Journal qt NeurochemistryLippincott—Raven Publishers, Philadelphia© 1998 International Society for Neurochemistry

Cell Type-Specific Regulation of CREB Gene Expression:Mutational Analysis of CREB Promoter Activity

Emily Coven, Yan Ni, Katherine L. Widnell, Jingshan Chen, *Willjam H. Walker,Ijoel F. Habener, and Eric J. Nestler

Laboratory of Molecular Psychiatry and Center for Genes and Behavior, Yale University School of Medicine, New Haven,Connecticut, * Department of Cell Biology, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, and

t Laboratory of Molecular Endocrinology, Massachusetts General Hospital, Howard Hughes Medical Instituteand Harvard Medical School, Boston, Massachusetts, U.S.A.

Abstract: Previous studies have shown that activation ofthecyclic AMP (cAMP) pathway down-regulates CREB ex-pression in CATH.a cells, an effect that appears to be medi-ated via inhibition of CREB gene transcription. In thecurrentstudy, wecompared this effect in CATH.a cells with regula-tion of CREB expression in anothercell line, 06 glioma cells.In contrast to the findings in CATH.a cells, activation of thecAMP pathway up-regulates CREB expression in 06 gliomacells. To determine whether these opposite effects can beexplained by regulation of CREB promoter activity, chioram-phenicol acetyltransferase (CAT) assays were performed inCATH.a and 06 glioma cells that were transiently trans-fected with a CREB promoter-CAT fusion plasmid. Activa-tion of the cAMP pathway decreased levels of CAT activityin transfected CATH.a cells but increased CAT activity intransfected 06 glioma cells. We next investigated the effectof mutations in the CREB promoter on such regulation inthese two cell lines. Mutations of single ORE orSpl bindingsites in the CREB promoter reduced basal levels of CATactivity but did not significantly attenuate regulation of thepromoter in CATH.a or 06 glioma cells. However, mutationor deletion of two ORE sites in the CREB promoter com-pletely abolished up-regulation of CAT activity in the 06glioma cells and abolished basal levels of CAT activity inCATH.a cells. CREB promoter activity was also studied incultured SHSY5Y cells and in primary cultures of striatalneurons as further comparisons. Activation of the cAMPpathway was found to increase CAT activity in both celltypes. In the striatal cultures, this effect was obliterated bymutation or deletion of either of the two OREs in the pro-moter. These findings demonstrate cell type-specific effectsof the cAMP pathway on CREB expression, which appearto be mediated via differential regulation of the CREB pro-moter. Key Words: CATH.a cells—C6 glioma cells—Stria-turn— Locus ceruleus—CREB gene transcription—CyclicAMP pathway—Cyclic AMP response elements.J. Neurochem. 71, 1865—1874 (1998).

CREB, cyclic AMP response element binding pro-tein, is a 43-kDa protein and member of the leucine

zipper family of transcription factors, which share cer-tain structural motifs and bind DNA as dimers. CREBregulates the transcription of many target genes bybinding to CREs (cyclic AMP response elements)present in the promoter regions of these genes. Thetranscriptional activity of CREB is activated upon itsphosphorylation on a single serine residue (Ser133) byany of several protein kinases, including cyclic AMP(cAMP)-dependent protein kinase (protein kinase A)(Gonzalez and Montminy, 1989; Meyer and Habener,1993; Richards et al., 1996; Xing et al., 1996).

Such phosphorylation was longbelieved to representthe only mechanism for CREB regulation becauseCREB expression was considered to be constitutive(see Gonzalez and Montminy, 1989; Molina et al.,1993). However, there is growing evidence that CREBexpression is not constitutive, indeed, that it is subjectto dynamic regulation. To date, regulation of CREBexpression has been demonstrated in neuronal cellsboth in vivo and in vitro (Brecht et al., 1994; Widnellet al., 1994, 1996a; Nibuya et al., 1996), in primaryrat Sertoli cells (Meyer et al., 1993; Walker et al.,1995), and in senescent human fibroblasts (Chin etal., 1996). That CREB gene expression is subject toregulation is not surprising because the CREB pro-

Received January 19, 1998; revised manuscript received April 30,1998; accepted May 7, 1998.

Address correspondence and reprint requests to Dr. E. J. Nestlerat Laboratory of Molecular Psychiatry and Center for Genes andBehavior, Yale University School of Medicine, 34 Park St., NewHaven, CT 06508, U.S.A.

The present address of Dr. K. L. Widnell is Department of Neurol-ogy, Johns Hopkins School of Medicine, Baltimore, MD 21205,U.S.A.

Abbreviations used: cAMP, cyclic AMP; CAT, chloramphenicolacetyltransferase; CRE, cyclic AMPresponse element; CREB, cyclicAMP response element binding protein; DMEM, Dulbecco’s modi-fied Eagle’s medium; IBMX, isobutylmethylxanthine; LC, locus cer-uleus; SDS, sodium dodecyl sulfate; VIP, vasoactive intestinal poly-peptide.

1865

1866 E. COVEN ET AL.

moter is known to contain three CRE sites and severalother types of DNA response elements (Walker et al.,1995). Together, these findings support the view thatregulation of CREB gene expression represents an ad-ditional mechanism by which CREB function is regu-lated under physiological conditions.

Support for this possibility, in the nervous system,comes from studies of the locus ceruleus (LC), themajor noradrenergic nucleus in brain that has beenimplicated in physical opiate dependence and with-drawal (see Nestler and Aghajanian, 1997). Chronicadministration of morphine up-regulates CREB ex-pression in this brain region (Widnell et al., 1994),and recent work has directly related this up-regulationto some of the functional changes that morphine expo-sure induces in these neurons (Lane-Ladd et al., 1997).Because morphine produces a sustained inhibition ofthe cAMP pathway in these neurons (Guitart and Nes-tler, 1989), and as a result inhibits CREB phosphoryla-tion (Guitart et al., 1992), our working hypothesishas been that the morphine-induced up-regulation ofCREB represents a compensatory homeostatic re-sponse to the inhibitory effects of the drug, which thenleads to other neuroadaptive changes (Nestler, 1996;Nestler and Aghajanian, 1997).

