INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

THE GENE ENCODING THE PEPTIDE hormone gastrin is produced by neuroendocrine cells of the adult stomach and is positively regulated by epidermal growth factor (EGF) receptor activation through an Sp1 binding site (16, 33). This raised the possibility that Sp1 transactivation may be modulated through inducible phosphorylation by the ras-extracellular signal-regulated kinase (Erk) pathway, as demonstrated for many other EGF-regulated promoters (43). Ras activation of the gastrin promoter is clinically relevant because of the elevated levels of gastrin mRNA in colon cancers containing an activating mutation of K-ras (41). Thus understanding how the ras-Erk pathway stimulates gastrin gene expression may further our understanding of gastrointestinal transformation. Consistent with the role of activated Ras in colonic transformation and gastrin gene expression, we recently showed that activation of the gastrin promoter through the GC-rich gastrin EGF response element (gERE) is ras-Erk dependent (34). Moreover, purified Erk 2 phosphorylates Sp1 and stimulates increased Sp1 binding and affinity for DNA. We also found that overexpression of Erk stimulates gastrin promoter activation and that the Mek 1 inhibitor PD-98059 prevents Sp1-dependent transactivation of the gastrin promoter (34).

Initial understanding of the role of Sp1 focused on its ability to regulate constitutively active genes (11) that mediate interaction with the basal transcription apparatus (45). However, recent evidence has emerged to indicate that the activity of Sp1 is modulated by cell growth and differentiation (5, 12, 37, 40, 47, 48). It appears that growth regulation of Sp1-dependent transcription correlates with a change in the state of Sp1 phosphorylation. Moreover, the kinases phosphorylating Sp1 as well as activated signaling pathways appear to be agonist, cell type, and promoter dependent. Thus Sp1 no longer represents a transcription factor unaffected by extracellular signals but exhibits complexities as intricate as other mitogen-inducible transcription factors, e.g., signal tranducers and activators of transcription (STAT), Elk, and Sap1A. This may be surprising, since many promoters contain GC-rich sites capable of binding Sp1. Yet, in many instances, these sites are bound not only by Sp1 and Sp1 family members but by other zinc finger transcription factors that may compete or cooperate with Sp1 (26, 27, 29, 35, 49, 50).

The focus of the present study was to determine whether Sp1 is a substrate for endogenously activated Erk alone or whether the "Sp1 kinase" activity represents additional EGF-activated kinases. We found that the major Ser/Thr kinase activity phosphorylating Sp1 targeted its NH₂-terminal domain and was distinct from Erk.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Plasmids. pGEX1-Sp1FLU or pGEX1N-Sp1ABC containing full-length human Sp1 or amino acids 1-530 of Sp1 cDNA (NH₂-ter_1-530 Sp1) were gifts from J. Horowitz (Duke University, Durham, NC) (51).

Cell culture. AGS cells (derived from a human gastric adenocarcinoma) (4) were purchased from American Type Culture Collection and cultured in DMEM (GIBCO-BRL) containing 10% FCS, 100 µg/ml penicillin, and 100 µg/ml streptomycin in a humidified atmosphere of 5% CO₂-95% air in 35-mm six-well culture dishes at 37°C. The cells were stably transfected with the 240 gastrin-luciferase reporter construct (240 GasLuc) or the 240 GasLuc construct containing a 4-bp mutation of the EGF response element (m240 GasLuc), selected in G418 (GIBCO-BRL), and pooled as previously described (33). The 240 GasLuc construct contained 240 bp of the human gastrin promoter, and the first exon ligated upstream of the luciferase reporter in pGL2 basic (Promega). Calcium coprecipitation (5 Prime-3 Prime) was used to create stable transfections into AGS cells. The cells were incubated for 48 h in Ham's F-12 nutrient mixture containing 100 µg/ml penicillin and 100 µg/ml streptomycin (GIBCO-BRL) without serum before treatment with 10 nM EGF.

