Epidermal Growth Factor and Okadaic Acid Stimulate Sp1 Proteolysis*

Sp1 nuclear levels have been shown todirectly correlate with the proliferative state of the cell. We therefore studied changes in the abundance of Sp1 in a rat pituitary cell line GH4 whose growth rate is regulated by epidermal growth factor (EGF). Nuclear extracts from GH4cells treated with 10 nm EGF for at least 16 h showed a 50% decrease in Sp1 binding to a GC-rich element present in the gastrin promoter. The decrease in binding correlated with a decrease in cell proliferation, a loss of nuclear Sp1 protein and a 50–60% decrease in Sp1-mediated transactivation through an Sp1 enhancer element in transfection assays. Okadaic acid, a phosphatase inhibitor, was synergistic with the effect of EGF on Sp1 protein levels suggesting that the loss of Sp1 was mediated by phosphorylation events. This result was confirmed by showing a 2-fold increase in orthophosphate-labeled Sp1 with EGF and okadaic acid. Cycloheximide prevented the expected loss of Sp1 mediated by EGF and okadaic acid suggesting that the synthesis of a protease may mediate these events. This hypothesis was tested directly by showing that the cysteine protease inhibitor leupeptin prevented Sp1 degradation. Using the PEST-FIND computer program, the computed PEST score for human and rat Sp1 is 10.4 and 13.7, respectively, indicating that Sp1 has a domain with a high concentration of proline, glutamic acid, serine, and threonine residues as reported for a number of proteins with inducible rates of degradation. Collectively, these results indicate that sustained stimulation of GH4 cells by EGF initiates a cascade of phosphorylation events that promotes Sp1 proteolysis, decreased Sp1 nuclear levels and decreased cellular proliferation.

While Sp1 is generally considered to be a constitutively active housekeeping gene, cellular Sp1 levels have been correlated with tumor cell mitotic rates, cellular proliferation, and cell cycle regulators (1,2). Sp1 mRNA levels increase in response to SV40 viral infection in CV-1 cells and in response to phorbol ester induction of T-lymphocytes (3). Sp1 protein expression in normal tissue varies markedly during development being highest in the thymus, lung, spleen, and variably expressed in the stomach (2). Post-translational modification of Sp1 by phosphorylation and glycosylation has been correlated with Sp1 transcriptional activation in vitro (4,5). Thus, Sp1 nuclear levels appear to be closely related to increased cellular proliferation.
Transcription factor Sp1 binds to GC-rich elements in the promoters of both cellular and viral genes and stimulates basal promoter activation through TATA-binding protein-associated factors (6,7). Overexpression of Sp1 transactivates genes suggesting that regulating the abundance of cellular Sp1 is one mechanism through which Sp1 regulates transcription. Changes in the abundance of Sp1 may occur by regulating Sp1 gene expression, mRNA stability, or post-translational events that result in altered DNA affinity or protein turnover. However, the relationship of these processes to gene activation has not been well characterized.
We have previously studied Sp1 binding to a GC-rich element (GGGGCGGGGTGGGGGG) designated gastrin EGF Response Element (gERE) 1 in the gastrin promoter (8). This element confers both epidermal growth factor (EGF) and phorbol ester responsiveness and consists of two overlapping domains. The 5Ј domain binds Sp1 and the 3Ј domain binds two additional complexes provisionally designated as the gastrin EGF responsive proteins 1 and 2 (gERP 1 and gERP 2) (9). EGF induction of the gastrin promoter occurs within the first 12 h and requires both the 5Ј and 3Ј half-sites. Although EGF induction of the gastrin promoter in GH 4 cells appear to require Sp1, there is no significant increase in the binding of this protein with EGF (9). Instead after 16 h of EGF treatment, Sp1 binding decreases. EGF treatment of GH 4 cells slows cellular proliferation which correlates inversely with cellular differentiation (10,11). Collectively, these results suggest that Sp1 nuclear levels correlate directly with cellular proliferation.
