating. The wells were gently washed (2X) with warm PBS to remove excess cells and then fixed in 4% formaldehyde for 30 min. The cells were then wash and stained with 0.25% crystal violet dye made up in 40% methanol for 30 mins. After extensive washing to remove residual dye, the cultures were dried and 2% SBS in PBS were added to release the crystal violet dye from the MEFs. The staining intensity was quantified by spectrophotometery (579 nm) using a plate reader. There were 4 replicates for each sample and time point analyzed.

Western Blot Analysis
Western blot analysis was performed according to methods that we described previously [54]. Briefly, PML 2/2 and PML +/+ MEFs were lysed in lysis buffer (50 mM NaCl, 20 mM Tris, pH 7.6, 1% NP-40, 1X protease inhibitor mixture) for 1 hr. The lysates were centrifuged at 16,0006g at 4uC for 10 min to remove the cell debris and insoluble proteins. Protein concentration was determined by using a Bio-Rad protein assay kit (Bio-Rad, USA). Protein lysate (30-50 mg) from each samples were loaded into the 10% SDS-PAGE for resolving. A Trans-Blot SD semi-dry electrophoretic transfer cell was used to electro-transfer the separated proteins onto Hybond NC membrane (GE Healthcare). The blotted membranes were then blocked with 5% skimmed milk for 1 hr and incubated with either PML (1:1000, Abcam #53773 or b-tubulin 1:1000 to 1500, Zymed Laboratories) primary antibodies. Bound antibodies were detected using the appropriate horseradish peroxidase-conjugated secondary antibodies (Southern biotechnology), followed by development with an ECL Western blotting Detection kit (GE Healthcare). The blots were analyzed using Quantity One software (Bio-Rad) and the intensity of the PML stained band was normalized against the b-tubulin band (internal control). Three replicates of each sample were studied.

Semi-quantitative RT-PCR Analysis
Total RNAs were isolated and purified from PML 2/2 and PML +/+ MEFs in TRIzol solution (Invitrogen Corporation, USA). Using an ImProm-II TM Reverse Transcription System kit (Promega, USA), 1mg of the total RNA was reverse transcribed into complementary DNA (cDNA). Twenty ml of PCR mixture containing 1 ml of cDNA, 2.5 ml of PCR 10X buffer, 0.75 ml of magnesium chloride solution, 1 ml of dNTP mix (10 mM, Promega Corporation, USA), 1 ml of forward primer, 1 ml of reverse primer, 0.25 ml of Taq polymerase (Bio-firm, Hong Kong) and DEPC-treated water in a eppendorf tube was placed into a PTC-100 thermal cycler (MJ Research, Watertown, MA, USA) for PCR amplification. A Primer3 software (version 0.4.0, Rozen and Skaletsky; http://frodo.wi.mit.edu) was used to design all of the primers and determined the PCR amplification conditions as listed in Table 3. Electrophoresis of the PCR products was performed on a 1.5% agarose gel. The produces were stained with GelRed TM dye and the intensities of the PCR product bands were measured using a GelDoc-It imaging system (UVP, BioImaging System, USA). b-actin was used as the internal control and for normalization. The band intensity of each gene expressed was determined and analysed using a MetaMorphH Imaging system software (Molecular Devices, United States). All experiments were repeated three times.

Immunofluorescent Microscopy
The PML +/+ and PML 2/2 MEFs were fixed in 10% formalin after the 24 hr culture. The samples were then washed with PBS for 10 mins (2X) and treated with 0.5% Triton X-100 for 15 mins to increase cell permeability. To block non-specific binding, the samples were washed with PBS and incubated for 1 hr in 1.5% host serum. The specimens were then incubated in mouse NDRG1 antibody (1:100, Abcam #ab124689) or SMAD3 (phospho S423+ S425, 1:100, Abcam # ab52903) overnight at room temperature. After washing with PBS (3X), the specimens were incubated in Cy3-conjugated anti-rabbit (1:300 dilutions) or anti-mouse QDot conjugated (1:100, Invitrogen, USA) secondary antibody for 1 hr at room temperature. The samples were further washed with PBS and the nuclei counterstained with 49, 6diamidine-29-phenylindole dihydrochloride (Roche diagnostics, Indianapolis). No primary antibodies were added to the negative control. The immunofluorescently stained PML +/+ and PML 2/2 MEFs were viewed under a BioRad 1024 L Sere Scanning Confocal Microscope (BioRad, USA) equipped with 40 X Zeiss PlanNeofluo objectives.

