INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The promyelocytic leukemia gene, PML, was first identified at the breakpoint of the t(15;17) translocation in acute promyelocytic leukemia (APL) (10, 14, 19, 35, 37). PML encodes a nuclear phosphoprotein that functions as a transcriptional regulator (9, 50, 58) and belongs to the RING family of proteins, which share a cysteine-rich motif at the N terminus. This motif is divided into a RING finger (C₃C₄ zinc binding) motif and two B-box (B1 and B2) motifs (18). This region is followed by a predicted -helical coiled-coil (dimerization) domain, which allows PML to homodimerize and form heterodimer complexes with the APL fusion protein PMLRAR and the promyelocytic leukemia zinc finger (PLZF) protein (37, 40). PML localizes to distinct domains in the nucleus called PML nuclear bodies, or PML oncogenic domains (PODs) (16, 60). In addition to PML, there are several other POD-associated factors, including SP100, the ubiquitin-like protein PIC1, and the interferon-stimulated 20-kDa gene product called ISG20 (3, 6, 20). PODs are frequently targeted and/or reorganized by viral proteins, such as the herpes simplex virus type 1 (HSV-1) gene product Vmw110 (17), the adenoviral proteins E1A and E4-ORF3 (8), the Epstein-Barr virus-encoded nuclear antigen EBNA-5 (53), and the human cytomegalovirus major immediate-early proteins IE1 and IE2 (1).

PMLRAR, which retains the cysteine-rich motif and the dimerization domain of PML and the DNA-binding and ligand-binding domains of retinoic acid receptor  (RAR), has been shown to play a direct role in POD morphology and hence APL leukemogenesis in vitro (16, 37, 60). Treatment of APL cells with all-trans-retinoic acid (ATRA) restores PML-containing PODs, apparently by degradation of the PMLRAR fusion protein and hence induces terminal differentiation (16, 60). Furthermore, in line with the in vitro studies, transgenic results have shown that the expression of PMLRAR plays a critical role in the development of leukemia in mice (7, 22, 26).

The results from our studies and others have shown that PML functions as a growth suppressor (2, 25, 39, 43, 47, 50), presumably by inducing G₁ cell cycle arrest and apoptosis (43). Interestingly, the domains of PML that mediate its association with PODs have also been found to be involved in its growth suppression function (44). Recently, the growth suppressor function of PML was conclusively demonstrated by PML gene knockout (59). Although the mechanism through which PML suppresses cellular growth and transformation is unknown, recent studies have shown that PML is involved in regulating transcription of certain genes in either a positive or negative manner. In particular, we have demonstrated previously that PML can repress transcription of the epidermal growth factor receptor (EGFR) and multidrug resistance 1 (MDR1) promoters (50, 58). Analysis of transcriptional repression of PML, by means of the GAL4 fusion assay, localized the repressive effects of PML mainly to the coiled-coil (dimerization) domain (58). PML has also been reported to enhance the transactivation properties of the progesterone receptor (24). Recently, we have found that PML is associated with the AP-1 complex and is able to upregulate Fos-mediated transcriptional activity. Although no direct interaction between PML and Fos was detected, it was found that the stimulation of transcriptional activity of Fos required the RING finger and the B1-box motifs of PML and the C-terminal domain of Fos (57). Moreover, PML was recently shown to interact with the retinoblastoma protein (pRb) in vivo and in vitro (2). Functional analysis of this PML-Rb interaction revealed that PML can inhibit Rb-mediated transactivation of the glucocorticoid receptor transcription, providing further evidence for the involvement of PML in regulation of transcription.

Our previous study demonstrated that PML suppresses the promoter activity of the EGFR gene (50). This promoter element is GC rich, contains multiple Sp1-binding sites, and lacks both TATA and CAAT boxes (28, 31). The promoter activity is regulated by a number of factors, including epidermal growth factor (EGF), cyclic AMP, and 12-O-tetradecanoyl-phorbol 13-acetate (27). Several transcription factors (e.g., p53 and Sp1) activate the promoter (13, 34), whereas others such as the ligand-activated thyroid hormone (T3R) and retinoic acid receptors (RARs) repress it (61). EGFR overexpression is associated with several malignancies (e.g., breast, colon, ovarian, and head and neck cancers), suggesting an important role for EGFR expression in growth and differentiation (48, 54).

