PTB-Associated Splicing Factor (PSF) Is a PPARγ-Binding Protein and Growth Regulator of Colon Cancer Cells

Peroxisome proliferator-activated receptor gamma (PPARγ) is a nuclear receptor that plays an essential role in cell proliferation, apoptosis, and inflammation. It is over-expressed in many types of cancer, including colon, stomach, breast, and lung cancer, suggesting that regulation of PPARγ might affect cancer pathogenesis. Here, using a proteomic approach, we identify PTB-associated splicing factor (PSF) as a novel PPARγ-interacting protein and demonstrate that PSF is involved in several important regulatory steps of colon cancer cell proliferation. To investigate the relationship between PSF and PPARγ in colon cancer, we evaluated the effects of PSF expression in DLD-1 and HT-29 colon cancer cell lines, which express low and high levels of PPARγ, respectively PSF affected the ability of PPARγ to bind, and expression of PSF siRNA significantly suppressed the proliferation of colon cancer cells. Furthermore, PSF knockdown induced apoptosis via activation of caspase-3. Interestingly, DLD-1 cells were more susceptible to PSF knockdown-induced cell death than HT-29 cells. Our data suggest that PSF is an important regulator of cell death that plays critical roles in the survival and growth of colon cancer cells. The PSF-PPARγ axis may play a role in the control of colorectal carcinogenesis. Taken together, this study is the first to describe the effects of PSF on cell proliferation, tumor growth, and cell signaling associated with PPARγ.


Introduction
Colon cancer continues to be a major public health problem. Worldwide, approximately 1 million new cases of colon cancer are diagnosed each year, with nearly 500,000 deaths attributed to this disease annually [1]. Most of these deaths occur as a consequence of late diagnosis. Although colon cancer develops in the colon and rectal tissues, the cancer cells can spread to other parts of the body, such as the liver, bone, brain, and lung, and form a new tumor. Because metastatic colon cancer is associated with high mortality [2][3], progression to metastasis is the critical point in colon cancer survival. Currently, chemotherapeutic agents are the main tools for treating colon cancer. However, most of these drugs are nonspecific or become less effective as tumor cells acquire multidrug resistance. Therefore, novel therapeutic options are needed to reduce colon cancer mortality.
PPARc is a member of the nuclear receptor super-family, whose members activate target gene transcription in a ligand-dependent manner [4][5]. Activation of PPARc by thiazolidinediones (TZDs) leads to an altered metabolism in adipose tissue, skeletal muscle cells, and liver that collectively results in insulin sensitization [6]. PPARc expression is increased in many types of cancer, including colon, lung, breast, and stomach cancer, suggesting that regulation of PPARc might affect cancer pathogenesis [7,8]. Although PPARc is expressed at significant levels in human colon cancer cells and tissue [8], the role of PPARc activation in colon cancer is still controversial [9]. Furthermore, the role of PPARc activation in cancer in general remains unclear. A number of high affinity synthetic agonists exist for PPARc, including rosiglitazone and troglitazone. It has been reported that these agonists inhibit the proliferation of a variety of human cancer cells. However, the mechanism of action in most cases points to receptor-independent effects [10]. Several studies describe the ability of a PPARa/c agonist, TZD18, to induce glioblastoma cell toxicity in a receptorindependent manner [11]. This compound induced apoptosis through cell cycle arrest. The apoptotic events were mediated by down-regulation of Bcl-2, up-regulation of Bax, and activation of caspase-3. These results suggest that TZDs can induce apoptosis independent of PPARc activation, primarily by activating the intrinsic apoptotic pathway.
PTB-associated splicing factor (PSF) is a multifunctional protein involved in transcription regulation, pre-mRNA processing, and DNA repair [12]. One of the most abundant nuclear proteins, it consists of a single polypeptide chain of ca. 76 kDa (determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis [SDS-PAGE]) [13]. The amino terminus is rich in proline and glutamine residues. PSF has multiple binding functions. A recent study revealed that PSF belongs to a family of putative tumor-suppressor proteins that contain an RNA-binding domain (RBD) and a DNAbinding domain (DBD) [14]. The DBD binds and represses the transcription of genes that have a PSF-binding site [14,15]. Thus, PSF is a highly complex protein that may be an important component of the transcriptional repression of many different genes involving different mechanisms. Recently, Wang et al. reported that PSF has a central role in the reversible regulation of mammalian cell proliferation and tumorigenesis [16]. Alteration in the expression of PSF and its binding partners may have potential as a therapeutic strategy against cancer [17]. However, how the various activities of PSF are regulated in colon cancer cells is not yet clear. We hypothesized that PSF interacts with PPARc. Therefore, the aim of the present study was to was to obtain evidence for a direct interaction between PSF and PPARc in colon carcinogenesis. Our results showed that PPARc interacts directly with PSF. To examine the PPARc-dependent effects of PSF, we also compared the HT-29 cell line, in which PPARc is highly expressed, with the DLD-1 cell line, in which PPARc is poorly expressed, under PSF knockdown conditions. The differential proteomic patterns of the two cell lines were assessed by LC-MS/ MS analysis. The level of PPARc in colon tissue is equal to or greater than that in adipose tissue [18]. This observation suggests the special role of PPARc in the colon, as reflected in part by the cell-or tissue-specific expression of the receptor [19]. The proteins differentially regulated in the two cell lines provide us with a better understanding of the events involved in colon cancer.