To study the effect of EGF on Sp1 in greater detail, changes in Sp1 binding and abundance were studied by EMSAs, immunoblots, and metabolically labeling endogenous Sp1. We found that Sp1 protein levels decreased in response to prolonged treatment with EGF. Since EGF receptor activation results in a cascade of phosphorylation events and Sp1 is known to be a phosphoprotein (4,12,13), we reasoned that EGF stimulation of Sp1 proteolysis may be related to its phosphorylation state. We therefore examined the effect of combining EGF with a phosphatase inhibitor, okadaic acid to sustain Sp1 phosphorylation. In addition, since the decrease in cellular Sp1 levels indicated that Sp1 may undergo proteolysis, we investigated whether this process required protein synthesis and was inhib-ited by protease inhibitors. The results indicate that EGF and okadaic acid increase Sp1 phosphorylation and an increase in Sp1 turnover mediated by a cysteine protease.
Orthophosphate Labeling of Sp1-GH 4 cells (10 6 cells) were incubated in 3 ml of phosphate-deficient DMEM with 10% dialyzed fetal calf serum (Life Technologies, Inc.) for 1 h prior to the addition of labeling medium. The cells were incubated for 4 h with 100 Ci/ml [ 32 P]orthophosphate (2500 -3500 Ci/mmol) in phosphate-deficient medium (labeling medium) that also contained EGF, okadaic acid, or both. The incubations were terminated by removing the labeling medium, rinsing the cells with PBS and preparing nuclear extracts by detergent lysis according to the method of Schreiber (15). Twenty-five micrograms of nuclear protein was then used to immunoprecipitate Sp1. Nonspecific protein was precleared using magnetized Dynabeads (Dynal Inc., Oslo, Norway) coated with rabbit IgG. The beads were washed twice in PBS containing 0.1% bovine serum albumin and the supernatants combined. Sp1 was immunoprecipitated from the combined precleared supernatants at 4°C for 2 h using Dynabeads precoated with anti-Sp1 antibody (Santa Cruz Biotechnology). The specificity of the immunoprecipitation assay was demonstrated by competing with Sp1 peptide (Santa Cruz Biotechnology). After collecting the pellet, the beads were washed three times in 0.1% bovine serum albumin dissolved in PBS and the labeled protein eluted by heating the sample to 65°C for 5 min in 30 l of Laemmli sample buffer and resolving on a 7.5% SDS-polyacrylamide gel.
Pulse-Chase Analysis-GH 4 cells were cultured to a density of 70% confluency on 75-cm 2 flasks (10 7 cells) in control DMEM with a total of 14% serum and 10 nM EGF, 50 nM okadaic acid, 2 M leupeptin, or combinations of these agents as indicated for 10 h. After culturing, the media was removed and the cells were rinsed once in DMEM minus methionine and cysteine with 10% dialyzed fetal calf serum. The cells were incubated for 1 h in the deficient medium then replaced with labeling medium. The cells were labeled in 10 ml of cysteine and methionine-deficient DMEM with 10% dialyzed fetal calf serum that contained ϳ30 Ci/ml [ 35 S]methionine (Ͼ1000 Ci/mmol) (Amersham). After 1 h, the cells were switched to complete medium containing 2 mM unlabeled methionine and 10 nM EGF, 50 nM okadaic acid, 10 nM EGF plus 50 nM okadaic acid, 2 M leupeptin or leupeptin plus EGF and okadaic acid. After time intervals up to 8 h in the complete medium, the cells were washed three times in PBS alone then lysed in ice-cold Nonidet P-40 lysis buffer containing 1% Nonidet P-40, 50 mM Tris, pH 8.0, 150 mM NaCl, 10 mM EDTA, 1 mM dithiothreitol, protease inhibi-tors, and phosphatase inhibitors as described above. The whole cell lysates were immediately frozen in liquid N 2 and stored at Ϫ80°C until use.