Filamentous Actin Staining
The PML +/+ and PML 2/2 MEFs were cultured on round glass coverslips for 24 hr, fixed and washed with PBS. The MEFs were then stained with fluorescein Isothiocyanate labeled phalloidin (1:100, Sigma #P5282) suspended in 0.5% Triton X-100 and PBS for 15 mins to reveal the presence of filamentous actin inside the cells. After incubation, the samples were rinsed in PBS (3X) for 30 mins and viewed under a BioRad 1024L Sere Scanning Confocal Microscope (BioRad, USA) equipped with 40 X Zeiss PlanNeofluo objectives.

Comparative Proteomic Analysis
First dimensional electrophosis (DE) was performed on an IPGphor IEF system using 11-cm long IPG electrode strip with pH 4-7 gradient (Amersham Biosciences, UK) and an Ettan IPGphor Strip Holder (Amersham Biosciences, UK). 150 mg of PML +/+ or PML 2/2 MEFs protein was applied for each IPG strip. The total volume of protein sample with rehydration buffer (8 M Urea, 2% CHAPS (w/v), 1% IPG buffer (v/v), 40 mM DTT) loaded onto the strip holder was 210 ml. 1 ml of IPG Cover Fluid was applied to each strip so as to minimize evaporation and urea crystallization. The rehydration step was done under voltage and followed by a separation process. The electrophoresis condition for step 1 was 30 V for 13 hrs; step 2 was 500 V for 1 hr; step 3 was 2000 V for 1 hr and step 4 was 5000 V for 20 hrs. The program was stopped when the total volt-hours reached 40000. After the first dimensional DE was completed, the IPG strips were removed from the strip holders. Each strip was then treated with 1% DTT in 6.5 ml of equilibration buffer (

Protein Identification by Mass Fingerprinting
All protein spots of interest were isolated from the gel and processed for silver destaining. The gel pieces were first washed in MilliQ water, immersed in 200 ml of destaining solution (15 mM potassium ferricyanide and 50 mM sodium thiosulphate) and then incubated at room temperature until they turned colorless. Each gel pieces was then washed in 400 ml of MilliQ water for 15 min (3X). The destained gel pieces were then equilibrated in 200 ml of 10 mM ammonium bicarbonate/50% acetonitrile for 15 min. The gel was dehydrated in 200 ml of acetonitrile for 15 min and dried at 30uC for 5 min. The gels were digested with 15mg/ml of trypsin in 40 mM ammonium bicarbonate/50% acetonitrile (v/v) at 35uC for 16 hrs. Three ml of extraction solution (50% acetonitrile (v/v) and 5% trifluoroacetic acid (Fluka Chemika, Switzerland) were used to stop the reaction. Three ml of reaction mixture was mixed with a-cyano-4-hydroxycinnamic acid matrix and then spotted onto a sample plate for ESI-MS/MS analysis (Bruker Daltonics, USA). The mass spectrums generated were analyzed using a Bruker Daltonics software and by mass fingerprinting, which were submitted to the SwissPort bioinformation stations using MASCOT 2.2.07 engine search.

Flow Cytometry
MEFs were stained with Propidium Iodide (PI) dye and processed for flow cytometry as described by Yau et al., 2010 [55]. Briefly, the MEF cultures were trypsinized, suspended and fixed in 70% ethanol overnight at 4uC. The samples were then washed with PBS (2x) and incubated in 20 mg/ml PI (Sigma, USA), 0.1% Trixton X-100 (Sigma, USA) and 100 mg/ml RNase A (Sigma, USA) in the dark for 1 hr. The DNA content analysis was performed in a BD FACSAria II Flow Cytometer ((Becton-Dickinson, Mountain View, CA) and the cell cycle profile was established using a ModFit LT TM software (Verity Software House, USA).