Preparation of Subcellular Fractions
The NE-PER Cell Fractionation Kit (Pierce Biotechnology, Rockford, IL, USA) was used to isolate the nuclear fraction from cells, according to the manufacturer's instructions. After the cytoplasmic fraction was separated, the nuclear fraction was subjected to brief centrifugation (1,0006g, 10 sec), and the interface was removed to reduce cytoplasmic contamination.

Pull-down Assay with a Metal Affinity Resin
Hexahistidine (66His)-tagged PPARc fusion proteins or empty vector controls were expressed in BL-21 (DE3) cells. Transformed BL-21 cells were induced with 0.5 mM isopropyl-1-b-D-galactopyranoside (IPTG) (Invitrogen) for 12 h at 25uC and collected by centrifugation. Next, 1 mL of supernatant was incubated with 20 mL of the TALON metal affinity resin (Takara) at 4uC for 1 h in lysis buffer. The resin was washed 5 times with wash buffer (20 mM MES pH 7.4, 150 mM NaCl), and the proteins were eluted with 150 mM imidazole in wash buffer. The amount of PPARc was quantified using the Protein Quantification Kit-Rapid (Dojindo, Kumamoto, Japan). For the pull-down assay, purified 66His-tagged PPARc (1 mg) was mixed with nuclear extracts from HT-29 cells in 50 mL of binding buffer containing 20 mM MES pH 7.4 and 150 mM NaCl; TALON resin was then added. After incubation for 2 h at 4uC, the resins were washed 5 times with 500 mL of wash buffer containing 20 mM MES pH 7.4 and 100 mM NaCl.

In-gel Digestion and Protein Identification by MALDI-TOF MS
In-gel digestion of gel bands was performed as previously described [20]. Briefly, protein spots, which were excised from the gel, were de-stained with 100 mM ammonium bicarbonate in 50% acetonitrile. The gel pieces were dried and digested with sequencing-grade modified trypsin (Promega). The peptide solution was recovered, and residual peptides were extracted by shaking with 5% trifluoroacetic acid (TFA) in 50% acetonitrile. The combined solutions were concentrated using a lyophilizer. The tryptic peptides, which were dissolved in 0.1% TFA, were desalted with Zip-Tip (Millipore, Billerica, MA, USA) according to the manufacturer's instructions, mixed with the equal volume of a matrix solution (10 mg/mL acyano-4-hydroxycinnamic acid in 50% acetonitrile/0.1% TFA), and applied to a target plate. MS/MS analyses were performed using the AB SCIEX TOF/TOF TM 5800 System (AB SCIEX, Foster City, CA, USA). Protein identification was performed through ProteinPilot TM software (AB Sciex, Framingham, MA, USA) using the UniProt database.