For the present study, the 5' proximal region of the EGFR promoter was characterized to search for factors that mediate transcriptional repression of the EGFR promoter activity by PML. Results from this study demonstrated that the repressive effects of PML are mapped to the sequences between 150 and 16 of the EGFR promoter, which comprise the majority of the basal promoter activity mediated mainly by the transcription factor Sp1 through the Sp1-binding sites (31, 34). We found that this Sp1-dependent activity was inhibited by PML in a dose-dependent manner. In vitro and in vivo binding assays demonstrated that PML and Sp1 are associated through specific domains. Furthermore, our study indicated that interaction of PML with Sp1 disrupted its ability to bind DNA. Thus, repression of EGFR promoter activity by PML is most likely caused by inhibition of its Sp1-dependent activity, which could represent a novel mechanism for negative regulation of the EGFR promoter.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Plasmids. Full-length and mutant GST-Sp1 in a pGex2TKMSC expression vector and pRCE2F1 were kindly provided by E. Wintersberger (Universität Wien, Vienna, Austria) (36). The pCDNA3/Sp1 expression vector used for in vitro translation was constructed by subcloning an XhoI-SmaI fragment from pGex2TKMCS/Sp1 into an XhoI-EcoRV-digested pCDNA3 plasmid (Invitrogen, San Diego, Calif.). The pPac0, pPac-Sp1, and pPac--gal vectors, which contain the Drosophila -actin promoter and polyadenylation sequences, were obtained from R. Tjian (33). pPac-PML was constructed by subcloning a BamHI-BglII fragment containing the full-length PML cDNA from the pFM211 vector (50) into the BamHI site in pPac0. pPac-E2F1 was generated by subcloning the full-length E2F1 cDNA as a BamHI-SalI (partial digestion) fragment from pVP16/E2F1 into the BamHI-XhoI site in pPac0. pVP16/E2F1 was a kind gift from H. Rotheneder (Universität Wien). PMLRAR cDNA expressed from a pSG5 plasmid under the control of the simian virus 40 (SV40) early promoter and enhancer was a gift from P. Chambon (37). PML mutants used in in vitro translations were constructed as described previously and were also expressed from pSG5 plasmid (44). The construction of EGFR promoter deletion mutants was described previously (34). The pDHFR-CAT reporter plasmid, which contains the hamster dihydrofolate reductase (DHFR) gene promoter linked to the chloramphenicol acetyltransferase (CAT) gene, was obtained from D. G. Johnson (5, 32). The pCMV-CAT reporter construct contains the cytomegalovirus minimal promoter in front of the CAT gene in a pCDNA3 vector (50). The p4x(UAS)-Luc vector, which contains four GAL4 DNA-binding sites (upstream activation sequences) in front of the thymidine kinase minimal promoter linked to the firefly luciferase gene, was obtained from M.-J. Tsai (Department of Cell Biology, Baylor College of Medicine, Houston, Tex.). The expression vectors used in two-hybrid assays were pK3VP16/PML and pHKG/Sp1. pK3VP16/PML, which expresses a VP16-PML fusion protein, was constructed by in-frame subcloning of PML cDNA as a BamHI-BglII fragment from the pFM211 vector (described above) into the BamHI site in pK3Vp16 vector, downstream of the sequences of the activation domain of the VP16 transcription factor. pHKG/Sp1 was constructed by subcloning the Sp1 cDNA as an XhoI-SmaI fragment (see above) into the SalI-XbaI (blunt ended) site in the pHKG4 vector, downstream of the DNA-binding domain of GAL4 transcription factor. pK3Vp16 and pHKG4 vectors were kindly provided by T. Kouzarides (Cambridge University, Cambridge, United Kingdom) (4).

Cell culture. SW13 human adenocortical carcinoma and U2OS human osteosarcoma cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U of penicillin per ml, and 100 µg of streptomycin per ml (GIBCO/BRL, Gaithersburg, Md.) in 5% CO₂ at 37°C. Drosophila melanogaster SL2 cells were cultured at 24°C in Schneider's Drosophila medium (GIBCO/BRL) supplemented with 10% heat-inactivated FBS and antibiotics as described above.

Transfection, mammalian two-hybrid assays, and CAT assays. For gene transfection experiments, SL2 cells were seeded at approximately 5 × 10⁶ per 60-mm-diameter dish 24 h before transfection. Cells were transfected by a calcium phosphate coprecipitation method as described previously (58). The quantities of plasmids used for transfections were as indicated in the legends to the figures. The total amount of DNA was adjusted to 15 µg with sheared and denatured salmon sperm DNA and sufficient amounts of pPac0 plasmid (containing the Drosophila -actin promoter) to maintain a constant promoter level. The pPac--gal plasmid (100 ng) (-galactosidase expression vector) was included in each experiment to monitor transfection efficiency. Twenty-four hours after transfection, cells were harvested by pipetting the medium up and down several times, transferred into 15-ml tubes, centrifuged at 1,000 × g for 5 min, washed two times in phosphate-buffered saline, transferred into Eppendorf tubes, and resuspended in 150 µl of 0.25 M Tris-HCl (pH 7.8). Cells were lysed by three cycles of freeze-thawing, and the clarified supernatants were used for -galactosidase and CAT assays as described previously (58). SW13 and U2OS cells were transfected at semiconfluence (50 to 70%) by calcium phosphate coprecipitation with the quantities of plasmids indicated in the legends to the figures. The total amounts of transfected DNA were adjusted to 20 µg with salmon sperm DNA and sufficient amounts of pSG5 vector. To monitor transfection efficiency, 1 µg of pCMV--galactosidase expression plasmid was included in each transfection. Approximately 16 h after transfection, the precipitate was washed. Cells were then fed with fresh medium for an additional 24 h, harvested, and lysed. The amounts of extracts used for CAT assays were then normalized with respect to the -galactosidase activity as described previously (58). The CAT activities were quantitated with a PhosphorImager (Bio-Rad Laboratories, Hercules, Calif.). In all experiments, each transfection was repeated at least twice. The CAT assay results presented are from typical experiments.