Support for this scheme has come from studies ofan LC-like cell line, CATH.a cells (Sun et al., 1993),wherein activation of the cAMP pathway down-regu-lates CREB protein and mRNA levels (Widnell et al.,1994). This down-regulation appears to be mediatedvia a decrease in CREB gene transcription based onnuclear run-on assays (Widnell et al., 1996b). More-over, activation of the cAMP pathway decreases theactivity of a 1 ,264-bp fragment of the 5’ promoter ofthe CREB gene in CATH.a cells.

However, these findings in CATH.a cells are inmarked contrast to those in certain other systems (e.g.,Sertoli cells), where activation of the cAMP pathwayhas been shown to increase CREB expression and pro-moter activity (Meyer et al., 1993; Walker et al.,1995). Thus, it would appear that perturbation of thecAMP pathway can exert opposite effects on CREBexpression in a cell type-specific manner. Indeed, thispossibility is consistent with the in vivo observationsthat although chronic morphine administration in-creases CREB expression in the LC, the same treat-ment decreases CREB expression in certain other brainregions and has no effect in still other regions (Widnelletal., 1996a).

The goal of the present study was to obtain a betterunderstanding of these phenomena by comparing regu-lation of CREB expression by perturbation of thecAMP pathway in CATH.a cells with that seen in C6glioma cells, where preliminary data demonstrated up-regulation of CREB in response to cAMP pathwayactivation. We then earned out a detailed mutationalanalysis of the regulation of the CREB promoter inthese two cell lines to identify putative response ele-

ments that could be responsible for this cell type-spe-cific regulation of CREB expression.

MATERIALS AND METHODS

Cell cultureCATH.a cells were a generous gift of D. M. Chikaraishi

(Duke University, Durham, NC, U.S.A.) and were culturedas described previously (Widnell et al., 1994). This cell linewas derived from a brainstem tumor of a tyrosine hydroxy-lase-SV4O T antigen transgenic mouse (Sun et al., 1993).CATH.a cells were plated at a density of I x iO~cells/cm2into 100-mm culture dishes. Medium was changed 24 h afterpassage, and cells were treated with vehicle or 5

1jM for-skolin 48 h after passage. This treatment maximally activatesthe cAMP pathway in this cell line (Widnell et al., 1994).

C6 glioma cells were a generous gift of P. Fishman(NINDS, NIH, Bethesda, MD, U.S.A.) and were cultured asdescribed previously (Hosoda et al., 1994). C6 glioma cellswere plated at a density of 1 X 106 cells/cm

2 into 100-mmculture dishes. Cells were treated with 1

1iM isoproterenoland 500 ~uMisobutylmethyixanthine (IBMX) 24 h after pas-sage. This treatment maximally activates the cAMP pathwayin this cell line (Hosoda et al., 1994; authors’ unpublishedobservations).

SHSY5Y cells were obtained from S. Brene (KarolinskaInstitute, Stockholm, Sweden) and were cultured as de-scribed previously (Boundy et al., 1998). Cells were platedat a density of I X io~cells in 100-mm culture dishes, andwere treated with 5 ~tM forskolin in fresh medium [Dulbec-co’s modified Eagle’s medium (DMEM)/10% fetal bovineserum] 24 h after passage.

Primary cultures of striatal neurons were prepared as de-scribed by Konradi et al. (1994) with modifications. In brief,striatawere dissectedfrom embryonic day 18 Sprague—Daw-ley rats. Tissue was suspended in DMEM/Fl2 medium(GibcoBRL) supplemented with (per liter) 20 ml of B27(GibcoBRL), 5 g of glucose, and 10 ml of 5,000 U/mlpenicillin and 5,000 ,ug/ml streptomycin. The tissue wasmechanically dissociated with a fire-narrowed Pasteur pi-pette, and cells were plated in six-well plates at a density of2 X 106 cells/well. Cells were treated with 5 ,uM forskolinin fresh medium, conditions known to activate the cAMPpathway in this cell type (Konradi et al., 1994).

Preparation of cell extracts, western blotting, andgel shift assays

Cells were extracted in a high-salt, high-detergent bufferexactly as described (Widnell et al., 1994) and were adjustedto contain final concentrations of 4% glycerol, 2% sodiumdodecyl sulfate (SDS), and 2% 2-mercaptoethanol, with bro-mophenol blue as a marker. Samples were boiled for 2 mmand aliquots were subjected to one-dimensional SDS—poly-acrylamide gel electrophoresis (8.0% acrylamide/0.3% bis-acrylamide). Proteins were transferred electrophoreticallyfrom gels to nitrocellulose papers (Towbin et al., 1979),which were then preblocked with 2% milk in a phosphate-buffered saline/Tween buffer containing 10 mM sodiumphosphate (pH 7.2), 140 mM NaCI, and 0.05% Tween 20(Sigma). The blots were incubated with anti-CREB antibody(1:10,000) raised against a TrpC—CREB fusion protein(kindly provided by D. Ginty, Johns Hopkins School ofMedicine, Baltimore, MD, and M. Greenberg, Harvard Mcd-

J. Neurochem., Vol. 71, No. 5, 1998

REGULATION OF CREB GENE PROMOTER 1867

ical School, Boston, MA, U.S.A.), and then incubated witha 1:4,000 dilution of goat anti-rabbit antibody conjugated tohorseradish peroxidase (Vector Labs). Blots were developedwith the enhanced chemiluminescence system of Amershamand exposed to Hyperfilm-ECL (Amersham) for I to 2 mm.