Co-immunoprecipitation and Western Blot
HT-29 cells and DLD-1 cells were resuspended in lysis buffer containing 10 mM Tris-HCl (pH 7.4), 100 mM NaCl, 50 mM KCl, 0.05% Tween-20, 10% glycerol, and the Halt Protease Inhibitor Cocktail (Takara). After 15-min incubation on ice, the cell lysate was sonicated and centrifuged at 16,0006g for 10 min at 4uC. The supernatant was collected as the whole-cell extract. The cell lysate was pre-cleared by adding 5 mL of Protein A/G Plus-Agarose (sc-2003, Santa Cruz Biotechnology) and incubated for 1 h at 4uC. The mouse monoclonal anti-PSF antibody (200 mg/ mL, sc-271796, Santa Cruz Biotechnology) and the cell extract were mixed and incubated at 4uC for 3 h. The sample was then mixed with 5 mL of IP matrix (ImmunoCruz TM IP/WB Optima B system, sc-45039, Santa Cruz Biotechnology) and incubated at 4uC overnight. After the incubation, the immunoprecipitates were washed 5 times with 0.5 mL of 10 mM Tris-HCl (pH 7.4), 0.5 M NaCl, 0.1 M KCl, and 0.025% Tween-20, and then eluted with SDS-PAGE reducing sample buffer. Samples were separated by 5-20% SDS-PAGE and western blotted. After washing, the membrane was incubated with a horseradish peroxidase-linked species-specific whole secondary antibody (anti-rabbit or -mouse IgG; GE Healthcare, Little Chalfont, UK) for 1 h at room temperature and then visualized with Pierce ECL Plus Western Blotting Substrate (Thermo Scientific, Pittsburgh, PA, USA) or EzWestLumi plus (ATTO, Tokyo, Japan).

Quantitative Real-time PCR Analysis
Total RNA was prepared from HT-29 and DLD-1 cells using NucleoSpinH RNA II (Takara). Then, 0.5 mg of total RNA was used for the subsequent synthesis of cDNA using the ReverTra Ace qPCR RT Kit (Toyobo, Osaka, Japan) as recommended by the manufacturer. Quantification of mRNA levels was measured by using an ECO Real-Time PCR system (Illumina, Inc., San Diego, CA, USA) and SYBR Green Realtime PCR Master Mix -Plus-(Toyobo) with the following primer pair sets: PSF, 59-TGCCATTCATGCTTCTATGCA-39 (F) and 59-GGCCTAGA-CACTCTCATGCTTTC-39 (R); 18S rRNA, 59-CAGCCACCC-GAGATTGAGCA- 39 (F) and 59-TAGTAGC-GACGGGCGGTGTG-39 (R). All PCRs were performed in a 10-mL volume using 48-well PCR plates (Illumina). The cycling conditions were 95uC for 10 min (polymerase activation), followed by 40 cycles of 95uC for 15 sec, 55uC for 15 sec, and 72uC for 30 sec. In order to determine which housekeeping genes were most suitable for the subsequent normalization of data, we initially selected 3 candidates: GAPDH, b-actin, and 18S-rRNA, commonly used internal controls in mammalian cells. After amplification, the samples were slowly heated from 55uC to 95uC with continuous reading of fluorescence to obtain a melting curve. The relative mRNA quantification was calculated by using the arithmetic formula 2 2DDCq , where DCq is the difference between the threshold cycle of a given target cDNA and an endogenous reference cDNA. Derivations of the formulas and validation tests have been described in Applied Biosystems User Bulletin No. 2.

Small Interfering RNA
PSF expression was inhibited in HT-29 and DLD-1 cells by transfection with a small interfering RNA (siRNA) targeting PSF (Santa Cruz Biotechnology), using Lipofectamine RNAiMAX (Invitrogen). Cells were plated onto 6-well plates (Iwaki, Tokyo, Japan) at a density of 5610 4 cells per well in DMEM containing 10% FBS. Cells were transfected with 100 pmol/mL of mRNAspecific siRNA or scrambled control siRNA. The reduction in PSF levels was confirmed by western blot analysis.