Immunoblots. Whole cell extracts prepared from AGS cells were lysed in 10 mM HEPES at pH 7.7, 30 mM NaCl, 2% glycerol, 0.5% Triton X-100, 3 mM MgCl₂, 0.2 mM EDTA, 0.2 mM Na₃VO₄, 2 mM NaF, 2 mM sodium pyrophosphate, and one Complete tablet per 10 ml (Boehringer-Mannheim), which contained a cocktail of protease inhibitors. The extracts were heat denatured in Laemmli sample buffer, resolved on a 7.5 or 10% Laemmli gel, and then electroblotted onto polyvinylidene difluoride membrane (Bio-Rad). After the sample was blocked for 1 h in 100 mM Tris · HCl, pH 7.5, 0.9% NaCl, and 0.05% Tween 20 containing 5% nonfat dry milk (TTBS), the blot was exposed to antibody diluted in TTBS for 1 h and then rinsed three times in TTBS. The blot was then incubated for 1 h in anti-goat IgG-horseradish peroxidase (1:1,000 dilution). After three additional 15-min rinses in TTBS, the resulting protein-antibody complexes were detected by chemiluminescence (SuperSignal, Pierce Biochemicals). Sp1 and Erk antibodies were purchased from Santa Cruz Biotechnology; phospho-Erk antibody was purchased from New England BioLabs. Rsk, p38, DNA-dependent protein kinase, and JNK antibodies were purchased from Santa Cruz Biotechnology.

Electrophoretic mobility shift assay. AGS whole cell extracts were prepared as described above, snap frozen, and stored at 80°C until use. A double-stranded oligonucleotide probe corresponding to the gERE GGGGCGGGGTGGGGGG was end-labeled using Klenow Exo-enzyme (New England BioLabs) and [-³²P]dCTP (Amersham). The probe was purified using Quick Spin columns (Boehringer Mannheim). Ten micrograms of whole cell extract were incubated in a final volume of 20 µl containing 10 mM Tris · HCl, pH 7.9, 1 mM ZnCl₂, 100 mM KCl, 1 mM EDTA, 300 ng dI · dC, 1 mM dithiothreitol (DTT), 5 mM MgCl₂, 10% glycerol, and radiolabeled probe (30,000 cpm/0.01 ng) at 25°C and then resolved on a 4% nondenaturing polyacrylamide gel containing 45 mM Tris base, 45 mM boric acid, and 1 mM EDTA. Antibodies to Sp1, Sp3, and Sp4 were purchased from Santa Cruz Biotechnology. Whole cell extracts were pretreated with 1-5 units of calf intestinal phosphatase (CIP) for 20 min at 30°C in electrophoretic mobility shift assay (EMSA) buffer before the probe was added.

Expression of fusion proteins. A 300-ml starter culture of transformed Escherichia coli strain BL21 (DE3) protease-deficient bacteria (Novagen) was grown overnight at 25°C and then used to inoculate 1 liter of fresh Luria broth containing 50 µg/ml ampicillin. The 1-liter culture was grown until log phase reached a density of A₆₀₀ = 0.4 optical density unit at 25°C, where A₆₀₀ is absorbance at 600 nm. Protein expression was initiated by the addition of a final concentration of 0.4 mM isopropyl--thiogalactopyranoside. After 2-3 h the cells were collected by centrifugation for 10 min at 8,000 rpm in a Beckman JA-10 rotor, resuspended in 15 ml of lysis buffer (1% Triton X-100, 1 mM DTT, 2 µg/ml aprotinin, 1 µg/ml leupeptin, 1 µg/ml pepstatin, 0.1 mM phenylmethylsulfonyl fluoride), and then sonicated twice for 15 s at 30% output power at 4°C. Bacterial debris was removed by centrifugation at 15,000 rpm for 30 min in a Beckman SW28 rotor. The fusion protein was isolated from the supernatant by batch chromatography with use of glutathione-Sepharose 4B beads at 25°C according to the manufacturer's instructions (Pharmacia). After the beads were washed three times in lysis buffer with protease inhibitors, the bound protein was stored in PBS with 0.02% sodium azide at 4°C until use (within 1 wk). The concentration of protein on the beads was determined by the method of Bradford (8), and the purity was confirmed by SDS gel electrophoresis and Coomassie blue staining. Glutathione S-transferase (GST) protein alone was prepared by expressing the pGEX vector (Pharmacia).

Column purification of extracts. Whole cell AGS extracts were prepared by detergent lysis in 0.5% Triton X-100, 50 mM -glycerophosphate, pH 7.2, 0.1 mM sodium orthovanadate, 2 mM MgCl₂, 1 mM EGTA, 10 µg/ml leupeptin, 2 µg/ml aprotinin, and 1 mM DTT, snap frozen, and stored at 80°C until use. MonoQ anion-exchange resin (1 ml; Pharmacia) was preequilibrated in column buffer consisting of 50 mM -glycerophosphate, pH 7.2, 0.1 mM sodium orthovanadate, 1 mM EGTA, and 1 mM DTT. The extracts were thawed, mixed, and loaded onto a preequilibrated MonoQ anion-exchange column three times. The column was washed three times in "column buffer" before elution of 1-ml fractions with an NaCl gradient of 180-560 mM. The fractions were collected in tubes containing 50 ml of 2 mM sodium orthovanadate and one Complete tablet per 2 ml of buffer. The extracts were analyzed by solid-phase kinase assays or by Western blots.

