MAP Kinase Phosphatase-1 Gene Transcription in Rat Neuroendocrine Cells Is Modulated by a Calcium-sensitive Block to Elongation in the First Exon*

Transcriptional elongation of many eukaryotic, pro-karyotic, and viral genes is tightly controlled, which contributes to gene regulation. Here we describe this phenomenon for the MAP kinase phosphatase 1 (MKP-1) immediate early gene. In rat GH4C1 pituitary cells, MKP-1 mRNA is rapidly and transiently induced by the thyrotropin-releasing hormone (TRH) and the epidermal growth factor EGF via transcriptional activation of the gene. Ca 2 (cid:1) signals are necessary for the induction of MKP-1 in response to TRH but not to EGF. Reporter gene analysis with the newly cloned rat promoter sequence shows only limited induction in response to various stimuli, including TRH or EGF. By nuclear run-on assays we demonstrate that in basal conditions, a strong block to elongation in the first exon regulates the MKP-1 gene and that stimulation with either TRH or EGF over-comes the block. Ca 2 (cid:1) signals are important to release the MKP-1 elongation block in a manner similar to the c- fos oncogene. These results suggest that a common mechanism of intragenic regulation may be conserved between MKP-1 and c- fos in mammalian cells. Long term Following t for EGF/EGTA (E/EE). Quantitation

Long term cellular processes such as proliferation, differentiation, and neuronal plasticity are controlled by extracellular stimuli and require the synthesis of new gene products. Following stimulation, expression of immediate early genes (IEGs) precedes the expression of late response genes, the latter encoding for proteins implicated in specific functions. The best known IEG products are transcription factors such as c-Fos, c-Jun, and c-Myc, which control the expression of late response genes (1,2). Not all IEGs encode for transcription factors. For instance, structural proteins like actin or tropomyosin, cytokines, and other regulatory proteins show a rapid and transient induction by growth factors (1).
Recently, a group of dual specificity phosphatases have been identified as being IEG products induced by various stimuli (growth factors, stress, neurotransmittors, etc.; reviewed in Ref. 3). These dual specificity (threonine/tyrosine) phosphatases have been named MAP kinase phosphatases (DSPs or MKPs), since they are effective in the inactivation of MAP kinases by dual dephosphorylation (4). MAP kinase phosphatase-1 (MKP-1/CL100/3CH134) is one example of this group of nuclear enzymes encoded by an IEG. Although MKP-1 gene transcription is activated by multiple signals, such as mitogens (5,6), cytokines (7), oxidative stress (8), heat shock (8), or hypoxia (9), the precise mode of gene regulation of this immediate early gene by such stimuli remains unclear. A comparison of the 5Ј-flanking sequence of the murine and the human genes revealed two conserved Ca 2ϩ /cAMP-responsive elements (CREs) and one E box motif in the promoter region of MKP-1. Recently, the upstream stimulatory factor, a member of the basic/helix-loop-helix/leucine zipper family has been shown to bind to the E box motif and transactivate MKP-1 expression in synergy with protein kinase A (10). Since multiple intracellular signals can target MKP-1 gene expression, it is likely that regulatory elements other than the CREs and the E box motifs may influence its transcription. Alternatively, additional regulation at the level of transcriptional elongation, termination, and/or mRNA stability may be important to control MKP-1 gene expression.
Regulation of gene expression at the level of transcriptional elongation is well established in prokaryotes and in an increasing number of eukaryotic genes (11,12). Earlier studies in mammalian cells have reported that the IEGs c-fos, c-myc, hsp70, and tumor necrosis factor-␣ display a strong block to transcriptional elongation in the promoter-proximal region (13)(14)(15)(16). Characterization of the block in vivo, by either nuclear run-on or KMnO 4 footprinting, showed that promoter-proximal pause sites in c-fos, c-myc, and hsp70 are important for controlling RNA polymerase II processivity (15,(17)(18)(19). In the case of c-fos, an additional transcriptional pause site is located in the first intron of this oncogene that is sensitive to Ca 2ϩ stimulation (17, 20 -23). The precise mechanism of intronic regulation of c-fos by Ca 2ϩ signals is not known.
Here we show a rapid and massive increase in MKP-1 mRNA triggered by either thyrotropin-releasing hormone (TRH) or epidermal growth factor (EGF) in GH4C1 neuroendocrine cells. Ca 2ϩ signals are necessary for the induction of MKP-1 gene expression in response to TRH but not to EGF. After cloning the rat MKP-1 genomic fragment, we demonstrate by a reporter gene assay that activity of the MKP-1 promoter alone cannot explain induction of MKP-1 expression in GH4C1 cells. Using nuclear run-on experiments, we show that control of MKP-1 transcription involves a block to elongation in the first exon. This block represents a decisive element in the regulation of MKP-1 gene expression.

EXPERIMENTAL PROCEDURES
Materials-The thyrotropin-releasing hormone TRH (Roche Molecular Biochemicals) and the epidermal growth factor EGF (Sigma) were diluted in H 2 O at 27.6 mM and stored in aliquots at Ϫ80°C. EGTA (Fluka Chemie) was diluted in H 2 O and stored at room temperature as 100 mM stock solution (pH 7.5). Actinomycin D (Sigma) and cycloheximide (Sigma) were diluted respectively in EtOH and Me 2 SO at 5 and 10 mg/ml and stored at 4°C. CPT-cAMP (Roche Molecular Biochemicals) was diluted in H 2 O, stored in aliquots at Ϫ20°C at a concentration of 10 mM. KCl (Fluka Chemie) was diluted in H 2 O and stored 4°C at a concentration of 3 M.
