PIASy Represses CCAAT/Enhancer-binding Protein δ (C/EBPδ) Transcriptional Activity by Sequestering C/EBPδ to the Nuclear Periphery*

CCAAT/enhancer binding proteinδ (C/EBPδ) plays a key role in mammary epithelial cell G0 growth arrest, and “loss of function” alterations in C/EBPδ have been reported in breast cancer and acute myeloid leukemia. C/EBPδ is regulated at the transcriptional, post-transcriptional, and post-translational levels, suggesting tight control of C/EBPδ content and function. Protein inhibitors of activated STATs (PIASs) regulate a growing number of transcription factors, including C/EBPs. HC11 nontransformed mammary epithelial cells express PIAS3, PIASxβ, and PIASy, and all three PIAS family members repress C/EBPδ transcriptional activity. PIASy is the most potent, however, repressing C/EBPδ transcriptional activity by >80%. PIASy repression of C/EBPδ transcriptional activity is dependent upon interaction between the highly conserved PIASy N-terminal nuclear matrix binding domain (SAPD) and the C/EBPδ transactivation domain (TAD). PIASy repression of C/EBPδ transcriptional activity is independent of histone deacetylase activity, PIASy E3 SUMO ligase activity, and C/EBPδ sumoylation status. PIASy expression is associated with C/EBPδ translocation from nuclear foci, where C/EBPδ co-localizes with p300, to the nuclear periphery. PIASy-mediated translocation of C/EBPδ is dependent upon the PIASy SAPD and C/EBPδ TAD. PIASy reduces the expression of C/EBPδ adhesion-related target genes and enhances repopulation of open areas within a cell monolayer in the in vitro “scratch” assay. These results demonstrate that PIASy represses C/EBPδ by a mechanism that requires interaction between the PIASy SAPD and C/EBPδ TAD and does not require PIASy SUMO ligase activity or C/EBPδ sumoylation. PIASy alters C/EBPδ nuclear localization, reduces C/EBPδ transcriptional activity, and enhances cell proliferation/migration.

preadipocytes prior to commitment to differentiation, and adipocyte differentiation is defective in C/EBP␦ knock-out mouse embryo fibroblasts (23)(24)(25). Recent work indicates that C/EBP␦ inhibits growth and promotes self-renewal in human limbic stem cells, suggesting a role for C/EBP␦ in the maintenance of stem cell pluripotency (26). In addition, "loss of function" alterations in C/EBP␦ have been reported in human breast cancer and acute myeloid leukemia (27)(28)(29)(30). Collectively, these results demonstrate that C/EBP␦ functions in growth control and cell fate determination, and alterations in C/EBP␦ are associated with tumorigenesis.
PIAS gene family members (PIAS1, PIAS3, PIASy, and PIASx) regulate transcription factor function by sumoylation, inhibition of DNA binding, HDAC recruitment, and sequestration in nuclear foci or in the nuclear periphery (31,32). PIAS proteins are characterized by a highly conserved domain structure that includes a SAP domain (SAPD) that contains an LXXLL motif and a conserved Ring finger-like domain (RFD) that is required for PIAS SUMO E3 ligase activity (31,32). PIAS proteins regulate C/EBP␣ and C/EBP⑀ sumoylation status and function (33)(34)(35); however, it is unknown if PIAS proteins regulate C/EBPs exclusively by sumoylation or if alternate mechanisms (i.e. HDAC recruitment or subnuclear sequestration) are also utilized. In this report, we demonstrate that PIASy represses C/EBP␦ transcriptional activity by sequestering C/EBP␦ in the nuclear periphery. PIASy-mediated sequestration of C/EBP␦ is associated with reduced expression of adhesion-related C/EBP␦ target genes and enhanced mammary epithelial cell proliferation/migration. These results suggest a potential role for PIASy-C/EBP␦ interactions in the control of mammary epithelial cell growth or migration.

EXPERIMENTAL PROCEDURES
Plasmid Construction-C/EBP␦ constructs (C/EBP␦ fulllength cDNA, C/EBP␦ ⌬TAD (amino acids 102-268), and K120R mutant) were cloned into the PCDNA3.1/V5-His-TOPO TA expression vector (Invitrogen). The FLAG-PIASy expression plasmid was constructed by PCR amplification of the PIASy coding sequence using a forward primer bearing a FLAG tag coding sequence, and the PCR product was inserted into the PCDNA3 expression vector. The ⌬SAP and ⌬RFD PIASy mutants were generated by site-directed mutagenesis, where amino acids 12-46 were deleted in ⌬SAP and amino acids 319 -363 were deleted in ⌬RFD. The GST-C/EBP␦ expression vector was generated by cloning the full-length C/EBP␦ cDNA into pGEX-4T-1 vector (Amersham Biosciences) for expression in the Escherichia coli strain DE3 (BL21). ⌬TAD-(102-268), DBLZ-(171-268), ⌬LZ-(1-233), and TAD-(1-102) GST fusion protein expression plasmids were constructed by PCR amplification of the corresponding cDNA sequences and cloned into the same vector. All expression vectors and inserts were verified by DNA sequencing.