Immunoprecipitation and immunoblotting. In immunoprecipitation experiments, approximately 400 µg of nuclear proteins from HeLa cells transiently overexpressing PML was diluted to 1 µg/µl in radioimmunoprecipitation assay (RIPA) buffer (140 mM NaCl, 27 mM KCl, 10 mM Na₂HPO₄, 18 mM KH₂PO₄, 1% Triton X-100, 13 mM sodium deoxycholate, 0.1% sodium dodecyl sulfate [SDS], 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 2 µg of leupeptin per ml, 1 µg of aprotinin per ml) and incubated with a polyclonal anti-PML antibody at 4°C overnight. The immunocomplexes were absorbed to protein A-Sepharose (Santa Cruz Biotechnology, Inc., Santa Cruz, Calif.) for 2 h at 4°C and washed three times in 1 ml of RIPA buffer. Associated proteins were then eluted in SDS sample buffer by boiling for 4 min. The released proteins were then resolved by 8% SDS-polyacrylamide gel electrophoresis (PAGE), transferred to nitrocellulose membranes (Schleicher & Schuell, Inc., Keene, N.H.), probed with anti-Sp1 antibody (Santa Cruz), and detected with the ECL (enhanced chemiluminescence) system (Amersham Life Sciences, Inc., Arlington Heights, Ill.). Immunoblot analyses of PML and PML mutants translated in vitro were performed by denaturing the proteins in SDS sample buffer and resolving by SDS-PAGE as described above.

In vitro transcription and translation. For in vitro translation experiments, the PML, PML mutants, E2F1, and cyclin A plasmids were transcribed in vitro with T7 RNA polymerase, while Sp1 plasmid was transcribed with SP6 RNA polymerase from the appropriate expression vectors (described above). The products were labeled with [³⁵S]methionine (NEN, Boston, Mass.) using the TNT coupled transcription-translation system (Promega Corp.). In the in vitro translation reactions, empty expression vectors were used as a control.

GST fusion proteins and GST pull-down assay. Full-length and mutated glutathione S-transferase (GST)-Sp1 proteins were expressed from pGex-2TK-MCS plasmids (described above), and the GST-PML protein was expressed from pGex3X plasmid as described previously (14). The GST fusion proteins produced in host bacteria were purified by standard procedures. The GST pull-down assays were performed essentially as described previously (57). Briefly, similar quantities of GST or GST fusion proteins immobilized on glutathione-Sepharose beads were washed in NETN buffer (100 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl [pH 8], 0.5% Nonidet P-40) and incubated with 5 µl of proteins translated in vitro and labeled with [³⁵S]methionine in 200 µl of NETN buffer for 2 h at 4°C on a rocker. Bound proteins were then washed three times in 0.5 ml of NETN buffer and eluted in SDS sample buffer by boiling for 5 min. The eluted proteins were subjected to SDS-PAGE as described above. The quantity and expression of GST and GST fusion proteins were determined by SDS-PAGE followed by Coomassie blue staining.

EMSAs. Nuclear proteins were isolated from HeLa cells and used in electrophoretic mobility shift assays (EMSAs) as described previously (58). Protein-DNA binding assays were performed by first preincubating 5 µg of HeLa nuclear proteins in a binding buffer (4% Ficoll 400, 20 mM HEPES [pH 7.9], 2 mM MgCl₂, 1 µg of salmon sperm DNA with the final concentration of KCl in the reaction mixture adjusted to 100 mM) for 10 min at room temperature in a total volume of 19 µl. In each reaction mixture, 10 fmol (1 µl) of the ³²P-labeled probe was then added, and the reaction mixture was then incubated at room temperature for an additional 30 min. Double-stranded Sp1 and E2F oligonucleotide probes were made by annealing the complementary strands of the following oligonucleotides: Sp1, 5'-CATTCGATCGGGGCGGGGCGAGC-3'; and E2F, 5'-TCCGTAGTTTTCGCGCTTAAATTTGAGAAAGGGCGCGAAACTAG TC-3'.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of the EGFR promoter domains responsive to PML. We have previously shown that PML can suppress the transcriptional activity of several promoters, including the EGFR and human MDR1 genes (50, 58). To identify the region or specific sequences responsive to the repressive activity of PML, the effects of PML on a series of deletion constructs of the EGFR promoter spanning the sequences from 1109 to 16 (the initiator ATG being +1) and linked to the CAT gene were examined (Fig. 1). The constructs were cotransfected into SW13 cells with or without PML. As expected, expression of PML significantly suppressed the activity of the EGFR promoter ERCAT(1109) (Fig. 1B). Deletion of sequences between 1109 and 150, although it reduced basal promoter activity, did not affect the repression of promoter activity by PML. Indeed, PML suppressed the activity of both ERCAT(150), containing sequences from 150 to 16, and the full-length construct ERCAT(1109) to a similar extent. This suggested that the region between 150 and 16 conferred most of the repressive effects of PML. Given the very low basal activity of ERCAT(105) and ERCAT(167/105), it was not feasible to further investigate the effects of PML on these deletion constructs. Under similar conditions, and in contrast to PML, the PMLRAR fusion protein caused a small inhibition (up to 20%) of ERCAT(1109) and ERCAT(911) activity but had little or no effect on EGFR mutants spanning sequences between 850 and 16 (Fig. 1B). The fact that the PMLRAR fusion protein had an altered transcriptional activity compared with that of the wild-type PML protein indicated the specificity of PML's effects.