Cell Culture and Stimulation-GH4C1 pituitary cells were maintained in Ham's F-10 medium (Life Technologies, Inc.) supplemented with 2.5% fetal bovine serum and 15% horse serum at 37°C in a humidified atmosphere with 5% CO 2 . Confluent GH4C1 cells were incubated in Ham's F-10 serum-free medium (SFM) containing 5 g/ liter transferrin for 24 h and then stimulated for the indicated time with either 100 nM TRH or 10 nM EGF. When indicated, 0.6 mM EGTA was added to the medium 5 min before TRH or EGF stimulation to chelate free extracellular calcium ([Ca 2ϩ ] e Ͻ0.1 M). Actinomycin D (5 g/ml) and cycloheximide (10 g/ml) were added in SFM 30 min before stimulation with either TRH or EGF; total RNA was isolated 30 min after TRH and EGF stimulation.
RNA Preparation, Northern Blot Analysis, and RNase Protection Assay-Total RNA was extracted from cells with an acid phenol-guanidinium reagent (TRI-Reagent, Molecular Research Center, Inc.) according to the manufacturer's instructions. RNA samples (10 g) were denaturated by incubation in glyoxal, subjected to electrophoresis in 1.2% agarose gels, and transferred to Nylon-N ϩ membranes (Amersham Pharmacia Biotech) by capillarity. mRNAs were detected by Northern blot hybridization in a Church buffer with [␣-32 P]CTP-labeled cDNA probes obtained by random priming. The probes used correspond to the SacII (position 40)-MunI (position 789) fragment of the rat MKP-1 cDNA (GenBank TM accession number X84004) subcloned in the plasmid pBSK(Ϫ) and to a fragment of 1.22 kilobase pairs of the mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA subcloned in pBSK(Ϫ).
For the RNase protection assay, 20 g of total RNA were hybridized overnight in hybridization buffer (Ambion, Inc.) with 1.5 ϫ 10 5 cpm of a riboprobe. After digestion with RNase A and T1, samples were treated with proteinase K, phenol-extracted, ethanol-precipitated, and analyzed on a 6% denaturating polyacrylamide urea sequencing gel. The riboprobe was prepared from the rat MKP-1 cDNA (positions 15-301) subcloned in pGEM-T-easy (Promega). Transcription with T7 of the SalI-linearized plasmid yielded a probe of 395 nucleotides (nt), and a 286-nt protected fragment. As an internal control for RNA loading, a GAPDH probe was included in the hybridation mixture. The GAPDH plasmid (24), digested with HinfI, produced a riboprobe of 185 nt and a 154-nt protected band after RNase digestion. Autoradiographic signals detected in Northern and RNase protection assays for MKP-1 were quantified in arbitrary units using a PhosphorImager (Molecular Dynamics, Inc.) and then normalized to the corresponding GAPDH values to correct variations in RNA loading.
Detection of the transcriptional initiation site for the rat MKP-1 gene was performed with a riboprobe prepared from the rat MKP-1 gene (positions 607-807; GenBank TM accession number AF357203) subcloned in pGEM-T-easy (Promega). Transcription with SP6 of the BamHI linearized plasmid results in a riboprobe of 287 bp. A sequencing ladder, corresponding to the antisense sequence of the probe starting at 807, was resolved in parallel with the protected fragment. The start site was determined relative to the probe sequence.
TaqMan RT-PCR-Quantification of the MKP-1 mRNA was performed by TaqMan RT-PCR (PerkinElmer Life Sciences) from the total RNAs extracted from 10 series of GH4C1 cells cultured in nonstimulated or stimulated conditions. Each RNA sample (diluted to 10 ng/l) was analyzed six times independently (replicas), and a standard curve was included in each plate. The standard curve was prepared from the total RNA of a TRH-treated sample with a high level of expression of MKP-1, diluted from 100 ng/l to 1 pg/l. Each dilution of the standard sample was amplified in triplicate. 18 S rRNA was simultaneously amplified in each tube for normalization.
Briefly, 5 l of each total RNA (corresponding to 50 ng of the unknown samples), primed with random hexamers, were retrotranscribed into cDNA in a final volume of 30 l using the TaqMan Gold RT-PCR Kit of PerkinElmer Life Sciences (PE N8080234), following the manufacturer's instructions. MKP-1 and 18 S rRNA were simultaneously amplified from each sample using 4 l of the previous cDNA, 1ϫ PerkinElmer's Universal PCR Master Mix (PE 4304437), 0.3ϫ 18 S rRNA Predeveloped Assay Reagent (PE 4310893E), 200 nM MKP-1 forward primer 5Ј-CGCGCTCCACTCAAGTCTTC-3Ј, 200 nM MKP-1 reverse primer 5Ј-GGTGGACTGTTTGCTGCACA-3Ј, and 250 nM MKP-1 TaqMan probe 5Ј-FAM (6-carboxyfluorescein)-AGCCGAAAACGCTTC-ATATCCTCCTTGGTAMRA (6-carboxytetramethylrhodamine)-3Ј. The primers and probe for the quantification of MKP-1 were designed using Primer Express 1.0 software from PerkinElmer Life Sciences. These primers amplified an 87-bp fragment of the rat's MKP-1 gene comprising the exon 1-2 boundary, to avoid amplification of genomic DNA. The amplification and quantification of MKP-1 and 18 S rRNA were performed using the ABI PRISM 7700 Sequence Detection System of PerkinElmer Life Sciences, using standard conditions. After fixing the base line in each experiment and the threshold for each amplification, two different values of threshold cycle were obtained for each replica: one for the amplification of MKP-1 and another for the amplification of 18 S rRNA. The relative quantities of MKP-1 and 18 S rRNA for each replica were obtained by interpolating each threshold cycle value in the corresponding standard curve. MKP-1 was then normalized to 18 S rRNA, and the mean relative quantity and S.D. for each set of replicas in each series were obtained. The mean relative quantity and S.D. of the 10 series were calculated, and values were expressed relative to basal levels.