Thawed lysates were clarified by centrifuging at 13,000 rpm for 30 min. Protein was determined on the supernatant using the method of Bradford (18). Immunoprecipitation was performed on 200 g of protein by incubating with 1.5 g of rabbit anti-Sp1 antibody and 30 l of protein A-agarose for 2 h at 4°C with gentle rocking. The labeled antigen-antibody complexes were washed three times in Nonidet P-40 lysis buffer, resuspended in 30 l of Laemmli sample buffer, heated to 65°C for 2 min, then resolved on a 7.5% SDS-polyacrylamide gel. The amount of labeled Sp1 protein immunoprecipitated was quantified on a PhosphorImager.
Immunoblot Analysis-Fifty micrograms of GH 4 cell nuclear extracts were resolved on a 7.5% SDS-polyacrylamide gel and electrophoretically transferred to polyvinylidene difluoride membrane (Bio-Rad). The membrane was blocked for 1 h in 100 mM Tris-HCl, pH 7.5, 0.9% NaCl, 0.1% Tween 20 (TTBS) containing 5% nonfat dry milk then probed for 1 h at 25°C with a 1:1000 dilution of anti-Sp1 antibody in the blocking solution. The membrane was rinsed in TTBS then probed for 1 h with a 1:1000 dilution of donkey anti-rabbit IgG linked to horseradish peroxidase in blocking solution. Antigen-antibody complexes were detected using enhanced chemiluminescence (ECL assay kit, Amersham).
Northern Blot Analysis-GH 4 cells were grown to 50% confluence on 150-mm culture dishes then treated with complete medium containing 10 nM EGF, 50 nM okadaic acid, or both. At the indicated times, total cellular RNA was prepared using TRI Reagent (Molecular Research Center, Cincinnati, OH). Twenty micrograms of total RNA was electrophoresed on 1% agarose gels containing 5.9% formaldehyde, 18 mM Na 2 HPO 4 , and 2 mM NaH 2 PO 4 and electrophoresed in the same buffer. RNA was transferred using a Vacuum-Blotter (Bio-Rad) to Hybond-N ϩ membrane (Amersham). The rat Sp1 cDNA subcloned into pRc/CMV (Invitrogen) was a kind gift from Guntram Suske (Institute of Molecular Tumor Biology, University of Marburg, Germany). An antisense rat Sp1 riboprobe was created by linearizing the plasmid with BamHI and priming from the Sp6 promoter. Blots were probed with a radiolabeled cyclophillin cDNA probe at 55°C then reprobed at 65°C in 50% formamide with a 600-base pair Sp1 riboprobe complementary to the terminal 600 base pairs of rat Sp1 (the zinc finger domain).
Cell Proliferation Assays-Cellular proliferation studies were performed on GH 4 cells in the presence or absence of 10 nM EGF using the CellTiter 96 AQ non-radioactive assay (Promega). GH 4 cells were plated in quadruplicates at a density of 5000 cells per microtiter well in serum-free DMEM without phenol red in the presence or absence of 10 nM EGF. At the time points indicated, cellular proliferation was measured by adding the tetrazolium compound (3-(4,5-dimethylthiazol-2yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) and measuring the formation of formazan after 2.5 h at an absorbance of 490 nm. The absorbance minus background was plotted as a function of time.