View larger version (38K): [in this window] [in a new window]   FIG. 1.   Functional analysis of the effects of PML and PMLRAR on EGFR promoter. (A) Schematic representation of EGFR deletion mutants linked to the CAT gene. The numbers indicate the sequences of the EGFR promoter relative to the translation start site (ATG +1). (B) Relative CAT activity in SW13 cells cotransfected with 5 µg of different deletion mutants of the EGFR promoter-CAT constructs as shown in panel A, together with 10 µg of PML and PMLRAR expression plasmids. The CAT activity of the longest mutant of the EGFR promoter, ERCAT(1109), in the absence of PML or PMLRAR, was arbitrarily set at 100, and the activities of the other constructs were calculated relatively. The mean CAT activity from three independent experiments is shown for each construct.

Inhibition of Sp1-mediated transcription of EGFR by PML. The study described above demonstrated that the 150-bp 5' element of the EGFR promoter confers repressive effects on PML. The transcriptional activity of the EGFR promoter is efficiently regulated by the transcription factor Sp1 through at least four Sp1 binding sites, and the 150-bp 5' element has been shown to be sufficient for the promoter activity (31, 34, 61). To examine whether the repression of EGFR transcription by PML was caused by inhibition of its Sp1-mediated activity, the effects of PML on Sp1-mediated activity of EGFR were examined in Drosophila melanogaster SL2 cells, which lack endogenous Sp1. To achieve sufficient protein expression in transient transfection assays, PML and Sp1 were expressed under the control of the Drosophila -actin promoter from pPac vector (provided by R. Tjian) (33). SL2 cells were transfected with ERCAT(150) alone or together with pPacSp1 and increasing amounts of pPacPML expression plasmids. As expected, cotransfection of ERCAT(150) with 250 ng of Sp1 expression vector resulted in about a 50-fold induction of promoter activity (Fig. 2B). Strikingly, cotransfection of the pPacPML expression vector at different concentrations (1, 2.5, and 5 µg) inhibited the Sp1-mediated stimulation of the promoter activity in a dose-dependent manner (Fig. 2B, lanes 3 to 5). When the expression of both Sp1 and PML in the transfected cells was examined by immunostaining, no significant changes in the expression of Sp1 were observed in the presence of different concentrations of PML. Similar to the case in mammalian cells, expression of PML in SL2 cells in the presence or absence of Sp1 produced the normal PML nuclear speckled pattern (15). In control transfections, PML had no significant effect on the basal promoter activity derived from a pCMV-CAT construct, indicating that the inhibition of Sp1 activity by PML was specific (Fig. 2C). Together, these results suggested that the suppression of EGFR promoter activity by PML was caused, at least in part, by inhibition of its Sp1-mediated activity.

View larger version (40K): [in this window] [in a new window]   FIG. 2.   Inhibition of Sp1-mediated transactivation of the EGFR promoter by PML. (A) Schematic representation of Sp1 binding sites in the 150-bp element of the EGFR promoter (28). (B) Effects of different concentrations of PML on Sp1-stimulated activity of the 150-bp region of EGFR. Drosophila SL2 cells were transfected with 3 µg of the ERCAT(150) reporter plasmid, pPacSp1, and increasing concentrations of pPacPML expression plasmids as indicated. In each transfection, 100 ng of pPac--gal (-galactosidase expression plasmid) was used to monitor transfection efficiency. The cells were harvested 24 h after transfection, and CAT activities were measured. The bar graph shows the fold CAT activity of each transfection relative to the activity exhibited by the ERCAT(150) reporter construct in the absence of the effector plasmids. (C) Effect of increasing concentrations of pPacPML on the basal activity of the pCMV-CAT vector in SL2 cells. In each transfection, cells were transfected with 5 µg of pCMV-CAT and increasing concentrations of pPacPML as indicated.

Association of PML and Sp1 in vivo. Given PML's specific inhibition of the Sp1-mediated activity of the EGFR promoter, experiments were done to see if an interaction between PML and Sp1 were possible. First, their association in vivo was examined by immunoprecipitation. For this, nuclear extracts prepared from HeLa cells transiently overexpressing the PML protein were used. In brief, the HeLa nuclear extracts were immunoprecipitated with a polyclonal anti-PML antibody, after which the immunocomplexes were absorbed to protein A-agarose, washed extensively, analyzed by SDS-PAGE, and detected by anti-Sp1 antibody (Santa Cruz). The results, as shown in Fig. 3A, indicated that PML and Sp1 did indeed associate in vivo. In control experiments, an unrelated antibody (anti-GAL4) produced no signal. The presence of Sp1 in the nuclear extracts was also confirmed by analyzing a sample of the extracts in a parallel lane by SDS-PAGE (Fig. 3A). The possible cross-reactivity of anti-PML antibody and Sp1 protein was ruled out, because the anti-PML antibody failed to precipitate either the purified or in vitro-translated Sp1 protein; likewise, the anti-Sp1 antibody was unable to detect PML protein in extracts from cells transiently overexpressing PML or the PML protein translated in vitro (not shown). PML could not be immunoprecipitated with the Sp1 antibody, but whether this was due to the antibody or to the nature of the Sp1-PML interaction, in which Sp1 epitopes detected by Sp1 antibody might be masked, was unclear.