Measurement of Cell Proliferation
PSF was knocked down in HT-29 and DLD-1 cells, which were seeded in 96-well culture plates (5610 3 cells/well) and incubated for 24 h. Cell proliferation was determined using the Cell Counting Kit-8 (Dojindo, Kumamoto, Japan): 10 mL of Cell Counting Kit-8 solution was added to the medium and incubated for 2 h in an incubator with 5% CO 2 ; the amount of orange formazan dye produced was calculated by measuring the absorbance at 450 nm in a microplate reader (Awareness Technology, Inc., Palm City, FL, USA).

Detection of Cytoplasmic Vacuolization
DLD-1 and HT-29 cells were grown on 96-well plates in DMEM for 24, 48, and 72 h after transfection with PSF siRNA. At these time points, cells were examined under an Olympus fluorescent microscope. Images were analyzed by counting the total number of cells and the number of vacuolated cells.
PPARc activation was determined in HT-29 or DLD-1 cells transfected with 125 ng of the pGL3-PPRE-acyl-CoA oxidase luciferase vector, 62.5 ng of the pcDNA3.1-PPARc vector, and 12.5 ng of the pSV-b-galactosidase (Promega) vector, which were constructed as previously reported [21,22]. Twenty-four hours after transfection, cells were treated with Opti-MEM (Invitrogen) containing the test compound dissolved in DMSO (up to 0.1%) and cultured for an additional 20 h. Luciferase activity was measured with the ONE-Glo Luciferase Assay System (Promega) using a LuMate microplate luminometer (Awareness Technology, Inc., Palm City, FL, USA).

Statistical Analysis
Student's t-test was used for statistical comparisons. Differences were considered significant when the P-value was below 0.05.

Protein-protein Interactions Assessed by Pull-down Experiments
Pull-down experiments with His-tagged fusion proteins attached to metal affinity beads are a screening technique for the identification of protein-protein interactions. Using the 66His-tagged PPARc as bait (Fig. 1A, right panel), we successfully captured a potential target protein (100 kDa) from HT-29 nuclear extracts (Fig. 1A). After extensive washing, bound proteins and the captured protein were excised from the gel, trypsin-digested, and analyzed by peptide mass fingerprinting with MALDI-MS. Tandem mass spectrometry (MS/MS) profiles identifying the PSF protein are shown in Fig. 1A. Because PSF is a nuclear protein, we then carried out cell fractionation, western blotting analysis, and immunostaining of PSF. As shown in Fig. 1B, in HT-29 and DLD-1 cell lines, PSF localized predominantly within the nuclear pellet. On the other hand, in HT-29 cells, PPARc localized within the cytosolic and nuclear fractions. To further investigate the interaction between PPARc and PSF, we performed co-immunoprecipitation (co-IP) experiments using nuclear extracts. As shown in Fig. 1C, PSF was detected with an anti-PSF antibody after immunoprecipitation of nuclear extracts from HT-29 and DLD-1 cells with an anti-PPARc antibody. Thus, PSF and PPARc interact within colon cancer cells.