Cell extracts were also subjectedto CRE gel shift analysisaccording to a published procedure (Widnell et al., 1994).A double-stranded synthetic I 8mer oligonucleotide derivedfrom the somatostatin gene promoter, which contains a con-sensus CRE site, was used. To confirm specificity of theCRE binding activity, a competition study was performedexactly as described (Widnell et al., 1994), using nonradio-labeled CRE probe or a mutant probe in which the CRE sitewas mutated. For the supershift experiments, I ~l of theanti-CREB antibody was incubatedwith thegel shift reactioncomponents for 30 mm at 4°Cbefore addition of the P-labeled probe.

Cell transfections and chloramphenicolacetyltransferase (CAT) assays

DNA transfections were performed as described by Chenand Okayama (1987) with the following modifications.CATH.a cells were preincubated for 1 h in DMEM supple-mented with 10% fetal bovine serum prior to transfectionwith a CREB-CAT construct. The CREB-CAT constructsand the mutant CREB-CAT constructs used contain se-quences of the CREB promoter region from 51— 1,264 bpupstream of the translation start site, as have been describedpreviously (Walker et al., 1995). CATH.a cells were trans-fected with 2.5 fig of CREB-CAT DNA at 37°Cin 3% CO2for 6 h, then incubated in RPMI medium containing vehicleor forskolin. C6 glioma cells were transfected with 2.5 ,ugof CREB-CAT DNA at 37°Cin 3% CO2 for 6 h, then incu-bated in serum-freeDMEM. After 18 h, C6 cells were treatedwith isoproterenol and IBMX. The cells were extracted atthe timepoints described, and CAT assays were performedusing Promega’s CAT Assay System, with the modificationthat [‘

4C]chloramphenicol was used at a final concentrationof 12 ~M. Levels of CAT activity were standardized to theamount of total protein as assessed by Bradford assay (Bio-Rad Laboratories, Richmond, CA, U.S.A.). The differenttransfection times used for CATH.a and C6 glioma cellswere based on initial results, which determined the optimaltransfection conditions for each cell line.

SHSY5Y cells and cultured striatal neurons were trans-fected with CREB-CAT constructs by use of LipofectAmine(GibcoBRL), according to the manufacturer’s specifica-tions. Cells were transfected for 24 h and then treated with5

1jM forskolin for an additional 24 h, prior to further analy-sis as described above.

RESULTS

Effect of activation of cAMP pathway on CREBexpression in CATH.a and C6 glioma cells

CATH.a cells and C6 glioma cells were treated withagents that are known to maximally stimulate thecAMP pathway: Forskolin (a direct activator of ade-nylyl cyclase) was used for CATH.a cells, whereasisoproterenol (a ,8-adrenergic receptor agonist) andIBMX (a phosphodiesterase inhibitor) were used forC6 glioma cells (Hosoda et al., 1994; Widnell et al.,1994). Vehicle- and drug-treated cells were then ana-

FIG. 1. Regulation of CREB immunoreactivity in CATH.a and 06glioma cells by cAMP pathway activation. A: CATH.a cells wereharvested 0, 2, 4, 6, 12, 24, or 48 h after incubation with vehicle(V) or 5 ©M forskolin (F) and were analyzed by CREB immu-noblotting. The graph shows the time-dependent forskolin regu-lation of CREB immunoreactivity in this cell type. Data are ex-pressed as percent change from vehicle (mean ± SEM). Eachexperiment (n = 6 per time point) was repeated two to four timeswith similar results. °p< 0.05 by Student’s t test. Inset: Lanesfrom a representative blot. B: 06 glioma cells were harvested 0,2, 6, or 24 h after incubation with vehicle (V) or 1 pM isoprotere-nol + 0.5mM IBMX (I) and were analyzed by CREB immunoblot-ting. The graph shows the time-dependent isoproterenol + IBMXregulation of CREB immunoreactivity in this cell type. Data areexpressed as percent change from vehicle (mean ±SEM). Eachexperiment (n = 6 per time point) was repeated two to four timeswith similar results. *p < 0.05 by Student’s t test. Inset: Lanesfrom a representative blot.

lyzed by western blotting for changes in CREB proteinexpression. In both cell lines, western blots revealed amajor band of 43 kDa, the reported molecular massfor CREB, which was specifically blocked by preincu-bation of the antibody with the TrpC—CREB fusionprotein (data not shown).

In CATH.a cells, levels of CREB immunoreactivityunderwent a transient increase followed by a sustaineddecrease in response to activation of the cAMP path-way (Fig. IA). The transient increase peaked after 4h of fonskolin treatment at 35% above control values.

J. Neurochein., Vol. 71, No. 5, 1998

1868 E. COVEN ET AL.

FIG. 2. Regulation of GRE binding activity in CATH.aand C6 glioma cells by cAMP pathway activation. A:CATH.a cell extract was analyzed by gel shift in theabsence of competing oligonucleotide probe (left-most lane) or in the presence of varying amounts ofcompeting probe, which was not radiolabeled. Twocompeting probes were used: the GRE probe itselfand the GRE probe containing a mutated GRE site(mutGRE), which is known to abolish GREB binding(see Materials and Methods). This experiment dem-onstrates the specificity of the GRE binding activitymeasured in this gel shift assay. B: GATH.a cellswere harvested 4 or 24 h after incubation with vehicle(V) or 5 pM forskolin (F) and analyzed for GRE bind-ing activity by gel shift analysis. The graph shows thetime-dependent forskolin regulation of GRE bindingactivity in this cell type. Data from 12 control andforskolin-treated samples are expressed as a per-cent change from vehicle (mean ±SEM). °p< 0.05by Student’s t test. Inset: Lanes from a representa-tive gel. C: 06 glioma cells were harvested 2 or 24h after incubation with vehicle (V) or 1 pM isoprotere-nol + 0.5 mM IBMX (I) and analyzed for GRE bindingactivity by gel shift analysis. The graph shows thetime-dependent isoproterenol + IBMX regulation ofGRE binding activity in this cell type. Data from 12control and forskolin-treated samples are expressedas a percent change from vehicle (mean ± SEM).°p< 0.05 by Student’s t test. Inset: Lanes from arepresentative gel.