View larger version (28K): [in this window] [in a new window]   FIG. 3.   In vivo analysis of the association of PML and Sp1. (A) Immunoprecipitation. Nuclear extracts from HeLa cells transiently overexpressing PML were subjected to immunoprecipitation with anti-PML antibody (-PML) or anti-GAL4 antibody (-GAL4). The immunocomplexes were absorbed to protein A-Sepharose, washed, resolved by SDS-PAGE (8% polyacrylamide), transferred to nitrocellulose membranes, and probed with anti-Sp1 antibody. In a parallel lane, a sample of nuclear extracts (lysate) was analyzed to indicate the position of the Sp1 band. (B) Mammalian two-hybrid assay. U2OS cells were cotransfected with 1 µg of GAL4-Sp1 and/or 3 µg of VP16-PML expression plasmids, as indicated, together with 4 µg of the luciferase reporter plasmid [4 × (UAS)-Luc]. In each transfection, cells were also transfected with 0.5 µg of pCMV--gal (-galactosidase expression plasmid) for monitoring transfection efficiency and normalization of luciferase activity. The luciferase activity of each sample was measured, was calculated relative to the activity exhibited by cells transfected with empty expression vector (control), and is shown as the fold increase.

PML-Sp1 interaction in vitro. The direct interaction of PML and Sp1 in solution was examined with the GST pull-down assay. In this assay, the in vitro-translated ³⁵S-labeled PML was incubated with the recombinant full-length protein or various truncated GST-Sp1 proteins immobilized on the glutathione-Sepharose beads, and the associated proteins were washed and analyzed by SDS-PAGE. As shown in Fig. 4B, the PML protein bound the full-length GST-Sp1 as well as the C-terminal Sp1 mutant GST-Sp1(622-788), but not the N-terminal Sp1 mutant GST-Sp1(1-293) or GST-Sp1(1-621). Little or no background binding with PML was detected in control experiments with the GST protein alone. Also, in assays run with control in vitro translation reaction mixtures (including those run with no DNA or empty expression vectors), no binding with the GST-Sp1 proteins was detected (not shown). In a parallel experiment, the binding of E2F1 to GST-Sp1 was examined as a positive control (36, 46). As expected, GST-Sp1 showed strong binding to E2F1 protein translated in vitro (Fig. 4C). Notably, the intensity of PML-Sp1 binding was comparable to the interaction observed between E2F1 and Sp1 when examined under the same conditions.

View larger version (43K): [in this window] [in a new window]   FIG. 4.   Analysis of in vitro binding of PML and Sp1. (A) Schematic diagram of Sp1 and Sp1 deletion mutants used in the GST pull-down assay. The letters A to D indicate different domains of the Sp1 protein. Three zinc finger DNA binding motifs of Sp1 present in the C-terminal region are shown. The numbers in parentheses indicate the amino acids of Sp1 linked to the GST protein. (B) GST and GST-Sp1 (full-length and deletion mutants) immobilized on glutathione-Sepharose beads were incubated with in vitro-translated, 35S-labeled PML. Bound proteins were then eluted and analyzed by SDS-PAGE (8% polyacrylamide) as described in Materials and Methods. As a control, a 1/10 volume of labeled PML protein used in the in vitro binding assays was resolved by SDS-PAGE. (C) GST and GST-Sp1 proteins were incubated with in vitro-translated 35S-labeled E2F protein as described for panel B. (D) GST and GST-PML proteins bound to glutathione-Sepharose beads were incubated with in vitro-translated Sp1 and resolved by SDS-PAGE (8% polyacrylamide). (E) In vitro-translated 35S-labeled PML, E2F1, and cyclin A (CA) were incubated with GST-PML protein as described for panel D.

Mediation of PML-Sp1 interaction by the coiled-coil domain. To identify domains of PML involved in its association with Sp1, several deletion mutants of the PML protein (Fig. 5A) were translated in vitro, labeled with [³⁵S]methionine, and subjected to the GST pull-down assay by using the full-length GST-Sp1 fusion protein. As shown in Fig. 5B, mutants of PML lacking the N-terminal proline-rich region (PMLpro), the proline-rich region plus the RING finger domain (PMLpr), or the C-terminal serine/proline-rich region (PMLsp) bound the Sp1 protein, whereas the deletion mutant of PML lacking the coiled-coil (dimerization) domain (PMLdim) could not. In control experiments, the incubation of the in vitro-translated, ³⁵S-labeled PML mutants with the GST protein alone produced no significant binding (not shown). Together, these results showed that the coiled-coil domain of PML was required for the association of PML with Sp1 in vitro.

View larger version (64K): [in this window] [in a new window]   FIG. 5.   Identification of domains of PML involved in its association with Sp1. (A) Schematic representation of PML and PML deletion mutants. Pro, proline-rich region; R, RING finger; B, B-box motifs (B1 and B2); coiled-coil, PML dimerization domain; Serine, serine/threonine-rich region. PMLpro, PMLpr, PMLdim, and PMLsp, respectively, lack the proline-rich region, the proline-rich region plus the RING finger, the dimerization domain, and the serine/proline-rich domain, as shown. (B) GST-Sp1 protein bound to glutathione-Sepharose beads was incubated with in vitro-translated, 35S-labeled PML and PML mutants. Bound proteins were then washed and analyzed by SDS-PAGE (8% polyacrylamide). (C) The in vitro-translated proteins (input) used in panel B (a 1/10 volume) were resolved by SDS-PAGE (8% polyacrylamide).