Interaction of pFN-PSF and pFN-PPARc Fusion Proteins in CV-1 Cells
To investigate the potential interaction between PSF and PPARc, we analyzed their interaction in a mammalian twohybrid assay in CV-1 cells. CV-1 cells were used because they do not express PPARc [21]. As expected from previous experiments, co-expression of PSF, fused to the GAL4 DNAbinding domain, and PPARc, fused to the VP16 activation domain, induced GAL4 promoter-driven luciferase expression (3.0-fold over that with empty vectors, Fig. 2A). The effect of rosiglitazone on the ability of PPARc-PSF to induce luciferase expression was analyzed as shown in Fig. 2B. PPARc activation did not significantly affect the PPARc-PSF association. Next, we determined the physical location of the interaction sites. PSF is composed of 707 amino acids (aa), has a molecular mass of 76 kDa, and consists of 2 structural and functional domains [23]. In order to investigate which of these domains are crucial for the interaction with PPARc, we constructed PSF deletion mutants. Interaction of chimeric Gal4-PSF deletion mutants with VP16-PPARc was assessed using the mammalian twohybrid reporter gene assay. As shown in Fig. 2C, loss of amino acids 1-290 of PSF had no effect on the interaction. Thus, the N-terminal domain is not essential for the interaction between these proteins. Loss of amino acids 291-370 of PSF disrupted the interaction between PSF and PPARc. Deletion of amino acids 371-450, 451-662, and 452-707 of PSF also disrupted the interaction with PPARc. Taken together, our results identified the first nucleotide binding domain (aa 291-370) as an important molecular site for PPARc binding.

PPARc Activation does not Regulate PSF Expression in HT-29 and DLD-1 Cells
To determine PPARc's role in regulating PSF expression, we examined the effect of a PPARc agonist, rosiglitazone (10 mM), on PSF expression in DLD-1 and HT-29 cells. As shown in Figs. 3A and B, in HT-29 cells, stimulation with rosiglitazone did not inhibit PSF mRNA and protein expression; however, the expression levels decreased in DLD-1 cells stimulated with rosiglitazone. The selective and irreversible PPARc antagonist GW9662 (10 mM) did not inhibit PSF expression in either cell line. Furthermore, addition of GW9662 and rosiglitazone did not change PSF mRNA and protein expression. These results suggest that PSF expression is PPARc-independent and indicate that mechanisms other than PPARc stimulation regulate the PPARc-PSF axis.

Knockdown of PSF Inhibits Cell Proliferation and Induces Vacuolation in DLD-1 Cells
To evaluate the effects of PSF on the proliferation of HT-29 and DLD-1 cells, PSF expression was knocked-down using siRNA. As shown in Fig. 4A, knockdown of PSF expression in HT-29 and DLD-1 cells using siRNA was effective, as evidenced by western blot analysis using an anti-PSF antibody. As shown in Fig. 4B, real-time quantitative RT-PCR analysis showed that PSF mRNA in siRNA transfected cells was knocked down by 80-90% compared to expression in untransfected (UT) control cells. DLD-1 cells appeared as empty, lucent spaces in phase contrast images at 48 h after siRNA transfection (Fig. 4C). At 48 and 72 h after transfection, approximately 30 and 40% of the total number of cells, respectively, showed extensive vacuolization of the cytoplasm. Cell vacuolation increased in number and size, occupying increasingly larger areas of the cytoplasm in a timedependent manner. Next, we determined the effect of PSF knockdown on cell proliferation by using a colorimetric assay. As shown in Fig. 4D, PSF knockdown severely inhibited cell proliferation in DLD-1 cells, which have a lower endogenous level of PPARc than HT-29 cells. Interestingly, PSF knockdown weakly inhibited cellular proliferation in HT-29 and LOVO cells, compared to proliferation in DLD-1 and Caco-2 cells. Thus, HT-29 cells appear to be more resistant to PSF knockdown-induced growth inhibition.