Solid-phase kinase assays. To study phosphorylation of Sp1, AGS cells grown to ~70% confluency were placed under serum-free conditions (Ham's F-12 nutrient mixture) for 48 h before the start of the experiment. Cells were treated for 1 h with 50 µM Mek 1 kinase inhibitor PD-98059 (BioMol Research Labs) before addition of 10 nM EGF. Whole cell extracts were prepared as previously described, and the protein concentration was determined by the method of Bradford (8). Thirty micrograms of GST-Sp1 fusion protein immobilized on beads or GST alone expressed from the empty pGEX vector were incubated with 200 µg of whole cell extracts with agitation for 3 h. Alternatively, 0.1 µg of recombinant GST-Erk 2 enzyme (Upstate Biotechnology) was incubated with 30 µg of GST-Sp1. Nonadherent protein was removed by centrifugation in an Eppendorf refrigerated microfuge for 5 min at 12,000 rpm and then washed three times in HEPES binding buffer (20 mM HEPES, pH 7.7, 50 mM NaCl, 2.5 mM MgCl₂, 0.1 mM EDTA, 0.05% Triton X-100). The beads were resuspended in 30 µl of kinase buffer (20 mM HEPES, pH 7.7, 20 mM MgCl₂, 20 mM p-nitrophenyl phosphate, 0.1 mM Na₃VO₄, 2 mM DTT, 20 mM ATP, 5 µCi of [-³²P]ATP) for 30 min at 30°C. The reaction was terminated on ice, and the beads were washed 3 times with 20 times the volume of kinase buffer without radiolabel. The protein was eluted in 30 µl of Laemmli sample buffer at 90°C for 2 min and resolved on a 10% SDS-polyacrylamide gel before autoradiography. Analysis of phosphate incorporation was determined on a Molecular Dynamics PhosphorImager. The gel was stained with Coomassie blue dye to verify loading.

In vivo phosphate labeling and immunoprecipitation of Sp1. AGS cells were cultured on 100-mm dishes, then preincubated in phosphate-free DMEM (GIBCO) for 1 h before the addition of labeling medium. The cells were incubated for 3 h in 3 ml of labeling medium consisting of phosphate-free DMEM with 100 µCi/ml [³²P]orthophosphate (Amersham) before the addition of PD-98059 for 1 h and then 10 nM EGF for 30 min. The cells were lysed on ice in 1 ml of RIPA buffer (1% Nonidet P-40, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 10 mM sodium phosphate, pH 7.2, 2 mM EDTA, 50 mM sodium fluoride, 0.2 mM sodium orthovanadate, 100 U/ml aprotinin). Sp1 protein was immunoprecipitated from cell extracts by use of rabbit polyclonal anti-Sp1 antibody and protein A-agarose (Santa Cruz). The immune complex was washed four times with RIPA buffer, then boiled for 3 min in 30 µl of 2× Laemmli sample buffer and resolved on a 10% SDS-polyacrylamide gel. Phosphorylated Sp1 was detected by autoradiography and analyzed on a PhosphorImager.

Phosphoamino acid analysis. Phosphorylated NH₂-ter_1-530 Sp1 fusion protein or in vivo radiolabeled immunoprecipitated Sp1 was separated on a 10% SDS-polyacrylamide gel and transferred to polyvinylidene difluoride membrane. The bands of interest were excised for analysis by two-dimensional electrophoresis, as described previously (7). The migration of phosphoserine, phosphothreonine, and phosphotyrosine was determined using standard mixtures.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

EGF stimulates the gastrin promoter through an Sp1 binding site. We previously showed that EGF stimulation of a human gastric cell line (AGS) increases gastrin promoter activity (16). Furthermore, EGF induction is mediated by a GC-rich element that binds several zinc finger factors, including Sp1 (16, 36). To study the time course of the induction, two lines of pooled stable transformants expressing the 240 gastrin reporter construct (240 GasLuc) and the same construct with a 4-bp mutation in the gERE element were treated over 6 h with 10 nM EGF. We found that significant EGF induction of the gastrin promoter was observed by 1 h and was maximal (~6-fold) within 4 h (Fig. 1). A 4-bp mutation that abolishes Sp1 binding was not induced by EGF over the same time period (16). We showed previously that the Mek 1 kinase inhibitor PD-98059 blocks EGF induction of the gastrin promoter by ~50% but has no effect on basal promoter activity (34).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Epidermal growth factor (EGF) induction of gastrin promoter maps to a GC-rich element. AGS cells were stably transfected with 240 gastrin-luciferase (GasLuc) construct or m240 GasLuc, which contains a 4-bp mutation within gastrin EGF response element (gERE). Pooled transformants were treated with 10 nM EGF for up to 6 h. Values (means ± SE for 3 independent experiments performed in triplicate) are expressed as order of magnitude induction of luciferase activity normalized to protein.