Disruption of Sp1-DNA binding by PML. The finding that PML and Sp1 could associate and that the C-terminal region of Sp1 and the coiled-coil domain of PML were involved in this association, combined with the knowledge that the C-terminal region of Sp1 mediates its binding to target DNA through three zinc finger motifs (33), raised the possibility that the binding of PML to Sp1 may interfere with its DNA-binding activity, thus providing a mechanism for the suppression of the EGFR promoter activity by PML. To examine the possible effects of PML on Sp1 DNA binding, EMSAs were done with nuclear extracts from HeLa cells incubated with labeled oligonucleotide probes bearing an Sp1 binding site in the presence of different concentrations of in vitro-translated PML protein. As shown in Fig. 6A, addition of in vitro-translated PML protein (1, 3, and 5 µl in lanes 8 to 10, respectively) to the EMSA reaction mixtures resulted in the disruption of low-mobility DNA-protein complexes corresponding to Sp1 (see below) in a manner dependent on PML concentration (compare lanes 8 to 10 with lane 5 in Fig. 6A). In control reactions, addition of rabbit reticulocyte lysate had no effect on Sp1 or other DNA-protein complexes retarded by the labeled Sp1 probe (lanes 11 and 12). An immunoblot analysis of the PML protein translated in vitro, representing amounts similar to those used in EMSA, was shown in Fig. 6C. In EMSA reactions, the specificity of the Sp1 complex was examined by oligonucleotide competition assay with a specific anti-Sp1 antibody not cross-reactive with Sp2, Sp3, or Sp4 (see Materials and Methods). As shown in Fig. 6A, in the presence of a 100-fold molar excess of nonlabeled probe, the specific DNA-protein complexes could be competed out (lanes 2 and 6). Incubation of extracts with anti-Sp1 antibody, but not nonimmune serum, resulted in an Sp1 supershift (lanes 4 and 7) that indicated the presence of Sp1 in the complexes disrupted by PML.

View larger version (54K): [in this window] [in a new window]   FIG. 6.   The effects of PML on Sp1 DNA binding. (A) EMSA of HeLa nuclear proteins. HeLa nuclear proteins were incubated with 32P-labeled Sp1 probe, and the Sp1-DNA complexes were resolved on a 4% native polyacrylamide gel. In lanes 2 and 5, specific complexes were competed out with 100-fold molar excesses of nonlabeled (cold) Sp1 probe to examine the specificity of the retarded complexes. The presence of Sp1 protein in the complexes was also tested with anti-Sp1 antibody (-Sp1) (lanes 4 and 7) and preimmune serum (lane 3). In lanes 8 to 10, the nuclear extracts were treated with 1, 3, and 5 µl of in vitro-translated PML protein, respectively. In lanes 11 and 12, 3 and 5 µl, respectively, of the lysate from the control reaction mixtures were used. (B) E2F DNA binding. In lane 1, HeLa nuclear proteins were incubated with 32P-labeled E2F probe, and the E2F-DNA complexes were subjected to electrophoresis as described for panel A. Lane 2 shows that the specific retarded complexes were competed out for binding by a 200-fold molar excess of nonlabeled E2F probe. In lanes 3 and 4, the extracts were treated with 3 and 5 µl of in vitro-translated PML protein, respectively, as described for panel A. In lane 5, nuclear proteins were treated with 5 µl of control lysate. (C) Analysis of the expression of the in vitro-translated PML proteins shown in panels A and B. In vitro-translated PML proteins (1, 3, and 5 µl) and control lysate (5 µl) were subjected to SDS-PAGE (8% polyacrylamide) and detected with anti-PML antibody.

Involvement of the PML coiled-coil domain in Sp1 DNA binding. To analyze the effects of PML on Sp1 DNA binding more directly, a purified Sp1 protein (0.25 footprinting unit/reaction; Promega) was used in EMSA reactions (Fig. 7A). As in the experiments containing HeLa nuclear extracts, the addition of the PML protein at increasing concentrations disrupted the Sp1 DNA-binding complex in a dose-dependent manner (Fig. 7A, lanes 4 to 6). In control reactions, addition of rabbit reticulocyte lysate had no effect on the complex (lane 7). These results further confirmed the specificity of the effect of PML on Sp1 DNA binding.

View larger version (73K): [in this window] [in a new window]   FIG. 7.   Effects of PML and PML deletion mutants on the DNA-binding activity of purified Sp1 protein. (A) Purified Sp1 protein (0.25 footprinting unit/reaction) was incubated with 32P-labeled Sp1 oligonucleotide probe, and the DNA-protein complexes were analyzed by electrophoresis as described for Fig. 6. Lane 1 shows the binding of Sp1 protein to the labeled Sp1 probe. The specificity of the Sp1 complex was examined by oligonucleotide competition assay using the nonlabeled Sp1 probe (lane 2) and anti-Sp1 antibody (lane 3). In lanes 4 to 6, 1, 3, and 5 µl of PML protein, respectively, were added to the Sp1-binding reaction mixtures. Lane 7 contains 5 µl of the control lysate. See text for details. (B) In vitro-translated PML and PML deletion mutants (3 µl) were incubated with labeled Sp1 probe, and the DNA-protein complexes were analyzed by electrophoresis as described for panel A. (C) HeLa nuclear extracts were incubated with PML deletion mutants as in panel B. In each of panels A to C, the Sp1 complex is indicated by an arrow. (D) The expression of PML and PML deletion mutants was then analyzed by immunoblotting. The abbreviations are as given for Fig. 5.