After 6 h of treatment, the CREB immunoreactivityreturned toward control values and then fell steadilyto 30% below control values after 24—48 h of treat-ment. This biphasic regulation of CREB expression issimilar to that observed previously in this cell line(Widnell et al., 1994).

In C6 glioma cells, the exact opposite effects wereobserved: Activation of the cAMP pathway caused atransient decrease in levels of CREB immunoreactiv-ity, which was followed by a sustained increase (Fig.IB). The transient decrease in CREB levels peakedafter 2 h of treatment with isoproterenol + IBMX at25% below control values. Levels of CREB immunore-activity then returned toward control values after 6 h oftreatment and thereafter climbed to 30% above controlvalues after 24 h. Further increases in CREB levelswere not seen with longer treatment times (data notshown).

Effect of activation of cAMP pathway on levels ofCRE binding activity in CATH.a and C6 gliomacells

To determine whether changes in CREB protein ex-pression are associated with changes in CREB func-tion, CRE DNA binding activity was measured inCATH.a and C6 glioma cells by gel shift assays.Assays were performed at representative time pointsat which significant changes in CREB immunoreactiv-ity had been observed (see Fig. 1). In both cell lines,CRE binding activity appeared as a single band thatwas specifically competed for by nonlabeled CREprobe but not by a mutated CRE probe (Fig. 2A). In

addition, the CRE binding activity in CATH.a and C6glioma cells was specifically “supershifted” by theanti-CREB antibody (data not shown), which furtherdemonstrates the specificity of this measure.

In CATH.a cells (Fig. 2B), after 4 h of treatmentwith forskolin, a 22% increase in CRE binding activitywas observed, which corresponds to the increase inCREB immunoreactivity at this time point. Likewise,after 24 h of treatment with forskolin, a 16% decreasein CRE binding was observed, which corresponds tothe decrease in CREB immunoreactivity at this latertime point. In C6 glioma cells (Fig. 2C), changes ob-served in CRE binding also corresponded to changesin CREB immunoreactivity. After 2 h of treatment withisoproterenol + IBMX, a small (and nonsignificant)decrease in CRE binding was observed, whereas after24 h, a 23% increase in CRE binding activity wasobserved.

Effect of activation of cAMP pathway on CREBpromoter activity in CATH.a and C6 glioma cells

We next studied the activity of the CREB promoterin CATH.a and C6 glioma cells. Cells were transientlytransfected with a construct of the CREB promoterlinked to a CAT reporter gene. Relative CAT activityin the cells was used as the measure of activation ofthe CREB promoter. In initial experiments, CATH.acells were transiently transfected with — I 264CREB-CAT, a construct that contains a 1 ,264-bp fragment ofthe CREB promoter (Walker et al., 1995). Cells weretransfected for 6 h, after which time they were exposedto forskolin (5 fiM) for 16, 24, 36, or 48 h and then

J. Neurochem., Vol. 71, No.5, /998

REGULATION OF CREB GENE PROMOTER 1869

FIG. 3. Regulation of CREB promoter activity in GATH.a cellsby cAMP pathway activation. A: GATH.a cells were transfectedwith a —1 ,264GREB-GAT promoter construct for 6 h and thentreated with vehicle or 5 pM forskolin for 16, 24, 36, or 48 h.The bar graph shows relative GAT activity in untransfected cells(U) and in vehicle-treated and forskolin-treated transfected cells.Data are expressed as percent change from vehicle (mean±SEM). Note that some SEMs are too small to be seen in thefigure. Each experiment (n = 3/group) was repeated at leastthree times with similar results. GAT activity was significantlygreater in vehicle-treated transfected cells relative to untrans-fected cells (~p< 0.05 by Student’s t test). Furthermore, GATactivity was significantly less in forskolin-treated transfectedcells relative to vehicle-treated transfected cells (°p< 0.05, °°p< 0.06 by Student’s t test). B: GATH.a cells were transfectedwith constructs containing the first 278, 537, 804, or 1,264 bpof the GREB promoter linked to a GAT reporter gene (see Fig.5A). Gells were transfected for 6 h and then treated with vehicleor 5 pM forskolin for 24 h. The bar graph shows relative GATactivity in vehicle- and forskolin-treated transfected cells. Dataare expressed as relative GAT activity (mean ± SEM). Eachexperiment (n = 3/group) was repeated at least three timeswith similar results. Despite considerable differences in the basalactivity of these various GREB promoters, a similar (=50%) de-gree of down-regulation was seen in response to forskolin (°p< 0.05 by Student’s t test).

harvested. At all time points, forskolin treatment re-sulted in a significant (‘=50%) down-regulation of rel-ative CAT activity (Fig. 3A). Based on these results,we used the 24-h time for all subsequent experiments.Next, forskolin regulation of CREB promoter activitywas studied in CATH.a cells that were transfec-

ted with — I 264CREB-CAT or with plasmids contain-ing one of three 5’ truncations of the 1,264-hppromoter (—8O4CREB-CAT, —537CREB-CAT, or—278CREB-CAT) (see Fig. 5A for schematic of pro-moter). As shown in Fig. 3B, although the basal levelof CREB promoter activity varied considerably amongthese deletion mutants, forskolin was found to re-duce promoter activity to approximately the same ex-tent with each construct. A similar (~50%)down-regulation of CREB promoter activity (using the—I264CREB-CAT construct) was observed upontreating CATH.a cells with vasoactive intestinal poly-peptide (VIP) for 16 h (control: 14,328 ±1,574 rela-tive CAT activity; VIP: 7,699 ±687; n = 6, p < 0.05by t test). As VIP is known to activate the cAMPpathway in these cells (Widnell et al., 1994), this find-ing indicates the physiological significance of the ef-fect of forskolin on this system.