PPARc Expression Level is Critical for Protection Against PSF Knockdown-induced Cell Growth Inhibition
As shown in Figs. 5A and B, we investigated PPARc and PSF mRNA and protein expression in 4 human colon cancer cell lines, HT-29, DLD-1, Caco-2, and LOVO. Total RNA was isolated from untreated cells. Real-time PCR analysis revealed that the relative level of PPARc mRNA in these cells was in the order HT-29. LOVO.Caco-2. DLD-1. Similarly, our previous report suggested that the PPARc protein level is high in HT-29 and LOVO cells and low in Caco-2 and DLD-1 cells [19]. This finding is also consistent with a report by Kitamura et al. [24]. Next, to test the functionality of PPARc, we transfected the cell lines with a luciferase reporter plasmid. HT-29 and LOVO cells were more responsive to rosiglitazone than DLD-1 and Caco-2 cells (Fig. 5C). Because we observed an inverse correlation between the level of PPARc expression and the sensitivity to PSF knockdown-induced inhibition of proliferation (see Fig. 4D), we reasoned that increasing the PPARc expression level in transfected colon cancer cells with naturally low levels of PPARc should reverse the PSF knockdown-induced effect on cell proliferation. To test this, we introduced the pcDNA3.1-FLAG-PPARc plasmid (Fig. 5D) into the 4 human colon cancer cell lines 24 h after transfection with PSF siRNA. As shown in Fig. 5E, cell proliferation was increased by PSF knockdown, and this inhibitory effect was reversed by PPARc overexpression. These data demonstrate that selective expression of PPARc reverses PSF knockdown-dependent cell growth inhibition.

Knockdown of PSF Expression by siRNA Induces Apoptosis in DLD-1 Cells
The decreased cell proliferation observed in conjunction with the morphological observations suggested that DLD-1 cells treated with PSF siRNA undergo apoptosis. To test this, cultures of DLD-1 and HT-29 cells were stained with Hoechst 33258 dye for 48 and 96 h. Hoechst 33258, a DNA sensitive fluorochrome, was used to assess changes in nuclear morphology following PSF knockdown. After knockdown of PSF for 96 h, DLD-1 but not HT-29 cells underwent morphologic changes typical of apoptosis, e.g., chromatin condensation and nuclear shrinkage (Figs. 6A and  B). To verify the type of cell death induced by PSF knockdown, western blot analysis was performed to confirm that caspase-3 was activated by PSF knockdown. Caspase-3 has a key role in apoptosis, being responsible for the proteolytic cleavage of many key proteins [25]. Caspase-3 was primarily present in its 35-kDa pro-form (Fig. 6C) in untreated DLD-1 cells. Following 24 h exposure to 5-fluorouracil, which was used as a positive control, the p17 fragment of cleaved, active caspase-3 was detected. The p17 fragment was also detected after treatment with PSF siRNA for 96 h. These results indicate that PSF knockdown induces apoptosis in DLD-1 cells but not in HT-29 cells and that decreasing PSF expression in DLD-1 cells can inhibit cell proliferation.

Figure 1. Physical interaction between PPARc and PSF in HT-29 cells. (A)
Pull-down affinity-binding assay with purified PPARc. Full-length PPARc expressed in E. coli as a 66His-tagged fusion protein was isolated and purified using TALON resin (upper right panel). The 66His-tagged PPARc protein was incubated with nuclear extracts isolated from HT-29 cells. After washing with wash buffer, the resin was collected by centrifugation, and SDS-PAGE was performed with a 5-20% (w/v) acrylamide gel. The separated protein bands were visualized by Coomassie Brilliant Blue. The protein band (a) was excised from the gel, digested with trypsin, and identified by mass fingerprinting. The number of peptides, percentage of sequence coverage, and the accession number for the protein are given in Table S1. (B) Verification of the localization of PSF in nuclear and cytosolic extracts from HT-29 and DLD-1 cells. Cytosolic extracts and nuclear extracts were prepared from cells and analyzed by immunoblotting using an antibody against human PSF. Immunofluorescence staining of formalin-fixed HT-29 and DLD-1 cells shows the nuclear localization of PSF (right panel). (C) HT-29 cells were lysed with lysis buffer and then analyzed by co-immunoprecipitation and western blotting with anti-PSF antibody. Beads alone and normal rabbit serum (IgG) were used as negative controls. Arrows show the position of PSF (100 kDa). doi:10.1371/journal.pone.0058749.g001