EMSA demonstrates Sp1 and Sp3 binding to gERE. We showed previously that Sp1 comprises the upper portion of the slower-migrating complex and that Sp3 and ZBP-89 comprise the lower two complexes (28, 34). To determine whether Sp1 binding to gERE increases with EGF treatment, whole cell extracts were prepared from AGS cells after treatment with EGF for various time periods. The results shown in Fig. 2 indicate that Sp1 and Sp3 binding increases within 5 min and is maximal by 30 min. However, only Sp1 mediates transactivation of the gastrin promoter (34). Therefore, to establish whether Sp1 binding is attenuated by pretreatment of the cells with PD-98059, AGS cells were pretreated with 50 µM PD-98059 before stimulation with EGF. The results show that PD-98059 effectively prevented Sp1 binding (Fig. 2A), suggesting that Sp1 binding is in part regulated by Mek 1 kinase activity, consistent with gastrin promoter induction requiring Mek 1. Furthermore, the results suggest that Sp1 phosphorylation occurs downstream of Mek 1 kinase activation. To confirm that the increase in binding was due to Sp1, antibody was incubated with the extracts before addition of the probe. The results show that the increase in binding was due primarily to Sp1 and Sp3 and not to the recruitment of novel transcription factors (Fig. 2B). An immunoblot using Sp1 antibody showed that Sp1 protein also increases with EGF stimulation within 30 min and is maximal by 4 h (Fig. 3). This increase was specific, since Erk 2 protein levels used as a control remained unchanged. Thus protein abundance may contribute to the sustained increase in Sp1 binding, whereas the initial increase in binding may be due to increased phosphorylation.

View larger version (95K): [in this window] [in a new window]   Fig. 2.   EGF stimulates Sp1 binding to gERE. Whole cell extracts were prepared from AGS cells treated with 10 nM EGF for up to 6 h. A: extracts from 48-h serum-starved cells (lane 1) or from cells treated with EGF for 5, 15, and 30 min and 1, 2, 3, 4, 5, and 6 h (lanes 2, 3, 4, 5, 6, 7, 8, 9, and 10) were resolved on a nondenaturing gel. Lanes 11-13, cells pretreated with PD-98059 (PD) and then treated with EGF for 1-3 h. B: untreated extracts (lane 1) or extracts from cells treated for 1-4 h (lanes 2-5) were compared with complexes generated with extracts incubated for 0-3 h with 1 µl of Sp1 antibody (lanes 6-9). Complexes were resolved on a nondenaturing gel electrophoresed for an extended period of time to resolve individual bands. *, Supershifted Sp1 protein. All extracts were incubated with radiolabeled gERE probe. Results are representative of duplicate experiments.

View larger version (68K): [in this window] [in a new window]   Fig. 3.   EGF stimulates an increase in Sp1 protein abundance. Whole cell extracts were prepared from AGS cells treated with 10 nM EGF for up to 6 h. Lane 1, extracts from 48-h serum-starved cells. Lanes 2, 3, 4, 5, 6, 7, and 8, extracts from cells treated with EGF for 15 and 30 min and 1, 2, 3, 4, and 6 h. Same blot was probed first with a 1:250 dilution of Sp1 antibody (top) and then with a 1:250 dilution of extracellular signal-regulated kinase (Erk) 2 antibody as a control for loading (bottom). Results are representative of 3 experiments.

Sp1 binding regulated by phosphorylation. We focused on the phosphorylation of Sp1, since Sp3 coexpression with the gastrin promoter in Drosophila Schneider cells has little effect on activation of the promoter, whereas Sp1 cotransfection stimulates the promoter nearly threefold (34). To examine whether phosphorylation contributed to Sp1 binding, whole cell extracts were treated with CIP before EMSAs were carried out. The results demonstrate that CIP treatment decreases binding without affecting the total amount of Sp1 protein (Fig. 4). Together these studies indicate that EGF stimulates Sp1 complex binding by two mechanisms: increased Sp1 abundance and increased binding due to phosphorylation.