We next studied CREB promoter activity in C6 gli-oma cells. In these experiments, cells were transfectedfor 6 h, 18 h later were treated with isoproterenol+ IBMX for an additional 24 h, and then harvested.In striking contrast to CATH.a cells, isoproterenol+ IBMX increased CREB promoter activity by abouttwofold in C6 glioma cells (see Fig. 4). A similardegree of activation was seen with the — I 264CREB-CAT and —278CREB-CAT constructs. It was alsofound that the — I264CREB-CAT, —8O4CREB-CAT,

FIG. 4. Regulation of GREB promoter activity in G6 glioma cellsby cAMP pathway activation. G6 cells were transfected withconstructs containing the first 278 or 1,264 bp of the GREBpromoter linked to a GAT reporter gene. Gells were transfectedfor 6 h, maintained in serum-free medium for 18 h, and thentreated with 1 pM isoproterenol (iso) + 0.5 mM IBMX for 24 h.The bar graph shows relative GAT activity in vehicle-treated andisoproterenol + IBMX-treated transfected cells. Since the datashown are compiled from several different trials, GAT activity isexpressed relative to that seen in untransfected cells, definedas 100% (mean ±SEM, n = 12). GAT activity was significantlygreater in vehicle-treated transfected cells relative to untrans-fected cells (~p< 0.05 by Student’s t test). In addition, GATactivity was further stimulated by isoproterenol + IBMX (*p<0.05 by Student’s ttest).

J. Neurochejn.. Vol. 7/, No.5, 1998

1870 E. COVEN ET AL.

—5 37CREB-CAT, and — 278CREB-CAT constructsyielded equivalent basal levels of CREB promoter ac-tivity in this cell line (data not shown).

Mutational analysis of regulation of CREBpromoter by cAMP pathway activation inCATH.a and C6 glioma cells

As the next step in characterizing regulation ofCREB promoter activity by the cAMP pathway,CATH.a and C6 glioma cells were transfected withplasmids containing one of several CREB promotermutants linked to a CAT reporter gene. The CREBpromoter mutants consist of the full-length (—1,264)CREB promoter mutated at one of two consensus Splsites, mutated at either or both of two consensus CREsites, or with both CRE sites deleted (Walker et al.,1995). These two Sp 1 and CRE sites are located withinthe proximal 278-hp region of the promoter (see Fig.5A). Mutations made to the SpI and CRE sites areknown to obliterate the functional activity of theseresponse elements (Walker et al., 1995).

In CATH.a cells, the predominant effectof the muta-tions upon CREB promoter activity is a strong decreasein basal levels. Thus, as can be seen in Fig. SB, muta-tion of either of two Spi sites or CRE sites resultedin a clear, and in some cases, marked reduction inbasal activity of the CREB promoter. However, in eachcase, forskolin treatment was found to further decreasepromoter activity by at least 50%, and in some casesthe degree of forskolin-induced inhibition was greaterthan that seen with the wild-type construct. Mutationor deletion of both CRE sites almost completely oblit-erated basal activity of the CREB promoter, whichmade it impossible to assess forskolin regulation underthese conditions.

Mutations in the CREB promoter result in similartrends in the C6 glioma cell line in that basal activityof the promoter was reduced by mutation of eitherCRE or Spl site and almost completely lost upon muta-tion or deletion of both CRE sites (Fig. SC). Moreover,activation of CREB promoter activity by isoproterenol+ IBMX, which was still apparent when one CRE sitewas mutated, was completely abolished in constructsin which both CRE sites were mutated or deleted. Inter-estingly, mutation of one SpI site (termed Spl-l) hadno effect on the induction of CREB promoter activityin response to isoproterenol + IBMX, whereas muta-tion of the other Spl site (termed Spl-2) completelyabolished this effect (Fig. SC).

Effect of activation of cAMP pathway on CREBpromoter activity in SHSY5Y cells and incultured striatal neurons

To determine whether inhibition versus activationof CREB promoter activity is the more typical responseof the CREB gene to activation of the cAMP pathwayin neural cells, we examined this phenomenon in twoother cell types: a human neuroblastoma cell line,SHSYSY cells, and primary cultures of late prenatal

rat striatal neurons. In both cases, the cells were trans-fected with the —278CREB-CAT promoter for 24 hand were then treated with 5 fiM forskolin for 24 h,after which time CAT expression was determined. Itwas found that forskolin treatment significantly in-creased CREB promoter activity both in SHSY5Y cells(62 ±6% increase; mean ±SEM increase; n = 6, p< 0.05 by Student’s t test) and in cultured striatalneurons (54 ±4% increase; n = 9, p < 0.05 byStudent’s t test). We also found that treatment of C6glioma cells with 5 ,uM forskolin for 24 h resulted inan 83 ±6% increase (n = 4, p < 0.05 by Student’st test) in CREB promoter activity.

A mutational analysis of this response was carriedout in the striatal neurons. As shown in Fig. 6, forskolinincreased the activity of the —1264CREB-CAT pro-moter to a comparable degree as seen for the—278CREB-CAT promoter. The basal activity of thepromoter was almost completely abolished by muta-tion of either of the two CRE sites. Such mutations alsocompletely abolished induction of promoter activity byforskolin.

DISCUSSION

One major finding of the present study is that regula-tion of CREB expression and of CRE binding activityin C6 glioma cells is antipodal to that previously ob-served in the CATH.a cell line. In the C6 cells, levelsof both CREB immunoreactivity and CRE binding firstdecrease (within 2 h) and then show a sustained in-crease in response to activation of the cAMP pathway.In contrast, in CATH.a cells, both of these measuresshow a transient increase, which is followed by a sus-tained decrease. The opposite regulation of CREB ex-pression seen in CATH.a and C6 cells after prolongedactivation of the cAMP pathway could be explainedby opposite regulation of CREB gene transcription.Thus, in CATH.a cells, the down-regulation of CREBexpression after prolonged cAMP pathway activationis associated with a reduction in CREB promoter activ-ity. In contrast, in C6 cells, CREB expression andCREB promoter activity are both up-regulated bycAMP pathway activation. Together, these data sub-stantiate the cell type-specific nature of the regulationof CREB gene expression. At the same time, our find-ings do not preclude the possibility that posttranscrip-tional mechanisms (e.g., regulation of CREB mRNAor protein stability) also contribute to the long-termregulation of CREB expression in the two cell lines.