Protein Abundance Changes Upon PSF Knockdown
Next, we carried out a comparative proteomic analysis of proteins identified after PSF knockdown in DLD-1 and HT-29 cells. For this study, crude whole cell pellets were isolated from cells and lysed using the freeze-thaw method followed by Dounce homogenization and centrifugation (13,0006g, 20 min, 4uC). Using mass spectrometry and proteomics analysis, we identified 25 distinct proteins whose levels were significantly altered following PSF knockdown in both cell lines (Table S1). We then identified candidate proteins potentially involved in the PPARc-PSF interaction and apoptosis. As expected from previous experiments (Figs. 4A-E), many of these proteins play a role in apoptosis and cell cycle regulation and act as molecular chaperones. Interestingly, in DLD-1 cells, voltage-dependent anion selective channel protein 2 (VDAC2) was up-regulated. We investigated VDAC2 mRNA and protein expression in DLD-1 cells after PSF knockdown. As shown in Fig. 7A, real-time PCR and western blot analysis confirmed that VDAC2 and Bax were upregulated under PSF knockdown conditions in DLD-1 cells. Next, we examined cells by fluorescence microscopy after staining with the mitochondria-specific dye rhodamine 123 [26] to determine whether there were changes in mitochondrial morphology after PSF knockdown. Cells were pre-incubated with rhodamine 123 for 30 min. Cells showed intense vacuolation after PSF knockdown (72 h), mostly in the perinuclear region (Fig. 7B). Large and medium size cells tended to be vacuolated. Vacuoles were never observed in mitochondria and nuclei. Next, we investigated whether PSF knockdown causes reactive oxygen species (ROS) formation in DLD-1 cells using 29,79-dichlorofluorescin diacetate (DCF) as a reporter of intracellular oxidant production. A DCF response was detected at 72 h post-transfection (Fig. 7C).