Cloning of the Rat Gene-The rat MKP-1 gene was amplified by PCR from rat genomic DNA of GH4C1 cells using primers directed toward two conserved regions in the promoter and exon 4 of the mouse and the human gene: 5Ј-TCT TGC AAC CCT CCT CCC TTT G-3Ј (sense) and 5Ј-GTG GAA CTC GGG AGG TGT TTG-3Ј (antisense). The resulting 2892-bp PCR fragment was cloned, sequenced, and aligned to the mouse and human MKP-1 gene. The sequence of the rat MKP-1 gene in GH4C1 was compared and corrected according to sequences amplified in rat pancreatic ␤ INS-1 and rat vascular smooth muscle cells. The cloned GH4C1 sequence is available in the GenBank TM data base with the accession number AF357203.
Southern blotting was performed with GH4C1 genomic DNA digested with either StuI or PvuII restriction enzymes known to cut in the exon 4. 20 g of each digested DNA and the size marker Smart Ladder (Genetech) were subjected to electrophoresis in a 0.7% agarose gel and transferred to Nylon-N ϩ membranes (Amersham Pharmacia Biotech) by capillarity. The membranes were hybridized in Church buffer with different random labeled [␣-32 P]CTP DNA probes amplified by PCR and corresponding to regions of the promoter, exon 1, or the introns in the 2892-bp PCR fragment: probe 1 (Ϫ631 to ϩ2), 2 (ϩ25 to ϩ464), 3 (ϩ464 to ϩ824), 4 (ϩ840 to ϩ1469), and 5 (ϩ1454 to ϩ1903).
Reporter Gene Assays-GH4C1 cells maintained in culture medium were detached from Petri dishes with trypsin, transiently transfected using FuGENE 6 transfection reagent (Roche Molecular Biochemicals), and seeded in 24-well multidishes (Falcon) at a density of 6 ϫ 10 4 cells/well. After 1 day, the medium was replaced by SFM 24 h before exposure to the various stimuli. Cells were stimulated with the addition of KCl, CPT-cAMP, TRH, and EGF prediluted into SFM. Stimulation was performed for 3 h at 37°C and stopped by removal of the medium. Firefly luciferase activity was determined according to Promega's Luciferase Reporter assay system instructions. Cells were co-transfected with Renilla luciferase expression plasmid to verify uniformity of transfection from experiment to experiment.
Plasmid pMKP-1 was generated by cloning the region of the rat MKP-1 gene extending from Ϫ631 to ϩ19 between the KpnI and NheI restriction sites of the pGL3 enhancer vector (Promega). Plasmid pSV40 was obtained by subcloning the SV40 promoter from the pGL3-control vector (ϩ49 to ϩ244) between the KpnI and NheI restriction sites of the pGL3 enhancer vector. The "c-fos/intron" construction was previously described elsewhere (25,26).
Nuclear Run-on Transcription Assays-GH4C1 cells in nonstimulated or stimulated conditions for 25 min, as indicated, were washed twice with phosphate-buffered saline (PBS) and detached with a rubber policeman in lysis buffer (10 mM Tris⅐Cl, pH 7.4, 10 mM NaCl, 3 mM MgCl 2 , 0.5% Nonidet P-40). Nuclei were isolated by centrifugation for 10 min at 300 ϫ g and 4°C, resuspended in lysis buffer, and purified by centrifugation through a 5-ml cushion of 30% (w/w) sucrose in lysis buffer. Isolated nuclei were resuspended in 100 l of storage buffer (50 mM Tris⅐HCl, pH 8.3, 40% glycerol, 5 mM MgCl 2 , 0.1 mM EDTA), quantified to obtain 5 ϫ 10 7 transcription-competent nuclei per aliquot and stored at Ϫ80°C until use. Nuclear run-on was carried out as described by Collart et al. (27). After in vitro elongation, the same amount of labeled RNA from each reaction was hybridized to the pBSK(Ϫ) plasmid containing different DNA inserts from the rat MKP-1 locus: A (position ϩ25 to SacII (ϩ464)), B (SacII (ϩ464) to SpeI (ϩ824)), C (ϩ840 to ϩ1469), D (ϩ1454 to ϩ1903), E (StuI (ϩ1854) to ϩ2237), which had been immobilized on a N ϩ -nylon membrane. Mouse c-fos oncogene fragments F (ϩ53 to ϩ764) and G (ϩ418 to ϩ1130), cloned in pBSK(Ϫ), were also blotted on the same membrane. pBSK(Ϫ) and GAPDH cDNA (ϩ1 to ϩ1220), subcloned in pBSK(Ϫ), were spotted and used, respectively, as a control for hybridization and an invariant internal control. Blots were developed by autoradiography and quantified with a Molecular Dynamics PhosphorImager.

Rapid Induction of MKP-1 mRNA by TRH and EGF in Pituitary GH4C1 Cells Depends on Transcriptional Activation-
TRH elicits both acute and long term responses in rat pituitary clonal growth hormone (GH) cell types, leading to enhanced secretion of both prolactin and GH (28,29). Activation of prolactin gene transcription and long term hormone secretion are preceded by the synthesis of the immediate early gene products, c-fos, junB, and c-jun, postulated to be involved in the regulation of the prolactin gene. TRH stimulates the transient induction of these immediate early genes via Ca 2ϩ -dependent mechanisms (30 -32).

FIG. 1. Rapid induction of MKP-1 mRNA by TRH and EGF in pituitary GH4C1 cells depends on transcriptional activation.
A, serum-depleted (24 h) GH4C1 cells were stimulated with TRH (100 nM) or EGF (10 nM) for the indicated times. 10 g of total RNA were analyzed by Northern blotting using [␣-32 P]CTP-labeled cDNA probes for MKP-1 and GAPDH, as described under "Experimental Procedures." B, time course of MKP-1 mRNA induction by TRH and EGF; Phospho-rImager data obtained from Northern blotting analysis performed as in A were normalized for GAPDH and presented as -fold induction over basal levels. C, effect of actinomycin D (Act D; 5 g/ml) and cycloheximide (CHX; 10 g/ml) added 30 min prior to stimulation by TRH and EGF on MKP-1 mRNA levels; recorded 30 min after stimulation by Northern blot analysis, as in A.