Whereas the mechanism for the sustained changesin CREB expression in CATH.a and C6 cells appearsto be at least in part transcriptional, the mechanismunderlying the transient changes in CREB levels seenin these cell lines remains unknown. The changes inCREB immunoreactivity seen within 2—4 h of cAMPpathway activation in these two cell lines occur inthe absence of detectable changes in levels of CREB

J. Neurocheni., Vol. 7/, No. 5, /998

REGULATION OF CREB GENE PROMOTER 1871

FIG. 5. Mutational analysis of the regulation ofCREB promoter activity in GATH.a and G6 gliomacells by cAMP pathway activation. A: Schematic il-lustration of the GREB promoter and the various mu-tations used in the present study. mt, mutant site;del, deleted site. Further information on these muta-tions is presented elsewhere (Walker et al., 1995).B: GATH.a cells were transfected with the variousCREB-GAT promoter constructs for 6 h and thentreated with vehicle or 5 pM forskolin for 24 h. Thebar graph shows relative GAT activity in vehicle-treated and forskolin-treated transfected GATH.acells. Data are expressed as percent control (mean±SEM). Note that some SEMs are too small to beseen in the figure. Each experiment (n = 3/group)was repeated at least three times with similar results.°p< 0.05 (statistically different from vehicle) by Stu-dent’s t test. C: G6 cells were transfected with amutant GREB-GATpromoter construct for 6 h, main-tained in serum-free medium for 18 h, and thentreated with 1 pM isoproterenol + 0.5 mM IBMX for24 h. The bar graph shows relative GAT activity invehicle-treated and isoproterenol + IBMX-treatedtransfected G6 cells. Since the data shown are com-piled from several different trials, GAT activity is ex-pressed relative to that seen in untransfected cells,defined as 100% (mean ± SEM, n = 12). *p < 0.05(statistically different from vehicle) by Student’stest.

mRNA (Widnell et al., 1994; authors’ unpublishedobservations). These findings suggest that posttransla-tional mechanisms may be involved, perhaps changesin the rate of proteolysis of the protein. The observationthat opposite effects are seen in the two cell lines sug-gests the possibility that a common mechanism maybe involved, but one that can be differentially regulated

depending on the cell type. In any event, these observa-tions highlight that levels of CREB protein and CREbinding activity can be rapidly modulated in cells inresponse to activation of the cAMP pathway.

Another major objective of the present study wasa mutational analysis of CREB promoter activity inCATH.a and C6 cell lines. It was found in initial stud-

J. Neurochem., Vol. 7/, No. .5. /998

1872 E. COVEN ET AL.

FIG. 6. Mutational analysis of the regulation of CREB promoteractivity in cultured striatal neurons by cAMP pathway activation.Striatal neurons were transfected with the various GREB-GATpromoter constructs for 24 h and then treated with vehicle or 5pM forskolin for anadditional 24 h. The bar graph shows relativeGAT activity in untransfected neurons and in vehicle-treated andforskolin-treated transfected neurons. mt, mutant site. Data areexpressed as relative GAT activity (mean ± SEM). Note thatsome SEMs are too small to be seen in the figure. Each experi-ment (n = 4/group) was repeated twice with similar results. °p< 0.05 (statistically different from vehicle) by Student’s t test.

ies that activation of the cAMP pathway caused similarregulation of CREB promoter activity (down-regula-tion in CATH.a cells and up-regulation in C6 cells) in aseries of S’truncated mutations of the I ,264-bp CREBpromoter. These results indicate that the DNA ele-ments governing opposite regulation of CREB pro-moter activity in CATH.a and C6 cells are presentwithin the first 278 bp of the promoter (see Fig. 5A).

We next sought to identify specific response ele-ments within the promoter responsible for this regula-tion. We focused on two CRE sites and two Spl siteslocated within the first 278 bp of the promoter, becauseprevious work in Sertoli cells showed that these sitesare important for the basal activity of the promoterand for its activation in response to cAMP pathwayactivation (Walker et al., 1995). In the present study,we obtained similar evidence for the importance ofthese sites. In CATH.a and C6 cells, mutation of eitherCRE site or of either Spl site significantly reducedbasal activity of the promoter. Moreover, mutation ordeletion of both CRE sites almost completely obliter-ated basal promoter activity. This loss of basal pro-moter activity made it impossible to assess the role ofthese CRE sites in the down-regulation of promoteractivity seen in CATH.a cells. However, we showedthat the loss of the two CRE sites caused completeloss of up-regulation of promoter activity in C6 cellsupon cAMP pathway activation. These two CRE sitesare downstream (3’) of the transcription initiation sitewithin the CREB promoter (Fig. SA). There is anadditional consensus CRE site in the CREB promoter,

>800 bp upstream of the others. However, this sitewould not appear to affect cAMP regulation of theCREB promoter, as similar regulation was seen (bothin CATH.a cells and in C6 cells) in promoters thatcontain this site (e.g., 1,264-bp promoter) and in thosethat do not (e.g., —278-bp promoter). Interestingly,one of the Spi sites (Spl-2) also appeared to be re-quired for the normal up-regulation of CREB promoteractivity in C6 cells. SpI sites were implicated in thecAMP-induced up-regulation of the CREB promoterin Sertoli cells, although in this case the Spl-1 sitewas shown to be involved (Walker et al., 1995).