Transient Cell Transfection-A minimal gastrin promoter construct was created by inserting into the BglII site of a promoterless luciferase vector (pGL2-basic, Promega) a 43-base pair oligonucleotide cassette that contained sequences Ϫ28 to ϩ9 of the human gastrin promoter and flanking 5Ј BglII and 3Ј BamHI restriction sites. Other constructs were created by inserting oligonucleotide cassettes corresponding to the wild type gERE (WT: GGGGCGGGGTGGGGGG), a gERE mutant (M6: GGGGCGGGGCGGGGCG), or the human metallothionein IIa Sp1 element (Sp1: CCCGGCCGGGGCGGGG) upstream of the gastrin promoter at a regenerated BglII site as described previously (9). All inserts were verified by restriction analysis and sequencing. Oligonucleotides were synthesized on an automated DNA synthesizer (Applied Biosystems, Inc.) employing ␤-cyanoethyl phosphoramide chemistry. Plasmids for transfection were prepared by a modified alkaline lysis procedure (Qiagen plasmid kit). Cells were transfected with 5 g of plasmid DNA for 15 min at 37°C in DMEM containing 400 g/ml DEAEdextran, 50 mM Tris-HCl followed by glycerol shock for 2 min. Luciferase assays were performed on an automated Luminometer Auto Lumat LB953 (Berthold) and normalized to nuclear protein. Protein was assayed by the method of Bradford (18).

EGF Stimulates a Decrease in Sp1
Binding and Protein-To assess whether EGF induces changes in Sp1 binding, GH 4 cells were treated with 10 nM EGF for 2, 4, or 16 h (Fig. 1A). In the presence of 1 mM ZnCl 2 , there was an increase in Sp1 and gERP 2 complex binding as described previously (9). No change in Sp1 or gERP binding was observed within 4 h. However, after at least 16 h of EGF treatment, Sp1 binding decreased without a significant change in the gERP 1 and 2 complexes. To assess whether the decrease in Sp1 binding was due to a change in Sp1 binding activity or a change in the level of nuclear Sp1, immunoblot analysis was performed using Sp1 antibody. Nuclear extracts prepared from GH 4 cells were treated for 16 h with EGF and resolved by SDS-PAGE. The results shown in Fig. 1B indicate that there was a decrease in the nuclear levels of Sp1. In addition, no Sp1 was detected by immunoblot analysis of cytosolic extracts after 16 h of EGF treatment indicating that the decrease in Sp1 was not due to retention of Sp1 in the cytoplasm (data not shown).
A Decrease in Sp1 Protein Correlates with a Decrease in Sp1-mediated Transcriptional Activation-To demonstrate that the lower levels of Sp1 generated by chronic EGF treatment also correlate with a decrease in Sp1 transactivation, several gastrin-reporter constructs were tested in GH 4 cells. GH 4 cells were pretreated with 10 nM EGF 24 h prior to transfection with gastrin-reporter constructs and compared with cells not pretreated with EGF prior to transfection. Three reporter constructs containing elements ligated upstream of a minimal gastrin promoter were transfected: the WT gERE element, a mutation of the gERE element that only binds Sp1 (M6) and an Sp1 consensus element from the human metallothionein promoter (Sp1). The binding specificity and activity of these constructs has been reported previously (9). The results shown in Table I demonstrate that pretreatment with EGF had little effect on the basal activity of the gastrin promoter in the presence of the WT element. However, those elements capable of binding only Sp1 were significantly affected by EGF pretreatment. In particular, both the mutated gERE element (M6) and the Sp1 element which normally confer higher basal activity than the WT element showed a 50 -60% decrease in basal promoter activity. Since these elements have a higher affinity for Sp1 compared with the gERP complexes (9), the observed decrease in basal activation was consistent with a decrease in the abundance of Sp1.
EGF Inhibits GH 4 Cell Proliferation-Tashjian and co-workers (10) showed previously that EGF inhibits GH 4 proliferation and DNA content within 72 h. Thus the decrease in Sp1 binding and abundance parallels a decrease in cellular proliferation. To confirm that EGF has a negative effect on GH 4 proliferation, the proliferative rate of GH 4 cells over 72 h was studied. Five thousand cells were plated per microtiter well with or without 10 nM EGF. Cellular proliferation was quantified at the times indicated and clearly showed that EGF negatively affected cellular proliferation (Fig. 2).