FIG. 2. Induction by TRH of MKP-1 mRNA depends upon Ca 2؉
signaling; induction by EGF does not. After stimulation of GH4C1 cells with TRH and EGF under experimental conditions that do or do not allow Ca 2ϩ signaling, MKP-1 mRNA levels were quantitatively assessed as follows. A, RNase protection assay of total RNA from GH4C1 cells isolated after 30 min of stimulation with either TRH or EGF. Where indicated, 0.6 mM EGTA was added 5 min before stimulation. 20 g of total RNA were co-hybridized to two [␣-32 P]UTP RNA probes for MKP-1 and GAPDH, as described under "Experimental Procedures" and digested with RNase A and T1. Free probes, protected fragments, and the RNA Century TM size marker (Ambion) were resolved on a 6% polyacrylamide sequencing gel. B, quantification by PhosphorImager of the RNA levels detected by RNase protection in A (ϮS.E., n ϭ 6). C, example of quantitative RT-PCR amplification records obtained for MKP-1 (upper panel) and 18 S mRNA (lower panel) under basal and TRH-stimulated conditions. Multiplex quantitative PCR amplification (TaqMan; PerkinElmer Life Sciences) was performed using sequence-specific primer pairs and probes corresponding to the junction of exons 1-2 of MKP-1 and to the transcribed region of the ribosomal 18 S RNA as described under "Experimental Procedures." Threshold cycle values are indicated by circles. D, average (ϮS.E., n ϭ 10) MKP-1 mRNA levels derived from threshold cycle values obtained in sextuplicate and plotted standard curves for MKP-1 were normalized for 18 S mRNA levels and then expressed as -fold induction over basal values.
MKP-1, originally identified as an immediate early gene product, is induced by mitogens (5, 6), heat shock, or oxidative stress (8) and also by Ca 2ϩ ionophore in cultured cells (33). As we show here by Northern blot analysis with a rat MKP-1 cDNA probe, TRH also stimulates the expression of MKP-1 (Fig. 1). TRH induction is as strong and rapid as with the mitogenic stimulus EGF, which was used as a positive control for MKP-1 induction. GH4C1 cells exposed to 100 nM TRH showed maximally enhanced levels of MKP-1 mRNA as early as 30 min, followed by a rapid decline to lower but still significantly elevated levels at 60 min which were sustained up to 6 h. Slightly slower induction kinetics were observed for EGF, with MKP-1 mRNA levels peaking at 45 min (Fig. 1, A and B). After more than 6 h of treatment, the level of MKP-1 mRNA was higher for TRH-stimulated cells compared with EGFtreated cells, where the MKP-1 message returned almost to the basal level (Fig. 1B). GAPDH mRNA levels were not changed by either TRH or EGF stimulation and can thus be used as an invariant internal control. Based on the kinetics shown in Fig.  1, MKP-1 expression was assessed in all subsequent experiments at 30 min after TRH or EGF stimulation.
To show that TRH-or EGF-induced MKP-1 mRNA accumulation is linked to transcriptional activation of the gene, we analyzed the effect of the transcriptional inhibitor actinomycin, added 30 min prior to a further 30-min exposure to TRH or EGF. Actinomycin D (Act D) completely suppressed MKP-1 mRNA induction by TRH and EGF (Fig. 1C, lanes 6 and 9). In addition, the protein synthesis inhibitor cycloheximide (CHX) did not prevent TRH or EGF induction of MKP-1, confirming that MKP-1 is an immediate early gene. In contrast to what was observed for other immediate early genes in GH cells (31), cycloheximide, per se, did not induce MKP-1. Nor did cycloheximide potentiate TRH or EGF induction of MKP-1 mRNA (Fig.  1C, lanes 3, 4, 7, and 10). Instead, cycloheximide treatment lowered MKP-1 mRNA levels under stimulated conditions (compare lanes 5 and 8 with lanes 7 and 10, respectively).
Induction of MKP-1 mRNA by TRH Depends upon Ca 2ϩ Signaling, and Induction by EGF Does Not-In GH4C1 cells, the increase of [Ca 2ϩ ] i stimulated by TRH consists of two phases: an initial spike, due to intracellular Ca 2ϩ store mobilization, followed by a "plateau phase" of enhanced action potential firing, causing spikes of [Ca 2ϩ ] i due to Ca 2ϩ influx during action potentials (34,35). Previous reports have shown that the prolonged phase of [Ca 2ϩ ] i spiking was essential to enhance and maintain proto-oncogene mRNA levels over time (31,32,36). To examine the role of Ca 2ϩ influx in MKP-1 expression induced by TRH, GH4C1 cells were preincubated for 5 min in serum-free medium containing 0.6 mM EGTA (free [Ca 2ϩ ] e Ͻ0.1 M). This procedure has little effect on the initial spike of [Ca 2ϩ ] i but prevents Ca 2ϩ action potentials and the FIG. 3. Cloning of the promoter and the introns of the rat MKP-1 gene and identification of the transcriptional initiation site. A, the rat MKP-1 gene was amplified by PCR, cloned, and sequenced (GenBank TM accession number AF357203), as described under "Experimental Procedures." The cloned fragment of 2.9 kilobase pairs was used to prepare random labeled DNA probes (probes 1-5) for Southern analysis. GH4C1 genomic DNA fragments obtained by either StuI or PvuII digestion were separated on a 0.7% agarose gel, blotted onto a Nylon N ϩ membrane, and hybridized to random labeled DNA probes corresponding to the promoter, exon 1, or intron 1, 2, or 3 of rat MKP-1. Smart Ladder (Eurogentec) size markers are shown in the first lane. B, conservation of the MKP-1 promoter sequences between rat, mouse, and human. Shown is multiple sequence alignment with ClustalW of the rat MKP-1 (GenBank TM accession number AF357203) with the mouse MKP-1 (GenBank TM accession number S64851) (36) and the human MKP-1 (GenBank TM accession number U01669) (6). Identical nucleotides are shown with asterisks, and the conserved DNA binding motif CREs, E box, and the TATA box are underlined in the sequence. The transcriptional start sites are indicated by arrows; gray arrows indicate the sites found in rat MKP-1 (Fig. 3B), the open arrow indicates the potential start site of the mouse MKP-1 (36), and filled arrows indicate the human MKP-1 sites mapped by Kwak et al. (6) and Sommer et al. (10). C, the transcriptional start site of the rat MKP-1 was mapped by RNase protection (lane 2). Total RNA from TRH-stimulated GH4C1 cells was hybridized to the rat antisense MKP-1 probe corresponding to the genomic sequence from ϩ172 to Ϫ34. The hybrid was digested by RNase A and T1, and the protected fragment was resolved on a 6% polyacrylamide sequencing gel run in parallel with the reaction products obtained from the MKP-1 antisense DNA sequenced with a primer starting at position ϩ172. The 200-base RNA Century TM size marker (Ambion) is shown in lane 1.