Luciferase Reporter Assay-HC11 cells were cultured in 12-well plates. Experiments were performed by co-transfecting each well with 100 ng of wild type C/EBP␦ full-length (FL) or C/EBP␦ K120R mutant plus 100 ng of C/EBP consensus site promoter-driven luciferase reporter vector (pC/EBP-Luc, (ATTGCGCAAT) 3 (catalog number 240112; Stratagene, La Jolla, CA), 200 ng of PIAS protein, and/or 100 ng of SENP1 expression vectors (Open Biosystems, Huntsville, AL). Cotransfection experiments were also performed with increasing amounts of SUMO-1, SUMO-2, and SUMO-3 (0, 50, and 100 ng). A Renilla luciferase expression plasmid was used as the transfection efficiency control (1 ng/well). Luciferase activity was analyzed 24 h after transfection using a dual luciferase reporter assay system (Promega, Madison, WI), and the luciferase results were normalized to Renilla luciferase activity. The normalized luciferase data for each independent experiment were derived from triplicate wells; experiments were replicated 2-3 times. Statistical analysis was performed with pooled data using one-way analysis of variance and post-test analysis by Dunnet's test with each PIAS treatment compared with the no PIAS control. Statistical significance was set at ␣ ϭ 0.05. Results from transfection experiments in which Trichostatin A was used are representative of three independent experiments with at least two replicates/experiment. Pooled data were analyzed with two-way analysis of variance with statistical significance set at ␣ ϭ 0.05.
In Vivo Sumoylation Assay-HC11 cells (cultured in 10-cm dishes) were co-transfected with HA-SUMO (SUMO-1, SUMO-2, or SUMO-3) expression constructs (2 g), C/EBP␦-V5-His (wild type or K120R mutant) (2 g), FLAG-PIASy (2 g), and SENP1 expression vectors (2 g). The total amount of DNA was equalized for each treatment by the addition of PCDNA3 vector DNA. Cells were harvested 24 h after transfection. Cell pellets were lysed in buffer A (8 M urea, 0.1 M Na 2 HPO 4 /NaH 2 PO 4 , 10 mM imidazole, pH 8.0), and lysate protein levels were quantitated by the Micro-BCA protein assay (Pierce). Equal amounts of lysate protein were incubated with Ni 2ϩ -nitrilotriacetic acid-agarose beads (Qiagen, Valencia, CA), and the beads were sequentially washed with buffer A, buffer A/TI (1 volume of buffer A, 3 volumes of buffer TI), and buffer TI (25 mM Tris-HCl, 20 mM imidazole, pH 6.3). Bound proteins were eluted in 40 l of 2ϫ Laemmli sample buffer by boiling for 5 min. Samples were analyzed by SDS-PAGE/Western blotting with anti-HA antibody (Cell Signaling, Danvers, MA).
GST Pull-down Assay-E. coli expressing GST-C/EBP␦ fusion proteins were lysed with lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM DTT, 1% Triton X-100, 1 mM EDTA, pH 7.5) supplemented with protease inhibitors, including 1 mM PMSF, 1 g/ml aprotinin, and 1 g/ml pepstatin. GST-C/EBP␦ fusion proteins were immobilized on glutathione-Sepharose 4B beads (Amersham Biosciences). HC11 cells expressing wild type or mutant FLAG-PIASy were lysed in the same lysis buffer, and lysates were incubated overnight with glutathione-Sepharose beads bound with GST-C/EBP␦ fusion proteins at 4°C. Glutathione-Sepharose beads were then collected by centrifugation, washed with lysis buffer, and boiled in 40 l of 2ϫ Laemmli sample buffer for 5 min. Samples were analyzed by SDS-PAGE/Western blotting with anti-FLAG antibody (Cell Signaling).
Immunofluorescent Staining and Confocal Microscope Imaging-HC11 cells were transfected with C/EBP␦-V5-His vector constructs (wild type or mutant) using the Lipofectamine Plus protocol and cultured on coverslips with or without FLAG-PIASy (wild type or mutant) expression vector. Approximately 24 h after transfection, cells were fixed with 2% paraformaldehyde/phosphate-buffered saline and permeabilized with 1% Triton X-100/phosphate-buffered saline. Cells were blocked in 10% normal goat serum in 0.1% Triton X-100/phosphate-buffered saline at room temperature for 1 h and then incubated with mouse anti-V5 and rabbit anti-p300 or rabbit anti-FLAG antibody at 4°C overnight in a humidified chamber. Cells were then incubated with Alexa Fluor 633 goat anti-mouse and Alexa Fluor 488 goat anti-rabbit secondary antibodies (Invitrogen) at room temperature for 1 h. Coverslips were mounted with a drop of Vectashield mounting media with 4Ј,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA). Protein localization was visualized using a Leica DM IRE2 confocal microscope system (Wetzlar, Germany).
HC11 Cell Proliferation and Migration; the "Scratch" Assay-The in vitro"scratch" assay was performed essentially as described by Liang et al. (36). Briefly, an open area was produced in ϳ90% HC11 cell monolayers using a 200-l pipette tip, and repopulation of the open area was assessed by crystal violet staining at 0, 24, and 48 h. Staining was performed with 0.25% crystal violet in 40% methanol at room temperature for 5 min followed by extensive washing and air drying.
To investigate the influence of C/EBP␦ on the capacity of cells to repopulate the open area produced by the scratch protocol, transfections were performed using the Lipofectamine transfection reagent and Plus reagent according to the manufacturer's instructions (Invitrogen). C/EBP␦ levels were reduced using the pSilencer TM Neo kit (Ambion, Inc., Austin TX). Briefly, HC11 cells were subjected to the Lipofectamine transfection protocol in which cells received either the pSilencer 2.1 vector (vector control), or a C/EBP␦ siRNA-pSilencer 2.1 construct (C/EBP␦ siRNA). Following transfection, the vector control and C/EBP␦ siRNA-treated HC11 cells were plated in CGM plus G418 (400 g/ml). Stable vector control and C/EBP␦ siRNA cell lines were selected and maintained in G418 (200 g/ml). To investigate the influence of PIASy expression on repopulation of the open area created in the monolayer by the scratch protocol, HC11 cells were subjected to transient transfection with the C/EBP␦ or C/EBP␦ plus PIASy expression constructs using the Lipofectamine transfection reagent and Plus reagent protocol according to the manufacturer's instructions (Invitrogen). Following transfection, HC11 cells were cultured in CGM and grown to a monolayer, and the scratch assay was performed as described above.