Okadaic Acid Augments the Effect of EGF on Sp1 Protein Levels-To determine whether the effect of EGF was related to the level of cellular phosphorylation, immunoblot analysis was performed on nuclear extracts prepared from cells treated with the phosphatase inhibitor okadaic acid or with both EGF and okadaic acid. The results shown in Fig. 3 demonstrate that treatment of the cells with okadaic acid alone resulted in a similar or greater decrease in Sp1 nuclear levels compared with EGF alone. Treatment of the cells with both EGF and okadaic acid had a synergistic effect, further depressing nuclear levels of Sp1. These results are consistent with studies showing that okadaic acid treatment of GH 4 cells inhibits proliferation in response to decreased phosphatase activity (19). Since okadaic acid alone and in concert enhanced the effect of EGF on the abundance of Sp1, these results suggested that the decrease in Sp1 is related to its phosphorylation state.
To determine whether the decrease in Sp1 protein was related to a decrease in Sp1 gene expression, Northern blot analysis was performed. The results shown in Fig. 5 indicate that treatment of cells with EGF and okadaic acid, which significantly decrease Sp1 protein levels, did not decrease Sp1 gene expression. Instead, okadaic acid alone and the combination of EGF and okadaic acid treatment stimulated Sp1 gene expression possibly due to a feedback regulatory loop that mediates activation of the Sp1 gene as a result of a significant decrease in Sp1 protein in the cell.
The effect of EGF and okadaic acid on Sp1 levels was reversible upon removal of these agents from the culture medium (Fig. 6). In complete medium, Sp1 binding activity increased with time (Fig. 6A). In contrast, the addition of EGF and okadaic acid for 16 h resulted in a Ͼ60% decrease in Sp1 binding activity that was detectable by 24 h. There were minimal changes in binding of the gERP factor complex and the ubiquitous transcription factor Oct-1 (Fig. 6B). The decreased ratio of Sp1 to Oct-1 binding was lowest at 24 h and returned to control levels after the removal of EGF and okadaic acid (Fig. 6C).
Cycloheximide Treatment Blocks Sp1 Degradation-To assess whether the decrease in Sp1 protein levels required protein synthesis, GH 4 cells were incubated with cycloheximide for 1 h prior to treatment with EGF and okadaic acid. Cycloheximide showed a dose-dependent inhibition of the expected decrease in cellular Sp1 levels (Fig. 7A). In addition to an inhibition of Sp1 degradation, cycloheximide treatment also decreased the appearance of several lower molecular weight species detected by Sp1 antibody (Fig. 7B). These presumably were proteolytic fragments of Sp1 since their appearance increased with EGF (see also Fig. 1B) or EGF and okadaic acid treatment and decreased with the inhibition of protein synthesis.
A Protease Inhibitor Prevents Sp1 Degradation-The dependence of Sp1 degradation upon protein synthesis suggested that Sp1 turnover may be related to increased protease activity. Therefore, to determine whether the decrease in Sp1 levels was reversed by specific protease inhibitors, GH 4 cells were pretreated with two membrane-permeable protease inhibitors MG-132 and leupeptin. MG-132 is a proteosome-restricted protease inhibitor; whereas leupeptin is a cysteine protease that inhibits lysosomal and calcium-activated proteases. In addi- tion, it has been shown that leupeptin inhibits a nuclear differentiation-associated cysteine protease that is unrelated to calpains (22). Fig. 8 shows that MG-132 at 1 M did not prevent the expected decrease in Sp1. This concentration was 10 -40fold lower than concentrations used to inhibit NFB processing by the proteosome (16). In contrast, the cysteine protease inhibitor leupeptin at 2 M was an effective inhibitor of Sp1 degradation, suggesting that the protease system targeting Sp1 turnover may not be related to the proteasome.