View larger version (51K): [in this window] [in a new window]   Fig. 4.   Phosphorylation regulates Sp1 binding. Whole cell extracts were treated with increasing amounts of calf intestinal phosphatase (CIP). Same extract was used in immunoblot, in which Sp1 antibody (1:250) was used (top), and in an electrophoretic mobility shift assay (bottom). Lane 1, 30 µg of whole cell extract; lane 2, 1 U of CIP; lane 3, 3 U of CIP; lane 4, 5 U of CIP.

Endogenous Erk activity. Because EGF stimulates gastrin promoter activity and Sp1 binding and PD-98059 modulates these effects (34), the time course of endogenous Erk activation was examined. Phospho-Erk antibodies were used to probe blots containing whole cell extracts prepared from AGS cells after EGF treatment. The results demonstrated rapid autophosphorylation of Erk kinases that was inhibited by PD-98059 (Fig. 5). Serum starvation was sufficient to inactivate endogenous Erk kinases (Fig. 5, lane 2). Unlike reports of transient activation of Erk in fibroblasts and PC-12 cells by EGF (15, 32, 39), Erk kinases in AGS cells remained phosphorylated for an extended period of time after EGF treatment. Apparently, in some cells, Erk activation is prolonged if growth factors stimulate the cells while they are adherent (46).

View larger version (47K): [in this window] [in a new window]   Fig. 5.   Immunoblot of endogenous Erk 2 in AGS cells. Whole cell extracts prepared from AGS cells were pretreated with PD-98059 and then treated for 15 min with 10 nM EGF (lane 1). Lane 2, extracts from 48-h serum-starved cells; lanes 3, 4, 5, 6, 7, 8, and 9, extracts from cells treated with EGF for 15 and 30 min and 1, 2, 3, 4, and 6 h. Same blot was probed first with a 1:250 dilution of phospho-Erk (Erk-PO4) antibody (top) and then with a 1:250 dilution of Erk 2 antibody (bottom).

Endogenous Sp1 kinase activity is downstream of Mek 1. To determine whether Sp1 is a target of EGF-activated kinases, a series of solid-phase kinase assays (20) were performed with extracts from EGF-treated AGS cells. Solid-phase kinase assays with whole cell lysates demonstrated that endogenous kinases activated by EGF phosphorylate the full-length form of Sp1 (Fig. 6A) and the NH₂-terminal truncated form containing amino acids 1-530 (NH₂-ter_1-530 Sp1) without the COOH-terminal zinc finger domain (Fig. 6B). Recombinant Erk 2 enzyme was used as a control for Sp1 phosphorylation in this assay system. Activation of Erk kinases may stimulate downstream kinases (6) or directly phosphorylate transcription factors, e.g., Elk and Sap (21). The other phosphoproteins migrating at ~200, 60, and 30 kDa represented autophosphorylation of endogenous AGS proteins present in the crude extract and not phosphorylation of residual bacterial proteins, since these phosphoproteins were not present in Fig. 6B, lane 11, which also contained bacterially expressed recombinant Sp1 without cell extract. The gel was stained with Coomassie blue to demonstrate equivalent loading of Sp1 in each lane. Sp1 phosphorylation was observed with the full-length and truncated forms, indicating EGF-induced phosphorylation of the NH₂ terminus. Within 30 min after EGF induction, a burst of kinase activity hyperphosphorylated Sp1 up to fourfold over basal levels, with prolonged phosphorylation averaging 2- to 2.5-fold over the 6-h time course (Fig. 6C). This follows a time course that parallels Sp1 binding (Fig. 2) but was somewhat delayed compared with the time course of endogenous Erk kinase activation (Fig. 5). Moreover, Sp1 was phosphorylated even in serum-free conditions when Erk kinases were minimally active.

View larger version (32K): [in this window] [in a new window]   Fig. 6.   EGF stimulates Sp1 kinase activity. Full-length Sp1 (A) or NH2-terminal acids 1-530 (NH2-ter1-530) form (B) was immobilized on glutathione-agarose beads, incubated with 200 µg of whole cell extracts from EGF-treated AGS cells, and then incubated with 5 µCi of [-32P]ATP for 30 min at 30°C. Phosphorylated Sp1 (Sp1-PO4) was resolved on a 10% SDS-polyacrylamide gel. Lane 1, cells incubated for 48 h in serum-free conditions; lanes 2, 3, 4, 5, 6, 7, 8, 9, and 10, cells treated for 5, 15, and 30 min and 1, 2, 3, 4, 5, and 6 h. Alternatively, 0.1 µg of active recombinant Erk enzyme (rErk 2) was incubated with glutathione S-transferase (GST)-Sp1 immobilized on glutathione-agarose beads (lane 11). Top: autoradiogram; bottom: same gel stained with Coomassie blue dye (rSp1). C: specific activity of phosphorylated Sp1 (in B) quantified on a PhosphorImager and plotted as a function of time for 3 independent experiments.