corresponding Ca 2ϩ influx (Refs. 32 and 36 and data not shown). To establish unequivocally a role for Ca 2ϩ signaling, the induction of MKP-1 by TRH and EGF in the presence and absence of EGTA were quantitatively assessed by RNase protection assays and by quantitative RT-PCR.
As shown in Fig. 2A, the MKP-1 riboprobe (395 nucleotides) protects a 286-nucleotide (nt) fragment of the MKP-1 mRNA in basal and stimulated conditions. The level of MKP-1 induction was determined after normalization of the MKP-1 signal to the signal of the protected fragment of GAPDH mRNA at 154 nt (Fig. 2B). RNase protection assays with antisense 18 S RNA and GAPDH confirmed that GAPDH mRNA levels did not change following stimulation with either TRH or EGF in low extracellular Ca 2ϩ (data not shown). TRH induced a 10-fold elevation of MKP-1 mRNA, which was reduced to 5-fold by inhibiting Ca 2ϩ influx. EGF, which raised MKP-1 mRNA to the same levels as TRH, could still markedly stimulate MKP-1 mRNA accumulation (8-fold) in the absence of free extracellular Ca 2ϩ .
These results were confirmed by the analysis of the same RNA samples by quantitative RT-PCR using primer pairs and probes specific for MKP-1 mRNA and the ribosomal 18 S RNA (see "Experimental Procedures") (Fig. 2C). The MKP-1 mRNA level was determined after normalization to the level of 18 S RNA and then plotted as -fold induction over basal conditions (Fig. 2D). Both methods of RNA analysis show very similar relative result when comparing data in Fig. 2, B and D. Taken together, the results show that a rise in [Ca 2ϩ ] i is essential for MKP-1 induction by TRH but is not important for EGF; this is consistent with the fact that Ca 2ϩ is not involved in the signaling pathways triggered by the EGF receptor.
Cloning of the Rat MKP-1 Gene and Identification of the Transcriptional Initiation Site-To elucidate the mechanisms involved in Ca 2ϩ -dependent and Ca 2ϩ -independent transcriptional activation of the MKP-1 gene GH4C1 cells, we cloned and sequenced the rat MKP-1 gene. The mouse and the human MKP-1 gene had been cloned previously in other studies (6,37). PCR was performed with GH4C1 genomic DNA and primers situated in two regions that are conserved in the promoter and the exon 4 of the mouse and the human genes (see "Experimental Procedures"). The sequence of the resulting 2892-bp PCR fragment was 66% identical to the mouse MKP-1 gene. To verify that this fragment corresponds to the rat MKP-1 gene and does not contain any repetitive sequences, Southern blotting was performed with GH4C1 genomic DNA digested with StuI or PvuII restriction enzymes (Fig. 3A). DNA probes corresponding to the region of the promoter or the introns in the 2892-bp PCR fragment hybridize to the same band as the exon 1 probe and do not cross-react with other genomic DNA fragments. This indicates that the different probes are specific for a single gene, namely the rat MKP-1 sequence. Alignment of the rat, mouse, and human MKP-1 sequences shows that the critical promoter elements identified in the mouse and human promoters are conserved in the rat promoter (Fig. 3B).
To map the transcriptional start site in the rat gene, an antisense RNA probe overlapping the TATA box and the MKP-1 ORF was designed. RNase protection assays with total RNA from TRH-stimulated GH4C1 cells showed two protected fragments: a major band at 183 nt and a minor band at 185 nt. The antisense DNA sequence starting at the MKP-1 ORF run in parallel with the protected fragments allows us to determine the position of initiation of transcription in the MKP-1 sequence (Fig. 3C). The two transcriptional start sites identified in the rat MKP-1 gene are located at the same positions as those determined for the mouse and the human gene, at 20 and 22 residues, respectively, downstream of the consensus TATA sequence (Fig. 3B).
Limited Stimulation of Reporter Gene Expression Driven by the Rat MKP-1 Promoter-The search for regulatory elements involved in Ca 2ϩ -dependent and Ca 2ϩ -independent activation of MKP-1 transcription was started using a reporter gene approach. The promoter region Ϫ631 to ϩ19 of the rat MKP-1 gene was inserted into a luciferase reporter vector. GH4C1 cells were transiently transfected and subsequently stimulated with a variety of agonists. As shown in Fig. 4, luciferase expression obtained with this construct was already high under unstimulated basal conditions and only marginally (Ͻ2-fold) increased following stimulation of the GH4C1 cells by KCl plus cAMP, TRH, or EGF. The lack of induction was not due to saturation of the system, since we obtained similar results in cells transfected with lower plasmid concentrations, and a linear relation was observed between plasmid concentration and luciferase expression in the range of concentrations used routinely (data not shown).