mRNA Isolation and Real Time PCR-Total mRNA was isolated using RNAzol B (Tel-Test, Inc., Friendswood, TX) according to the manufacturer's instructions. Total mRNA (1 g) was used for reverse transcription using the reverse transcriptase kit (Invitrogen). The reverse transcription products were amplified by real time PCR using the Light-Cycler 480 real time PCR system (Roche Applied Science). Amplification was performed in a total volume of 20 l containing 10 l of 2ϫ SYBR Green PCR master mix, 0.2 l of forward and reverse primers, and 1 l of cDNA in each reaction. PCR specificity was verified by assessing the melting curves of each amplification product. Real time PCR data were normalized to the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA control. The following primers were used to assess C/EBP␦ cell adhesion related target gene mRNA levels: glycoprotein V (GP5) sense, 5Ј-CGCCAGCCTGT-CGTTCT; GP5 antisense, 5Ј-GCCTGTTATTGGGACTTT-CAC; integrin ␤8 (ITGB8) sense, 5Ј-TTCTCCTGTCCCTATCT-CCA; ITGB8 antisense, 5Ј-TGAGACAGATTGTGAGGGTG; protocadherin 9 (PCDH9) sense, 5Ј-ACAGCCACCACGGTCC-TCTA; PCDH9 antisense, 5Ј-CCCTTGTTGTTCCCGCTCAC; GAPDH sense, 5ЈCTCACTGGCATGGCCTTCCG; GAPDH antisense (37), 5ЈACCACCCTGTTGCTGTAGCC. The -fold change in specific mRNA levels was calculated using the comparative CT (⌬⌬CT) method (38). Results are presented as mean Ϯ S.E. of the -fold changes derived from three experiments with triplicate analyses performed for each treatment. Student's t test was used to analyze the real time PCR results, and the -fold change in individual C/EBP␦ target gene mRNA levels was considered significant at p Ͻ 0.05.

PIAS Family Members Repress C/EBP␦ Transcriptional
Activity-Previous reports have demonstrated that PIAS mediated sumoylation of the conserved regulatory domain motif (RDM) of C/EBP⑀ increases C/EBP⑀ transcriptional activity (33,34,39). These results suggest that PIAS proteins can regulate the function of C/EBPs in cells that co-express PIAS and C/EBP family members. Kim and co-workers reported that C/EBP␦ contains an RDM and that C/EBP␦ can be sumoylated; however, the role of specific PIAS proteins in C/EBP␦ sumoylation was not defined (39). C/EBP␦ gene expression is highly induced in G 0 growth-arrested human and mouse mammary epithelial cells, and C/EBP␦ protein levels are essentially undetectable in growing cells (11, 12, 14, 15, 18 -21). We hypothesized that PIAS family members expressed in G 0 growth-arrested cells could influence C/EBP␦ transcriptional activity and alter cell function. Since the influence of growth status on the expression of PIAS family members in mammary epithelial cells had not been previously reported, we assessed the expression of three PIAS family members (PIAS3, PIASx␤, and PIASy) in exponentially growing and G 0 growth-arrested HC11 mouse mammary epithelial cells. The results demonstrated that the mRNA levels of all three PIAS family members were present at relatively constitutive levels in growing and G 0 growtharrested HC11 cells (Fig. 1A). The constitutive expression of PIAS family members suggests that PIAS proteins function in a broad range of cellular contexts.
Having determined that PIAS family members were expressed in the HC11 mammary epithelial cell line, we next investigated the influence of PIAS proteins on C/EBP␦ tran-scriptional activity. Full-length PIAS3, PIASx␤, and PIASy cDNAs cloned into the pcDNA3 expression vector were co-transfected into HC11 cells with a pC/EBP-Luciferase (pC/EBP-Luc) reporter construct, and luciferase activity was assessed. The results demonstrated that each of the three PIAS family members significantly repressed C/EBP␦ transcriptional activity; however, PIASy was the most potent, inhibiting C/EBP␦ transcriptional activity by ϳ80% (Fig.  1B). In addition, PIASy inhibited C/EBP␦ transcriptional activity in a dose-dependent manner, demonstrating a direct correlation between PIASy levels and inhibition of C/EBP␦ transcriptional activity (Fig. 1C).
PIASy inhibits transcriptional activity by a multiple mechanisms, including histone deacetylase (HDAC) recruitment (37,40). To investigate HDAC recruitment as the mechanism underlying PIASymediated inhibition of C/EBP␦ transcriptional activity, HC11 cells were co-transfected with PIASy plus the pC/EBP-Luc reporter construct in the presence or absence of trichostatin A, an HDAC inhibitor (37). PIASy expression significantly reduced C/EBP␦ transcriptional activation of the pC/EBP-Luc reporter construct (Fig. 1D). However, treatment with increasing doses of trichostatin A (0 -330 nM) had no effect on PIASy-mediated inhibition of C/EBP␦ transcriptional activation of the pC/EBP-Luc reporter construct, as assessed by luciferase activity (Fig. 1D). These results suggest that HDAC recruitment does not play a major role in PIASy-mediated inhibition of C/EBP␦ transcriptional activity.