View larger version (28K): [in this window] [in a new window]   Fig. 7.   EGF-induced Sp1 phosphorylation inhibited by PD-98059. Solid-phase kinase assays were performed on extracts from cells pretreated with PD-98059 and then treated with 10 nM EGF for 1 h. A: lanes 1-4 contain recombinant Sp1 incubated with whole cell extracts, and lanes 2 and 4 contain extracts pretreated with PD-98059 (*). Top: autoradiogram of Coomassie blue-stained gel (bottom). B: phosphate-labeled Sp1 was quantified on a PhosphorImager. Values (means ± SE for 4 independent experiments) represent percent inhibition by PD-98059 at 1 h. ** P < 0.001. C: in vivo radiolabeled Sp1 immunoprecipitated from untreated cells (lane 1), cells treated with PD-98059 alone (lane 2), cells treated with EGF alone for 30 min (lane 3), and cells treated with EGF after pretreatment with PD-98059 for 1 h (lane 4). Top: autoradiogram of SDS-polyacrylamide gel; bottom: Coomassie blue-stained gel. D: amount of Pi incorporated into immunoprecipitated Sp1 quantified on a PhosphorImager. Results are representative of 2 experiments.

View larger version (56K): [in this window] [in a new window]   Fig. 8.   Sp1 kinase activity distinct from Erk. A: whole cell extracts were prepared from AGS cells treated without or with 10 nM EGF for 30 min. Extracts were fractionated on a MonoQ anion-exchange column and eluted in 1-ml fractions with 180-360 mM NaCl. Aliquots were evaluated by immunoblot for Erk activation. Anti-Erk 2 antibody detects primarily Erk 2 protein; anti-PO4-Erk 1,2 antibody detects only phospho-Erk proteins. B: autoradiograms of solid-phase kinase assays performed with eluted fractions with glutathione S-transferase (GST)-Sp1 as substrate (Sp1-PO4).

View larger version (65K): [in this window] [in a new window]   Fig. 9.   Sp1 kinase activity is novel. A: whole cell extracts were prepared from AGS cells treated without or with 10 nM EGF for 30 min. Extracts were fractionated on a MonoQ column, eluted in 1-ml fractions with 300-540 mM NaCl, and used in solid-phase kinase reactions with GST-Sp1 as substrate (lanes 1-13). Lane 14, 0.1 µg of recombinant Erk 2 incubated with GST-Sp1. Bottom: corresponding Coomassie blue-stained gel detecting recombinant Sp1 (rSp1). B: quantification of solid-phase kinase assay from 300 to 540 mM NaCl with GST or GST-Sp1 as substrate. C: whole cell extracts were prepared from AGS cells treated without or with 10 nM EGF for 30 min. Extracts were fractionated on a MonoQ column, eluted in 1-ml fractions with 300-540 mM NaCl, and used in solid-phase kinase reactions with beads but without immobilized recombinant Sp1 as substrate (lanes 1-13). D: full-length Sp1 (lanes 1-5) or NH2-ter1-530 Sp1 (lanes 6-10) was used in a solid-phase kinase assay with peak MonoQ fraction isolated from AGS cells before or after treatment with 10 nM EGF for 10 min and with or without PD-98059 pretreatment for 1 h. Lanes 1 and 6, no treatment; lanes 2 and 7, treatment with EGF for 10 min; lanes 3 and 8, treatment with EGF for 10 min and pretreatment with PD-98059; lanes 4 and 9, treatment with PD-98059 alone; lanes 5 and 10, phosphorylation of either form of Sp1 with Erk 2 enzyme.

View larger version (30K): [in this window] [in a new window]   Fig. 10.   EGF stimulates phosphorylation of Sp1 on Ser and Thr in vivo. Immunoprecipitated [32P]orthophosphate-labeled endogenous Sp1 was resolved on 10% SDS-polyacrylamide gel and then transferred to polyvinylidene difluoride membrane. Sp1 band was excised and then submitted for phosphoamino acid analysis. Autoradiography was performed after 2-dimensional electrophoresis. A: phosphoamino acid analysis of immunoprecipitated Sp1 from unstimulated cells. B: analysis after EGF stimulation. C: radioactive spots containing phosphoserine and phosphothreonine were analyzed on a PhosphorImager.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