As a further control, we compared the MKP-1 reporter construct with an SV40 promoter construct and with a "c-fos/ intron" construct that contains the sequence from the promoter to exon 2 fused to the ORF of the luciferase gene (25,26). These reporter constructs were compared in basal and stimulated conditions with KCl plus cAMP, TRH, or EGF (Table I). As reported previously (25), expression of luciferase from the c-fos construct is highly induced in all three stimulated conditions, whereas with the MKP-1 promoter, stimulation is limited to a 1.5-1.7-fold increase. Comparing the reporter data from the MKP-1 promoter with the SV40 promoter constructs suggests that the MKP-1 promoter behaves very much like a constitutive promoter in this cell line. This may indicate that the 650-bp fragment of the rat MKP-1 gene contains sufficient promoter elements to drive the expression of the reporter gene . Cell extracts were analyzed for firefly and Renilla luciferase activities using the corresponding cofactors. Reporter luciferase activity was normalized to constitutive Renilla luciferase expression in the same extracts and is shown as the average ratio between the RLUs obtained (ϮS.E., n ϭ 3). The relative -fold induction of this construct in different conditions of stimulation is showed in Table I. but lacks control elements that restrict basal expression and thereby permit strong activation both by Ca 2ϩ -dependent (KCl plus cAMP; TRH) and Ca 2ϩ -independent (EGF) stimuli. The discrepancy between the highly controlled endogenous expression of MKP-1 (Fig. 2) and the lack of control by the same stimuli with the reporter gene expression system (Fig. 4) suggests that important regulatory steps may require elements that are not situated in the promoter region, defined here between Ϫ631 and ϩ19.

MKP-1 Transcription Is Controlled by a Calcium-sensitive Elongation Block in the First
Exon-To analyze in more detail the transcriptional control of MKP-1 and determine if the gene is also regulated at steps different from initiation (e.g. elongation), we performed nuclear run-on experiments with nuclei isolated from GH4C1 cells after stimulation with TRH or EGF (Fig. 5). Nascent RNAs transcripts in nuclei isolated from stimulated and nonstimulated cells were labeled by in vitro elongation with [␣-32 P]UTP, purified, and hybridized to unlabeled DNA fragments spotted on a membrane. These fragments include sequences spread out along the MKP-1 gene (Fig. 5, A-E), two fragments (F and G) corresponding to exon 1 and intron 1 of the mouse c-fos oncogene, the GAPDH gene as a control of transcription, and pBSK as a negative control (see "Experimental Procedures"). As shown in Fig. 5, MKP-1 transcription is initiated under basal conditions (signal in A, lane 1) but does not proceed to the end of the gene. The polymerase is blocked at the exon 1-intron 1 junction. When cells are stimulated with TRH or EGF (lanes 2 and 4), the block to elongation is released, and MKP-1 gene transcription is completed (signal in E). To determine the effect of Ca 2ϩ signals on the block of elongation in MKP-1 gene transcription, nuclear run-on assays were performed with cells stimulated with TRH or EGF in low extracellular Ca 2ϩ . The addition of 0.6 mM of EGTA prior to TRH stimulation decreases transcription efficiency at the level of the first intron (probes B and D) more than in the first exon (probe A); the ratios B/A and D/A are significantly reduced by EGTA in TRH-stimulated conditions (Table II, bottom). This indicates that TRH-induced Ca 2ϩ signals are important to release the block of elongation. Stimulation with EGF in the presence or absence of EGTA both activate MKP-1 transcription at the level of initiation and elongation. A quantitative analysis of these results is shown in Table II. The signals for fragments A-E were corrected for U content and normalized to the signal of GAPDH (Table II, top). By comparing the signal intensities (as -fold induction over basal condition) observed in nuclear run-on assays, for the probes A, B, and D, with the level of induction of MKP-1 messenger measured by RNase protection assays or TaqMan (Fig. 2), we show that for any condition of stimulation, the level of induction for the probes B and D is similar to the one observed for the MKP-1 mRNA (Table II, bottom). The rise of initiation of transcription under stimulated conditions (monitored on signal A) is not sufficient to account for the level of induction of the MKP-1 mRNA. Additional activation of elongation is necessary to maximally stimulate MKP-1 gene expression. These results together with those reported in Fig. 4 suggest that MKP-1 transcription is controlled by a constitutively active promoter and that a major contribution to gene transcription is conferred by a strong block of elongation within the gene.
A similar result was obtained with the c-fos oncogene in nuclear run-on assays performed with GH4C1 nuclei (probes F and G, Fig. 5). The calcium dependent block to elongation in the intron 1 of c-fos was previously described in neuroendocrine cells (25) and other cell lines (20 -23).
To localize more precisely the block of elongation in the MKP-1 gene, we prepared restriction fragments of the exon 1-intron 1 region and used these probes to map the pause site by nuclear run-on assays. Digested products were separated by gel electrophoresis, transferred to a membrane, and hybridized with labeled RNAs isolated from nuclear run-on reactions performed with basal or TRH-stimulated cells. DNA fragment that does not hybridize in basal conditions but does hybridize in the stimulated conditions indicates that the polymerase was blocked before reaching the sequence corresponding to that When indicated, 0.6 mM EGTA was added 5 min prior to stimulation to chelate-free calcium in the medium. After isolation of the nuclei and nuclear run-on transcription, labeled RNAs were hybridized to DNA fragments spotted on a membrane. Fragments A-E correspond to the rat MKP-1 gene; fragments F and G correspond to the c-fos oncogene, pBSK, and GAPDH cDNA. Extension of the probes A-G is presented on the MKP-1 and c-fos genomic maps; a precise description is given under "Experimental Procedures." Shown is a typical experiment, repeated five times for basal, TRH, and TRH/EGTA conditions and three times for EGF and EGF/EGTA conditions. Quantitative analysis of these data is presented in Table II.