Discussion
In the present study, we showed that PSF interacts with PPARc in colon cancer cells. The interaction was originally revealed by using a mammalian two-hybrid assay and was subsequently confirmed in cell cultures by pull-down assays and co-immunoprecipitation experiments. PSF is a multifunctional protein that functions as transcriptional repressor for several nuclear receptors [27]. Increased expression of PSF in tumor cells suppresses tumorigenesis [14]. This finding suggests that PSF has a central role in the regulation of cell proliferation and tumorigenesis and therefore presents a potential therapeutic strategy for cancer. However, the function of PSF in regulating colon cancer cells has not been reported.
To date, a limited number of direct targets for PPARc have been identified in studies using colon cancer cells. PPARc has been found in cells from various lineages, e.g., colon cancer [19], stomach cancer [28], breast cancer [29], and prostate cancer [30]. PPARc is recognized as a transcription factor that participates in the regulation of adipocyte differentiation. PPARc agonists are currently in clinical use for the treatment of Type II diabetes [31]. While previous studies demonstrated that some PPARc agonists inhibit the growth of cancer cells [32], many reports show that PPARc ligand-mediated growth inhibition seems to vary depending on the cancer cell type. In colon cancer cells, the growthsuppressing effect of PPARc ligands evident in in vitro studies was not clearly confirmed by in vivo studies [33]. Activation of PPARc increases colonic polyps in the APC +/min mouse model of colon carcinogenesis [34,35]. These results may be due, in part, to PPARc-dependent and -independent pathways.
The results of this study and those presented in a previous report [19] suggest that PPARc overexpression in DLD-1 cells impedes cell growth inhibition. The effect on growth inhibition may depend on the quantity of PPARc protein present and its interaction with PSF. In this study, our results showed that PPARc mRNA and protein were expressed at various levels in 4 colon cancer cell lines. The effects of PSF expression in these cell lines varied, in a manner that correlated with the level of PPARc expression. The proliferation of DLD-1 and Caco-2 cells, which express a low level of PPARc, was significantly inhibited by knockdown of PSF, whereas the proliferation of HT-29 and LOVO cells, which express a higher level of PPARc, was inhibited weakly or not at all. In DLD-1 cells, but not HT-29 cells, PSF knockdown also induced morphological changes associated with apoptosis, i.e., cell shrinkage and condensation of nuclear chromatin. Cornillon et al. reported that cells undergo extensive vacuolation as they proceed towards apoptosis [36]. Consistently, PSF knockdown did not induce vacuolation in HT-29 cells, whereas increased cell vacuolation was observed after PSF knockdown in DLD-1 cells. Thus, we observed distinct cell type-specific differences associated with the PPARc-PSF interaction. In DLD-1 cells, but not HT-29 cells, PSF knockdown also induced morphological changes associated with apoptosis, i.e., cell shrinkage and condensation of nuclear chromatin. Furthermore, PSF knockdown did not induce vacuolation in HT-29 cells, whereas increased cell vacuolation was observed after PSF knockdown in DLD-1 cells. Thus, we observed distinct cell typespecific differences associated with the PPARc-PSF interaction. Cornillon et al. reported that cells undergo extensive vacuolation as they proceed towards apoptosis [36]. We suggest that the process of cytoplasmic vacuolation can lead to a particular and distinctive form of cell death.
The main hypothesis driving this study is that activation of apoptosis plays a pivotal role in the PSF-PPARc axis in colon cancer cells. During the induction of apoptosis, the permeability of the mitochondrial membrane changes, cytochrome c leaks into the cytoplasm, and caspases are activated [37]. Bax-induced caspase activation, via the release of cytochrome c from the mitochondria through VDAC, promotes the apoptosis pathway. Increased expression of mitochondrial VDAC and subcellular co-localization of VDAC/Bax increases mitochondrial permeability and apoptosis [38]. Thus, a particularly important outcome of the present study was the discovery that down-regulation of PSF stimulated apoptosis and markedly increased VDAC2 levels in DLD-1 cells. VDAC2 forms the pores of the outer mitochondrial membrane, and its involvement in mitochondrial-dependent apoptosis has been studied previously [39,40]. It has been reported that VDAC2 normally inhibits the proapoptotic activity of Bak and that apoptotic signals induce the dissociation of Bak from VADC2 [41]. Chandra et al. suggested that an increase in VDAC2 complex formation in stimulated HCT116 colon cancer cells might be a pro-survival mechanism activated by apoptotic stimuli [41]. VDAC isoforms are also important regulators of mitochondrial metabolic activity, which is required for ROS production [42]. An extensive number of reports indicate that an increase in intracellular ROS production induces apoptosis through the mitochondrial pathway [43]. However, it is unclear whether ROS produced in the mitochondria are important in the regulation of cell death. Interestingly, in this study, we detected ROS production after PSF knockdown in DLD-1 cells. We propose that elevated levels of ROS may induce PSF-PPARc signaling and regulate the apoptotic machinery. Although the mechanism of growth inhibition via the PPARc-PSF axis in colon cancer cells has not been fully elucidated, our present study demonstrated that the PSF expression level is an important regulatory element for colon cancer cell growth. Understanding the diverse molecular interactions between PSF and it targets in the cancer system will provide insight into the pathogenesis of colon cancer. Therapies directed at PPARc expression or its binding partners may lead to novel approaches to treat colon cancer. The effect of PSF on PPARc targets and their contributions to PSFmediated cellular processes requires further investigation ( Fig. S1 and Fig. 8). PPARc-interaction partners may provide insight into the biological functions of PPARc and provide us with a better understanding of the events involved in colon cancer.  their interaction in a mammalian two-hybrid assay in CV-1 cells. Treatment with rosiglitazone did not further stabilized PSF-PPARc complex. Real-time PCR measurement of PPARc mRNA expression under PSF knockdown conditions in HT-29 and DLD-1 cells (C) The relative PPARc levels normalized to 18S rRNA are expressed as mean 6 SEM (n = 3), **P,0.01.

(EPS)
Table S1 MALDI-TOF MS analysis for PSF siRNA transfected HT-29 and DLD-1 cells. Analyses were performed using the AB SCIEX TOF/TOF TM 5800 System (AB SCIEX, Foster City, CA, USA). Protein identification was performed through ProteinPilot TM software (AB Sciex, Framingham, MA, USA) using the UniProt database. We identified 25 distinct proteins whose levels were significantly altered following PSF knockdown in both cell lines. (XLS)