EGF-dependent phosphorylation of Sp1 was studied by partially purifying distinct Sp1 kinase activity from a human gastric cell line. We showed recently that EGF stimulates the gastrin promoter through Mek 1-dependent and -independent pathways (34). In addition, overexpression of Ras and Erk proteins stimulates the gastrin promoter through an Sp1 binding site. The role of Ras activation in the regulation of the gastrin promoter is significant, since some colon and pancreatic tumors overexpress gastrin, which correlates with activated ras mutations. Gastrinomas are malignant pancreatic islet tumors in which gastrin is overproduced (52). The frequency of ras mutations in this tumor subtype is unknown; however, 80% of nonislet pancreatic cancers have K-ras mutations (2). Many of these gastrinomas are the result of a mutation in the tumor suppresser gene menin (9), which encodes a corepressor that binds to JunD, a component of activator protein (AP)-1 (1). It is well known that ras-Erk signaling targets the Fos component of the activator protein type 1 (AP-1) complex, and Fos is required for EGF-dependent activation of the gastrin gene (31). Collectively, these results imply that aberrant regulation of AP-1 in gastrointestinal tumors modulates gastrin gene expression. Moreover, K-ras mutations are found frequently in colon cancers (17) and have recently been correlated with elevated gastrin gene expression in these tumors (41).

The studies reported here show that gastrin promoter induction by EGF correlates with an increase in Sp1 binding to gERE, an event that is partially Mek 1 dependent. Prior studies with purified Erk indicate that, like other mitogen-inducible transcription factors, e.g., Elk and Sap (18, 53), Erk is a candidate kinase directly mediating phosphorylation of Sp1. Indeed, Sp1 contains several consensus Erk phosphorylation motifs (x-x-S/T-P) (13); there is precedence for Erk phosphorylation of nuclear proteins, and purified Erk phosphorylates Sp1 in vitro (this report) (34). However, Erk kinase activity did not partition on an anion-exchange column with the fractions exhibiting maximal Sp1 kinase activity. This result may reflect the fact that phosphorylation of Sp1 by recombinant Erk 2 depends on the level of enzyme activity and concentration. For example, Erk 1 and Erk 2 can phosphorylate c-Jun, but not as effectively as Jun (NH₂-terminal) kinase (JNK) activity (20). Regulation of the gastrin promoter by overexpression of Erk 1 or Erk 2 as expression vectors represents promoter activation in the presence of supraphysiological levels (34). Indeed, supraphysiological levels of Erk may not reflect events in vivo (22). However, expression of kinase-deficient Erk 1 and Erk 2 prevented gastrin promoter activation, presumably by abolishing endogenous Erk activity (34). The Sp1 kinase activity and Erk 2 are Ser/Thr kinases and are Mek 1 dependent. Collectively, these studies suggest that the Sp1 kinase activity lies downstream of Erk 2, although other protein targets of Erk that bind to upstream elements cannot be ruled out (Fig. 11). The Rsk 1-3 kinases are known to reside downstream of Erk and phosphorylate transcription factors (6, 54). However, we found that Rsk 1-3 kinases, p38, JNK, DNA-dependent kinase, as well as Erk 1 and Erk 2, also do not copurify with the Sp1 kinase activity (data not shown). Another Erk-independent kinase has been reported that mediates the feedback regulation of the guanyl nucleotide exchange protein SOS (22). This novel but unidentified kinase is activated by insulin, is also Mek 1 dependent, and regulates the dissociation of the Grb2-SOS complex. Therefore, precedence exists for novel Mek 1-dependent subsets of growth-regulated kinases.

View larger version (12K): [in this window] [in a new window]   Fig. 11.   Proposed role of Sp1 kinase activity in regulation of gastrin gene expression. Ras and Erk expression stimulate gastrin promoter activity. Inhibition of mitogen-activated protein kinase kinase (Mek) 1 blocks Sp1 kinase activity, demonstrating that this activity lies downstream of Mek. Erk 2 phosphorylates Sp1 and possibly targets elsewhere in promoter; however, Sp1 kinase activity exists primarily in fractions devoid of Erk 2. Therefore, primary "Sp1 kinase" activity is Mek 1 dependent and likely Erk dependent. Mek 1-dependent, Erk-independent Sp1 kinase activity has not been excluded.