EGF and Okadaic Acid Increase the Rate of Sp1 Degradation-Both the binding assays and immunoblots showed that EGF and okadaic acid decrease the abundance of Sp1 in the cell. Moreover, EGF and okadaic acid together stimulate an increase in Sp1 phosphorylation. To correlate these findings with an increase in the rate of Sp1 degradation, the fate of metabolically labeled Sp1 was followed over time during each treatment. GH 4 cells were pretreated with EGF, okadaic acid, leupeptin, or various combinations for 10 h then labeled with [ 35 S]methionine in methionine-deficient medium. The label was chased for intervals up to 8 h in the presence of the agonists and 2 mM unlabeled methionine in DMEM. The halflife of Sp1 was 6 h under control conditions. The results shown in Fig. 9A reveal that EGF and okadaic acid alone or in com-bination increased the loss of radiolabeled Sp1. However, the most significant loss was observed with okadaic acid. The maximum effect of each treatment on Sp1 degradation rates was detected 6 h after exposure to the pulse of label and indicated that a minimum of 16 h with EGF and okadaic acid was necessary to observe changes (Fig. 9B). The calcium-dependent protease inhibitor slowed the normal turnover of Sp1 protein in the cell and the expected loss of Sp1 accelerated by EGF and okadaic acid treatment. These results correlated with the loss of total nuclear Sp1 observed on immunoblots and the time lag required for Sp1 degradation.
PEST Sequence Identification-Several targets of inducible proteolysis contain a region with an unusually high concentration of the amino acids proline (P), glutamic (E), or aspartic acid (D), serine (S), and threonine (T); whereas fewer that 5% of a random survey of the GENEPRO data bank contain such PEST sequences (23,24). PEST-FIND, a program ranking these amino acid domains within proteins produced significant scores (Ͼ5.0) for a number of proteins with rapid rates of degradation: cdc25 (15.6), IB␣ (5.1), and c-Fos (10.1). In contrast, proteins with low degradation rates produced low scores, e.g. adenylate kinase (Ϫ11.8). We evaluated the amino acid sequence of Sp1 to determine if it also contained a PEST domain. Human Sp1 produced a score of 10.4 and rat Sp1 FIG. 6. Reversibility of EGF and okadaic acid (Ok) effects on Sp1 binding. A, EMSA of nuclear extracts using WT-gERE oligonucleotide in the presence of 1 mM Zn 2ϩ to assess Sp1 (arrowhead) and gERP binding (arrows). B, EMSA using Oct-1 probe (open arrow). C, EMSA binding quantified by PhosphorImager analysis then expressed as a ratio of Sp1 to Oct-1 binding over time. Hatched bars represent the Sp1/Oct-1 ratio of untreated cells; solid bars represent the ratio of EGF/okadaic acid-treated cells. Control cells were grown in standard medium and compared with cells treated with 10 nM EGF and 50 nM okadaic acid. Cells were harvested at the times indicated and nuclear extracts were prepared. The maximum time exposed to EGF and okadaic was 16 h. The medium was then removed from both groups, the cells were washed with PBS and replaced with fresh medium without EGF and okadaic acid.  (lanes 3-7). Panel A indicates a 105-kDa protein corresponding to the size of intact Sp1 (arrowhead). Panel B is an overexposure of the same blot that reveals additional lower molecular weight bands recognized by the anti-Sp1 antibody (arrows). produced a score of 13.7, consistent with an internal PEST sequence between amino acids 437 and 458, (Fig. 10). Targeted degradation of proteins containing PEST sequences is a mechanism for rapidly controlling transcription factor activity (23). Thus proteolysis of Sp1 correlated with the presence of a PEST domain.

DISCUSSION
The present study shows that stimulation of GH 4 cells with physiologic concentrations of EGF results in a decrease in the cellular content of Sp1 and that okadaic acid augments this effect. These results reflect a decrease in both immunoreactive Sp1, metabolically-labeled Sp1, as well as Sp1 DNA binding and transcriptional activity. Okadaic acid specifically binds to and inhibits the catalytic subunits of serine/threonine phosphatases 1 and 2A with different potencies (13). It therefore has been used to study the effect of unopposed kinase activity on transcription and DNA-protein interactions (25).