These findings support a positive autoregulatoryscheme for CREB gene expression in C6 cells, similarto that proposed for Sertoli cells: Activation of thecAMP pathway leads to phosphorylation and activa-tion of CREB, which then—via the CREs in the CREBpromoter—activate CREB gene transcription. Themechanism by which cAMP pathway activation leadsto down-regulation of CREB promoter activity inCATH.a cells remains more obscure. The down-regu-lation could be mediated by the CREs in the promoter,although this was not possible to demonstrate becausebasal promoter activity was obliterated in this cell line.If the CREs are involved, it will be important in futurestudies to determine whether the down-regulation ismediated via CREB itself or some other CREB-likeprotein(s), such as ICER (inducible cAMP early re-pressor), as has been observed recently in Sertoli cells(Walker et al., 1998). Another possibility, which nowwarrants further analysis, is that distinct response ele-ments—contained within the 278-hp proximal pro-moter region—mediate the down-regulation of CREBexpression in CATH.a cells. It is also conceivable thatother differences in the regulatory properties of thesecell lines, for example, differences in the cAMP path-way or some other intracellular signaling pathway,may be responsible for the opposite regulation ofCREB expression observed.

Whereas activation of the cAMP pathway can leadto up- or down-regulation of CREB gene expressiondepending on the cell type, the predominant responseappears to be up-regulation. Thus, up-regulation ofCREB promoter activity, which was first reported forSertoli cells (Walker et al., 1995), is demonstrated inthe present study in C6 cells, SHSY5Y cells, and cul-tured striatal neurons. In the striatal neurons, like theC6 cells, this up-regulation is dependent on the twoCREs present in the CREB promoter. These resultshighlight the unique regulatory properties of CREBexpression in CATH.a cells, where a down-regulationis seen. CATH.a cells show many properties of LCneurons and indeed were derived from a brainstemtumor of a tyrosine hydroxylase-SV4O T antigentransgenic mouse (Sun et al., 1993). Interestingly,chronic exposure of rats to morphine, which acutelyinhibits the cAMP pathway, causes an up-regulationof CREB expression in the LC. This up-regulation has

J. Neurochem., Vol. 71, No. 5, 1998

REGULATION OF CREB GENE PROMOTER 1873

been directly related to biochemical and electrophysio-logical adaptations to morphine observed in these neu-rons and to behavioral aspects of morphine dependence(Lane-Ladd et al., 1997;Nestler and Aghajanian, 1997;Olson et al., 1998; see also Maldonado et al., 1996).Interestingly, up-regulation of CREB in response tochronic morphine administration does not occur inother brain regions analyzed; in fact, in some regionsa down-regulation has been observed (Widnell et al.,1996a). These findings highlight the importance ofdetermining the unique molecular properties of LCneurons that are responsible for this distinct responseto morphine. It is possible that CATH.a cells will pro-vide a useful model system in which to characterizethese mechanisms.

Morphine addiction is just one of many forms ofneural plasticity in which a role for CREB has beenimplicated (see Bourtchuladze et al., 1994; Carew,1996; Martin and Kandel, 1996; Yin and Tully, 1996;Kornhauser and Greenberg, 1997). As the precisemechanisms by which CREB exerts these effects areidentified, it will be important to determine whetheralterations in CREB expression are also involved inthese other phenomena. Indeed, it has been shown thatchronic exposure to every major class of antidepressanttreatment up-regulates CREB expression in hippocam-pus (Nibuya et al., 1996). In this regard, it is strikingthat the proximal region of the CREB gene promoterin the mouse, rat, and human each contains the twoCRE and two SpI sites analyzed in this present study.This extraordinary conservation in noncoding, regula-tory sequence in the CREB gene across species is con-sistent with a vital role played by the CREB promoter,and by these sites in particular, in the regulation ofCREB expression and function. This further highlightsthe importance of better understanding the detailed mo-lecular mechanisms governing the striking cell typespecificity of the regulation of CREB gene expression.

Acknowledgment: This work was supported by grantsfrom the National Institute on Drug Abuse and NationalInstitute of Mental Health and by the Abraham RibicoffResearch Facilities of the Connecticut Mental Health Center,State of Connecticut Department of Mental Health and Ad-diction Services.

REFERENCES

Boundy V. A., Chen J. S., and Nestler E. J. (1998) Regulation ofcAMP-dependent protein kinase subunit expression in CATH.aand SH-SY5Y cells. J. Pharmacol. Exp. Ther. (in press).

Bourtchuladze R., Frenguelli B., Blendy J., Cioffi D., Schutz G.,and Silva A. J. (1994) Deficient long-term memory in mice witha targeted mutation of the cAMP-responsive element-bindingprotein. Cell 79, 59—68.

Brecht S., Gass P., Anton F., Bravo R., Zimmerman M., and Herde-gen T. (1994) Induction of c-Jun and suppression of CREBtranscription factor proteins in axotomizedneurons of substantianigra and covariation with tyrosine hydroxylase. Mol. Cell. Neu-rose’,. 5, 431 —441.

Carew T. J. (1996) Molecular enhancement of memory formation.Neuron 16, 5—8.

Chen C. and Okayama H. (1987) High efficiency transformation ofmammalian cells by plasmid DNA. Mol. Cell. Biol. 7, 2745—2752.

Chin J. H., Okazaki M., Frazier J. S., Hu Z. W., and Hoffman B. B.(1996) Impaired cAMP-mediated gene expression and de-creased cAMP response element binding protein in senescentcells. Am. I. Physiol. 271, C362—C371.

Gonzalez G. A. and Montminy M. R. (1989) Cyclic AMP stimulatessomatostatin transcription by phosphorylation of CREB at ser-inc 133. Cell 59, 675—680.

Guitart X. and Nestler E. J. (1989) Identification of morphine- andcyclic AMP-regulated phosphoproteins (MARPPs) in the locuscoeruleus and other regions of rat brain: regulation by acuteand chronic niorphine. J. Neuro.cci. 9, 4371 —4387.