During preparation of this report, Black et al. (5) reported that Sp1 is phosphorylated in vivo during the G₀-to-G₁ transition in serum-activated fibroblasts. The growth-regulated Sp1 kinase is distinct from previously identified kinases. Phosphorylation occurred exclusively on Ser in the zinc finger domain between amino acids 612 and 678, which resulted in a decrease in Sp1 binding to a GC-box within the dihydrofolate promoter. Phosphorylation within the DNA-binding domain may interfere with binding because of steric hindrance or charge repulsion (23). Therefore, the observed decrease in Sp1 binding in response to a growth signal was difficult to reconcile with the Sp1-dependent stimulation of the dihydrofolate promoter. The authors suggested that Sp1 may cooperate with other coregulators to mediate overall transcriptional activation, despite a decrease in Sp1 binding. Although it was not evaluated in their study, another possibility is that increased phosphorylation directly regulates the level of transcriptional activity. However, the absence of phosphorylation in the Ser/Thr-rich transactivation domain makes this function less likely.

Another unusual feature of the Sp1 kinase activity reported by Black et al. (5) is the significant delay in Sp1 phosphorylation (maximal at 6 h) compared with increased Sp1 abundance, which was maximal within 2 h. An alternative explanation is that the late burst in phosphorylation of the zinc finger domain represents a mechanism to decrease binding and shut off transcription of the dihydrofolate promoter at the end of the G₁ phase. Phosphorylation of Sp1 has been shown to activate proteosomal and nonproteosomal degradation of Sp1, which inhibits transcription in some instances (19, 30, 38, 44). The Sp1 kinase activity reported here and by Black et al. may also represent two different enzymes: a kinase activated early that targets the Sp1 transactivation domain to activate transcription and a kinase activated late that targets the DNA-binding domain to inhibit transcription. The Sp1 kinase activity described in the present study correlates with transcriptional activation. We examined early time points that revealed a rapid increase in phosphorylation that was maximal within 30 min and was sustained for 6 h. The increase in Sp1 abundance was not significant until 30 min after treatment and was not maximal until ~4 h, indicating some delay in altering Sp1 abundance compared with modulating the level of phosphorylation. Moreover, the two kinase activities reported by each group may be distinct from each other simply because of differences in cell types, ligands, and promoters.

Other investigators have shown that kinase activity affects Sp1 binding and, presumably, transactivation. Chun et al. (10) showed that human immunodeficiency virus type 1 TAT protein forms a complex with Sp1 that augments DNA-dependent protein kinase phosphorylation of Sp1. Ser-131 within the NH₂-terminal domain appears to be a target of the kinase and required for TAT-dependent transactivation. In regenerating liver, casein kinase 2 activity phosphorylates the zinc finger domain of Sp1 (Thr-579) and decreases Sp1 binding (3). In glucose-responsive promoters, glucose-inducible Sp1 binding is prevented by casein kinase 2 phosphorylation (55). Apparently, carbohydrate dynamically regulates the level of O-glycosylation on amino acid hydroxyl groups, which in turn are targets for phosphorylating enzymes. Dephosphorylation of DNA binding domains and subsequent O-glycosylation stimulates binding and transcription and prevents Sp1 proteolysis (19, 25). In contrast, Sp1 transactivation and binding appear to be enhanced if phosphorylation occurs outside the zinc finger domain. Increased protein kinase A activity in HL-60 cells stimulates Sp1 phosphorylation upstream of the zinc finger domain and subsequent binding, possibly through the protein kinase A consensus site at Thr-366 (47). In contrast, protein kinase C translocates to the nucleus in the absence of the von Hippel-Lindau tumor suppressor gene product, phosphorylating Sp1 in the zinc finger domain and stimulating the vascular EGF promoter (42). Collectively, these studies demonstrate that Sp1 is regulated in a dynamic fashion by phosphorylation at several different protein domains by a wide range of protein kinases.

The initial report on Sp1 phosphorylation demonstrated that the modification was mediated by a DNA-dependent kinase that requires Sp1 to be prebound to DNA (24). Growth-regulated kinase phosphorylation of Sp1 occurs in the presence of ethidium bromide used to disrupt the protein-DNA interaction (5). Moreover, the signal transduction pathway(s) regulating DNA-dependent kinase activity remains obscure (10). Thus DNA-dependent kinase does not appear to be the major kinase regulating receptor-mediated induction of Sp1 phosphorylation.

In summary, EGF regulation of the gastrin promoter is mediated in part through activation of ras and Mek 1 (Fig. 11). A novel Sp1 kinase activated by EGF lies downstream of Mek 1 and, possibly, Erk. There are >100 putative Ser and Thr phosphorylation sites within the NH₂-terminal domain of Sp1. These sites will be evaluated once the Sp1 kinase(s) is cloned to determine which of the potential Sp1 phosphorylation motifs are bona fide targets of this kinase and to determine whether they contribute to Sp1-dependent transcription.