SV40 viral infection of CV1-L cells stimulates Sp1 phosphorylation by a DNA-dependent protein kinase (4). This increase in Sp1 phosphorylation may also be related to viral inhibition of an opposing phosphatase since the small T antigen of SV40 interacts with protein phosphatase 2A (26), the same phosphatase inhibited by okadaic acid. In GH 4 cells, 10 nM okadaic acid inhibits GH 4 cell proliferation and interfers with the formation of the mitotic spindle, thereby slowing cellular progression through mitosis (19). Similarly, EGF inhibits cellular proliferation of GH 4 cells (10) suggesting that both EGF and okadaic acid may function through overlapping pathways. One pathway of overlap appears to be the ability of both agents to affect the abundance of Sp1. The loss of Sp1 was observed with EGF and augmented by okadaic acid, suggesting that activation of the EGF receptor phosphorylates Sp1 through increased intracellular serine/threonine protein kinase activity. No decrease in Sp1 mRNA was observed with stimulation indicating that the loss of Sp1 was post-transcriptional. Since phosphorylation of Sp1 can be detected within 4 h well before significant degradation is detected, the results suggest that sustained Sp1 phosphorylation targets the protein for turnover. The loss of Sp1 results from increased proteolytic activity since a cysteine protease inhibitor, leupeptin, blocked the degradation expected. In addition, degradation was also inhibited in a dosedependent fashion by cycloheximide at concentrations that have been shown to inhibit protein synthesis. The cycloheximide result is consistent with induction of Sp1 degradation and the time required to synthesize an Sp1-specific protease. The effect of EGF and okadaic acid on Sp1 was specific since a constitutive transcription factor Oct-1 remained unchanged, as did the gERP complexes.
While the effect of EGF stimulation on a general transcription factor in GH 4 cells appear paradoxical there is clear precedence. Phorbol esters, EGF, and okadaic acid were previously shown to be antiproliferative in GH 4 cells resulting in phenotypic changes consistent with differentiation (10,11,19). Van Doulah et al. (19) relates this antiproliferative effect to the slowed progression through mitosis and sustained phosphorylation of the retinoblastoma gene product Rb. Subsequent studies in GH 4 cells have shown that the ability to dissociate the differentiating effects of EGF from its mitogenic response is dependent upon the concentration of other serum factors (27). Okadaic acid mimics EGF or phorbol ester activation of c-fos gene expression, AP-1 binding, and NFB transactivation (20, 28,29). Thus increasing protein phosphorylation or decreasing phosphoprotein phosphatase activity similarly affect transcriptional regulation and have been implicated as common pathways for cell transformation and tumor promotion. In GH 4 cells, these extracellular regulators inhibit proliferation and promote differentiation.
Sp1 abundance and activity are clearly related to proliferation perhaps through its ability to cooperate with the retinoblastoma gene product Rb and the ability to regulate Rb gene expression. Rb stimulates the transcription of several growthrelated genes (e.g. insulin growth factor-2, c-fos, transforming growth factor-␤1) through complexes that contain Sp1 (30). Moreover, the promoter elements regulated by Rb are also binding sites for Sp1 (31). One potential mechanism by which Rb cooperates with Sp1 is that Rb releases a negative coregulator from Sp1 (32). Sp1 may also regulate Rb gene expression since the Rb promoter contains an important Sp1 site that when mutated to abolish binding contributes to low-penetrance hereditary retinoblastoma (33).