The Lys 120 Residue within the C/EBP␦ RDM ((I/V/L)KXEP) is the Site for Post-translational SUMO Modification; however, Sumoylation Does Not Influence C/EBP␦ Transcriptional
Activity-Having determined that PIASy significantly inhibits C/EBP␦ transcriptional activity, we sought to determine if PIASy expression increased C/EBP␦ sumoylation. Using transient transfection and co-expression/pull-down assays, we found that C/EBP␦ is sumoylated by SUMO-1 ( Fig. 2A), SUMO-2, and SUMO-3 (data not shown). PIASy co-expression, however, did not enhance C/EBP␦ sumoylation ( Fig.  2A, lane 2 versus lane 3). Co-expression of SENP1, an isopeptidase that cleaves the SUMO moiety from protein sub-FIGURE 1. PIASy represses C/EBP␦ transcriptional activity. A, total RNA was isolated from subconfluent, growing HC11 cells (Grow) and 48 h confluent, G 0 growth-arrested HC11 cells (GA) and reverse transcribed (RT), and the RT products were amplified with primers specific for PIASy, PIASx␤, PIAS3, and GAPDH (constitutively expressed mRNA control). Amplified products were separated by agarose gel electrophoresis and stained with EtBr. B, HC11 cells were co-transfected with C/EBP␦, pC/EBP-luciferase, and individual PIAS family members. Cells were harvested 24 h post-transfection, and luciferase activity was determined and normalized to Renilla luciferase activity (transfection efficiency control). Transfection of HC11 cells with the empty expression vector was used as the negative control for C/EBP␦ and PIAS expression and is designated as His (pcDNA3.1/V5-His-TOPO TA). The CON lane denotes transfection with C/EBP␦ plus an empty expression vector as a control for luciferase activity in the absence of PIAS family gene expression. C/EBP␦ expression levels plus the ␤-actin protein loading control are shown in the bottom panels. C, HC11 cells were co-transfected with C/EBP␦, increasing amounts of PIASy (0, 25, 50, 100, 200 ng) plus pC/EBP-Luc. Cells were harvested 24 h post-transfection, and luciferase activity was determined and normalized to the Renilla transfection efficiency control. FLAG-PIASy and C/EBP␦-V5 expression levels plus the ␤-actin protein loading control are shown in the bottom panels. D, HC11 cells were transfected with or without PIASy and treated with increasing amounts of trichostatin A (TSA) (0, 165, and 330 nM), and luciferase activity was determined. Representative C/EBP␦ levels plus the ␤-actin protein loading control are shown in the bottom panels.
strates, significantly reduced the detectable level of C/EBP␦/ SUMO conjugates ( Fig. 2A, lanes 2 and 3 versus lane 4). These results demonstrate that C/EBP␦ is a sumoylation target, but PIASy expression does not significantly enhance C/EBP␦ sumoylation.
C/EBP family members, including C/EBP␦, contain a conserved transcriptional inhibitory domain ((I/V/L)KXEP) or reg-ulatory domain motif (RDM) that is a target for sumoylation (33,39). We produced a C/EBP␦ K120R RDM mutant construct and found that the C/EBP␦ K120R mutant was not sumoylated (Fig. 2A, lane 5). These results demonstrate that the conserved Lys 120 consensus site within the C/EBP␦ RDM ((I/V/L)KXEP) is the site for post-translational SUMO modification.
Williams and co-workers (33) demonstrated that conjugation of SUMO-1 to the conserved lysine residue within the RDM in C/EBP⑀ releases the transcriptional inhibition mediated by the unmodified RDM, indicating that sumoylation is associated with increased transcriptional activation of C/EBP⑀. In contrast, Wang et al. (41) reported that sumoylation of NF-IL6␤ (C/EBP␦) reduced transcriptional activation. To gain new insights into the influence of sumoylation on C/EBP␦ transcriptional activity, HC11 cells were co-transfected with C/EBP␦ or the C/EBP␦ K120R mutant with increasing amounts of SUMO-1, SUMO-2, and SUMO-3. Consistent with previous reports, we found that the transcriptional activity of the C/EBP␦ K120R mutant was ϳ30% higher than the C/EBP␦ wild type construct (Fig. 2B) (39). However, increasing SUMO-1 levels by transient transfection had no effect on the transcriptional activity of either the wild type C/EBP␦ or the C/EBP␦ K120R mutant (Fig. 2B). Similarly, co-expression of SUMO-2 or SUMO-3 had no effect on C/EBP␦ transcriptional activity (data not shown). These results confirm the inhibitory role of the C/EBP␦ RDM on C/EBP␦ transcriptional activity and demonstrate that expression of SUMO family members does not influence C/EBP␦ transcriptional activity.
PIASy Repression of C/EBP␦ Transcriptional Activity Is Independent of C/EBP␦ Sumoylation Status-We next investigated the effect of PIASy on the transcriptional activity of wild type C/EBP␦ and the sumoylation-defective C/EBP␦ K120R mutant by transfecting HC11 cells with PIAS and C/EBP␦ expression constructs plus the pC/EBP-Luc reporter construct. The results indicate that PIASy significantly inhibits the transcriptional activity of both wild type (WT) C/EBP␦ and the C/EBP␦ K120R mutant (Fig. 3A). In addition, the results confirm the finding that the C/EBP␦ K120R mutant transcriptional activity is ϳ30% higher than that of wild type C/EBP␦ (see Figs. 2B and 3A). Co-expression of PIASy plus the SENP1 isopeptidase, which cleaves SUMO-1 from substrate proteins, slightly increased C/EBP␦ wild type and C/EBP␦ K120R transcriptional activity; however, both remained at ϳ50% of the control (CON) levels (Fig. 3A). Western blot analysis demonstrated that the protein levels of the expressed constructs (i.e. C/EBP␦ WT/C/EBP␦ K120R mutant and PIASy) were relatively even. This indicates that PIASy-mediated inhibition of wild type C/EBP␦ and C/EBP␦ K120R transcriptional activity was not due to artifactual differences in protein levels (Fig. 3B). The combined data from the PIASy, C/EBP␦ wild type, and C/EBP␦ K120R mutant co-expression experiments and the SENP1 isopeptidase treatments indicate that PIASy inhibition of C/EBP␦ transcriptional activity is sumoylation-independent.