Abrogation of Sp1-mediated DHFR promoter activity by PML. Given the specific inhibition of Sp1 DNA-binding activity and transactivation by PML, it was important to examine whether PML could repress transcription from another Sp1-regulated promoter. Most of the growth-regulated promoters contain binding sites for Sp1 and E2F. Among these promoters, thymidine kinase and DHFR gene promoters have recently been shown to be regulated by Sp1 (36, 46). We therefore investigated the effects of PML on the Sp1-dependent activity of DHFR in SL2 cells, in which the activity of DHFR is dependent on exogenous Sp1 (46). For this purpose, the hamster DHFR promoter driving expression of the CAT gene (DHFR-CAT) was transfected into SL2 cells either alone or together with pPacSp1 and different concentrations of the pPacPML expression vectors. The hamster DHFR promoter contains four Sp1 binding sites and two overlapping binding sites for the transcription factor E2F1 (5). As expected, the DHFR target promoter was almost silent in SL2 cells, and its activity depended upon the Sp1 expression (Fig. 8A, compare lanes 1 and 2). Transfection of 250 ng of pPacSp1 resulted in a marked (35-fold) induction of CAT expression from the DHFR target promoter. Similar to the effects of PML on Sp1-mediated activity of the EGFR target promoter (Fig. 2A), cotransfection of pPacPML suppressed Sp1-stimulated activity of the DHFR reporter gene in a dose-dependent manner (Fig. 8A). Moreover, PML could suppress the transcriptional activity of DHFR and thymidine kinase in mammalian cells (unpublished data). In control transfection experiments, expression of PML did not significantly affect the E2F1-stimulated activity of DHFR target promoter, indicating the specificity of PML's effect (Fig. 8B, lanes 2 and 3). Together, these results showed that the Sp1-mediated transcriptional activity of the DHFR promoter could be inhibited by PML, suggesting that DHFR is another target of Sp1-PML.

View larger version (48K): [in this window] [in a new window]   FIG. 8.   Inhibition of Sp1-mediated transactivation of DHFR promoter by PML. (A) Dose-dependent inhibition of Sp1-mediated activity of DHFR by PML. Drosophila SL2 cells were transfected with 3 µg of DHFR-CAT, promoter pPacSp1, and increasing concentrations of pPacPML as indicated. In each transfection, cells were also transfected with 100 ng of pPac--gal (-galactosidase expression plasmid) for monitoring of transfection efficiency and normalization of CAT activities. (B) Effect of PML on E2F activity of DHFR promoter. SL2 cells were transfected as in panel A, except that in lanes 2 and 3, pPacE2F1 was included as indicated. In each panel, the CAT activity exhibited by the DHFR-CAT in the absence of Sp1 or E2F1 was set to 1, and the activities in other experiments were calculated in relation to that setting.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We report here that the PML growth suppressor interacts with Sp1 and inhibits the Sp1-mediated transcriptional activity of the EGFR gene promoter. This finding is based on the observations that (i) PML repressed the transcription of the EGFR promoter by targeting a 150-bp 5' element of the promoter that is driven by the transcription factor Sp1; (ii) PML and Sp1 interacted both in vitro and in vivo through specific domains; (iii) PML specifically disrupted the binding of Sp1 to target DNA; and (iv) expression of exogenous PML inhibited Sp1-mediated transcription of Sp1 target promoters in Sp1-negative Drosophila melanogaster SL2 cells. The inhibition of Sp1-mediated transcriptional activity by PML may therefore represent a novel mechanism for inhibiting EGFR promoter activity. Furthermore, cotransfection of PML inhibited the Sp1-mediated activity of the DHFR promoter, indicating that the repressive effects of PML are not limited to EGFR and that PML may function as an inhibitor of Sp1-targeted promoters.

The repressive effects of PML on EGFR transcription are specific because the 150-bp element 5' to the initiation site is responsible for this activity. Under similar conditions, the PMLRAR fusion protein failed to inhibit the promoter activity at this site. One possible explanation could be its altered nuclear localization from a speckled pattern (PODs) to a diffused pattern (16, 60). The recent findings that PODs are possible sites for transcription support this notion (29, 42). Moreover, PMLRAR, which contains most of the functional domains of PML and RAR, has also been shown to have different functional properties compared with those of PML and RAR (14, 35, 58).