Guitart X., Thompson M. A., Mirante C. K., Greenberg M. E., andNestler E. J. (1992) Regulation of cyclic AMP response ele-ment-binding protein (CREB) phosphorylation by acute andchronic morphine in the rat locus coeruleus. J. Neurochem. 58,1168—1171.

Hosoda K., Feussner G. K., Rydelek-Fitzgerald L., Fishman P. H.,and Duman R. 5. (1994) Agonist and cyclic AMP—mediatedregulation of ~

1-adrenergic receptor mRNA and gene transcrip-tion in rat C6 glioma cells. J. Neurochem. 63, 1635—1645.

Konradi C., Cole R. L., Heckers S., and Hyman S. E. (1994) Am-phetamine regulates gene expression in rat striatum via tran-scription factor CREB. J. Neuro,sci. 14, 5623—5634.

Kornhauser J. and Greenberg M. E. (1997) A kinase to remember:dual roles for MAP kinase in long-term memory. Neuron 18,839—842.

Lane-Ladd S. B., Pineda J.. Boundy V., Pfeuffer T., Krupinski J.,Aghajanian G. K., and Nestler E. J. (1997) CREB in the locuscoeruleus: biochemical, physiological, and behavioral evidencefor a role in opiate dependence. J. Neurosci. 17, 7890—7901.

Maldonado R., Blendy J. A., Tzavara E., Gass P., Roques B. P.,Hanoune J., and SchUtz G. (1996) Reduction of morphine absti-nence in mice with a mutation in the gene encoding CREB.Science 273, 657—659.

Martin K. C. and Kandel E. R. (1996) Cell adhesion molecules,CREB, and the formation of new synaptic connections. Neuron17, 567—570.

Meyer T. E. and Habener J. F. (1993) Cyclic adenosine 3 ‘,5 ‘-mono-phosphate response element binding protein (CREB) and re-lated transcription-activating deoxyribonucleic acid-bindingproteins. Endocr. Rev. 14, 269—290.

Meyer T. E., Waeber G., Lin J., Beckmann W., and Habener J. F.(1993) The promoter of the gene encoding 3 ‘,5 ‘-cyclic adeno-sine monophosphate (cAMP) response element binding proteincontains cAMP response elements: evidence for positive auto-regulation of gene transcription. Endocrinology 132, 770—780.

Molina C. A., Foulkes N. S., Lally E., and Sassone-Corsi P. (1993)lnducibility and negative autoregulation of CREB: an alterna-tive promoter directs the expression of ICER, an early responserepressor. Cell 75, 875—886.

Nestler E. J. (1996) Under siege: the brain on opiates. Neuron 16,897—900.

Nestler E. J. and Aghajanian G. K. (1997) Molecular and cellularbasis of addiction. Science 278, 58—63.

Nibuya M., Nestler E. J., and Duman R. S. (1996) Chronic antide-pressant administration increases the expression of CREB in rathippocampus. J. Neurosci. 16, 2365—2372.

Olson V. G., Punch L. J., Lane-Ladd S. B., Neve R. L., Taylor J. R.,Carlezon W. A., and Nestler E. J. (1998) Severity of naloxone-precipitated morphine withdrawal depends upon functional ac-tivity of CREB in locus coeruleus. Soc. Neurosci. Ah,str. (inpress).

Richards J. P., Bachinger H. P., Goodman R. H., and Brennan R. G.(1996) Analysis of the structural properties of cAMP-respon-

J. Neurochern., Vol. 7/, No. 5, /998

1874 E. COVEN ET AL.

sive element-binding protein (CREB) and phosphorylatedCREB. J. Biol. Chem. 271, 13716—13723.

Sun C., Fung B. P., Tischler A. S., and Chikaraishi D. M. (1993)Catecholaminergic cell lines from the brain and adrenal glandsof tyrosine hydroxylase-5V40 T antigen transgenic mice. I.Neurosci. 13, 1280—1291.

Towbin H., Staehlin T., and Gordon J. (1979) Electrophoretic trans-fer of proteins from polyacrylamide gels to nitrocellulose sheets:procedure and some applications. Proc. Nati. Acad. Sci. USA76, 4350—4354.

Walker W. H., Fucci L., and Habener J. F. (1995) Expression ofthe gene encoding transcription factor cyclic adenosine 3 ‘,5monophosphate (cAMP) response element-binding protein(CREB): regulation by follicle stimulating hormone-inducedcAMP signaling in primary rat Sertoli cells. Endocrinology 136,3534—3545.

Walker W. H., Daniel P. B., and Habener J. F. (1998) InduciblecAMP early repressor ICER down-regulation of CREB geneexpression in Sertoli cells. Mol. Cell. Endocrinol. (in press).

Widnell K. L., Russell D. S., and Nestler E. J. (1994) Regulation ofexpression of cAMP response element-binding protein in thelocus-coeruleus in vivo and in a locus coeruleus-like cell linein vitro. Proc. NatI. Acad. Sci. USA 91, 10947—10951.

Widnell K. L., Self D. W., Lane S. B., Russell D. S., Vaidya V. A.,Miserendino M. J. D., Rubin C. S., Duman R. S., and NestlerE. J. (1996a) Regulation of CREB expression: in vivo evidencefor a functional role in morphine action in the nucleus accum-bens. J. Pharmacol. E.rp. Ther. 276, 306—315.

Widnell K. L., Chen J. S., Iredale P. A., Walker W. H., Duman R. S.,HabenerJ. F., and Nestler E. J. (1996h) Transcriptional regula-tion of CREB (cyclic AMP response element-binding protein)expression in CATH.a cells. J. Neurochem. 66, 1770—1773.

Xing J., Ginty D. D., and Greenberg M. F. (1996) Coupling of theRAS-MAPK pathway to gene activation by RSK2, a growthfactor-regulated CREB kinase. Science 273, 959—963.

Yin J. C. and Tully T. (1996) CREB and the formation of long-term memory. Curr. Opin. Neurohiol. 6, 264—268.

J. Neuroche,n., Vol. 71, No. 5, 1998