Additional support for the inverse correlation of Sp1 nuclear levels with differentiation has been observed in embryonal carcinoma cells that differentiate into parietal endoderm (22). In this system, Sp1 nuclear levels decreased with retinoic acid induction of differentiation. The decrease was due to a cysteine protease localized in the nucleus since leupeptin and antipain prevented the loss of Sp1 protein. Thus, as reported here, anti-proliferative effects or differentiation increases nuclear protease activity and decreases the level of specific transcription factors including Sp1. Moreover, both the current study and the experiments by Scholtz et al. (22) indicate that a cysteine protease mediates Sp1 degradation. There are no studies reporting Sp1 ubiquitination; thus, ubiquitin-mediated degradation of Sp1 cannot be excluded. Even so, it is not clear that all ubiquitinated proteins are degraded by the proteosome (34,35).
Regulation of transcription from promoters containing Sp1binding sites is thought to involve several poorly characterized mechanisms including: modulation of Sp1 glycosylation (5), inducible Sp1 phosphorylation (4), control of Sp1 abundance (36), cooperative interactions with tissue-specific transcription factors (31,37), or control of Sp1 affinity for DNA and transcription factors within the TFIID complex (38). Previous reports support a role for tissue specific changes in Sp1 levels during development (2), but this was in a heterogeneous cell population and may represent relative increases in only some cell types. A recent study reported that decreased Sp1 binding activity in terminally differentiated liver tissue correlates with protein phosphorylation (39). However, this study did not assess total immunoreactive Sp1 levels. The present study examines both Sp1 DNA binding and protein levels in a homogenous tissue culture system. The results indicate that Sp1 degradation represents another means of regulating its transcriptional activity.
Inducible proteolysis as a means of regulating transcription factor activity has been established in other systems. Two models in eukaryotes are: 1) protease induced translocation of a protein with gain of DNA-binding function, and 2) protease degradation of a DNA-binding protein with loss of function. NFB is sequestered in the cytoplasm until cytokine induced phosphorylation of the inhibitory IB subunit results in ubiquitinylation, proteosomal degradation, and subsequent translocation of the Rel homology domain to the nucleus (16). A similar mechanism regulates the sterol regulatory element 1 within the 5Ј-flanking region of the low density lipoproteinreceptor gene. SREBP-1, a nuclear envelope/endoplasmic reticulum-bound transcription factor, is released by proteolysis in sterol-depleted cells, translocates to the nucleus, and activates transcription via cooperation with a weak adjacent Sp1-binding site (40,41). An example of the second model is seen in the termination of the activity of the yeast transcription factor Gcn4, a regulator of amino acid and purine biosynthesis. Turnover of Gcn4 is regulated by extracellular amino acid concentration, with high amino acid levels resulting in ubiquitin-dependent degradation of Gcn4 (42). Likewise, characterization of the turnover of c-Jun showed that the ␦-domain is necessary but not sufficient to mediate ubiquitin-dependent degradation of the transcription factor (43). The ␦-domain is absent from the unregulated retroviral homologue v-Jun suggesting that the inability to degrade the viral homologue may represent the means for viral protein escape from cellular control.
In summary, EGF stimulation of GH 4 cells results in the proteolysis of Sp1. Sp1 binding to gERE has been shown to be required for hormone induction of the gastrin promoter in this cell line (9). The time course determined for Sp1 degradation suggests that this may represent one mechanism for the desensitization of EGF hormone stimulation. Alternatively, it may represent a part of the process affecting an antiproliferative and differentiated phenotype for these cells in response to EGF. The dysregulation of cellular proteases has been identified in a number of disease states and proteolysis is important in the control of differentiation (22,44,45). Phenotypic changes may result from inducible degradation of transcription factors resulting in the alteration of cell fate (16,43,46). In GH 4 cells, inducible expression of the gastrin gene coincides with the differentiated state of these cells since prolactin levels are also FIG. 10. Location of Sp1 PEST domain. The program PEST-FIND was used to predict the location of all significant PEST domains within the primary sequence of human Sp1. Only a single domain with a PEST score greater than 5.0 was identified. This sequence is shown relative to the location of zinc finger, acidic, and basic domains. induced with EGF (10). Thus, the results reported here link these well documented observations to a series of specific molecular events that will likely be applicable to other biologic systems.