The EGFR promoter is GC rich and lacks both TATA and CAAT boxes (28). Several binding sites for Sp1 have been identified in the EGFR promoter, four of which have been shown to bind Sp1 (34). Accordingly, in Drosophila SL2 cells, which are negative for Sp1, the activity of the EGFR promoter is entirely dependent on the exogenous Sp1 (61) (Fig. 2A). In mammalian cells, several transcription factors have been shown to exert their effects on the promoter activity directly or indirectly through interaction with Sp1. For example, T3R and RAR have been shown to inhibit the EGFR promoter activity by competing with Sp1 to bind an overlapping binding site present in the proximal 36-bp segment (between 112 and 77) of the promoter (27, 61). Moreover, p53, which functions as an activator of EGFR promoter, has been shown to form complexes with Sp1 and to stimulate its binding to DNA (13, 23). These findings support the notion that Sp1 functions as a crucial regulator of EGFR promoter activity. Moreover, the present study showed that the inhibition of EGFR transcription by PML is caused, in part, by inhibition of its Sp1-mediated activity and indicated that PML could form complexes with Sp1 in vivo and in vitro and efficiently repress the Sp1-mediated activity of the EGFR promoter in Sp1-negative SL2 cells.

The physical interaction between PML and Sp1 provides further direct support for the inhibition of EGFR activity by PML. In particular, our in vitro binding assays showed that the C-terminal region of Sp1 is required for interaction with PML, which contains three zinc fingers that mediate Sp1 DNA-binding activity. Our EMSAs demonstrated that addition of PML protein disrupted the binding of Sp1 to target DNA, regardless of whether the crude nuclear extracts or the purified Sp1 protein was used. Furthermore, our attempts to detect PML-Sp1 complexes that bind DNA were unsuccessful (not shown). Together, these data suggest that PML interferes with Sp1 DNA-binding activity most likely by forming complexes that do not bind DNA.

An increasing number of transcription factors have been found that interact with Sp1. One set of these factors includes E2F1, GATA1, and YY1, which act synergistically with Sp1 on DNA to increase transcriptional activity (21, 36, 45, 46). Another set of Sp1-interacting transcription factors that impair Sp1-mediated transcriptional activity includes the von Hipple-Lindau (VHL) tumor suppressor protein, p107, and Sp1-I. The VHL interacts with Sp1 and inhibits its activation of the vascular endothelial growth factor (VEGF) promoter (51). Strikingly similar to PML in activity, VHL was originally found to inhibit VEGF promoter activity through a GC-rich element whose activity is regulated by Sp1. This suggests that VEGF may also be an Sp1-PML target. The cell cycle-regulatory protein p107, a member of the Rb family, also associates with Sp1 and inhibits its transcriptional activity (12), but the mechanism by which it does so is unclear. However, as with PML, no Sp1-p107 complex that bound DNA was detected, suggesting that p107 may also interfere with Sp1 DNA binding. Sp1-I, a 20-kDa Rb-associated factor (11), impairs Sp1 transcriptional activity by association with the pocket domain of Rb. Rb has been reported to synergistically stimulate transcriptional activity of Sp1 (55, 56). Because no direct interaction between Rb and Sp1 has been detected, it has been hypothesized that Rb expression sequesters or liberates inhibitory factors associated with Sp1, such as Sp1-I. It has recently been reported that Rb interacts with PML both in vitro and in vivo through its pocket domain (2). The Rb pocket has also been shown to mediate its stimulation of Sp1-mediated transcriptional activity (38). It is therefore tempting to hypothesize that PML may be another Sp1-associated inhibitor that is targeted by Rb. Therefore, it would be interesting to examine whether Sp1 activity inhibited by PML can be restored by Rb in a set of squelching experiments.

Several studies from our laboratory and others have shown that PML is a growth and transformation suppressor (24, 41, 43, 46, 48, 49, 59). We have recently found that overexpression of PML can significantly suppress the growth and tumorigenicity of breast cancer cells by inducing G₁ arrest and apoptosis (43). Accordingly, ectopic expression of PML in normal human fibroblasts results in the induction of G₁ arrest (unpublished results). Furthermore, we have recently shown that the stable expression of PML in HeLa cells can lengthen the G₁ phase of the cell cycle (49). Together, these studies further demonstrate that the effect of PML on G₁ cell cycle progression correlates with an alteration of Rb phosphorylation and expression of a number of cell cycle-related proteins, such as cyclin E, cyclin D1, Cdk2, p27, p21, and p53 (43, 49). These findings in turn suggest that PML most likely exerts its suppressive effects on cell growth by targeting the protein factors involved at the G₁/S transition checkpoint. A large group of genes whose products are involved in cell growth (e.g., DHFR and thymidine kinase) are activated during progression through G₁/S and contain binding sites for Sp1 as well as E2F in their promoter regions (5, 46).

In this study, we have shown that the repressive effects of PML on Sp1 activity are not limited to EGFR and that PML efficiently suppresses the Sp1-stimulated activity of the DHFR promoter in Drosophila SL2 cells. The region of Sp1 implicated in inhibition by PML also mediates Sp1-E2F1 interaction (36, 46). Thus, the present findings imply that PML inactivation of Sp1 during the G₁-S transition could be a major mechanism for rendering PML's growth suppressor function. Recently, it has been demonstrated that Sp1 functions as a critical factor in cell growth and control of DHFR expression during cell cycle progression after serum stimulation of quiescent cells (52). In addition, detailed analysis of the DHFR promoter has revealed that the Sp1 sites, but not the E2F site, of the promoter mediate its transcriptional activity during the late G₁/S phase of the cell cycle (30). Together, these studies strongly support the importance of Sp1 in the regulation of cell growth and cell cycle progression. In this light, investigations into the effect of PML on Sp1-mediated transcription of other genes involved in the G₁/S checkpoint are now under way in our laboratory.