The PIASy SAPD Interacts with the C/EBP␦ TAD-To gain a better understanding of the mechanism by which PIASy inhibits C/EBP␦ transcriptional activity, we investigated the structural domains that mediate PIASy-C/EBP␦ protein-protein interactions. The following GST tagged C/EBP␦ constructs  JULY 18, 2008 • VOLUME 283 • NUMBER 29 ( Fig. 4A) were used in pull-down assays with full-length FLAG-PIASy: C/EBP␦ FL, C/EBP␦ ⌬TAD (TAD deleted), C/EBP␦ DBLZ (TAD and regulatory domain deleted), and C/EBP␦ ⌬LZ (LZ deleted). The initial results demonstrated that the GST-C/ EBP␦ FL construct pulled down the full-length FLAG-tagged PIASy protein from HC11 cell lysates (Fig. 4B). To identify the C/EBP␦ domains that interact with PIASy, additional pulldown experiments were performed with GST-C/EBP␦ domain constructs. The results demonstrated that GST-C/EBP␦ constructs that contain the TAD (i.e. C/EBP␦ FL and C/EBP␦ ⌬LZ) pulled down the full-length FLAG-tagged PIASy protein (Fig.  4C). In contrast, GST-C/EBP␦ constructs lacking the TAD (⌬TAD, DBLZ) were ineffective in pulling down PIASy (Fig.  4C). We next performed the pull-down assays with the C/EBP␦ TAD (TAD alone, regulatory domain, DB, and LZ deleted) and demonstrated that the C/EBP␦ TAD alone is capable of pulling down PIASy (Fig. 4C, right). These results demonstrate that the C/EBP␦ TAD, a region that encompasses amino acids 1-102, is required for C/EBP␦-PIASy interaction.

PIASy Represses C/EBP␦ by Nuclear Sequestration
To investigate the domains of the PIASy protein that directly or indirectly interact with C/EBP␦, we performed pull-down experiments with GST-full-length C/EBP␦ FL and PIASy ⌬SAP (lacking the SAPD) and PIASy ⌬RFD (lacking the RFD) constructs (Fig. 4A). The results demonstrated that C/EBP␦ FL pulled down the PIASy ⌬RFD (SAPD present) construct but was ineffective in pulling down the PIASy ⌬SAP (SAPD deleted) construct (Fig. 4D). These results are consistent with a direct or indirect interaction between the C/EBP␦ TAD and the PIASy SAPD. These results also indicate that the RFD domain, which has been shown in previous reports to mediate substrate recognition and to catalyze SUMO conjugation (42,43), does not play a major role in the interaction between PIASy and C/EBP␦.
PIASy Inhibits C/EBP␦ Transcriptional Activity-Having demonstrated that full-length PIASy inhibits C/EBP␦ transcriptional activity (Fig. 1, B-D) and that full-length PIASy and PIASy ⌬RFD (SAPD intact) interact with C/EBP␦ (Fig. 4, B-D), we next investigated the influence of PIASy ⌬RFD and ⌬SAP mutant constructs on C/EBP␦ transcriptional activity. As expected, the full-length, WT PIASy significantly (ϳ70%) inhibited C/EBP␦ transcriptional activity (compare Fig. 1A and  Fig. 5A). The PIASy ⌬SAP deletion mutant, which did not inter-act with C/EBP␦ in pull-down assays (Fig. 4D), also did not inhibit C/EBP␦ transcriptional activity (Fig.  5A). However, the PIASy ⌬RFD deletion mutant, which did interact with C/EBP␦ in in vitro pull-down assays (Fig. 4D), did not inhibit C/EBP␦ transcriptional activity (Fig.  5A). Western blot analysis of whole cell lysates demonstrated that C/EBP␦ protein levels were relatively even across all treatments (Fig. 5B, top). However, the PIASy WT and PIASy ⌬SAP protein levels were ϳ5-fold higher than the PIASy ⌬RFD level (Fig. 5B, middle). The rationale for the reduced PIASy ⌬RFD levels was initially unclear; however, subsequent experiments demonstrated that PIASy ⌬RFD does not localize to the nucleus (see below; Fig. 6, F and K) despite the presence of a nuclear localization signal within the PIASy ⌬RFD construct. The inability of PIASy ⌬RFD to localize to the nucleus is consistent with the lack of effect of PIASy ⌬RFD on C/EBP␦ transcriptional activity (Fig. 5A), and the persistent presence of PIASy ⌬RFD in the cytoplasm could result in reduced PIASy ⌬RFD protein stability, resulting in reduced levels of the PIASy ⌬RFD protein (Fig. 5B).
PIASy Translocates C/EBP␦ from Discrete Nuclear Foci to the Nuclear Periphery-We next used confocal microscopy to investigate the intracellular interaction between PIASy and C/EBP␦ and the influence of this interaction on C/EBP␦ subnuclear localization in HC11 mammary epithelial cells. We previously reported that C/EBP␦ is almost exclusively localized to the nucleus in mammary epithelial cells using cell lysis protocols that separate nuclear and cytoplasmic compartments with C/EBP␦ detection by Western blot (13,18). In this report, we used confocal microscopy to extend these findings, and the results indicated that full-length C/EBP␦, C/EBP␦ ⌬TAD, and the C/EBP␦ K120R (sumoylation-defective) mutant are all localized to discrete foci within the nucleus (Fig. 6, A-C). In addition, confocal image merging demonstrated that all three C/EBP␦ constructs co-localize with p300, a key co-activator of RNA polymerase II-mediated transcription (Fig. 6, A-C). As a critical transcriptional co-activator, p300 is localized in subnuclear regions associated with active gene transcription (44,45). These results demonstrate that nuclear localized C/EBP␦ constructs colocalize with p300 in transcriptionally active nuclear foci (Fig. 6, A-C). We next used confocal analyses to demonstrate that PIASy (full-length) is localized to the nuclear periphery (Fig. 6D), PIASy ⌬SAP is diffusely dispersed within the nucleus (Fig. 6E), and PIASy ⌬RFD does not enter the nucleus at all, localizing in the cytoplasm (Fig. 6F).