Scanning Electron Microscopy
Morphology of PML 2/2 and PML +/+ MEFs were examined under the scanning electron microscope (SEM). Briefly, the cells were cultured on 13 mm 2 round glass coverslip for 24 hr and then fixed in 4% paraformaldehyde and 2.5% glutaldehyde for 24 hrs. The cultures were washed with PBS (3X) and treated with 1% Osmium tetroxide in 0.1 M phosphate buffer. After washing with PBS, the specimens were dehydrated, critical point dried and coated with Gold. The cells were viewed under a SU6600 Variable Pressure Schottky FE-SEM (Hitachi Technologies, Japan).

In vitro Scratch Assay
The scratch assay was performed according to methods described by Liang et al. [56]. PML +/+ and PML 2/2 MEFs (0.5 ml of 1610 5 cells/ml) were first seeded onto 4-well culture plates. After 16 hr incubation, the cells formed a monolayer with 80% confluence. Before a gap was created in the culture, the cells were treated with 2 mg/ml mitomycin C (Sigma, USA) for 1 hr to prevent further cell proliferation (which could confound our interpretation of the cell migration analysis). After the treatment, a sterile p10 pipet tip was used to scrap off cells from the center of the monolayer to create a gap in the culture. Reference lines were etched onto the bottom of the plastic culture dishes to define the position of the gap/wound. Photos were taken of the PML +/+ and PML 2/2 MEFs migrating into the wound area at 0, 4, 8 and 24 hr incubation. The photographic images captured were quantitatively analyzed by establishing the number of cells that have migrated into the gap.

Chemotactic Assay
Chemotactic cell migration assay was performed according to modified methods described by Webb and Lee [57]. A Neuro multi-chemotaxis chamber (Neuro Probe, Inc., Gaithersbury, USA) consisting of a sheet of porous polycarbonate membrane (25680 mm with 8.0 mm pores) sandwich between a series of upper and lower wells was used to perform the assay. An appropriate volume of DMEM medium (control), DMEM +0.5% FBS and DMEM +10% FBS were introduced into the lower wells. In the top wells were added 1610 5 cells/ml PML 2/2 or PML +/+ MEFs (suspended in DMEM). The multi-chemotaxis chamber was then incubated at 37uC and 5% CO 2 for 4 hrs. After incubation, all the cells on the top wells were removed and the upper surface of the membrane wiped clean of all attached cells using a cotton swab. The membrane was then fixed in 10% formalin for 1 hr, stained with Hematoxylin and Eosin dyes and mounted with 10% glycerol. The total number of migratory cells found on the bottom surface of the polycarbonate membrane were determined and statistically analyzed.

Silencing NDRG1 Expression in MEFs
A synthetic 21 base-pair long siRNA (Santa Cruz, USA) was used to target and silence NDRG1 expression. The method used for delivering the NDRG1-siRNAs and CTL-siRNA (control) into PML +/+ MEFs has been described by Tang et al., 2006 [58]. Briefly, 0.25 ml of 1610 5 PML +/+ MEFs were seeded onto each well of a 4-well culture plate and cultured for 16 hrs until approximately 60% confluent. The culture medium was then replaced with Opti-MEMH I Reduced Serum Medium (Invitrogen, USA) containing 2% FBS for 1 hr. One ml of LipofectamineH RNAiMAX Transfection Reagent (Invitrogen, USA) and 20 nM of either NDRG1or CTL-siRNAs were used for transfection. The MEFs were transfected twice (at 6 hrs intervals) and cultured for 24 hrs. The cells were then collected for semi-quantitative RT-PCR, flow cytometry analysis and immunofluorescent staining.

Statistics
The data were analyzed using two-tailed, paired student's t-test. P,0.05 was considered to be statistically significant. All statistical analysis was performed using a SPSS software.