TABLE I
Quantification of MKP-1 promoter activity in GH4C1: comparison with the SV40 promoter and with the c-fos/intron constructs Luciferase activities measured on Fig. 4 are plotted as -fold induction over basal conditions for each construct (ϮS.E., n ϭ 4). In the last column, the basal luciferase activity of the three constructs is shown normalized to constitutive renilla luciferase expression in the same extracts (RLU luciferase/renilla Ϯ S.E., n ϭ 4). The same amount of both plasmids was transfected in all cases (see "Experimental Procedures"). The pMKP-1 construct contains the region Ϫ631 to ϩ19 of the rat MKP-1 gene; pSV40 contains the promoter SV40 of the pGL3-control vector (Promega); and the c-fos/intron construct comprises the region of the mouse c-fos oncogene from the promoter to exon 2 fused to the ORF of the luciferase reporter gene. The transcriptional activity of this construct in GH4C1 was previously described elsewhere (25 6A, lane 1) showed, after nuclear run-on in basal conditions, a hybridized fragment SacII-NheI but not SacII-HindIII (Fig.  6B, lane 1), indicating that the polymerase was blocked before reaching the intron 1. DNA digested by NcoI, NheI, and Hin-dIII (Fig. 6A, lane 2) showed, after nuclear run-on, that the three fragments NcoI-NheI, NcoI-NcoI, and NcoI-HindIII are all hybridized by nascent RNA in basal conditions (Fig. 6B,  lane 2). The weak signal corresponding to the NcoI-HindIII fragment in basal condition suggests that RNA polymerase II has transcribed part of the sequence between NcoI and SacII but was blocked before entering intron 1. Taken together, these results show that the block of elongation in MKP-1 is located in the exon 1 possibly in the first 300 -400 nt of the 5Ј-end of the gene. Further fine mapping by KMnO 4 footprinting (18) and nuclear run-on performed with antisense oligonucleotide probes (38) will permit us to determine if the block to transcriptional elongation correlates with promoter-proximal pauses sites and/or intrinsic sites of premature termination in the exon 1 of MKP-1.

DISCUSSION
This study identified a calcium-sensitive block to elongation within the first exon of the rat MKP-1 gene as an important element for transcription regulation. TRH strongly stimulates MKP-1 transcription by enhancing initiation and elongation by Ca 2ϩ -sensitive mechanisms. Stimulation by EGF resulted in a similar rise in MKP-1 transcription, based again on enhanced initiation and elongation but (in contrast to TRH) without the need for Ca 2ϩ signaling. The involvement of a calcium-sensitive elongation block in MKP-1 gene expression is reminiscent of c-fos expression for which a block to elongation is an essential element (Fig. 6 and Refs. 20 -23 and 25). This suggests that common mechanisms of intragenic regulation may be conserved between MKP-1 and c-fos.
Change in MKP-1 and c-fos gene expression level, via potential deregulation of the block to elongation, may have physiological importance in tumorigenesis. For c-fos, it is well known that this oncogene is up-regulated in several tumor cells (e.g. 39 -41). For MKP-1, several studies have shown an overexpression of both the MKP-1 mRNA and protein at different stages of breast and prostate carcinoma (42)(43)(44)(45). These latter reports have suggested that up-regulation of MKP-1 may contribute to tumor growth by inactivating preferentially stress-activated MAP kinases (stress-activated protein kinase/Jun N-terminal kinases, p38) that can promote apoptosis in tumor cells exposed to toxic stimuli.
The dual specificity MAP kinase phosphatases, DSPs or MKPs, inactivate MAP kinases by dual dephosphorylation. In turn, MKPs can be controlled in at least three ways: first by subtype-specific enhancement of their catalytic activity triggered by direct interaction with MAP kinases (46,47); second by MAP kinase phosphorylation (in the case of MKP-1), which stabilizes the otherwise very labile protein (48); and third by transcriptional activation of some MAP kinase phosphatases (e.g. MKP-1 and MKP-2) encoding for immediate early genes (4).
Transcription of the MKP-1 gene can be induced rapidly by multiple intracellular signals, including active mitogenic or stress-activated MAP kinase cascades, protein kinase C, cAMP, and Ca 2ϩ signals (5)(6)(7)(8)(9)(10)33). Prior to our study, the induction of the human MKP-1 gene in HeLa cells, by the mitogenic stimuli EGF, fibroblast growth factor, or platelet-derived growth factor, was thought to be essentially mediated by activation through the promoter of the gene (6).