Having determined that nuclear localized C/EBP␦ constructs are concentrated in nuclear foci in association with p300, we next investigated the influence of full-length PIASy and the PIASy ⌬SAP and ⌬RFD constructs on C/EBP␦ nuclear localization. Co-expression of full-length C/EBP␦ plus full-length PIASy resulted in the dramatic translocation of C/EBP␦ from

. PIASy represses C/EBP␦ transcriptional activity independent of C/EBP␦ sumoylation status.
A, HC11 cells were co-transfected with C/EBP␦ WT or C/EBP␦ Lys 120 mutant (K120R) plus pC/EBP-luc plus empty expression vector (pcDNA3.1/V5-his) (His) (control; CON), PIASy, or PIASy plus SENP1. Luciferase activity was assessed and normalized to Renilla luciferase control activity. Analysis of variance with pooled data demonstrated a significant effect of PIASy on C/EBP␦ transcriptional activity compared with control (␣ ϭ 0.05). B, Western blot analysis of C/EBP␦ WT or C/EBP␦ Lys 120 mutant, PIASy, and ␤-actin (loading control) protein levels. Western blot results shown are representative of three independent experiments. discrete nuclear foci to the nuclear periphery (Fig. 6G). Confocal image merging demonstrated that C/EBP␦ and PIASy colocalized to the nuclear periphery (Fig. 6G). Co-expression of the C/EBP␦ ⌬TAD mutant plus full-length PIASy resulted in differential localization of both proteins within the nucleus with the C/EBP␦ ⌬TAD mutant present in nuclear foci and the full-length PIASy localized to the nuclear periphery (Fig. 6H). Co-expression of the sumoylation-defective C/EBP␦ K120R mutant plus full-length PIASy resulted in the co-localization of both proteins to the nuclear periphery (Fig. 6I). However, coexpression of full-length C/EBP␦ plus the PIASy ⌬SAP (SAP domain-deleted) mutant did not alter C/EBP␦ subnuclear localization from nuclear foci (Fig. 6J). Co-expression of fulllength C/EBP␦ plus the PIASy ⌬RFD (RFD domain-deleted) mutant resulted in a complex localization pattern in which C/EBP␦ remained within nuclear foci and the PIASy ⌬RFD mutant localized to the cytoplasm, exhibiting an apparent defect in nuclear import despite the presence of a nuclear localization signal (Fig. 6K).
Previous reports have shown that C/EBP␤ and C/EBP␦ interact with CBP/p300, and this interaction is associated with increased transcriptional activity (46,47). To determine if PIASy expression induced p300 subnuclear localization in a manner similar to C/EBP␦, we expressed PIASy and assessed endogenous p300 nuclear localization. The results demonstrated that p300 remained localized in discrete nuclear foci and did not co-localize to the nuclear periphery with PIASy (Fig. 6L). This indicates that PIASy does not translocate intact C/EBP␦/p300 transcription complexes and suggests that PIASy may interact with and translocate "free" or uncomplexed C/EBP␦ prior to or after interaction with p300 on target gene promoters. Finally, the specificity of the primary and secondary antibodies used in these experiments was assessed by confocal microscopy, and the results were negative, indicating that no antibody binding occurs in the absence of specific target proteins (data not shown).

PIASy Decreases the Expression of C/EBP␦ Cell Adhesionrelated Target Genes and Enhances HC11 Cell Proliferation/
Migration-Having demonstrated that PIASy sequesters C/EBP␦ in the nuclear periphery and reduces C/EBP␦ transcriptional activity, we hypothesized that PIASy expression would influence biological responses associated with C/EBP␦, such as growth control. We implemented the in vitro "scratch" assay to investigate the biological effects of PIASy-C/EBP␦ interactions (36). The scratch assay assesses the capacity of cells to repopulate an open area created in a confluent cell monolayer by the combined effects of cell proliferation and migration (36,48). We hypothesized that PIASy mediated sequestration of C/EBP␦ would mimic C/EBP␦ siRNA treatment, since both reduce C/EBP␦ function. To determine if C/EBP␦ siRNA treatment influenced mammary epithelial cell proliferation/migration, we performed "scratch" assays, using parental HC11 cells (nontransfected), HC11 cells stably trans- fected with the pSilencer vector (vector control), and HC11 cells stably transfected with C/EBP␦ siRNA-pSilencer vector (C/EBP␦ siRNA) (Fig. 7A). The results demonstrated that repopulation of the open area created by the "scratch" was markedly enhanced in C/EBP␦ siRNA-treated HC11 cells compared with nontransfected parental and vector control-transfected HC11 cells (Fig. 7A). The efficiency of reduction in C/EBP␦ protein levels by the siRNA treatment was confirmed by Western blot analysis of HC11 cells at ϳ90% confluence (determined by visual inspection, t ϭ 0) and 24, 48, and 72 h after reaching confluence (Fig.  7B). The results demonstrated that C/EBP␦ expression was induced in parental HC11 cells and vector control HC11 cells within 24 h of reaching confluence and that C/EBP␦ protein levels remained elevated 48 and 72 h after reaching confluence (Fig. 7B). In contrast, C/EBP␦ protein levels were minimally detectable in C/EBP␦ siRNA-treated HC11 cells at all time points, indicating  that the siRNA treatment was effective in reducing C/EBP␦ protein levels (Fig. 7B). C/EBPs are well established transcriptional activators (1); therefore, we hypothesized that reducing C/EBP␦ content by specific siRNA or repression of C/EBP␦ transcriptional activity by PIASy expression would reduce C/EBP␦ target gene expression. In experiments using "ChIP-chip" assays, we have identified C/EBP␦ target genes that function in transcriptional regu- B, Western blot analysis of C/EBP␦ protein levels from whole cell lysates isolated from cell lines used in A at 0, 24, 48, and 72 h postconfluence. Blots were probed with a ␤-actin antibody as the protein loading control. C, total RNA was isolated from exponentially growing and 48-h confluent HC11 cells. RNA was reverse transcribed (RT), and the RT products were amplified using primers specific for C/EBP␦ target genes, GP5, PCDH9, and ITGB8, using the Roche LightCyclerா 480 real time PCR system. C/EBP␦ target gene mRNA levels were quantified using the relative quantification method following normalization to the GAPDH housekeeping gene mRNA control. D, total RNA was isolated from 48-h confluent HC11 cells transiently transfected with scrambled siRNA or C/EBP␦-specific siRNA. Real time PCR analysis was performed as described for C. E, HC11 cells were transiently transfected with C/EBP␦ and C/EBP␦ ϩ PIASy expression constructs and grown to confluence (90%), and an open area was created in the monolayer using a 200-l pipette tip (0 h). Repopulation of the open area was monitored at 0, 24, and 48 h by crystal violet staining. F, total RNA was isolated from confluent monolayers of HC11 cells transiently transfected with C/EBP␦ and C/EBP␦ plus PIASy expression constructs. Real time PCR analysis was performed as described for C. Student's t tests were used for statistical evaluation of the data in C, D, and F. The -fold change in individual C/EBP␦ target gene mRNA levels was considered significant at ␣ ϭ 0.05. lation, cell adhesion, signal transduction, apoptosis, DNA repair, and intermediary metabolism. 4 Having demonstrated in this study that cellular confluence is associated with the persistent induction of C/EBP␦ protein levels (Fig. 7B), we assessed the influence of cellular confluence on the expression of a subset of C/EBP␦ target genes that function in cell adhesion (i.e. glycoprotein V (GP5) (platelet), protocadherin 9 (PCDH9), and integrin, ␤8 (ITGB8). The rationale for analyzing these three adhesion-related C/EBP␦ target genes is that contact-mediated growth inhibition plays a major role in cell growth control and tissue homeostasis, and C/EBP␦ has been identified as a regulatory gene that is highly induced in contact inhibition-mediated growth arrest (49). The results indicated that cellular confluence is associated with a dramatic (ϳ15-30-fold) increase in the mRNA levels of all three adhesion-related C/EBP␦ target genes (GP5, PCDH9, and ITGB8) compared with growing cells (Fig. 7C). To determine if reducing C/EBP␦ levels also reduced C/EBP␦ cell adhesion-related target gene expression, HC11 cells were transiently transfected with C/EBP␦ siRNA and a scrambled siRNA control. The results demonstrated that transient C/EBP␦ siRNA treatment reduced GP5, PCDH9, and ITGB8 mRNA levels, but only ITGB8 reduction, were statistically significant (Fig. 7D).
To test the hypothesis that PIASy-mediated sequestration of C/EBP␦ would mimic the effects of C/EBP␦ siRNA treatment, the scratch assay was performed using HC11 cells transiently transfected with C/EBP␦ and C/EBP␦ plus PIASy expression constructs. The results demonstrated that PIASy plus C/EBP␦expressing HC11 cells exhibited enhanced repopulation of the open area created by the scratch at 24 and 48 h compared with C/EBP␦-expressing controls (Fig. 7E). We next investigated the hypothesis that PIASy expression would reduce the expression of C/EBP␦ cell adhesion-related target genes. The results indicated that PIASy expression significantly reduced PCDH9 and ITGB8 mRNA levels (Fig. 7F). These results indicate that PIASy expression reduces C/EBP␦ target gene expression and increases cell proliferation/migration in the in vitro scratch assay.

DISCUSSION
This report demonstrates that PIASy is a potent inhibitor of C/EBP␦ transcriptional activity. Mechanistic studies indicate that PIASy translocates C/EBP␦ from discrete transcriptionally active nuclear foci to the nuclear periphery, a region generally associated with reduced transcriptional activity (50). PIASymediated inhibition of C/EBP␦ transcriptional activity is independent of HDAC recruitment, C/EBP␦ sumoylation status, and PIASy E3 SUMO ligase activity. PIAS proteins alter transcription by a number of mechanisms, including post-translational modification with SUMO family members, SUMO-1, SUMO-2, and SUMO-3 (32,51). In this study, we found that co-expression of SUMO-1, SUMO-2, or SUMO-3 resulted in sumoylation of wild type C/EBP␦; however, expression of SUMOs at levels up to ϳ4 times those used in previous studies did not affect C/EBP␦ transcriptional activity ( Fig. 2B) (data not shown). Although co-expression of SUMO family members had little effect on C/EBP␦ transcriptional activity, PIASy significantly inhibited the transcriptional activity of both the C/EBP␦ wild type and the sumoylationdefective C/EBP␦ K120R mutant constructs. This inhibition was mediated by interaction between the PIASy N-terminal SAP domain and the C/EBP␦ N-terminal TAD domain. These findings are consistent with accumulating evidence indicating that the PIASy SAP domain interacts with transcription factors and represses transcription factor activity by multiple mechanisms, including interaction with the nuclear matrix and ATrich DNA (32,51).