In our quantitative study, we show a clear discrepancy between the levels of induction for the MKP-1 mRNA and reporter gene assays with the promoter of MKP-1. We observed an 18-fold stimulation of the MKP-1 mRNA by TRH and EGF (Fig. 2D), which contrasts with the 1.5-and 1.6-fold stimulation, respectively, of MKP-1 reporter gene expression from the promoter (Table I). This difference can have two possible explanations: (i) chromatin structure influences the basal activity of the gene and may tightly be regulated following stimulation, and (ii) further 5Ј-or 3Ј-end sequences not present in the promoter construct may be important to control MKP-1 gene transcription. We showed (Fig. 5) that a controlled release of a block to elongation in the exon 1 may explain part of this  Fig. 5 were quantified using a PhosphoImager (top). Values for basal, TRH, and TRH/EGTA conditions (n ϭ 5) and for EGF and EGF/EGTA conditions (n ϭ 3) were corrected for uridine (U) content for each fragment and normalized to the GAPDH signal (fixed here at 10 arbitrary units (a.u.)). Results are presented as mean levels Ϯ S.E. Hybridization signals from fragments A, B, and D were plotted as -fold induction over basal conditions for each probe (ϮS.E., n ϭ 5 or 3) (bottom). The ratios B/A and D/A are indicative of relative levels of enhancement of transcriptional elongation versus initiation in the MKP-1 gene for stimulated conditions. Student's t tests were used for statistical evaluation of the population TRH versus TRH/EGTA (T/TE) and EGF versus EGF/EGTA (E/EE). Differences were considered to be significant at p Յ 0.05. NS, not significant. Quantitation of the transcriptional activity within the c-fos locus (fragment F and G, Fig. 5 Table II); this result is again compatible with an important contribution of elongation control to the transcriptional activation. The transient transcriptional activation of MKP-1 by TRH and EGF results in a significant enhancement of MKP-1 mRNA at 30 min ( Fig. 1) but also at the protein level 1 h after stimulation (data not shown). It becomes thus evident that a tight control of MKP-1 gene transcription at the level of initiation and elongation, combined with its differential binding and catalytic activation by specific MAP kinases, provides a sophisticated mechanism for rapid and targeted inactivation of selected MAP kinase cascades.
Calcium is a critical modulator of immediate early gene expression at different levels in mammalian cells. In contrast to a number of studies that examine the role of Ca 2ϩ signaling by using Ca 2ϩ ionophores and other nonphysiologic stimuli, we describe here the role of physiological Ca 2ϩ signaling triggered by the releasing factor TRH in MKP-1 gene expression. TRH stimulates the release of Ca 2ϩ from internal stores and provokes a prolonged precisely patterned Ca 2ϩ influx, due to Ca 2ϩ action potentials (34,35). GH4C1 cells stimulated by TRH in medium with low free extracellular Ca 2ϩ result in the suppression of sustained intracellular Ca 2ϩ signals. (32,36). We demonstrated marked differences in MKP-1 gene expression in the absence or presence of extracellular Ca 2ϩ after TRH stimulation, which defines a physiological role for calcium in MKP-1 induction associated with the release of the block to elongation. In addition to Ca 2ϩ signaling, TRH activates other pathways, notably through cAMP and MAP kinases. The residual TRH effects seen for MKP-1 transcription (Table II) in the absence of sustained Ca 2ϩ signaling are probably due to such additional pathways.
How does Ca 2ϩ control MKP-1 gene transcription? Since very few reports have studied the regulation of MKP-1 gene expression, we can only speculate on responsive elements in the MKP-1 gene and relate them to known functional determinants in other Ca 2ϩ -regulated genes (e.g. c-fos). First, considering homologies between the rat, mouse, and human MKP-1 promoters, we find two conserved CRE elements and one E box motif. The CRE element and its binding protein CREB are known to mediate calcium signals in the nucleus for several genes including the c-fos oncogene. Ca 2ϩ -activated gene transcription is mediated through phosphorylation of CREB and the CREB-binding protein by CaM kinase IV. CREB phosphorylation is also a target for multiple signal transduction pathways, which involve different activated protein kinases (49). Thus Ca 2ϩ -, mitogenic, or stress-activated MAP kinases may control MKP-1 transcription, as in c-fos, through phosphorylation of CREB transcription factors at the level of initiation. Consistent with this idea, Sommer et al. (10) have reported that the upstream stimulatory factor that binds the E-box motif in the promoter of MKP-1 cooperates with signals that stimulate CRE-dependent transcription.
Other regulatory elements may be involved to control MKP-1 transcription by calcium. Recently, a Ca 2ϩ -regulated transcriptional repressor (DREAM) that binds to the DRE element has been show to control the transcriptional activity of the c-fos oncogene (50). The DRE element in the human c-fos oncogene is located at 20 nt downstream of the transcriptional initiation site. A similar consensus DRE element is located in the first 10 bp of the rat, mouse, and human MKP-1 gene that could be a target for DREAM-mediated transcriptional repression. Alternatively, Ca 2ϩ regulation of MKP-1 gene transcription may  (2) were transferred on a Nylon-N ϩ membrane. The membranes were hybridized with labeled RNAs obtained by in vitro elongation using nuclei isolated from GH4C1 cells stimulated or not (basal) with 100 nM TRH as in nuclear run-on experiments (Fig. 5). Hybridization signals obtained in basal conditions, indicating RNA transcripts upstream of the block to elongation (indicated with solid arrows) correspond to NcoI-NcoI, NcoI-NheI, and SacII-NheI fragments; hybridization signals from SacII-HindIII and NcoI-HindIII fragments (indicated by dashed arrows) are only obtained following release of the elongation block by TRH and indicate transcripts downstream of the elongation block site. Quantification of the hybridization signals by film densitometry and normalized for U content gave the following arbitrary values: SacII-NheI (1450 versus 2460), SacII-HindIII (260 versus 1320); NheI-NcoI plus NcoI-NcoI (720 versus 1067), and NcoI-HindIII (150 versus 1220) for basal versus TRH stimulated cell nuclei, respectively. depend on transcription factors translocated to the nucleus by the calcium-activated phosphatase calcineurin (51,52) or on an alternative Ca 2ϩ signaling pathway sensitive to the calmodulin antagonist W7. Coulon et al. (23) have reported that this latter calcium signaling pathway affects the intragenic regulation of c-fos expression via suppression of a transcriptional pause site.
In summary, we have identified a novel gene, MKP-1, which is regulated at the level of elongation of transcription in a manner similar to the c-fos oncogene. While the mechanisms that control transcriptional elongation remain relatively unclear, MKP-1 gene expression studies conducted in parallel with the analysis of a similar phenomenon in c-fos expression are certainly useful to further advance our knowledge of the regulation of transcriptional elongation in eukaryotes and on immediate early gene expression in mammalian cells.