The essential role of the PIASy SAP domain in PIASy-mediated inhibition of C/EBP␦ transcriptional activity contrasts with the apparently limited role of the Ring finger domain. The PIASy RFD is homologous to the Ring finger domain of ubiquitin E3 ligases and functions in substrate protein recognition and SUMO conjugation (32). The PIASy ⌬RFD construct interacts with C/EBP␦ in in vitro GST pull-down experiments (Fig.  4D), but the PIASy ⌬RFD construct does not inhibit C/EBP␦ transcriptional activity in cell-based luciferase assays (Fig. 5A). Confocal microscopy demonstrated that PIASy ⌬RFD does not enter the nucleus, localizing in the cytoplasm (Fig. 6, F and K). This indicates that nuclear import or nuclear retention of the PIASy ⌬RFD is defective, despite the presence of the PIASy NLS. These results suggest that PIASy SUMO ligase activity may play a role in PIASy nucleocytoplasmic transport. Sumoylation has been shown to regulate nuclear pore function and nucleocytoplasmic trafficking (52). In addition, our results demonstrate that the PIASy N-terminal SAP domain, which is highly conserved among PIAS proteins, plays a major role in PIAS-mediated C/EBP␦ repression by subnuclear translocation and sequestration.
The mechanism underlying PIASy sequestration of C/EBP␦ within the nuclear periphery is under investigation. The nuclear periphery is associated with inner nuclear membrane proteins that have been linked to transcriptional repression including LBR (lamin B receptor), LAP2␤ (lamina-associated polypeptide 2␤), and emerin (50). Sachdev et al. (43) reported that PIASy represses LEF1 transcriptional activity by sumoylation-independent sequestration of LEF1 in promyelocytic leukemia nuclear bodies. PIASy also represses the transcriptional activation of Nurr1 and Ets-1 by targeting to the nuclear matrix and this repression is also independent of sumoylation status (53,54). The present results are similar to these observations in that we also found that PIASy-mediated repression of C/EBP␦ transcriptional activity is independent of sumoylation. One area in which our results differ from previous studies is that although PIASy repression and sequestration of LEF1, Nurr1, and ETS is sumoylation-independent, PIASy enhances LEF1, Nurr1, and ETS sumoylation (43,53,54). Our results demonstrate that PIASy does not significantly enhance sumoylation of C/EBP␦. These collective findings highlight the complexity of the role of PIASy and sumoylation in the regulation of individual transcription factors. In addition, it has recently been reported that C/EBP␣ transcriptional activity may be regulated by sequestration in transcriptionally inactive pericentromeric heterochromatin (55). Since the PIASy SAP domain binds ATrich DNA present in scaffold attachment regions, also called matrix attachment regions, it is possible that subnuclear sequestration could include the binding of C/EBP␦ to AT-rich or repetitive DNA present at the nuclear periphery (31,56). C/EBP␦ and C/EBP␤ have been reported to bind to satellite DNA, and this binding was associated with reduced C/EBP␦ and C/EBP␤ transcriptional activity (57).
In studies investigating the biological significance of PIASy subnuclear sequestration of C/EBP␦, we found that PIASy-expressing HC11 cells exhibited reduced expression of C/EBP␦ cell adhesion related target genes (GP5, PCDH9, and ITGB) and an increased capacity to repopulate open areas in the cell monolayer induced by the "scratch." These responses paralleled those observed in C/EBP␦ siRNA-treated HC11 cells, suggesting that PIASy could alter mammary epithelial cell growth and migration through interactions with C/EBP␦. However, PIASy interacts with a broad range of transcriptional control proteins, and these interactions may also impact mammary epithelial cell growth control and migration (32). In addition, C/EBP␦ activates a broad range of target genes that may be altered by PIASy expression. Therefore, these findings both identify new regulatory interactions and raise new questions regarding the broad biological significance of these interactions.
From the perspective of mammary gland biology, these results suggest that PIASy-C/EBP␦ interactions could influence mammary gland development, gestation, differentiation, or involution by regulating C/EBP␦ transcriptional activity and target gene expression. From a breast cancer perspective, adhesion molecules function in epithelial cell contact inhibition, and disruption of this critical function by aberrant PIASy expression could promote tumorigenesis (58). The inhibitory effects of PIASy on ITGB8 may be particularly relevant to mammary gland biology and breast cancer. ITGB8 functions in epithelial cell growth inhibition by a unique mechanism involving activation of the latent form of TGF-␤ and reduced ITGB8 levels are associated with increased mammary epithelial cell proliferation in vivo and in vitro (59 -62).
Finally, the results from this work indicate that the expression of PIAS gene family members, including PIASy, is independent of mammary epithelial cell growth status. This suggests that biological responses that are influenced by PIAS proteins may be determined by the cellular content of individual PIAS-interacting proteins, such as C/EBP␦. Recent results indicate that PIASy induces cellular senescence or apoptosis, depending on the cellular content of p53 and Rb (63). The results from the present study suggest that the role of PIASy in cellular senescence and apoptosis may also be influenced by interactions with C/EBP␦. In addition, it has recently been shown that C/EBP␦ induces G 0 growth arrest of chronic myelogenous leukemia (CML) cell lines, and this arrest is associated with increased detection of C/EBP␦⅐Rb complexes (64). These findings raise the possibility that interactions between Rb, C/EBP␦, and PIASy could play a major role in cell fate determination. Future work will investigate the functional interactions between Rb, C/EBP␦, and PIASy as a potential convergence point in cell fate determination.